sched.h revision 716a42348cdaf04534b15fbdc9c83e25baebfed5
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/* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */ 42#define SCHED_RESET_ON_FORK 0x40000000 43 44#ifdef __KERNEL__ 45 46struct sched_param { 47 int sched_priority; 48}; 49 50#include <asm/param.h> /* for HZ */ 51 52#include <linux/capability.h> 53#include <linux/threads.h> 54#include <linux/kernel.h> 55#include <linux/types.h> 56#include <linux/timex.h> 57#include <linux/jiffies.h> 58#include <linux/rbtree.h> 59#include <linux/thread_info.h> 60#include <linux/cpumask.h> 61#include <linux/errno.h> 62#include <linux/nodemask.h> 63#include <linux/mm_types.h> 64 65#include <asm/system.h> 66#include <asm/page.h> 67#include <asm/ptrace.h> 68#include <asm/cputime.h> 69 70#include <linux/smp.h> 71#include <linux/sem.h> 72#include <linux/signal.h> 73#include <linux/path.h> 74#include <linux/compiler.h> 75#include <linux/completion.h> 76#include <linux/pid.h> 77#include <linux/percpu.h> 78#include <linux/topology.h> 79#include <linux/proportions.h> 80#include <linux/seccomp.h> 81#include <linux/rcupdate.h> 82#include <linux/rculist.h> 83#include <linux/rtmutex.h> 84 85#include <linux/time.h> 86#include <linux/param.h> 87#include <linux/resource.h> 88#include <linux/timer.h> 89#include <linux/hrtimer.h> 90#include <linux/task_io_accounting.h> 91#include <linux/kobject.h> 92#include <linux/latencytop.h> 93#include <linux/cred.h> 94 95#include <asm/processor.h> 96 97struct exec_domain; 98struct futex_pi_state; 99struct robust_list_head; 100struct bio; 101struct fs_struct; 102struct bts_context; 103struct perf_counter_context; 104 105/* 106 * List of flags we want to share for kernel threads, 107 * if only because they are not used by them anyway. 108 */ 109#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 110 111/* 112 * These are the constant used to fake the fixed-point load-average 113 * counting. Some notes: 114 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 115 * a load-average precision of 10 bits integer + 11 bits fractional 116 * - if you want to count load-averages more often, you need more 117 * precision, or rounding will get you. With 2-second counting freq, 118 * the EXP_n values would be 1981, 2034 and 2043 if still using only 119 * 11 bit fractions. 120 */ 121extern unsigned long avenrun[]; /* Load averages */ 122extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift); 123 124#define FSHIFT 11 /* nr of bits of precision */ 125#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 126#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 127#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 128#define EXP_5 2014 /* 1/exp(5sec/5min) */ 129#define EXP_15 2037 /* 1/exp(5sec/15min) */ 130 131#define CALC_LOAD(load,exp,n) \ 132 load *= exp; \ 133 load += n*(FIXED_1-exp); \ 134 load >>= FSHIFT; 135 136extern unsigned long total_forks; 137extern int nr_threads; 138DECLARE_PER_CPU(unsigned long, process_counts); 139extern int nr_processes(void); 140extern unsigned long nr_running(void); 141extern unsigned long nr_uninterruptible(void); 142extern unsigned long nr_iowait(void); 143extern void calc_global_load(void); 144extern u64 cpu_nr_migrations(int cpu); 145 146extern unsigned long get_parent_ip(unsigned long addr); 147 148struct seq_file; 149struct cfs_rq; 150struct task_group; 151#ifdef CONFIG_SCHED_DEBUG 152extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 153extern void proc_sched_set_task(struct task_struct *p); 154extern void 155print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 156#else 157static inline void 158proc_sched_show_task(struct task_struct *p, struct seq_file *m) 159{ 160} 161static inline void proc_sched_set_task(struct task_struct *p) 162{ 163} 164static inline void 165print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 166{ 167} 168#endif 169 170extern unsigned long long time_sync_thresh; 171 172/* 173 * Task state bitmask. NOTE! These bits are also 174 * encoded in fs/proc/array.c: get_task_state(). 175 * 176 * We have two separate sets of flags: task->state 177 * is about runnability, while task->exit_state are 178 * about the task exiting. Confusing, but this way 179 * modifying one set can't modify the other one by 180 * mistake. 181 */ 182#define TASK_RUNNING 0 183#define TASK_INTERRUPTIBLE 1 184#define TASK_UNINTERRUPTIBLE 2 185#define __TASK_STOPPED 4 186#define __TASK_TRACED 8 187/* in tsk->exit_state */ 188#define EXIT_ZOMBIE 16 189#define EXIT_DEAD 32 190/* in tsk->state again */ 191#define TASK_DEAD 64 192#define TASK_WAKEKILL 128 193 194/* Convenience macros for the sake of set_task_state */ 195#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 196#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 197#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 198 199/* Convenience macros for the sake of wake_up */ 200#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 201#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 202 203/* get_task_state() */ 204#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 205 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 206 __TASK_TRACED) 207 208#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 209#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 210#define task_is_stopped_or_traced(task) \ 211 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 212#define task_contributes_to_load(task) \ 213 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ 214 (task->flags & PF_FROZEN) == 0) 215 216#define __set_task_state(tsk, state_value) \ 217 do { (tsk)->state = (state_value); } while (0) 218#define set_task_state(tsk, state_value) \ 219 set_mb((tsk)->state, (state_value)) 220 221/* 222 * set_current_state() includes a barrier so that the write of current->state 223 * is correctly serialised wrt the caller's subsequent test of whether to 224 * actually sleep: 225 * 226 * set_current_state(TASK_UNINTERRUPTIBLE); 227 * if (do_i_need_to_sleep()) 228 * schedule(); 229 * 230 * If the caller does not need such serialisation then use __set_current_state() 231 */ 232#define __set_current_state(state_value) \ 233 do { current->state = (state_value); } while (0) 234#define set_current_state(state_value) \ 235 set_mb(current->state, (state_value)) 236 237/* Task command name length */ 238#define TASK_COMM_LEN 16 239 240#include <linux/spinlock.h> 241 242/* 243 * This serializes "schedule()" and also protects 244 * the run-queue from deletions/modifications (but 245 * _adding_ to the beginning of the run-queue has 246 * a separate lock). 247 */ 248extern rwlock_t tasklist_lock; 249extern spinlock_t mmlist_lock; 250 251struct task_struct; 252 253extern void sched_init(void); 254extern void sched_init_smp(void); 255extern asmlinkage void schedule_tail(struct task_struct *prev); 256extern void init_idle(struct task_struct *idle, int cpu); 257extern void init_idle_bootup_task(struct task_struct *idle); 258 259extern int runqueue_is_locked(void); 260extern void task_rq_unlock_wait(struct task_struct *p); 261 262extern cpumask_var_t nohz_cpu_mask; 263#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 264extern int select_nohz_load_balancer(int cpu); 265extern int get_nohz_load_balancer(void); 266#else 267static inline int select_nohz_load_balancer(int cpu) 268{ 269 return 0; 270} 271#endif 272 273/* 274 * Only dump TASK_* tasks. (0 for all tasks) 275 */ 276extern void show_state_filter(unsigned long state_filter); 277 278static inline void show_state(void) 279{ 280 show_state_filter(0); 281} 282 283extern void show_regs(struct pt_regs *); 284 285/* 286 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 287 * task), SP is the stack pointer of the first frame that should be shown in the back 288 * trace (or NULL if the entire call-chain of the task should be shown). 289 */ 290extern void show_stack(struct task_struct *task, unsigned long *sp); 291 292void io_schedule(void); 293long io_schedule_timeout(long timeout); 294 295extern void cpu_init (void); 296extern void trap_init(void); 297extern void update_process_times(int user); 298extern void scheduler_tick(void); 299 300extern void sched_show_task(struct task_struct *p); 301 302#ifdef CONFIG_DETECT_SOFTLOCKUP 303extern void softlockup_tick(void); 304extern void touch_softlockup_watchdog(void); 305extern void touch_all_softlockup_watchdogs(void); 306extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write, 307 struct file *filp, void __user *buffer, 308 size_t *lenp, loff_t *ppos); 309extern unsigned int softlockup_panic; 310extern int softlockup_thresh; 311#else 312static inline void softlockup_tick(void) 313{ 314} 315static inline void touch_softlockup_watchdog(void) 316{ 317} 318static inline void touch_all_softlockup_watchdogs(void) 319{ 320} 321#endif 322 323#ifdef CONFIG_DETECT_HUNG_TASK 324extern unsigned int sysctl_hung_task_panic; 325extern unsigned long sysctl_hung_task_check_count; 326extern unsigned long sysctl_hung_task_timeout_secs; 327extern unsigned long sysctl_hung_task_warnings; 328extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write, 329 struct file *filp, void __user *buffer, 330 size_t *lenp, loff_t *ppos); 331#endif 332 333/* Attach to any functions which should be ignored in wchan output. */ 334#define __sched __attribute__((__section__(".sched.text"))) 335 336/* Linker adds these: start and end of __sched functions */ 337extern char __sched_text_start[], __sched_text_end[]; 338 339/* Is this address in the __sched functions? */ 340extern int in_sched_functions(unsigned long addr); 341 342#define MAX_SCHEDULE_TIMEOUT LONG_MAX 343extern signed long schedule_timeout(signed long timeout); 344extern signed long schedule_timeout_interruptible(signed long timeout); 345extern signed long schedule_timeout_killable(signed long timeout); 346extern signed long schedule_timeout_uninterruptible(signed long timeout); 347asmlinkage void __schedule(void); 348asmlinkage void schedule(void); 349extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner); 350 351struct nsproxy; 352struct user_namespace; 353 354/* 355 * Default maximum number of active map areas, this limits the number of vmas 356 * per mm struct. Users can overwrite this number by sysctl but there is a 357 * problem. 