sched.h revision 09faef11df8c559a23e2405d123cb2683733a79a
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_list; 101struct fs_struct; 102struct perf_event_context; 103 104/* 105 * List of flags we want to share for kernel threads, 106 * if only because they are not used by them anyway. 107 */ 108#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 109 110/* 111 * These are the constant used to fake the fixed-point load-average 112 * counting. Some notes: 113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 114 * a load-average precision of 10 bits integer + 11 bits fractional 115 * - if you want to count load-averages more often, you need more 116 * precision, or rounding will get you. With 2-second counting freq, 117 * the EXP_n values would be 1981, 2034 and 2043 if still using only 118 * 11 bit fractions. 119 */ 120extern unsigned long avenrun[]; /* Load averages */ 121extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift); 122 123#define FSHIFT 11 /* nr of bits of precision */ 124#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 125#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 126#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 127#define EXP_5 2014 /* 1/exp(5sec/5min) */ 128#define EXP_15 2037 /* 1/exp(5sec/15min) */ 129 130#define CALC_LOAD(load,exp,n) \ 131 load *= exp; \ 132 load += n*(FIXED_1-exp); \ 133 load >>= FSHIFT; 134 135extern unsigned long total_forks; 136extern int nr_threads; 137DECLARE_PER_CPU(unsigned long, process_counts); 138extern int nr_processes(void); 139extern unsigned long nr_running(void); 140extern unsigned long nr_uninterruptible(void); 141extern unsigned long nr_iowait(void); 142extern unsigned long nr_iowait_cpu(void); 143extern unsigned long this_cpu_load(void); 144 145 146extern void calc_global_load(void); 147 148extern unsigned long get_parent_ip(unsigned long addr); 149 150struct seq_file; 151struct cfs_rq; 152struct task_group; 153#ifdef CONFIG_SCHED_DEBUG 154extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 155extern void proc_sched_set_task(struct task_struct *p); 156extern void 157print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 158#else 159static inline void 160proc_sched_show_task(struct task_struct *p, struct seq_file *m) 161{ 162} 163static inline void proc_sched_set_task(struct task_struct *p) 164{ 165} 166static inline void 167print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 168{ 169} 170#endif 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#define TASK_WAKING 256 194#define TASK_STATE_MAX 512 195 196#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW" 197 198extern char ___assert_task_state[1 - 2*!!( 199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)]; 200 201/* Convenience macros for the sake of set_task_state */ 202#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 203#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 204#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 205 206/* Convenience macros for the sake of wake_up */ 207#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 208#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 209 210/* get_task_state() */ 211#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 213 __TASK_TRACED) 214 215#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 216#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 217#define task_is_stopped_or_traced(task) \ 218 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 219#define task_contributes_to_load(task) \ 220 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ 221 (task->flags & PF_FREEZING) == 0) 222 223#define __set_task_state(tsk, state_value) \ 224 do { (tsk)->state = (state_value); } while (0) 225#define set_task_state(tsk, state_value) \ 226 set_mb((tsk)->state, (state_value)) 227 228/* 229 * set_current_state() includes a barrier so that the write of current->state 230 * is correctly serialised wrt the caller's subsequent test of whether to 231 * actually sleep: 232 * 233 * set_current_state(TASK_UNINTERRUPTIBLE); 234 * if (do_i_need_to_sleep()) 235 * schedule(); 236 * 237 * If the caller does not need such serialisation then use __set_current_state() 238 */ 239#define __set_current_state(state_value) \ 240 do { current->state = (state_value); } while (0) 241#define set_current_state(state_value) \ 242 set_mb(current->state, (state_value)) 243 244/* Task command name length */ 245#define TASK_COMM_LEN 16 246 247#include <linux/spinlock.h> 248 249/* 250 * This serializes "schedule()" and also protects 251 * the run-queue from deletions/modifications (but 252 * _adding_ to the beginning of the run-queue has 253 * a separate lock). 254 */ 255extern rwlock_t tasklist_lock; 256extern spinlock_t mmlist_lock; 257 258struct task_struct; 259 260#ifdef CONFIG_PROVE_RCU 261extern int lockdep_tasklist_lock_is_held(void); 262#endif /* #ifdef CONFIG_PROVE_RCU */ 263 264extern void sched_init(void); 265extern void sched_init_smp(void); 266extern asmlinkage void schedule_tail(struct task_struct *prev); 267extern void init_idle(struct task_struct *idle, int cpu); 268extern void init_idle_bootup_task(struct task_struct *idle); 269 270extern int runqueue_is_locked(int cpu); 271extern void task_rq_unlock_wait(struct task_struct *p); 272 273extern cpumask_var_t nohz_cpu_mask; 274#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 275extern int select_nohz_load_balancer(int cpu); 276extern int get_nohz_load_balancer(void); 277extern int nohz_ratelimit(int cpu); 278#else 279static inline int select_nohz_load_balancer(int cpu) 280{ 281 return 0; 282} 283 284static inline int nohz_ratelimit(int cpu) 285{ 286 return 0; 287} 288#endif 289 290/* 291 * Only dump TASK_* tasks. (0 for all tasks) 292 */ 293extern void show_state_filter(unsigned long state_filter); 294 295static inline void show_state(void) 296{ 297 show_state_filter(0); 298} 299 300extern void show_regs(struct pt_regs *); 301 302/* 303 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 304 * task), SP is the stack pointer of the first frame that should be shown in the back 305 * trace (or NULL if the entire call-chain of the task should be shown). 