358 * 359 * When a program's coredump is generated as ELF format, a section is created 360 * per a vma. In ELF, the number of sections is represented in unsigned short. 361 * This means the number of sections should be smaller than 65535 at coredump. 362 * Because the kernel adds some informative sections to a image of program at 363 * generating coredump, we need some margin. The number of extra sections is 364 * 1-3 now and depends on arch. We use "5" as safe margin, here. 365 */ 366#define MAPCOUNT_ELF_CORE_MARGIN (5) 367#define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN) 368 369extern int sysctl_max_map_count; 370 371#include <linux/aio.h> 372 373extern unsigned long 374arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 375 unsigned long, unsigned long); 376extern unsigned long 377arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 378 unsigned long len, unsigned long pgoff, 379 unsigned long flags); 380extern void arch_unmap_area(struct mm_struct *, unsigned long); 381extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 382 383#if USE_SPLIT_PTLOCKS 384/* 385 * The mm counters are not protected by its page_table_lock, 386 * so must be incremented atomically. 387 */ 388#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 389#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 390#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 391#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 392#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 393 394#else /* !USE_SPLIT_PTLOCKS */ 395/* 396 * The mm counters are protected by its page_table_lock, 397 * so can be incremented directly. 398 */ 399#define set_mm_counter(mm, member, value) (mm)->_##member = (value) 400#define get_mm_counter(mm, member) ((mm)->_##member) 401#define add_mm_counter(mm, member, value) (mm)->_##member += (value) 402#define inc_mm_counter(mm, member) (mm)->_##member++ 403#define dec_mm_counter(mm, member) (mm)->_##member-- 404 405#endif /* !USE_SPLIT_PTLOCKS */ 406 407#define get_mm_rss(mm) \ 408 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 409#define update_hiwater_rss(mm) do { \ 410 unsigned long _rss = get_mm_rss(mm); \ 411 if ((mm)->hiwater_rss < _rss) \ 412 (mm)->hiwater_rss = _rss; \ 413} while (0) 414#define update_hiwater_vm(mm) do { \ 415 if ((mm)->hiwater_vm < (mm)->total_vm) \ 416 (mm)->hiwater_vm = (mm)->total_vm; \ 417} while (0) 418 419static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) 420{ 421 return max(mm->hiwater_rss, get_mm_rss(mm)); 422} 423 424static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) 425{ 426 return max(mm->hiwater_vm, mm->total_vm); 427} 428 429extern void set_dumpable(struct mm_struct *mm, int value); 430extern int get_dumpable(struct mm_struct *mm); 431 432/* mm flags */ 433/* dumpable bits */ 434#define MMF_DUMPABLE 0 /* core dump is permitted */ 435#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 436#define MMF_DUMPABLE_BITS 2 437 438/* coredump filter bits */ 439#define MMF_DUMP_ANON_PRIVATE 2 440#define MMF_DUMP_ANON_SHARED 3 441#define MMF_DUMP_MAPPED_PRIVATE 4 442#define MMF_DUMP_MAPPED_SHARED 5 443#define MMF_DUMP_ELF_HEADERS 6 444#define MMF_DUMP_HUGETLB_PRIVATE 7 445#define MMF_DUMP_HUGETLB_SHARED 8 446#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 447#define MMF_DUMP_FILTER_BITS 7 448#define MMF_DUMP_FILTER_MASK \ 449 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 450#define MMF_DUMP_FILTER_DEFAULT \ 451 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ 452 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) 453 454#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS 455# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) 456#else 457# define MMF_DUMP_MASK_DEFAULT_ELF 0 458#endif 459 460struct sighand_struct { 461 atomic_t count; 462 struct k_sigaction action[_NSIG]; 463 spinlock_t siglock; 464 wait_queue_head_t signalfd_wqh; 465}; 466 467struct pacct_struct { 468 int ac_flag; 469 long ac_exitcode; 470 unsigned long ac_mem; 471 cputime_t ac_utime, ac_stime; 472 unsigned long ac_minflt, ac_majflt; 473}; 474 475/** 476 * struct task_cputime - collected CPU time counts 477 * @utime: time spent in user mode, in &cputime_t units 478 * @stime: time spent in kernel mode, in &cputime_t units 479 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds 480 * 481 * This structure groups together three kinds of CPU time that are 482 * tracked for threads and thread groups. Most things considering 483 * CPU time want to group these counts together and treat all three 484 * of them in parallel. 485 */ 486struct task_cputime { 487 cputime_t utime; 488 cputime_t stime; 489 unsigned long long sum_exec_runtime; 490}; 491/* Alternate field names when used to cache expirations. */ 492#define prof_exp stime 493#define virt_exp utime 494#define sched_exp sum_exec_runtime 495 496#define INIT_CPUTIME \ 497 (struct task_cputime) { \ 498 .utime = cputime_zero, \ 499 .stime = cputime_zero, \ 500 .sum_exec_runtime = 0, \ 501 } 502 503/* 504 * Disable preemption until the scheduler is running. 505 * Reset by start_kernel()->sched_init()->init_idle(). 506 * 507 * We include PREEMPT_ACTIVE to avoid cond_resched() from working 508 * before the scheduler is active -- see should_resched(). 509 */ 510#define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE) 511 512/** 513 * struct thread_group_cputimer - thread group interval timer counts 514 * @cputime: thread group interval timers. 515 * @running: non-zero when there are timers running and 516 * @cputime receives updates. 517 * @lock: lock for fields in this struct. 518 * 519 * This structure contains the version of task_cputime, above, that is 520 * used for thread group CPU timer calculations. 521 */ 522struct thread_group_cputimer { 523 struct task_cputime cputime; 524 int running; 525 spinlock_t lock; 526}; 527 528/* 529 * NOTE! "signal_struct" does not have it's own 530 * locking, because a shared signal_struct always 531 * implies a shared sighand_struct, so locking 532 * sighand_struct is always a proper superset of 533 * the locking of signal_struct. 534 */ 535struct signal_struct { 536 atomic_t count; 537 atomic_t live; 538 539 wait_queue_head_t wait_chldexit; /* for wait4() */ 540 541 /* current thread group signal load-balancing target: */ 542 struct task_struct *curr_target; 543 544 /* shared signal handling: */ 545 struct sigpending shared_pending; 546 547 /* thread group exit support */ 548 int group_exit_code; 549 /* overloaded: 550 * - notify group_exit_task when ->count is equal to notify_count 551 * - everyone except group_exit_task is stopped during signal delivery 552 * of fatal signals, group_exit_task processes the signal. 553 */ 554 int notify_count; 555 struct task_struct *group_exit_task; 556 557 /* thread group stop support, overloads group_exit_code too */ 558 int group_stop_count; 559 unsigned int flags; /* see SIGNAL_* flags below */ 560 561 /* POSIX.1b Interval Timers */ 562 struct list_head posix_timers; 563 564 /* ITIMER_REAL timer for the process */ 565 struct hrtimer real_timer; 566 struct pid *leader_pid; 567 ktime_t it_real_incr; 568 569 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 570 cputime_t it_prof_expires, it_virt_expires; 571 cputime_t it_prof_incr, it_virt_incr; 572 573 /* 574 * Thread group totals for process CPU timers. 575 * See thread_group_cputimer(), et al, for details. 576 */ 577 struct thread_group_cputimer cputimer; 578 579 /* Earliest-expiration cache. */ 580 struct task_cputime cputime_expires; 581 582 struct list_head cpu_timers[3]; 583 584 struct pid *tty_old_pgrp; 585 586 /* boolean value for session group leader */ 587 int leader; 588 589 struct tty_struct *tty; /* NULL if no tty */ 590 591 /* 592 * Cumulative resource counters for dead threads in the group, 593 * and for reaped dead child processes forked by this group. 594 * Live threads maintain their own counters and add to these 595 * in __exit_signal, except for the group leader. 596 */ 597 cputime_t utime, stime, cutime, cstime; 598 cputime_t gtime; 599 cputime_t cgtime; 600 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 601 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 602 unsigned long inblock, oublock, cinblock, coublock; 603 struct task_io_accounting ioac; 604 605 /* 606 * Cumulative ns of schedule CPU time fo dead threads in the 607 * group, not including a zombie group leader, (This only differs 608 * from jiffies_to_ns(utime + stime) if sched_clock uses something 609 * other than jiffies.) 610 */ 611 unsigned long long sum_sched_runtime; 612 613 /* 614 * We don't bother to synchronize most readers of this at all, 615 * because there is no reader checking a limit that actually needs 616 * to get both rlim_cur and rlim_max atomically, and either one 617 * alone is a single word that can safely be read normally. 618 * getrlimit/setrlimit use task_lock(current->group_leader) to 619 * protect this instead of the siglock, because they really 620 * have no need to disable irqs. 621 */ 622 struct rlimit rlim[RLIM_NLIMITS]; 623 624#ifdef CONFIG_BSD_PROCESS_ACCT 625 struct pacct_struct pacct; /* per-process accounting information */ 626#endif 627#ifdef CONFIG_TASKSTATS 628 struct taskstats *stats; 629#endif 630#ifdef CONFIG_AUDIT 631 unsigned audit_tty; 632 struct tty_audit_buf *tty_audit_buf; 633#endif 634}; 635 636/* Context switch must be unlocked if interrupts are to be enabled */ 637#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 638# define __ARCH_WANT_UNLOCKED_CTXSW 639#endif 640 641/* 642 * Bits in flags field of signal_struct. 643 */ 644#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 645#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 646#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 647#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 648/* 649 * Pending notifications to parent. 650 */ 651#define SIGNAL_CLD_STOPPED 0x00000010 652#define SIGNAL_CLD_CONTINUED 0x00000020 653#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 654 655#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 656 657/* If true, all threads except ->group_exit_task have pending SIGKILL */ 658static inline int signal_group_exit(const struct signal_struct *sig) 659{ 660 return (sig->flags & SIGNAL_GROUP_EXIT) || 661 (sig->group_exit_task != NULL); 662} 663 664/* 665 * Some day this will be a full-fledged user tracking system.. 666 */ 667struct user_struct { 668 atomic_t __count; /* reference count */ 669 atomic_t processes; /* How many processes does this user have? */ 670 atomic_t files; /* How many open files does this user have? */ 671 atomic_t sigpending; /* How many pending signals does this user have? */ 672#ifdef CONFIG_INOTIFY_USER 673 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 674 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 675#endif 676#ifdef CONFIG_EPOLL 677 atomic_t epoll_watches; /* The number of file descriptors currently watched */ 678#endif 679#ifdef CONFIG_POSIX_MQUEUE 680 /* protected by mq_lock */ 681 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 682#endif 683 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 684 685#ifdef CONFIG_KEYS 686 struct key *uid_keyring; /* UID specific keyring */ 687 struct key *session_keyring; /* UID's default session keyring */ 688#endif 689 690 /* Hash table maintenance information */ 691 struct hlist_node uidhash_node; 692 uid_t uid; 693 struct user_namespace *user_ns; 694 695#ifdef CONFIG_USER_SCHED 696 struct task_group *tg; 697#ifdef CONFIG_SYSFS 698 struct kobject kobj; 699 struct delayed_work work; 700#endif 701#endif 702 703#ifdef CONFIG_PERF_COUNTERS 704 atomic_long_t locked_vm; 705#endif 706}; 707 708extern int uids_sysfs_init(void); 709 710extern struct user_struct *find_user(uid_t); 711 712extern struct user_struct root_user; 713#define INIT_USER (&root_user) 714 715 716struct backing_dev_info; 717struct reclaim_state; 718 719#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 720struct sched_info { 721 /* cumulative counters */ 722 unsigned long pcount; /* # of times run on this cpu */ 723 unsigned long long run_delay; /* time spent waiting on a runqueue */ 724 725 /* timestamps */ 726 unsigned long long last_arrival,/* when we last ran on a cpu */ 727 last_queued; /* when we were last queued to run */ 728#ifdef CONFIG_SCHEDSTATS 729 /* BKL stats */ 730 unsigned int bkl_count; 731#endif 732}; 733#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 734 735#ifdef CONFIG_TASK_DELAY_ACCT 736struct task_delay_info { 737 spinlock_t lock; 738 unsigned int flags; /* Private per-task flags */ 739 740 /* For each stat XXX, add following, aligned appropriately 741 * 742 * struct timespec XXX_start, XXX_end; 743 * u64 XXX_delay; 744 * u32 XXX_count; 745 * 746 * Atomicity of updates to XXX_delay, XXX_count protected by 747 * single lock above (split into XXX_lock if contention is an issue). 748 */ 749 750 /* 751 * XXX_count is incremented on every XXX operation, the delay 752 * associated with the operation is added to XXX_delay. 753 * XXX_delay contains the accumulated delay time in nanoseconds. 754 */ 755 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 756 u64 blkio_delay; /* wait for sync block io completion */ 757 u64 swapin_delay; /* wait for swapin block io completion */ 758 u32 blkio_count; /* total count of the number of sync block */ 759 /* io operations performed */ 760 u32 swapin_count; /* total count of the number of swapin block */ 761 /* io operations performed */ 762 763 struct timespec freepages_start, freepages_end; 764 u64 freepages_delay; /* wait for memory reclaim */ 765 u32 freepages_count; /* total count of memory reclaim */ 766}; 767#endif /* CONFIG_TASK_DELAY_ACCT */ 768 769static inline int sched_info_on(void) 770{ 771#ifdef CONFIG_SCHEDSTATS 772 return 1; 773#elif defined(CONFIG_TASK_DELAY_ACCT) 774 extern int delayacct_on; 775 return delayacct_on; 776#else 777 return 0; 778#endif 779} 780 781enum cpu_idle_type { 782 CPU_IDLE, 783 CPU_NOT_IDLE, 784 CPU_NEWLY_IDLE, 785 CPU_MAX_IDLE_TYPES 786}; 787 788/* 789 * sched-domains (multiprocessor balancing) declarations: 790 */ 791 792/* 793 * Increase resolution of nice-level calculations: 794 */ 795#define SCHED_LOAD_SHIFT 10 796#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 797 798#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 799 800#ifdef CONFIG_SMP 801#define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 802#define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 803#define SD_BALANCE_EXEC 4 /* Balance on exec */ 804#define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 805#define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 806#define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 807#define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 808#define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 809#define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 810#define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 811#define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 812#define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 813 814enum powersavings_balance_level { 815 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */ 816 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package 817 * first for long running threads 818 */ 819 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle 820 * cpu package for power savings 821 */ 822 MAX_POWERSAVINGS_BALANCE_LEVELS 823}; 824 825extern int sched_mc_power_savings, sched_smt_power_savings; 826 827static inline int sd_balance_for_mc_power(void) 828{ 829 if (sched_smt_power_savings) 830 return SD_POWERSAVINGS_BALANCE; 831 832 return 0; 833} 834 835static inline int sd_balance_for_package_power(void) 836{ 837 if (sched_mc_power_savings | sched_smt_power_savings) 838 return SD_POWERSAVINGS_BALANCE; 839 840 return 0; 841} 842 843/* 844 * Optimise SD flags for power savings: 845 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings. 846 * Keep default SD flags if sched_{smt,mc}_power_saving=0 847 */ 848 849static inline int sd_power_saving_flags(void) 850{ 851 if (sched_mc_power_savings | sched_smt_power_savings) 852 return SD_BALANCE_NEWIDLE; 853 854 return 0; 855} 856 857struct sched_group { 858 struct sched_group *next; /* Must be a circular list */ 859 860 /* 861 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 862 * single CPU. This is read only (except for setup, hotplug CPU). 863 * Note : Never change cpu_power without recompute its reciprocal 864 */ 865 unsigned int __cpu_power; 866 /* 867 * reciprocal value of cpu_power to avoid expensive divides 868 * (see include/linux/reciprocal_div.h) 869 */ 870 u32 reciprocal_cpu_power; 871 872 /* 873 * The CPUs this group covers. 874 * 875 * NOTE: this field is variable length. (Allocated dynamically 876 * by attaching extra space to the end of the structure, 877 * depending on how many CPUs the kernel has booted up with) 878 * 879 * It is also be embedded into static data structures at build 880 * time. (See 'struct static_sched_group' in kernel/sched.c) 881 */ 882 unsigned long cpumask[0]; 883}; 884 885static inline struct cpumask *sched_group_cpus(struct sched_group *sg) 886{ 887 return to_cpumask(sg->cpumask); 888} 889 890enum sched_domain_level { 891 SD_LV_NONE = 0, 892 SD_LV_SIBLING, 893 SD_LV_MC, 894 SD_LV_CPU, 895 SD_LV_NODE, 896 SD_LV_ALLNODES, 897 SD_LV_MAX 898}; 899 900struct sched_domain_attr { 901 int relax_domain_level; 902}; 903 904#define SD_ATTR_INIT (struct sched_domain_attr) { \ 905 .relax_domain_level = -1, \ 906} 907 908struct sched_domain { 909 /* These fields must be setup */ 910 struct sched_domain *parent; /* top domain must be null terminated */ 911 struct sched_domain *child; /* bottom domain must be null terminated */ 912 struct sched_group *groups; /* the balancing groups of the domain */ 913 unsigned long min_interval; /* Minimum balance interval ms */ 914 unsigned long max_interval; /* Maximum balance interval ms */ 915 unsigned int busy_factor; /* less balancing by factor if busy */ 916 unsigned int imbalance_pct; /* No balance until over watermark */ 917 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 918 unsigned int busy_idx; 919 unsigned int idle_idx; 920 unsigned int newidle_idx; 921 unsigned int wake_idx; 922 unsigned int forkexec_idx; 923 int flags; /* See SD_* */ 924 enum sched_domain_level level; 925 926 /* Runtime fields. */ 927 unsigned long last_balance; /* init to jiffies. units in jiffies */ 928 unsigned int balance_interval; /* initialise to 1. units in ms. */ 929 unsigned int nr_balance_failed; /* initialise to 0 */ 930 931 u64 last_update; 932 933#ifdef CONFIG_SCHEDSTATS 934 /* load_balance() stats */ 935 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 936 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 937 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 938 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 939 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 940 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 941 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 942 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 943 944 /* Active load balancing */ 945 unsigned int alb_count; 946 unsigned int alb_failed; 947 unsigned int alb_pushed; 948 949 /* SD_BALANCE_EXEC stats */ 950 unsigned int sbe_count; 951 unsigned int sbe_balanced; 952 unsigned int sbe_pushed; 953 954 /* SD_BALANCE_FORK stats */ 955 unsigned int sbf_count; 956 unsigned int sbf_balanced; 957 unsigned int sbf_pushed; 958 959 /* try_to_wake_up() stats */ 960 unsigned int ttwu_wake_remote; 961 unsigned int ttwu_move_affine; 962 unsigned int ttwu_move_balance; 963#endif 964#ifdef CONFIG_SCHED_DEBUG 965 char *name; 966#endif 967 968 /* 969 * Span of all CPUs in this domain. 