306 */ 307extern void show_stack(struct task_struct *task, unsigned long *sp); 308 309void io_schedule(void); 310long io_schedule_timeout(long timeout); 311 312extern void cpu_init (void); 313extern void trap_init(void); 314extern void update_process_times(int user); 315extern void scheduler_tick(void); 316 317extern void sched_show_task(struct task_struct *p); 318 319#ifdef CONFIG_DETECT_SOFTLOCKUP 320extern void softlockup_tick(void); 321extern void touch_softlockup_watchdog(void); 322extern void touch_softlockup_watchdog_sync(void); 323extern void touch_all_softlockup_watchdogs(void); 324extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write, 325 void __user *buffer, 326 size_t *lenp, loff_t *ppos); 327extern unsigned int softlockup_panic; 328extern int softlockup_thresh; 329#else 330static inline void softlockup_tick(void) 331{ 332} 333static inline void touch_softlockup_watchdog(void) 334{ 335} 336static inline void touch_softlockup_watchdog_sync(void) 337{ 338} 339static inline void touch_all_softlockup_watchdogs(void) 340{ 341} 342#endif 343 344#ifdef CONFIG_DETECT_HUNG_TASK 345extern unsigned int sysctl_hung_task_panic; 346extern unsigned long sysctl_hung_task_check_count; 347extern unsigned long sysctl_hung_task_timeout_secs; 348extern unsigned long sysctl_hung_task_warnings; 349extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write, 350 void __user *buffer, 351 size_t *lenp, loff_t *ppos); 352#endif 353 354/* Attach to any functions which should be ignored in wchan output. */ 355#define __sched __attribute__((__section__(".sched.text"))) 356 357/* Linker adds these: start and end of __sched functions */ 358extern char __sched_text_start[], __sched_text_end[]; 359 360/* Is this address in the __sched functions? */ 361extern int in_sched_functions(unsigned long addr); 362 363#define MAX_SCHEDULE_TIMEOUT LONG_MAX 364extern signed long schedule_timeout(signed long timeout); 365extern signed long schedule_timeout_interruptible(signed long timeout); 366extern signed long schedule_timeout_killable(signed long timeout); 367extern signed long schedule_timeout_uninterruptible(signed long timeout); 368asmlinkage void schedule(void); 369extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner); 370 371struct nsproxy; 372struct user_namespace; 373 374/* 375 * Default maximum number of active map areas, this limits the number of vmas 376 * per mm struct. Users can overwrite this number by sysctl but there is a 377 * problem. 378 * 379 * When a program's coredump is generated as ELF format, a section is created 380 * per a vma. In ELF, the number of sections is represented in unsigned short. 381 * This means the number of sections should be smaller than 65535 at coredump. 382 * Because the kernel adds some informative sections to a image of program at 383 * generating coredump, we need some margin. The number of extra sections is 384 * 1-3 now and depends on arch. We use "5" as safe margin, here. 385 */ 386#define MAPCOUNT_ELF_CORE_MARGIN (5) 387#define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN) 388 389extern int sysctl_max_map_count; 390 391#include <linux/aio.h> 392 393#ifdef CONFIG_MMU 394extern void arch_pick_mmap_layout(struct mm_struct *mm); 395extern unsigned long 396arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 397 unsigned long, unsigned long); 398extern unsigned long 399arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 400 unsigned long len, unsigned long pgoff, 401 unsigned long flags); 402extern void arch_unmap_area(struct mm_struct *, unsigned long); 403extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 404#else 405static inline void arch_pick_mmap_layout(struct mm_struct *mm) {} 406#endif 407 408 409extern void set_dumpable(struct mm_struct *mm, int value); 410extern int get_dumpable(struct mm_struct *mm); 411 412/* mm flags */ 413/* dumpable bits */ 414#define MMF_DUMPABLE 0 /* core dump is permitted */ 415#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 416 417#define MMF_DUMPABLE_BITS 2 418#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1) 419 420/* coredump filter bits */ 421#define MMF_DUMP_ANON_PRIVATE 2 422#define MMF_DUMP_ANON_SHARED 3 423#define MMF_DUMP_MAPPED_PRIVATE 4 424#define MMF_DUMP_MAPPED_SHARED 5 425#define MMF_DUMP_ELF_HEADERS 6 426#define MMF_DUMP_HUGETLB_PRIVATE 7 427#define MMF_DUMP_HUGETLB_SHARED 8 428 429#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 430#define MMF_DUMP_FILTER_BITS 7 431#define MMF_DUMP_FILTER_MASK \ 432 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 433#define MMF_DUMP_FILTER_DEFAULT \ 434 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ 435 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) 436 437#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS 438# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) 439#else 440# define MMF_DUMP_MASK_DEFAULT_ELF 0 441#endif 442 /* leave room for more dump flags */ 443#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */ 444 445#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK) 446 447struct sighand_struct { 448 atomic_t count; 449 struct k_sigaction action[_NSIG]; 450 spinlock_t siglock; 451 wait_queue_head_t signalfd_wqh; 452}; 453 454struct pacct_struct { 455 int ac_flag; 456 long ac_exitcode; 457 unsigned long ac_mem; 458 cputime_t ac_utime, ac_stime; 459 unsigned long ac_minflt, ac_majflt; 460}; 461 462struct cpu_itimer { 463 cputime_t expires; 464 cputime_t incr; 465 u32 error; 466 u32 incr_error; 467}; 468 469/** 470 * struct task_cputime - collected CPU time counts 471 * @utime: time spent in user mode, in &cputime_t units 472 * @stime: time spent in kernel mode, in &cputime_t units 473 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds 474 * 475 * This structure groups together three kinds of CPU time that are 476 * tracked for threads and thread groups. Most things considering 477 * CPU time want to group these counts together and treat all three 478 * of them in parallel. 