970 * 971 * NOTE: this field is variable length. (Allocated dynamically 972 * by attaching extra space to the end of the structure, 973 * depending on how many CPUs the kernel has booted up with) 974 * 975 * It is also be embedded into static data structures at build 976 * time. (See 'struct static_sched_domain' in kernel/sched.c) 977 */ 978 unsigned long span[0]; 979}; 980 981static inline struct cpumask *sched_domain_span(struct sched_domain *sd) 982{ 983 return to_cpumask(sd->span); 984} 985 986extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new, 987 struct sched_domain_attr *dattr_new); 988 989/* Test a flag in parent sched domain */ 990static inline int test_sd_parent(struct sched_domain *sd, int flag) 991{ 992 if (sd->parent && (sd->parent->flags & flag)) 993 return 1; 994 995 return 0; 996} 997 998#else /* CONFIG_SMP */ 999 1000struct sched_domain_attr; 1001 1002static inline void 1003partition_sched_domains(int ndoms_new, struct cpumask *doms_new, 1004 struct sched_domain_attr *dattr_new) 1005{ 1006} 1007#endif /* !CONFIG_SMP */ 1008 1009struct io_context; /* See blkdev.h */ 1010 1011 1012#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 1013extern void prefetch_stack(struct task_struct *t); 1014#else 1015static inline void prefetch_stack(struct task_struct *t) { } 1016#endif 1017 1018struct audit_context; /* See audit.c */ 1019struct mempolicy; 1020struct pipe_inode_info; 1021struct uts_namespace; 1022 1023struct rq; 1024struct sched_domain; 1025 1026struct sched_class { 1027 const struct sched_class *next; 1028 1029 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 1030 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 1031 void (*yield_task) (struct rq *rq); 1032 1033 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync); 1034 1035 struct task_struct * (*pick_next_task) (struct rq *rq); 1036 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 1037 1038#ifdef CONFIG_SMP 1039 int (*select_task_rq)(struct task_struct *p, int sync); 1040 1041 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 1042 struct rq *busiest, unsigned long max_load_move, 1043 struct sched_domain *sd, enum cpu_idle_type idle, 1044 int *all_pinned, int *this_best_prio); 1045 1046 int (*move_one_task) (struct rq *this_rq, int this_cpu, 1047 struct rq *busiest, struct sched_domain *sd, 1048 enum cpu_idle_type idle); 1049 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 1050 int (*needs_post_schedule) (struct rq *this_rq); 1051 void (*post_schedule) (struct rq *this_rq); 1052 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 1053 1054 void (*set_cpus_allowed)(struct task_struct *p, 1055 const struct cpumask *newmask); 1056 1057 void (*rq_online)(struct rq *rq); 1058 void (*rq_offline)(struct rq *rq); 1059#endif 1060 1061 void (*set_curr_task) (struct rq *rq); 1062 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 1063 void (*task_new) (struct rq *rq, struct task_struct *p); 1064 1065 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 1066 int running); 1067 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 1068 int running); 1069 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 1070 int oldprio, int running); 1071 1072#ifdef CONFIG_FAIR_GROUP_SCHED 1073 void (*moved_group) (struct task_struct *p); 1074#endif 1075}; 1076 1077struct load_weight { 1078 unsigned long weight, inv_weight; 1079}; 1080 1081/* 1082 * CFS stats for a schedulable entity (task, task-group etc) 1083 * 1084 * Current field usage histogram: 1085 * 1086 * 4 se->block_start 1087 * 4 se->run_node 1088 * 4 se->sleep_start 1089 * 6 se->load.weight 1090 */ 1091struct sched_entity { 1092 struct load_weight load; /* for load-balancing */ 1093 struct rb_node run_node; 1094 struct list_head group_node; 1095 unsigned int on_rq; 1096 1097 u64 exec_start; 1098 u64 sum_exec_runtime; 1099 u64 vruntime; 1100 u64 prev_sum_exec_runtime; 1101 1102 u64 last_wakeup; 1103 u64 avg_overlap; 1104 1105 u64 nr_migrations; 1106 1107 u64 start_runtime; 1108 u64 avg_wakeup; 1109 1110#ifdef CONFIG_SCHEDSTATS 1111 u64 wait_start; 1112 u64 wait_max; 1113 u64 wait_count; 1114 u64 wait_sum; 1115 1116 u64 sleep_start; 1117 u64 sleep_max; 1118 s64 sum_sleep_runtime; 1119 1120 u64 block_start; 1121 u64 block_max; 1122 u64 exec_max; 1123 u64 slice_max; 1124 1125 u64 nr_migrations_cold; 1126 u64 nr_failed_migrations_affine; 1127 u64 nr_failed_migrations_running; 1128 u64 nr_failed_migrations_hot; 1129 u64 nr_forced_migrations; 1130 u64 nr_forced2_migrations; 1131 1132 u64 nr_wakeups; 1133 u64 nr_wakeups_sync; 1134 u64 nr_wakeups_migrate; 1135 u64 nr_wakeups_local; 1136 u64 nr_wakeups_remote; 1137 u64 nr_wakeups_affine; 1138 u64 nr_wakeups_affine_attempts; 1139 u64 nr_wakeups_passive; 1140 u64 nr_wakeups_idle; 1141#endif 1142 1143#ifdef CONFIG_FAIR_GROUP_SCHED 1144 struct sched_entity *parent; 1145 /* rq on which this entity is (to be) queued: */ 1146 struct cfs_rq *cfs_rq; 1147 /* rq "owned" by this entity/group: */ 1148 struct cfs_rq *my_q; 1149#endif 1150}; 1151 1152struct sched_rt_entity { 1153 struct list_head run_list; 1154 unsigned long timeout; 1155 unsigned int time_slice; 1156 int nr_cpus_allowed; 1157 1158 struct sched_rt_entity *back; 1159#ifdef CONFIG_RT_GROUP_SCHED 1160 struct sched_rt_entity *parent; 1161 /* rq on which this entity is (to be) queued: */ 1162 struct rt_rq *rt_rq; 1163 /* rq "owned" by this entity/group: */ 1164 struct rt_rq *my_q; 1165#endif 1166}; 1167 1168struct task_struct { 1169 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1170 void *stack; 1171 atomic_t usage; 1172 unsigned int flags; /* per process flags, defined below */ 1173 unsigned int ptrace; 1174 1175 int lock_depth; /* BKL lock depth */ 1176 1177#ifdef CONFIG_SMP 1178#ifdef __ARCH_WANT_UNLOCKED_CTXSW 1179 int oncpu; 1180#endif 1181#endif 1182 1183 int prio, static_prio, normal_prio; 1184 unsigned int rt_priority; 1185 const struct sched_class *sched_class; 1186 struct sched_entity se; 1187 struct sched_rt_entity rt; 1188 1189#ifdef CONFIG_PREEMPT_NOTIFIERS 1190 /* list of struct preempt_notifier: */ 1191 struct hlist_head preempt_notifiers; 1192#endif 1193 1194 /* 1195 * fpu_counter contains the number of consecutive context switches 1196 * that the FPU is used. If this is over a threshold, the lazy fpu 1197 * saving becomes unlazy to save the trap. This is an unsigned char 1198 * so that after 256 times the counter wraps and the behavior turns 1199 * lazy again; this to deal with bursty apps that only use FPU for 1200 * a short time 1201 */ 1202 unsigned char fpu_counter; 1203#ifdef CONFIG_BLK_DEV_IO_TRACE 1204 unsigned int btrace_seq; 1205#endif 1206 1207 unsigned int policy; 1208 cpumask_t cpus_allowed; 1209 1210#ifdef CONFIG_PREEMPT_RCU 1211 int rcu_read_lock_nesting; 1212 int rcu_flipctr_idx; 1213#endif /* #ifdef CONFIG_PREEMPT_RCU */ 1214 1215#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1216 struct sched_info sched_info; 1217#endif 1218 1219 struct list_head tasks; 1220 struct plist_node pushable_tasks; 1221 1222 struct mm_struct *mm, *active_mm; 1223 1224/* task state */ 1225 struct linux_binfmt *binfmt; 1226 int exit_state; 1227 int exit_code, exit_signal; 1228 int pdeath_signal; /* The signal sent when the parent dies */ 1229 /* ??? */ 1230 unsigned int personality; 1231 unsigned did_exec:1; 1232 unsigned in_execve:1; /* Tell the LSMs that the process is doing an 1233 * execve */ 1234 1235 /* Revert to default priority/policy when forking */ 1236 unsigned sched_reset_on_fork:1; 1237 1238 pid_t pid; 1239 pid_t tgid; 1240 1241 /* Canary value for the -fstack-protector gcc feature */ 1242 unsigned long stack_canary; 1243 1244 /* 1245 * pointers to (original) parent process, youngest child, younger sibling, 1246 * older sibling, respectively. (p->father can be replaced with 1247 * p->real_parent->pid) 1248 */ 1249 struct task_struct *real_parent; /* real parent process */ 1250 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1251 /* 1252 * children/sibling forms the list of my natural children 1253 */ 1254 struct list_head children; /* list of my children */ 1255 struct list_head sibling; /* linkage in my parent's children list */ 1256 struct task_struct *group_leader; /* threadgroup leader */ 1257 1258 /* 1259 * ptraced is the list of tasks this task is using ptrace on. 1260 * This includes both natural children and PTRACE_ATTACH targets. 1261 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1262 */ 1263 struct list_head ptraced; 1264 struct list_head ptrace_entry; 1265 1266 /* 1267 * This is the tracer handle for the ptrace BTS extension. 1268 * This field actually belongs to the ptracer task. 1269 */ 1270 struct bts_context *bts; 1271 1272 /* PID/PID hash table linkage. */ 1273 struct pid_link pids[PIDTYPE_MAX]; 1274 struct list_head thread_group; 1275 1276 struct completion *vfork_done; /* for vfork() */ 1277 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1278 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1279 1280 cputime_t utime, stime, utimescaled, stimescaled; 1281 cputime_t gtime; 1282 cputime_t prev_utime, prev_stime; 1283 unsigned long nvcsw, nivcsw; /* context switch counts */ 1284 struct timespec start_time; /* monotonic time */ 1285 struct timespec real_start_time; /* boot based time */ 1286/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1287 unsigned long min_flt, maj_flt; 1288 1289 struct task_cputime cputime_expires; 1290 struct list_head cpu_timers[3]; 1291 1292/* process credentials */ 1293 const struct cred *real_cred; /* objective and real subjective task 1294 * credentials (COW) */ 1295 const struct cred *cred; /* effective (overridable) subjective task 1296 * credentials (COW) */ 1297 struct mutex cred_guard_mutex; /* guard against foreign influences on 1298 * credential calculations 1299 * (notably. ptrace) */ 1300 1301 char comm[TASK_COMM_LEN]; /* executable name excluding path 1302 - access with [gs]et_task_comm (which lock 1303 it with task_lock()) 1304 - initialized normally by flush_old_exec */ 1305/* file system info */ 1306 int link_count, total_link_count; 1307#ifdef CONFIG_SYSVIPC 1308/* ipc stuff */ 1309 struct sysv_sem sysvsem; 1310#endif 1311#ifdef CONFIG_DETECT_HUNG_TASK 1312/* hung task detection */ 1313 unsigned long last_switch_count; 1314#endif 1315/* CPU-specific state of this task */ 1316 struct thread_struct thread; 1317/* filesystem information */ 1318 struct fs_struct *fs; 1319/* open file information */ 1320 struct files_struct *files; 1321/* namespaces */ 1322 struct nsproxy *nsproxy; 1323/* signal handlers */ 1324 struct signal_struct *signal; 1325 struct sighand_struct *sighand; 1326 1327 sigset_t blocked, real_blocked; 1328 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1329 struct sigpending pending; 1330 1331 unsigned long sas_ss_sp; 1332 size_t sas_ss_size; 1333 int (*notifier)(void *priv); 1334 void *notifier_data; 1335 sigset_t *notifier_mask; 1336 struct audit_context *audit_context; 1337#ifdef CONFIG_AUDITSYSCALL 1338 uid_t loginuid; 1339 unsigned int sessionid; 1340#endif 1341 seccomp_t seccomp; 1342 1343/* Thread group tracking */ 1344 u32 parent_exec_id; 1345 u32 self_exec_id; 1346/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, 1347 * mempolicy */ 1348 spinlock_t alloc_lock; 1349 1350#ifdef CONFIG_GENERIC_HARDIRQS 1351 /* IRQ handler threads */ 1352 struct irqaction *irqaction; 1353#endif 1354 1355 /* Protection of the PI data structures: */ 1356 spinlock_t pi_lock; 1357 1358#ifdef CONFIG_RT_MUTEXES 1359 /* PI waiters blocked on a rt_mutex held by this task */ 1360 struct plist_head pi_waiters; 1361 /* Deadlock detection and priority inheritance handling */ 1362 struct rt_mutex_waiter *pi_blocked_on; 1363#endif 1364 1365#ifdef CONFIG_DEBUG_MUTEXES 1366 /* mutex deadlock detection */ 1367 struct mutex_waiter *blocked_on; 1368#endif 1369#ifdef CONFIG_TRACE_IRQFLAGS 1370 unsigned int irq_events; 1371 int hardirqs_enabled; 1372 unsigned long hardirq_enable_ip; 1373 unsigned int hardirq_enable_event; 1374 unsigned long hardirq_disable_ip; 1375 unsigned int hardirq_disable_event; 1376 int softirqs_enabled; 1377 unsigned long softirq_disable_ip; 1378 unsigned int softirq_disable_event; 1379 unsigned long softirq_enable_ip; 1380 unsigned int softirq_enable_event; 1381 int hardirq_context; 1382 int softirq_context; 1383#endif 1384#ifdef CONFIG_LOCKDEP 1385# define MAX_LOCK_DEPTH 48UL 1386 u64 curr_chain_key; 1387 int lockdep_depth; 1388 unsigned int lockdep_recursion; 1389 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1390 gfp_t lockdep_reclaim_gfp; 1391#endif 1392 1393/* journalling filesystem info */ 1394 void *journal_info; 1395 1396/* stacked block device info */ 1397 struct bio *bio_list, **bio_tail; 1398 1399/* VM state */ 1400 struct reclaim_state *reclaim_state; 1401 1402 struct backing_dev_info *backing_dev_info; 1403 1404 struct io_context *io_context; 1405 1406 unsigned long ptrace_message; 1407 siginfo_t *last_siginfo; /* For ptrace use. */ 1408 struct task_io_accounting ioac; 1409#if defined(CONFIG_TASK_XACCT) 1410 u64 acct_rss_mem1; /* accumulated rss usage */ 1411 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1412 cputime_t acct_timexpd; /* stime + utime since last update */ 1413#endif 1414#ifdef CONFIG_CPUSETS 1415 nodemask_t mems_allowed; /* Protected by alloc_lock */ 1416 int cpuset_mem_spread_rotor; 1417#endif 1418#ifdef CONFIG_CGROUPS 1419 /* Control Group info protected by css_set_lock */ 1420 struct css_set *cgroups; 1421 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1422 struct list_head cg_list; 1423#endif 1424#ifdef CONFIG_FUTEX 1425 struct robust_list_head __user *robust_list; 1426#ifdef CONFIG_COMPAT 1427 struct compat_robust_list_head __user *compat_robust_list; 1428#endif 1429 struct list_head pi_state_list; 1430 struct futex_pi_state *pi_state_cache; 1431#endif 1432#ifdef CONFIG_PERF_COUNTERS 1433 struct perf_counter_context *perf_counter_ctxp; 1434 struct mutex perf_counter_mutex; 1435 struct list_head perf_counter_list; 1436#endif 1437#ifdef CONFIG_NUMA 1438 struct mempolicy *mempolicy; /* Protected by alloc_lock */ 1439 short il_next; 1440#endif 1441 atomic_t fs_excl; /* holding fs exclusive resources */ 1442 struct rcu_head rcu; 1443 1444 /* 1445 * cache last used pipe for splice 1446 */ 1447 struct pipe_inode_info *splice_pipe; 1448#ifdef CONFIG_TASK_DELAY_ACCT 1449 struct task_delay_info *delays; 1450#endif 1451#ifdef CONFIG_FAULT_INJECTION 1452 int make_it_fail; 1453#endif 1454 struct prop_local_single dirties; 1455#ifdef CONFIG_LATENCYTOP 1456 int latency_record_count; 1457 struct latency_record latency_record[LT_SAVECOUNT]; 1458#endif 1459 /* 1460 * time slack values; these are used to round up poll() and 1461 * select() etc timeout values. These are in nanoseconds. 1462 */ 1463 unsigned long timer_slack_ns; 1464 unsigned long default_timer_slack_ns; 1465 1466 struct list_head *scm_work_list; 1467#ifdef CONFIG_FUNCTION_GRAPH_TRACER 1468 /* Index of current stored adress in ret_stack */ 1469 int curr_ret_stack; 1470 /* Stack of return addresses for return function tracing */ 1471 struct ftrace_ret_stack *ret_stack; 1472 /* time stamp for last schedule */ 1473 unsigned long long ftrace_timestamp; 1474 /* 1475 * Number of functions that haven't been traced 1476 * because of depth overrun. 1477 */ 1478 atomic_t trace_overrun; 1479 /* Pause for the tracing */ 1480 atomic_t tracing_graph_pause; 1481#endif 1482#ifdef CONFIG_TRACING 1483 /* state flags for use by tracers */ 1484 unsigned long trace; 1485 /* bitmask of trace recursion */ 1486 unsigned long trace_recursion; 1487#endif /* CONFIG_TRACING */ 1488}; 1489 1490/* Future-safe accessor for struct task_struct's cpus_allowed. */ 1491#define tsk_cpumask(tsk) (&(tsk)->cpus_allowed) 1492 1493/* 1494 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1495 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1496 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1497 * values are inverted: lower p->prio value means higher priority. 1498 * 1499 * The MAX_USER_RT_PRIO value allows the actual maximum 1500 * RT priority to be separate from the value exported to 1501 * user-space. This allows kernel threads to set their 1502 * priority to a value higher than any user task. Note: 1503 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1504 */ 1505 1506#define MAX_USER_RT_PRIO 100 1507#define MAX_RT_PRIO MAX_USER_RT_PRIO 1508 1509#define MAX_PRIO (MAX_RT_PRIO + 40) 1510#define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1511 1512static inline int rt_prio(int prio) 1513{ 1514 if (unlikely(prio < MAX_RT_PRIO)) 1515 return 1; 1516 return 0; 1517} 1518 1519static inline int rt_task(struct task_struct *p) 1520{ 1521 return rt_prio(p->prio); 1522} 1523 1524static inline struct pid *task_pid(struct task_struct *task) 1525{ 1526 return task->pids[PIDTYPE_PID].pid; 1527} 1528 1529static inline struct pid *task_tgid(struct task_struct *task) 1530{ 1531 return task->group_leader->pids[PIDTYPE_PID].pid; 1532} 1533 1534/* 1535 * Without tasklist or rcu lock it is not safe to dereference 1536 * the result of task_pgrp/task_session even if task == current, 1537 * we can race with another thread doing sys_setsid/sys_setpgid. 1538 */ 1539static inline struct pid *task_pgrp(struct task_struct *task) 1540{ 1541 return task->group_leader->pids[PIDTYPE_PGID].pid; 1542} 1543 1544static inline struct pid *task_session(struct task_struct *task) 1545{ 1546 return task->group_leader->pids[PIDTYPE_SID].pid; 1547} 1548 1549struct pid_namespace; 1550 1551/* 1552 * the helpers to get the task's different pids as they are seen 1553 * from various namespaces 1554 * 1555 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1556 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1557 * current. 