479 */ 480struct task_cputime { 481 cputime_t utime; 482 cputime_t stime; 483 unsigned long long sum_exec_runtime; 484}; 485/* Alternate field names when used to cache expirations. */ 486#define prof_exp stime 487#define virt_exp utime 488#define sched_exp sum_exec_runtime 489 490#define INIT_CPUTIME \ 491 (struct task_cputime) { \ 492 .utime = cputime_zero, \ 493 .stime = cputime_zero, \ 494 .sum_exec_runtime = 0, \ 495 } 496 497/* 498 * Disable preemption until the scheduler is running. 499 * Reset by start_kernel()->sched_init()->init_idle(). 500 * 501 * We include PREEMPT_ACTIVE to avoid cond_resched() from working 502 * before the scheduler is active -- see should_resched(). 503 */ 504#define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE) 505 506/** 507 * struct thread_group_cputimer - thread group interval timer counts 508 * @cputime: thread group interval timers. 509 * @running: non-zero when there are timers running and 510 * @cputime receives updates. 511 * @lock: lock for fields in this struct. 512 * 513 * This structure contains the version of task_cputime, above, that is 514 * used for thread group CPU timer calculations. 515 */ 516struct thread_group_cputimer { 517 struct task_cputime cputime; 518 int running; 519 spinlock_t lock; 520}; 521 522/* 523 * NOTE! "signal_struct" does not have it's own 524 * locking, because a shared signal_struct always 525 * implies a shared sighand_struct, so locking 526 * sighand_struct is always a proper superset of 527 * the locking of signal_struct. 528 */ 529struct signal_struct { 530 atomic_t count; 531 atomic_t live; 532 533 wait_queue_head_t wait_chldexit; /* for wait4() */ 534 535 /* current thread group signal load-balancing target: */ 536 struct task_struct *curr_target; 537 538 /* shared signal handling: */ 539 struct sigpending shared_pending; 540 541 /* thread group exit support */ 542 int group_exit_code; 543 /* overloaded: 544 * - notify group_exit_task when ->count is equal to notify_count 545 * - everyone except group_exit_task is stopped during signal delivery 546 * of fatal signals, group_exit_task processes the signal. 547 */ 548 int notify_count; 549 struct task_struct *group_exit_task; 550 551 /* thread group stop support, overloads group_exit_code too */ 552 int group_stop_count; 553 unsigned int flags; /* see SIGNAL_* flags below */ 554 555 /* POSIX.1b Interval Timers */ 556 struct list_head posix_timers; 557 558 /* ITIMER_REAL timer for the process */ 559 struct hrtimer real_timer; 560 struct pid *leader_pid; 561 ktime_t it_real_incr; 562 563 /* 564 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use 565 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these 566 * values are defined to 0 and 1 respectively 567 */ 568 struct cpu_itimer it[2]; 569 570 /* 571 * Thread group totals for process CPU timers. 572 * See thread_group_cputimer(), et al, for details. 573 */ 574 struct thread_group_cputimer cputimer; 575 576 /* Earliest-expiration cache. */ 577 struct task_cputime cputime_expires; 578 579 struct list_head cpu_timers[3]; 580 581 struct pid *tty_old_pgrp; 582 583 /* boolean value for session group leader */ 584 int leader; 585 586 struct tty_struct *tty; /* NULL if no tty */ 587 588 /* 589 * Cumulative resource counters for dead threads in the group, 590 * and for reaped dead child processes forked by this group. 591 * Live threads maintain their own counters and add to these 592 * in __exit_signal, except for the group leader. 593 */ 594 cputime_t utime, stime, cutime, cstime; 595 cputime_t gtime; 596 cputime_t cgtime; 597#ifndef CONFIG_VIRT_CPU_ACCOUNTING 598 cputime_t prev_utime, prev_stime; 599#endif 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 unsigned long maxrss, cmaxrss; 604 struct task_io_accounting ioac; 605 606 /* 607 * Cumulative ns of schedule CPU time fo dead threads in the 608 * group, not including a zombie group leader, (This only differs 609 * from jiffies_to_ns(utime + stime) if sched_clock uses something 610 * other than jiffies.) 611 */ 612 unsigned long long sum_sched_runtime; 613 614 /* 615 * We don't bother to synchronize most readers of this at all, 616 * because there is no reader checking a limit that actually needs 617 * to get both rlim_cur and rlim_max atomically, and either one 618 * alone is a single word that can safely be read normally. 619 * getrlimit/setrlimit use task_lock(current->group_leader) to 620 * protect this instead of the siglock, because they really 621 * have no need to disable irqs. 622 */ 623 struct rlimit rlim[RLIM_NLIMITS]; 624 625#ifdef CONFIG_BSD_PROCESS_ACCT 626 struct pacct_struct pacct; /* per-process accounting information */ 627#endif 628#ifdef CONFIG_TASKSTATS 629 struct taskstats *stats; 630#endif 631#ifdef CONFIG_AUDIT 632 unsigned audit_tty; 633 struct tty_audit_buf *tty_audit_buf; 634#endif 635 636 int oom_adj; /* OOM kill score adjustment (bit shift) */ 637}; 638 639/* Context switch must be unlocked if interrupts are to be enabled */ 640#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 641# define __ARCH_WANT_UNLOCKED_CTXSW 642#endif 643 644/* 645 * Bits in flags field of signal_struct. 646 */ 647#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 648#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 649#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 650#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 651/* 652 * Pending notifications to parent. 653 */ 654#define SIGNAL_CLD_STOPPED 0x00000010 655#define SIGNAL_CLD_CONTINUED 0x00000020 656#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 657 658#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 659 660/* If true, all threads except ->group_exit_task have pending SIGKILL */ 661static inline int signal_group_exit(const struct signal_struct *sig) 662{ 663 return (sig->flags & SIGNAL_GROUP_EXIT) || 664 (sig->group_exit_task != NULL); 665} 666 667/* 668 * Some day this will be a full-fledged user tracking system.. 