1558 * task_xid_nr_ns() : id seen from the ns specified; 1559 * 1560 * set_task_vxid() : assigns a virtual id to a task; 1561 * 1562 * see also pid_nr() etc in include/linux/pid.h 1563 */ 1564pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, 1565 struct pid_namespace *ns); 1566 1567static inline pid_t task_pid_nr(struct task_struct *tsk) 1568{ 1569 return tsk->pid; 1570} 1571 1572static inline pid_t task_pid_nr_ns(struct task_struct *tsk, 1573 struct pid_namespace *ns) 1574{ 1575 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); 1576} 1577 1578static inline pid_t task_pid_vnr(struct task_struct *tsk) 1579{ 1580 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); 1581} 1582 1583 1584static inline pid_t task_tgid_nr(struct task_struct *tsk) 1585{ 1586 return tsk->tgid; 1587} 1588 1589pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1590 1591static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1592{ 1593 return pid_vnr(task_tgid(tsk)); 1594} 1595 1596 1597static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, 1598 struct pid_namespace *ns) 1599{ 1600 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); 1601} 1602 1603static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1604{ 1605 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); 1606} 1607 1608 1609static inline pid_t task_session_nr_ns(struct task_struct *tsk, 1610 struct pid_namespace *ns) 1611{ 1612 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); 1613} 1614 1615static inline pid_t task_session_vnr(struct task_struct *tsk) 1616{ 1617 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); 1618} 1619 1620/* obsolete, do not use */ 1621static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1622{ 1623 return task_pgrp_nr_ns(tsk, &init_pid_ns); 1624} 1625 1626/** 1627 * pid_alive - check that a task structure is not stale 1628 * @p: Task structure to be checked. 1629 * 1630 * Test if a process is not yet dead (at most zombie state) 1631 * If pid_alive fails, then pointers within the task structure 1632 * can be stale and must not be dereferenced. 1633 */ 1634static inline int pid_alive(struct task_struct *p) 1635{ 1636 return p->pids[PIDTYPE_PID].pid != NULL; 1637} 1638 1639/** 1640 * is_global_init - check if a task structure is init 1641 * @tsk: Task structure to be checked. 1642 * 1643 * Check if a task structure is the first user space task the kernel created. 1644 */ 1645static inline int is_global_init(struct task_struct *tsk) 1646{ 1647 return tsk->pid == 1; 1648} 1649 1650/* 1651 * is_container_init: 1652 * check whether in the task is init in its own pid namespace. 1653 */ 1654extern int is_container_init(struct task_struct *tsk); 1655 1656extern struct pid *cad_pid; 1657 1658extern void free_task(struct task_struct *tsk); 1659#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1660 1661extern void __put_task_struct(struct task_struct *t); 1662 1663static inline void put_task_struct(struct task_struct *t) 1664{ 1665 if (atomic_dec_and_test(&t->usage)) 1666 __put_task_struct(t); 1667} 1668 1669extern cputime_t task_utime(struct task_struct *p); 1670extern cputime_t task_stime(struct task_struct *p); 1671extern cputime_t task_gtime(struct task_struct *p); 1672 1673/* 1674 * Per process flags 1675 */ 1676#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1677 /* Not implemented yet, only for 486*/ 1678#define PF_STARTING 0x00000002 /* being created */ 1679#define PF_EXITING 0x00000004 /* getting shut down */ 1680#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1681#define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1682#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1683#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1684#define PF_DUMPCORE 0x00000200 /* dumped core */ 1685#define PF_SIGNALED 0x00000400 /* killed by a signal */ 1686#define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1687#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1688#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1689#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1690#define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1691#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1692#define PF_KSWAPD 0x00040000 /* I am kswapd */ 1693#define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1694#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1695#define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1696#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1697#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1698#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1699#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1700#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1701#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1702#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1703#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1704#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1705 1706/* 1707 * Only the _current_ task can read/write to tsk->flags, but other 1708 * tasks can access tsk->flags in readonly mode for example 1709 * with tsk_used_math (like during threaded core dumping). 1710 * There is however an exception to this rule during ptrace 1711 * or during fork: the ptracer task is allowed to write to the 1712 * child->flags of its traced child (same goes for fork, the parent 1713 * can write to the child->flags), because we're guaranteed the 1714 * child is not running and in turn not changing child->flags 1715 * at the same time the parent does it. 1716 */ 1717#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1718#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1719#define clear_used_math() clear_stopped_child_used_math(current) 1720#define set_used_math() set_stopped_child_used_math(current) 1721#define conditional_stopped_child_used_math(condition, child) \ 1722 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1723#define conditional_used_math(condition) \ 1724 conditional_stopped_child_used_math(condition, current) 1725#define copy_to_stopped_child_used_math(child) \ 1726 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1727/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1728#define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1729#define used_math() tsk_used_math(current) 1730 1731#ifdef CONFIG_SMP 1732extern int set_cpus_allowed_ptr(struct task_struct *p, 1733 const struct cpumask *new_mask); 1734#else 1735static inline int set_cpus_allowed_ptr(struct task_struct *p, 1736 const struct cpumask *new_mask) 1737{ 1738 if (!cpumask_test_cpu(0, new_mask)) 1739 return -EINVAL; 1740 return 0; 1741} 1742#endif 1743static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1744{ 1745 return set_cpus_allowed_ptr(p, &new_mask); 1746} 1747 1748/* 1749 * Architectures can set this to 1 if they have specified 1750 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, 1751 * but then during bootup it turns out that sched_clock() 1752 * is reliable after all: 1753 */ 1754#ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1755extern int sched_clock_stable; 1756#endif 1757 1758extern unsigned long long sched_clock(void); 1759 1760extern void sched_clock_init(void); 1761extern u64 sched_clock_cpu(int cpu); 1762 1763#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1764static inline void sched_clock_tick(void) 1765{ 1766} 1767 1768static inline void sched_clock_idle_sleep_event(void) 1769{ 1770} 1771 1772static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1773{ 1774} 1775#else 1776extern void sched_clock_tick(void); 1777extern void sched_clock_idle_sleep_event(void); 1778extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1779#endif 1780 1781/* 1782 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1783 * clock constructed from sched_clock(): 1784 */ 1785extern unsigned long long cpu_clock(int cpu); 1786 1787extern unsigned long long 1788task_sched_runtime(struct task_struct *task); 1789extern unsigned long long thread_group_sched_runtime(struct task_struct *task); 1790 1791/* sched_exec is called by processes performing an exec */ 1792#ifdef CONFIG_SMP 1793extern void sched_exec(void); 1794#else 1795#define sched_exec() {} 1796#endif 1797 1798extern void sched_clock_idle_sleep_event(void); 1799extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1800 1801#ifdef CONFIG_HOTPLUG_CPU 1802extern void idle_task_exit(void); 1803#else 1804static inline void idle_task_exit(void) {} 1805#endif 1806 1807extern void sched_idle_next(void); 1808 1809#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1810extern void wake_up_idle_cpu(int cpu); 1811#else 1812static inline void wake_up_idle_cpu(int cpu) { } 1813#endif 1814 1815extern unsigned int sysctl_sched_latency; 1816extern unsigned int sysctl_sched_min_granularity; 1817extern unsigned int sysctl_sched_wakeup_granularity; 1818extern unsigned int sysctl_sched_shares_ratelimit; 1819extern unsigned int sysctl_sched_shares_thresh; 1820#ifdef CONFIG_SCHED_DEBUG 1821extern unsigned int sysctl_sched_child_runs_first; 1822extern unsigned int sysctl_sched_features; 1823extern unsigned int sysctl_sched_migration_cost; 1824extern unsigned int sysctl_sched_nr_migrate; 1825extern unsigned int sysctl_timer_migration; 1826 1827int sched_nr_latency_handler(struct ctl_table *table, int write, 1828 struct file *file, void __user *buffer, size_t *length, 1829 loff_t *ppos); 1830#endif 1831#ifdef CONFIG_SCHED_DEBUG 1832static inline unsigned int get_sysctl_timer_migration(void) 1833{ 1834 return sysctl_timer_migration; 1835} 1836#else 1837static inline unsigned int get_sysctl_timer_migration(void) 1838{ 1839 return 1; 1840} 1841#endif 1842extern unsigned int sysctl_sched_rt_period; 1843extern int sysctl_sched_rt_runtime; 1844 1845int sched_rt_handler(struct ctl_table *table, int write, 1846 struct file *filp, void __user *buffer, size_t *lenp, 1847 loff_t *ppos); 1848 1849extern unsigned int sysctl_sched_compat_yield; 1850 1851#ifdef CONFIG_RT_MUTEXES 1852extern int rt_mutex_getprio(struct task_struct *p); 1853extern void rt_mutex_setprio(struct task_struct *p, int prio); 1854extern void rt_mutex_adjust_pi(struct task_struct *p); 1855#else 1856static inline int rt_mutex_getprio(struct task_struct *p) 1857{ 1858 return p->normal_prio; 1859} 1860# define rt_mutex_adjust_pi(p) do { } while (0) 1861#endif 1862 1863extern void set_user_nice(struct task_struct *p, long nice); 1864extern int task_prio(const struct task_struct *p); 1865extern int task_nice(const struct task_struct *p); 1866extern int can_nice(const struct task_struct *p, const int nice); 1867extern int task_curr(const struct task_struct *p); 1868extern int idle_cpu(int cpu); 1869extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1870extern int sched_setscheduler_nocheck(struct task_struct *, int, 1871 struct sched_param *); 1872extern struct task_struct *idle_task(int cpu); 1873extern struct task_struct *curr_task(int cpu); 1874extern void set_curr_task(int cpu, struct task_struct *p); 1875 1876void yield(void); 1877 1878/* 1879 * The default (Linux) execution domain. 