669 */ 670struct user_struct { 671 atomic_t __count; /* reference count */ 672 atomic_t processes; /* How many processes does this user have? */ 673 atomic_t files; /* How many open files does this user have? */ 674 atomic_t sigpending; /* How many pending signals does this user have? */ 675#ifdef CONFIG_INOTIFY_USER 676 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 677 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 678#endif 679#ifdef CONFIG_EPOLL 680 atomic_t epoll_watches; /* The number of file descriptors currently watched */ 681#endif 682#ifdef CONFIG_POSIX_MQUEUE 683 /* protected by mq_lock */ 684 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 685#endif 686 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 687 688#ifdef CONFIG_KEYS 689 struct key *uid_keyring; /* UID specific keyring */ 690 struct key *session_keyring; /* UID's default session keyring */ 691#endif 692 693 /* Hash table maintenance information */ 694 struct hlist_node uidhash_node; 695 uid_t uid; 696 struct user_namespace *user_ns; 697 698#ifdef CONFIG_PERF_EVENTS 699 atomic_long_t locked_vm; 700#endif 701}; 702 703extern int uids_sysfs_init(void); 704 705extern struct user_struct *find_user(uid_t); 706 707extern struct user_struct root_user; 708#define INIT_USER (&root_user) 709 710 711struct backing_dev_info; 712struct reclaim_state; 713 714#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 715struct sched_info { 716 /* cumulative counters */ 717 unsigned long pcount; /* # of times run on this cpu */ 718 unsigned long long run_delay; /* time spent waiting on a runqueue */ 719 720 /* timestamps */ 721 unsigned long long last_arrival,/* when we last ran on a cpu */ 722 last_queued; /* when we were last queued to run */ 723#ifdef CONFIG_SCHEDSTATS 724 /* BKL stats */ 725 unsigned int bkl_count; 726#endif 727}; 728#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 729 730#ifdef CONFIG_TASK_DELAY_ACCT 731struct task_delay_info { 732 spinlock_t lock; 733 unsigned int flags; /* Private per-task flags */ 734 735 /* For each stat XXX, add following, aligned appropriately 736 * 737 * struct timespec XXX_start, XXX_end; 738 * u64 XXX_delay; 739 * u32 XXX_count; 740 * 741 * Atomicity of updates to XXX_delay, XXX_count protected by 742 * single lock above (split into XXX_lock if contention is an issue). 743 */ 744 745 /* 746 * XXX_count is incremented on every XXX operation, the delay 747 * associated with the operation is added to XXX_delay. 748 * XXX_delay contains the accumulated delay time in nanoseconds. 749 */ 750 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 751 u64 blkio_delay; /* wait for sync block io completion */ 752 u64 swapin_delay; /* wait for swapin block io completion */ 753 u32 blkio_count; /* total count of the number of sync block */ 754 /* io operations performed */ 755 u32 swapin_count; /* total count of the number of swapin block */ 756 /* io operations performed */ 757 758 struct timespec freepages_start, freepages_end; 759 u64 freepages_delay; /* wait for memory reclaim */ 760 u32 freepages_count; /* total count of memory reclaim */ 761}; 762#endif /* CONFIG_TASK_DELAY_ACCT */ 763 764static inline int sched_info_on(void) 765{ 766#ifdef CONFIG_SCHEDSTATS 767 return 1; 768#elif defined(CONFIG_TASK_DELAY_ACCT) 769 extern int delayacct_on; 770 return delayacct_on; 771#else 772 return 0; 773#endif 774} 775 776enum cpu_idle_type { 777 CPU_IDLE, 778 CPU_NOT_IDLE, 779 CPU_NEWLY_IDLE, 780 CPU_MAX_IDLE_TYPES 781}; 782 783/* 784 * sched-domains (multiprocessor balancing) declarations: 785 */ 786 787/* 788 * Increase resolution of nice-level calculations: 789 */ 790#define SCHED_LOAD_SHIFT 10 791#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 792 793#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 794 795#ifdef CONFIG_SMP 796#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */ 797#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */ 798#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */ 799#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */ 800#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */ 801#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */ 802#define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */ 803#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */ 804#define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */ 805#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */ 806#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */ 807 808#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */ 809 810enum powersavings_balance_level { 811 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */ 812 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package 813 * first for long running threads 814 */ 815 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle 816 * cpu package for power savings 817 */ 818 MAX_POWERSAVINGS_BALANCE_LEVELS 819}; 820 821extern int sched_mc_power_savings, sched_smt_power_savings; 822 823static inline int sd_balance_for_mc_power(void) 824{ 825 if (sched_smt_power_savings) 826 return SD_POWERSAVINGS_BALANCE; 827 828 if (!sched_mc_power_savings) 829 return SD_PREFER_SIBLING; 830 831 return 0; 832} 833 834static inline int sd_balance_for_package_power(void) 835{ 836 if (sched_mc_power_savings | sched_smt_power_savings) 837 return SD_POWERSAVINGS_BALANCE; 838 839 return SD_PREFER_SIBLING; 840} 841 842/* 843 * Optimise SD flags for power savings: 844 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings. 845 * Keep default SD flags if sched_{smt,mc}_power_saving=0 846 */ 847 848static inline int sd_power_saving_flags(void) 849{ 850 if (sched_mc_power_savings | sched_smt_power_savings) 851 return SD_BALANCE_NEWIDLE; 852 853 return 0; 854} 855 856struct sched_group { 857 struct sched_group *next; /* Must be a circular list */ 858 859 /* 860 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 861 * single CPU. 862 */ 863 unsigned int cpu_power; 864 865 /* 866 * The CPUs this group covers. 867 * 868 * NOTE: this field is variable length. (Allocated dynamically 869 * by attaching extra space to the end of the structure, 870 * depending on how many CPUs the kernel has booted up with) 871 * 872 * It is also be embedded into static data structures at build 873 * time. (See 'struct static_sched_group' in kernel/sched.c) 874 */ 875 unsigned long cpumask[0]; 876}; 877 878static inline struct cpumask *sched_group_cpus(struct sched_group *sg) 879{ 880 return to_cpumask(sg->cpumask); 881} 882 883enum sched_domain_level { 884 SD_LV_NONE = 0, 885 SD_LV_SIBLING, 886 SD_LV_MC, 887 SD_LV_CPU, 888 SD_LV_NODE, 889 SD_LV_ALLNODES, 890 SD_LV_MAX 891}; 892 893struct sched_domain_attr { 894 int relax_domain_level; 895}; 896 897#define SD_ATTR_INIT (struct sched_domain_attr) { \ 898 .relax_domain_level = -1, \ 899} 900 901struct sched_domain { 902 /* These fields must be setup */ 903 struct sched_domain *parent; /* top domain must be null terminated */ 904 struct sched_domain *child; /* bottom domain must be null terminated */ 905 struct sched_group *groups; /* the balancing groups of the domain */ 906 unsigned long min_interval; /* Minimum balance interval ms */ 907 unsigned long max_interval; /* Maximum balance interval ms */ 908 unsigned int busy_factor; /* less balancing by factor if busy */ 909 unsigned int imbalance_pct; /* No balance until over watermark */ 910 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 911 unsigned int busy_idx; 912 unsigned int idle_idx; 913 unsigned int newidle_idx; 914 unsigned int wake_idx; 915 unsigned int forkexec_idx; 916 unsigned int smt_gain; 917 int flags; /* See SD_* */ 918 enum sched_domain_level level; 919 920 /* Runtime fields. */ 921 unsigned long last_balance; /* init to jiffies. units in jiffies */ 922 unsigned int balance_interval; /* initialise to 1. units in ms. */ 923 unsigned int nr_balance_failed; /* initialise to 0 */ 924 925 u64 last_update; 926 927#ifdef CONFIG_SCHEDSTATS 928 /* load_balance() stats */ 929 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 930 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 931 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 932 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 933 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 934 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 935 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 936 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 937 938 /* Active load balancing */ 939 unsigned int alb_count; 940 unsigned int alb_failed; 941 unsigned int alb_pushed; 942 943 /* SD_BALANCE_EXEC stats */ 944 unsigned int sbe_count; 945 unsigned int sbe_balanced; 946 unsigned int sbe_pushed; 947 948 /* SD_BALANCE_FORK stats */ 949 unsigned int sbf_count; 950 unsigned int sbf_balanced; 951 unsigned int sbf_pushed; 952 953 /* try_to_wake_up() stats */ 954 unsigned int ttwu_wake_remote; 955 unsigned int ttwu_move_affine; 956 unsigned int ttwu_move_balance; 957#endif 958#ifdef CONFIG_SCHED_DEBUG 959 char *name; 960#endif 961 962 unsigned int span_weight; 963 /* 964 * Span of all CPUs in this domain. 965 * 966 * NOTE: this field is variable length. (Allocated dynamically 967 * by attaching extra space to the end of the structure, 968 * depending on how many CPUs the kernel has booted up with) 969 * 970 * It is also be embedded into static data structures at build 971 * time. (See 'struct static_sched_domain' in kernel/sched.c) 972 */ 973 unsigned long span[0]; 974}; 975 976static inline struct cpumask *sched_domain_span(struct sched_domain *sd) 977{ 978 return to_cpumask(sd->span); 979} 980 981extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], 982 struct sched_domain_attr *dattr_new); 983 984/* Allocate an array of sched domains, for partition_sched_domains(). */ 985cpumask_var_t *alloc_sched_domains(unsigned int ndoms); 986void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms); 987 988/* Test a flag in parent sched domain */ 989static inline int test_sd_parent(struct sched_domain *sd, int flag) 990{ 991 if (sd->parent && (sd->parent->flags & flag)) 992 return 1; 993 994 return 0; 995} 996 997unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu); 998unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu); 999 1000#else /* CONFIG_SMP */ 1001 1002struct sched_domain_attr; 1003 1004static inline void 1005partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], 1006 struct sched_domain_attr *dattr_new) 1007{ 1008} 1009#endif /* !CONFIG_SMP */ 1010 1011 1012struct io_context; /* See blkdev.h */ 1013 1014 1015#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 1016extern void prefetch_stack(struct task_struct *t); 1017#else 1018static inline void prefetch_stack(struct task_struct *t) { } 1019#endif 1020 1021struct audit_context; /* See audit.c */ 1022struct mempolicy; 1023struct pipe_inode_info; 1024struct uts_namespace; 1025 1026struct rq; 1027struct sched_domain; 1028 1029/* 1030 * wake flags 1031 */ 1032#define WF_SYNC 0x01 /* waker goes to sleep after wakup */ 1033#define WF_FORK 0x02 /* child wakeup after fork */ 1034 1035#define ENQUEUE_WAKEUP 1 1036#define ENQUEUE_WAKING 2 1037#define ENQUEUE_HEAD 4 1038 1039#define DEQUEUE_SLEEP 1 1040 1041struct sched_class { 1042 const struct sched_class *next; 1043 1044 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags); 1045 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags); 1046 void (*yield_task) (struct rq *rq); 1047 1048 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags); 1049 1050 struct task_struct * (*pick_next_task) (struct rq *rq); 1051 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 1052 1053#ifdef CONFIG_SMP 1054 int (*select_task_rq)(struct rq *rq, struct task_struct *p, 1055 int sd_flag, int flags); 1056 1057 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 1058 void (*post_schedule) (struct rq *this_rq); 1059 void (*task_waking) (struct rq *this_rq, struct task_struct *task); 1060 void (*task_woken) (struct rq *this_rq, struct task_struct *task); 