1880 */ 1881extern struct exec_domain default_exec_domain; 1882 1883union thread_union { 1884 struct thread_info thread_info; 1885 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1886}; 1887 1888#ifndef __HAVE_ARCH_KSTACK_END 1889static inline int kstack_end(void *addr) 1890{ 1891 /* Reliable end of stack detection: 1892 * Some APM bios versions misalign the stack 1893 */ 1894 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1895} 1896#endif 1897 1898extern union thread_union init_thread_union; 1899extern struct task_struct init_task; 1900 1901extern struct mm_struct init_mm; 1902 1903extern struct pid_namespace init_pid_ns; 1904 1905/* 1906 * find a task by one of its numerical ids 1907 * 1908 * find_task_by_pid_ns(): 1909 * finds a task by its pid in the specified namespace 1910 * find_task_by_vpid(): 1911 * finds a task by its virtual pid 1912 * 1913 * see also find_vpid() etc in include/linux/pid.h 1914 */ 1915 1916extern struct task_struct *find_task_by_vpid(pid_t nr); 1917extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1918 struct pid_namespace *ns); 1919 1920extern void __set_special_pids(struct pid *pid); 1921 1922/* per-UID process charging. */ 1923extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1924static inline struct user_struct *get_uid(struct user_struct *u) 1925{ 1926 atomic_inc(&u->__count); 1927 return u; 1928} 1929extern void free_uid(struct user_struct *); 1930extern void release_uids(struct user_namespace *ns); 1931 1932#include <asm/current.h> 1933 1934extern void do_timer(unsigned long ticks); 1935 1936extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1937extern int wake_up_process(struct task_struct *tsk); 1938extern void wake_up_new_task(struct task_struct *tsk, 1939 unsigned long clone_flags); 1940#ifdef CONFIG_SMP 1941 extern void kick_process(struct task_struct *tsk); 1942#else 1943 static inline void kick_process(struct task_struct *tsk) { } 1944#endif 1945extern void sched_fork(struct task_struct *p, int clone_flags); 1946extern void sched_dead(struct task_struct *p); 1947 1948extern void proc_caches_init(void); 1949extern void flush_signals(struct task_struct *); 1950extern void __flush_signals(struct task_struct *); 1951extern void ignore_signals(struct task_struct *); 1952extern void flush_signal_handlers(struct task_struct *, int force_default); 1953extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1954 1955static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1956{ 1957 unsigned long flags; 1958 int ret; 1959 1960 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1961 ret = dequeue_signal(tsk, mask, info); 1962 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1963 1964 return ret; 1965} 1966 1967extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1968 sigset_t *mask); 1969extern void unblock_all_signals(void); 1970extern void release_task(struct task_struct * p); 1971extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1972extern int force_sigsegv(int, struct task_struct *); 1973extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1974extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1975extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1976extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1977extern int kill_pgrp(struct pid *pid, int sig, int priv); 1978extern int kill_pid(struct pid *pid, int sig, int priv); 1979extern int kill_proc_info(int, struct siginfo *, pid_t); 1980extern int do_notify_parent(struct task_struct *, int); 1981extern void force_sig(int, struct task_struct *); 1982extern void force_sig_specific(int, struct task_struct *); 1983extern int send_sig(int, struct task_struct *, int); 1984extern void zap_other_threads(struct task_struct *p); 1985extern struct sigqueue *sigqueue_alloc(void); 1986extern void sigqueue_free(struct sigqueue *); 1987extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 1988extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1989extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1990 1991static inline int kill_cad_pid(int sig, int priv) 1992{ 1993 return kill_pid(cad_pid, sig, priv); 1994} 1995 1996/* These can be the second arg to send_sig_info/send_group_sig_info. */ 1997#define SEND_SIG_NOINFO ((struct siginfo *) 0) 1998#define SEND_SIG_PRIV ((struct siginfo *) 1) 1999#define SEND_SIG_FORCED ((struct siginfo *) 2) 2000 2001static inline int is_si_special(const struct siginfo *info) 2002{ 2003 return info <= SEND_SIG_FORCED; 2004} 2005 2006/* True if we are on the alternate signal stack. */ 2007 2008static inline int on_sig_stack(unsigned long sp) 2009{ 2010 return (sp - current->sas_ss_sp < current->sas_ss_size); 2011} 2012 2013static inline int sas_ss_flags(unsigned long sp) 2014{ 2015 return (current->sas_ss_size == 0 ? SS_DISABLE 2016 : on_sig_stack(sp) ? SS_ONSTACK : 0); 2017} 2018 2019/* 2020 * Routines for handling mm_structs 2021 */ 2022extern struct mm_struct * mm_alloc(void); 2023 2024/* mmdrop drops the mm and the page tables */ 2025extern void __mmdrop(struct mm_struct *); 2026static inline void mmdrop(struct mm_struct * mm) 2027{ 2028 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 2029 __mmdrop(mm); 2030} 2031 2032/* mmput gets rid of the mappings and all user-space */ 2033extern void mmput(struct mm_struct *); 2034/* Grab a reference to a task's mm, if it is not already going away */ 2035extern struct mm_struct *get_task_mm(struct task_struct *task); 2036/* Remove the current tasks stale references to the old mm_struct */ 2037extern void mm_release(struct task_struct *, struct mm_struct *); 2038/* Allocate a new mm structure and copy contents from tsk->mm */ 2039extern struct mm_struct *dup_mm(struct task_struct *tsk); 2040 2041extern int copy_thread(unsigned long, unsigned long, unsigned long, 2042 struct task_struct *, struct pt_regs *); 2043extern void flush_thread(void); 2044extern void exit_thread(void); 2045 2046extern void exit_files(struct task_struct *); 2047extern void __cleanup_signal(struct signal_struct *); 2048extern void __cleanup_sighand(struct sighand_struct *); 2049 2050extern void exit_itimers(struct signal_struct *); 2051extern void flush_itimer_signals(void); 2052 2053extern NORET_TYPE void do_group_exit(int); 2054 2055extern void daemonize(const char *, ...); 2056extern int allow_signal(int); 2057extern int disallow_signal(int); 2058 2059extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 2060extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 2061struct task_struct *fork_idle(int); 2062 2063extern void set_task_comm(struct task_struct *tsk, char *from); 2064extern char *get_task_comm(char *to, struct task_struct *tsk); 2065 2066#ifdef CONFIG_SMP 2067extern void wait_task_context_switch(struct task_struct *p); 2068extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 2069#else 2070static inline void wait_task_context_switch(struct task_struct *p) {} 2071static inline unsigned long wait_task_inactive(struct task_struct *p, 2072 long match_state) 2073{ 2074 return 1; 2075} 2076#endif 2077 2078#define next_task(p) \ 2079 list_entry_rcu((p)->tasks.next, struct task_struct, tasks) 2080 2081#define for_each_process(p) \ 2082 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 2083 2084extern bool is_single_threaded(struct task_struct *); 2085 2086/* 2087 * Careful: do_each_thread/while_each_thread is a double loop so 2088 * 'break' will not work as expected - use goto instead. 2089 */ 2090#define do_each_thread(g, t) \ 2091 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 2092 2093#define while_each_thread(g, t) \ 2094 while ((t = next_thread(t)) != g) 2095 2096/* de_thread depends on thread_group_leader not being a pid based check */ 2097#define thread_group_leader(p) (p == p->group_leader) 2098 2099/* Do to the insanities of de_thread it is possible for a process 2100 * to have the pid of the thread group leader without actually being 2101 * the thread group leader. For iteration through the pids in proc 2102 * all we care about is that we have a task with the appropriate 2103 * pid, we don't actually care if we have the right task. 2104 */ 2105static inline int has_group_leader_pid(struct task_struct *p) 2106{ 2107 return p->pid == p->tgid; 2108} 2109 2110static inline 2111int same_thread_group(struct task_struct *p1, struct task_struct *p2) 2112{ 2113 return p1->tgid == p2->tgid; 2114} 2115 2116static inline struct task_struct *next_thread(const struct task_struct *p) 2117{ 2118 return list_entry_rcu(p->thread_group.next, 2119 struct task_struct, thread_group); 2120} 2121 2122static inline int thread_group_empty(struct task_struct *p) 2123{ 2124 return list_empty(&p->thread_group); 2125} 2126 2127#define delay_group_leader(p) \ 2128 (thread_group_leader(p) && !thread_group_empty(p)) 2129 2130static inline int task_detached(struct task_struct *p) 2131{ 2132 return p->exit_signal == -1; 2133} 2134 2135/* 2136 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 2137 * subscriptions and synchronises with wait4(). Also used in procfs. Also 2138 * pins the final release of task.io_context. Also protects ->cpuset and 2139 * ->cgroup.subsys[]. 2140 * 2141 * Nests both inside and outside of read_lock(&tasklist_lock). 