1061 1062 void (*set_cpus_allowed)(struct task_struct *p, 1063 const struct cpumask *newmask); 1064 1065 void (*rq_online)(struct rq *rq); 1066 void (*rq_offline)(struct rq *rq); 1067#endif 1068 1069 void (*set_curr_task) (struct rq *rq); 1070 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 1071 void (*task_fork) (struct task_struct *p); 1072 1073 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 1074 int running); 1075 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 1076 int running); 1077 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 1078 int oldprio, int running); 1079 1080 unsigned int (*get_rr_interval) (struct rq *rq, 1081 struct task_struct *task); 1082 1083#ifdef CONFIG_FAIR_GROUP_SCHED 1084 void (*moved_group) (struct task_struct *p, int on_rq); 1085#endif 1086}; 1087 1088struct load_weight { 1089 unsigned long weight, inv_weight; 1090}; 1091 1092#ifdef CONFIG_SCHEDSTATS 1093struct sched_statistics { 1094 u64 wait_start; 1095 u64 wait_max; 1096 u64 wait_count; 1097 u64 wait_sum; 1098 u64 iowait_count; 1099 u64 iowait_sum; 1100 1101 u64 sleep_start; 1102 u64 sleep_max; 1103 s64 sum_sleep_runtime; 1104 1105 u64 block_start; 1106 u64 block_max; 1107 u64 exec_max; 1108 u64 slice_max; 1109 1110 u64 nr_migrations_cold; 1111 u64 nr_failed_migrations_affine; 1112 u64 nr_failed_migrations_running; 1113 u64 nr_failed_migrations_hot; 1114 u64 nr_forced_migrations; 1115 1116 u64 nr_wakeups; 1117 u64 nr_wakeups_sync; 1118 u64 nr_wakeups_migrate; 1119 u64 nr_wakeups_local; 1120 u64 nr_wakeups_remote; 1121 u64 nr_wakeups_affine; 1122 u64 nr_wakeups_affine_attempts; 1123 u64 nr_wakeups_passive; 1124 u64 nr_wakeups_idle; 1125}; 1126#endif 1127 1128struct sched_entity { 1129 struct load_weight load; /* for load-balancing */ 1130 struct rb_node run_node; 1131 struct list_head group_node; 1132 unsigned int on_rq; 1133 1134 u64 exec_start; 1135 u64 sum_exec_runtime; 1136 u64 vruntime; 1137 u64 prev_sum_exec_runtime; 1138 1139 u64 nr_migrations; 1140 1141#ifdef CONFIG_SCHEDSTATS 1142 struct sched_statistics statistics; 1143#endif 1144 1145#ifdef CONFIG_FAIR_GROUP_SCHED 1146 struct sched_entity *parent; 1147 /* rq on which this entity is (to be) queued: */ 1148 struct cfs_rq *cfs_rq; 1149 /* rq "owned" by this entity/group: */ 1150 struct cfs_rq *my_q; 1151#endif 1152}; 1153 1154struct sched_rt_entity { 1155 struct list_head run_list; 1156 unsigned long timeout; 1157 unsigned int time_slice; 1158 int nr_cpus_allowed; 1159 1160 struct sched_rt_entity *back; 1161#ifdef CONFIG_RT_GROUP_SCHED 1162 struct sched_rt_entity *parent; 1163 /* rq on which this entity is (to be) queued: */ 1164 struct rt_rq *rt_rq; 1165 /* rq "owned" by this entity/group: */ 1166 struct rt_rq *my_q; 1167#endif 1168}; 1169 1170struct rcu_node; 1171 1172struct task_struct { 1173 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1174 void *stack; 1175 atomic_t usage; 1176 unsigned int flags; /* per process flags, defined below */ 1177 unsigned int ptrace; 1178 1179 int lock_depth; /* BKL lock depth */ 1180 1181#ifdef CONFIG_SMP 1182#ifdef __ARCH_WANT_UNLOCKED_CTXSW 1183 int oncpu; 1184#endif 1185#endif 1186 1187 int prio, static_prio, normal_prio; 1188 unsigned int rt_priority; 1189 const struct sched_class *sched_class; 1190 struct sched_entity se; 1191 struct sched_rt_entity rt; 1192 1193#ifdef CONFIG_PREEMPT_NOTIFIERS 1194 /* list of struct preempt_notifier: */ 1195 struct hlist_head preempt_notifiers; 1196#endif 1197 1198 /* 1199 * fpu_counter contains the number of consecutive context switches 1200 * that the FPU is used. If this is over a threshold, the lazy fpu 1201 * saving becomes unlazy to save the trap. This is an unsigned char 1202 * so that after 256 times the counter wraps and the behavior turns 1203 * lazy again; this to deal with bursty apps that only use FPU for 1204 * a short time 1205 */ 1206 unsigned char fpu_counter; 1207#ifdef CONFIG_BLK_DEV_IO_TRACE 1208 unsigned int btrace_seq; 1209#endif 1210 1211 unsigned int policy; 1212 cpumask_t cpus_allowed; 1213 1214#ifdef CONFIG_TREE_PREEMPT_RCU 1215 int rcu_read_lock_nesting; 1216 char rcu_read_unlock_special; 1217 struct rcu_node *rcu_blocked_node; 1218 struct list_head rcu_node_entry; 1219#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */ 1220 1221#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1222 struct sched_info sched_info; 1223#endif 1224 1225 struct list_head tasks; 1226 struct plist_node pushable_tasks; 1227 1228 struct mm_struct *mm, *active_mm; 1229#if defined(SPLIT_RSS_COUNTING) 1230 struct task_rss_stat rss_stat; 1231#endif 1232/* task state */ 1233 int exit_state; 1234 int exit_code, exit_signal; 1235 int pdeath_signal; /* The signal sent when the parent dies */ 1236 /* ??? */ 1237 unsigned int personality; 1238 unsigned did_exec:1; 1239 unsigned in_execve:1; /* Tell the LSMs that the process is doing an 1240 * execve */ 1241 unsigned in_iowait:1; 1242 1243 1244 /* Revert to default priority/policy when forking */ 1245 unsigned sched_reset_on_fork:1; 1246 1247 pid_t pid; 1248 pid_t tgid; 1249 1250#ifdef CONFIG_CC_STACKPROTECTOR 1251 /* Canary value for the -fstack-protector gcc feature */ 1252 unsigned long stack_canary; 1253#endif 1254 1255 /* 1256 * pointers to (original) parent process, youngest child, younger sibling, 1257 * older sibling, respectively. (p->father can be replaced with 1258 * p->real_parent->pid) 1259 */ 1260 struct task_struct *real_parent; /* real parent process */ 1261 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1262 /* 1263 * children/sibling forms the list of my natural children 1264 */ 1265 struct list_head children; /* list of my children */ 1266 struct list_head sibling; /* linkage in my parent's children list */ 1267 struct task_struct *group_leader; /* threadgroup leader */ 1268 1269 /* 1270 * ptraced is the list of tasks this task is using ptrace on. 1271 * This includes both natural children and PTRACE_ATTACH targets. 1272 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1273 */ 1274 struct list_head ptraced; 1275 struct list_head ptrace_entry; 1276 1277 /* PID/PID hash table linkage. */ 1278 struct pid_link pids[PIDTYPE_MAX]; 1279 struct list_head thread_group; 1280 1281 struct completion *vfork_done; /* for vfork() */ 1282 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1283 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1284 1285 cputime_t utime, stime, utimescaled, stimescaled; 1286 cputime_t gtime; 1287#ifndef CONFIG_VIRT_CPU_ACCOUNTING 1288 cputime_t prev_utime, prev_stime; 1289#endif 1290 unsigned long nvcsw, nivcsw; /* context switch counts */ 1291 struct timespec start_time; /* monotonic time */ 1292 struct timespec real_start_time; /* boot based time */ 1293/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1294 unsigned long min_flt, maj_flt; 1295 1296 struct task_cputime cputime_expires; 1297 struct list_head cpu_timers[3]; 1298 1299/* process credentials */ 1300 const struct cred *real_cred; /* objective and real subjective task 1301 * credentials (COW) */ 1302 const struct cred *cred; /* effective (overridable) subjective task 1303 * credentials (COW) */ 1304 struct mutex cred_guard_mutex; /* guard against foreign influences on 1305 * credential calculations 1306 * (notably. ptrace) */ 1307 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */ 1308 1309 char comm[TASK_COMM_LEN]; /* executable name excluding path 1310 - access with [gs]et_task_comm (which lock 1311 it with task_lock()) 1312 - initialized normally by setup_new_exec */ 1313/* file system info */ 1314 int link_count, total_link_count; 1315#ifdef CONFIG_SYSVIPC 1316/* ipc stuff */ 1317 struct sysv_sem sysvsem; 1318#endif 1319#ifdef CONFIG_DETECT_HUNG_TASK 1320/* hung task detection */ 1321 unsigned long last_switch_count; 1322#endif 1323/* CPU-specific state of this task */ 1324 struct thread_struct thread; 1325/* filesystem information */ 1326 struct fs_struct *fs; 1327/* open file information */ 1328 struct files_struct *files; 1329/* namespaces */ 1330 struct nsproxy *nsproxy; 1331/* signal handlers */ 1332 struct signal_struct *signal; 1333 struct sighand_struct *sighand; 1334 1335 sigset_t blocked, real_blocked; 1336 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1337 struct sigpending pending; 1338 1339 unsigned long sas_ss_sp; 1340 size_t sas_ss_size; 1341 int (*notifier)(void *priv); 1342 void *notifier_data; 1343 sigset_t *notifier_mask; 1344 struct audit_context *audit_context; 1345#ifdef CONFIG_AUDITSYSCALL 1346 uid_t loginuid; 1347 unsigned int sessionid; 1348#endif 1349 seccomp_t seccomp; 1350 1351/* Thread group tracking */ 1352 u32 parent_exec_id; 1353 u32 self_exec_id; 1354/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, 1355 * mempolicy */ 1356 spinlock_t alloc_lock; 1357 1358#ifdef CONFIG_GENERIC_HARDIRQS 1359 /* IRQ handler threads */ 1360 struct irqaction *irqaction; 1361#endif 1362 1363 /* Protection of the PI data structures: */ 1364 raw_spinlock_t pi_lock; 1365 1366#ifdef CONFIG_RT_MUTEXES 1367 /* PI waiters blocked on a rt_mutex held by this task */ 1368 struct plist_head pi_waiters; 1369 /* Deadlock detection and priority inheritance handling */ 1370 struct rt_mutex_waiter *pi_blocked_on; 1371#endif 1372 1373#ifdef CONFIG_DEBUG_MUTEXES 1374 /* mutex deadlock detection */ 1375 struct mutex_waiter *blocked_on; 1376#endif 1377#ifdef CONFIG_TRACE_IRQFLAGS 1378 unsigned int irq_events; 1379 unsigned long hardirq_enable_ip; 1380 unsigned long hardirq_disable_ip; 1381 unsigned int hardirq_enable_event; 1382 unsigned int hardirq_disable_event; 1383 int hardirqs_enabled; 1384 int hardirq_context; 1385 unsigned long softirq_disable_ip; 1386 unsigned long softirq_enable_ip; 1387 unsigned int softirq_disable_event; 1388 unsigned int softirq_enable_event; 1389 int softirqs_enabled; 1390 int softirq_context; 1391#endif 1392#ifdef CONFIG_LOCKDEP 1393# define MAX_LOCK_DEPTH 48UL 1394 u64 curr_chain_key; 1395 int lockdep_depth; 1396 unsigned int lockdep_recursion; 1397 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1398 gfp_t lockdep_reclaim_gfp; 1399#endif 1400 1401/* journalling filesystem info */ 1402 void *journal_info; 1403 1404/* stacked block device info */ 1405 struct bio_list *bio_list; 1406 1407/* VM state */ 1408 struct reclaim_state *reclaim_state; 1409 1410 struct backing_dev_info *backing_dev_info; 1411 1412 struct io_context *io_context; 1413 1414 unsigned long ptrace_message; 1415 siginfo_t *last_siginfo; /* For ptrace use. */ 1416 struct task_io_accounting ioac; 1417#if defined(CONFIG_TASK_XACCT) 1418 u64 acct_rss_mem1; /* accumulated rss usage */ 1419 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1420 cputime_t acct_timexpd; /* stime + utime since last update */ 1421#endif 1422#ifdef CONFIG_CPUSETS 1423 nodemask_t mems_allowed; /* Protected by alloc_lock */ 1424 int mems_allowed_change_disable; 1425 int cpuset_mem_spread_rotor; 1426 int cpuset_slab_spread_rotor; 1427#endif 1428#ifdef CONFIG_CGROUPS 1429 /* Control Group info protected by css_set_lock */ 1430 struct css_set *cgroups; 1431 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1432 struct list_head cg_list; 1433#endif 1434#ifdef CONFIG_FUTEX 1435 struct robust_list_head __user *robust_list; 1436#ifdef CONFIG_COMPAT 1437 struct compat_robust_list_head __user *compat_robust_list; 1438#endif 1439 struct list_head pi_state_list; 1440 struct futex_pi_state *pi_state_cache; 1441#endif 1442#ifdef CONFIG_PERF_EVENTS 1443 struct perf_event_context *perf_event_ctxp; 1444 struct mutex perf_event_mutex; 1445 struct list_head perf_event_list; 1446#endif 1447#ifdef CONFIG_NUMA 1448 struct mempolicy *mempolicy; /* Protected by alloc_lock */ 1449 short il_next; 1450#endif 1451 atomic_t fs_excl; /* holding fs exclusive resources */ 1452 struct rcu_head rcu; 1453 1454 /* 1455 * cache last used pipe for splice 1456 */ 1457 struct pipe_inode_info *splice_pipe; 1458#ifdef CONFIG_TASK_DELAY_ACCT 1459 struct task_delay_info *delays; 1460#endif 1461#ifdef CONFIG_FAULT_INJECTION 1462 int make_it_fail; 1463#endif 1464 struct prop_local_single dirties; 1465#ifdef CONFIG_LATENCYTOP 1466 int latency_record_count; 1467 struct latency_record latency_record[LT_SAVECOUNT]; 1468#endif 1469 /* 1470 * time slack values; these are used to round up poll() and 1471 * select() etc timeout values. These are in nanoseconds. 