2142 * It must not be nested with write_lock_irq(&tasklist_lock), 2143 * neither inside nor outside. 2144 */ 2145static inline void task_lock(struct task_struct *p) 2146{ 2147 spin_lock(&p->alloc_lock); 2148} 2149 2150static inline void task_unlock(struct task_struct *p) 2151{ 2152 spin_unlock(&p->alloc_lock); 2153} 2154 2155extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 2156 unsigned long *flags); 2157 2158static inline void unlock_task_sighand(struct task_struct *tsk, 2159 unsigned long *flags) 2160{ 2161 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 2162} 2163 2164#ifndef __HAVE_THREAD_FUNCTIONS 2165 2166#define task_thread_info(task) ((struct thread_info *)(task)->stack) 2167#define task_stack_page(task) ((task)->stack) 2168 2169static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 2170{ 2171 *task_thread_info(p) = *task_thread_info(org); 2172 task_thread_info(p)->task = p; 2173} 2174 2175static inline unsigned long *end_of_stack(struct task_struct *p) 2176{ 2177 return (unsigned long *)(task_thread_info(p) + 1); 2178} 2179 2180#endif 2181 2182static inline int object_is_on_stack(void *obj) 2183{ 2184 void *stack = task_stack_page(current); 2185 2186 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 2187} 2188 2189extern void thread_info_cache_init(void); 2190 2191#ifdef CONFIG_DEBUG_STACK_USAGE 2192static inline unsigned long stack_not_used(struct task_struct *p) 2193{ 2194 unsigned long *n = end_of_stack(p); 2195 2196 do { /* Skip over canary */ 2197 n++; 2198 } while (!*n); 2199 2200 return (unsigned long)n - (unsigned long)end_of_stack(p); 2201} 2202#endif 2203 2204/* set thread flags in other task's structures 2205 * - see asm/thread_info.h for TIF_xxxx flags available 2206 */ 2207static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 2208{ 2209 set_ti_thread_flag(task_thread_info(tsk), flag); 2210} 2211 2212static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2213{ 2214 clear_ti_thread_flag(task_thread_info(tsk), flag); 2215} 2216 2217static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2218{ 2219 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2220} 2221 2222static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2223{ 2224 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2225} 2226 2227static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2228{ 2229 return test_ti_thread_flag(task_thread_info(tsk), flag); 2230} 2231 2232static inline void set_tsk_need_resched(struct task_struct *tsk) 2233{ 2234 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2235} 2236 2237static inline void clear_tsk_need_resched(struct task_struct *tsk) 2238{ 2239 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2240} 2241 2242static inline int test_tsk_need_resched(struct task_struct *tsk) 2243{ 2244 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2245} 2246 2247static inline int restart_syscall(void) 2248{ 2249 set_tsk_thread_flag(current, TIF_SIGPENDING); 2250 return -ERESTARTNOINTR; 2251} 2252 2253static inline int signal_pending(struct task_struct *p) 2254{ 2255 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2256} 2257 2258extern int __fatal_signal_pending(struct task_struct *p); 2259 2260static inline int fatal_signal_pending(struct task_struct *p) 2261{ 2262 return signal_pending(p) && __fatal_signal_pending(p); 2263} 2264 2265static inline int signal_pending_state(long state, struct task_struct *p) 2266{ 2267 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2268 return 0; 2269 if (!signal_pending(p)) 2270 return 0; 2271 2272 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2273} 2274 2275static inline int need_resched(void) 2276{ 2277 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2278} 2279 2280/* 2281 * cond_resched() and cond_resched_lock(): latency reduction via 2282 * explicit rescheduling in places that are safe. The return 2283 * value indicates whether a reschedule was done in fact. 2284 * cond_resched_lock() will drop the spinlock before scheduling, 2285 * cond_resched_softirq() will enable bhs before scheduling. 2286 */ 2287extern int _cond_resched(void); 2288 2289#define cond_resched() ({ \ 2290 __might_sleep(__FILE__, __LINE__, 0); \ 2291 _cond_resched(); \ 2292}) 2293 2294extern int __cond_resched_lock(spinlock_t *lock); 2295 2296#ifdef CONFIG_PREEMPT 2297#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET 2298#else 2299#define PREEMPT_LOCK_OFFSET 0 2300#endif 2301 2302#define cond_resched_lock(lock) ({ \ 2303 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \ 2304 __cond_resched_lock(lock); \ 2305}) 2306 2307extern int __cond_resched_softirq(void); 2308 2309#define cond_resched_softirq() ({ \ 2310 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \ 2311 __cond_resched_softirq(); \ 2312}) 2313 2314/* 2315 * Does a critical section need to be broken due to another 2316 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2317 * but a general need for low latency) 2318 */ 2319static inline int spin_needbreak(spinlock_t *lock) 2320{ 2321#ifdef CONFIG_PREEMPT 2322 return spin_is_contended(lock); 2323#else 2324 return 0; 2325#endif 2326} 2327 2328/* 2329 * Thread group CPU time accounting. 2330 */ 2331void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); 2332void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); 2333 2334static inline void thread_group_cputime_init(struct signal_struct *sig) 2335{ 2336 sig->cputimer.cputime = INIT_CPUTIME; 2337 spin_lock_init(&sig->cputimer.lock); 2338 sig->cputimer.running = 0; 2339} 2340 2341static inline void thread_group_cputime_free(struct signal_struct *sig) 2342{ 2343} 2344 2345/* 2346 * Reevaluate whether the task has signals pending delivery. 2347 * Wake the task if so. 2348 * This is required every time the blocked sigset_t changes. 2349 * callers must hold sighand->siglock. 2350 */ 2351extern void recalc_sigpending_and_wake(struct task_struct *t); 2352extern void recalc_sigpending(void); 2353 2354extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2355 2356/* 2357 * Wrappers for p->thread_info->cpu access. No-op on UP. 2358 */ 2359#ifdef CONFIG_SMP 2360 2361static inline unsigned int task_cpu(const struct task_struct *p) 2362{ 2363 return task_thread_info(p)->cpu; 2364} 2365 2366extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2367 2368#else 2369 2370static inline unsigned int task_cpu(const struct task_struct *p) 2371{ 2372 return 0; 2373} 2374 2375static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2376{ 2377} 2378 2379#endif /* CONFIG_SMP */ 2380 2381extern void arch_pick_mmap_layout(struct mm_struct *mm); 2382 2383#ifdef CONFIG_TRACING 2384extern void 2385__trace_special(void *__tr, void *__data, 2386 unsigned long arg1, unsigned long arg2, unsigned long arg3); 2387#else 2388static inline void 2389__trace_special(void *__tr, void *__data, 2390 unsigned long arg1, unsigned long arg2, unsigned long arg3) 2391{ 2392} 2393#endif 2394 2395extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); 2396extern long sched_getaffinity(pid_t pid, struct cpumask *mask); 2397 2398extern void normalize_rt_tasks(void); 2399 2400#ifdef CONFIG_GROUP_SCHED 2401 2402extern struct task_group init_task_group; 2403#ifdef CONFIG_USER_SCHED 2404extern struct task_group root_task_group; 2405extern void set_tg_uid(struct user_struct *user); 2406#endif 2407 2408extern struct task_group *sched_create_group(struct task_group *parent); 2409extern void sched_destroy_group(struct task_group *tg); 2410extern void sched_move_task(struct task_struct *tsk); 2411#ifdef CONFIG_FAIR_GROUP_SCHED 2412extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2413extern unsigned long sched_group_shares(struct task_group *tg); 2414#endif 2415#ifdef CONFIG_RT_GROUP_SCHED 2416extern int sched_group_set_rt_runtime(struct task_group *tg, 2417 long rt_runtime_us); 2418extern long sched_group_rt_runtime(struct task_group *tg); 2419extern int sched_group_set_rt_period(struct task_group *tg, 2420 long rt_period_us); 2421extern long sched_group_rt_period(struct task_group *tg); 2422extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk); 2423#endif 2424#endif 2425 2426extern int task_can_switch_user(struct user_struct *up, 2427 struct task_struct *tsk); 2428 2429#ifdef CONFIG_TASK_XACCT 2430static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2431{ 2432 tsk->ioac.rchar += amt; 2433} 2434 2435static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2436{ 2437 tsk->ioac.wchar += amt; 2438} 2439 2440static inline void inc_syscr(struct task_struct *tsk) 2441{ 2442 tsk->ioac.syscr++; 2443} 2444 2445static inline void inc_syscw(struct task_struct *tsk) 2446{ 2447 tsk->ioac.syscw++; 2448} 2449#else 2450static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2451{ 2452} 2453 2454static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2455{ 2456} 2457 2458static inline void inc_syscr(struct task_struct *tsk) 2459{ 2460} 2461 2462static inline void inc_syscw(struct task_struct *tsk) 2463{ 2464} 2465#endif 2466 2467#ifndef TASK_SIZE_OF 2468#define TASK_SIZE_OF(tsk) TASK_SIZE 2469#endif 2470 2471/* 2472 * Call the function if the target task is executing on a CPU right now: 2473 */ 2474extern void task_oncpu_function_call(struct task_struct *p, 2475 void (*func) (void *info), void *info); 2476 2477 2478#ifdef CONFIG_MM_OWNER 2479extern void mm_update_next_owner(struct mm_struct *mm); 2480extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2481#else 2482static inline void mm_update_next_owner(struct mm_struct *mm) 2483{ 2484} 2485 2486static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2487{ 2488} 2489#endif /* CONFIG_MM_OWNER */ 2490 2491#define TASK_STATE_TO_CHAR_STR "RSDTtZX" 2492 2493#endif /* __KERNEL__ */ 2494 2495#endif 2496