1472 */ 1473 unsigned long timer_slack_ns; 1474 unsigned long default_timer_slack_ns; 1475 1476 struct list_head *scm_work_list; 1477#ifdef CONFIG_FUNCTION_GRAPH_TRACER 1478 /* Index of current stored address in ret_stack */ 1479 int curr_ret_stack; 1480 /* Stack of return addresses for return function tracing */ 1481 struct ftrace_ret_stack *ret_stack; 1482 /* time stamp for last schedule */ 1483 unsigned long long ftrace_timestamp; 1484 /* 1485 * Number of functions that haven't been traced 1486 * because of depth overrun. 1487 */ 1488 atomic_t trace_overrun; 1489 /* Pause for the tracing */ 1490 atomic_t tracing_graph_pause; 1491#endif 1492#ifdef CONFIG_TRACING 1493 /* state flags for use by tracers */ 1494 unsigned long trace; 1495 /* bitmask of trace recursion */ 1496 unsigned long trace_recursion; 1497#endif /* CONFIG_TRACING */ 1498#ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */ 1499 struct memcg_batch_info { 1500 int do_batch; /* incremented when batch uncharge started */ 1501 struct mem_cgroup *memcg; /* target memcg of uncharge */ 1502 unsigned long bytes; /* uncharged usage */ 1503 unsigned long memsw_bytes; /* uncharged mem+swap usage */ 1504 } memcg_batch; 1505#endif 1506}; 1507 1508/* Future-safe accessor for struct task_struct's cpus_allowed. */ 1509#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed) 1510 1511/* 1512 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1513 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1514 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1515 * values are inverted: lower p->prio value means higher priority. 1516 * 1517 * The MAX_USER_RT_PRIO value allows the actual maximum 1518 * RT priority to be separate from the value exported to 1519 * user-space. This allows kernel threads to set their 1520 * priority to a value higher than any user task. Note: 1521 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1522 */ 1523 1524#define MAX_USER_RT_PRIO 100 1525#define MAX_RT_PRIO MAX_USER_RT_PRIO 1526 1527#define MAX_PRIO (MAX_RT_PRIO + 40) 1528#define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1529 1530static inline int rt_prio(int prio) 1531{ 1532 if (unlikely(prio < MAX_RT_PRIO)) 1533 return 1; 1534 return 0; 1535} 1536 1537static inline int rt_task(struct task_struct *p) 1538{ 1539 return rt_prio(p->prio); 1540} 1541 1542static inline struct pid *task_pid(struct task_struct *task) 1543{ 1544 return task->pids[PIDTYPE_PID].pid; 1545} 1546 1547static inline struct pid *task_tgid(struct task_struct *task) 1548{ 1549 return task->group_leader->pids[PIDTYPE_PID].pid; 1550} 1551 1552/* 1553 * Without tasklist or rcu lock it is not safe to dereference 1554 * the result of task_pgrp/task_session even if task == current, 1555 * we can race with another thread doing sys_setsid/sys_setpgid. 1556 */ 1557static inline struct pid *task_pgrp(struct task_struct *task) 1558{ 1559 return task->group_leader->pids[PIDTYPE_PGID].pid; 1560} 1561 1562static inline struct pid *task_session(struct task_struct *task) 1563{ 1564 return task->group_leader->pids[PIDTYPE_SID].pid; 1565} 1566 1567struct pid_namespace; 1568 1569/* 1570 * the helpers to get the task's different pids as they are seen 1571 * from various namespaces 1572 * 1573 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1574 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1575 * current. 1576 * task_xid_nr_ns() : id seen from the ns specified; 1577 * 1578 * set_task_vxid() : assigns a virtual id to a task; 1579 * 1580 * see also pid_nr() etc in include/linux/pid.h 1581 */ 1582pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, 1583 struct pid_namespace *ns); 1584 1585static inline pid_t task_pid_nr(struct task_struct *tsk) 1586{ 1587 return tsk->pid; 1588} 1589 1590static inline pid_t task_pid_nr_ns(struct task_struct *tsk, 1591 struct pid_namespace *ns) 1592{ 1593 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); 1594} 1595 1596static inline pid_t task_pid_vnr(struct task_struct *tsk) 1597{ 1598 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); 1599} 1600 1601 1602static inline pid_t task_tgid_nr(struct task_struct *tsk) 1603{ 1604 return tsk->tgid; 1605} 1606 1607pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1608 1609static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1610{ 1611 return pid_vnr(task_tgid(tsk)); 1612} 1613 1614 1615static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, 1616 struct pid_namespace *ns) 1617{ 1618 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); 1619} 1620 1621static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1622{ 1623 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); 1624} 1625 1626 1627static inline pid_t task_session_nr_ns(struct task_struct *tsk, 1628 struct pid_namespace *ns) 1629{ 1630 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); 1631} 1632 1633static inline pid_t task_session_vnr(struct task_struct *tsk) 1634{ 1635 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); 1636} 1637 1638/* obsolete, do not use */ 1639static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1640{ 1641 return task_pgrp_nr_ns(tsk, &init_pid_ns); 1642} 1643 1644/** 1645 * pid_alive - check that a task structure is not stale 1646 * @p: Task structure to be checked. 1647 * 1648 * Test if a process is not yet dead (at most zombie state) 1649 * If pid_alive fails, then pointers within the task structure 1650 * can be stale and must not be dereferenced. 1651 */ 1652static inline int pid_alive(struct task_struct *p) 1653{ 1654 return p->pids[PIDTYPE_PID].pid != NULL; 1655} 1656 1657/** 1658 * is_global_init - check if a task structure is init 1659 * @tsk: Task structure to be checked. 1660 * 1661 * Check if a task structure is the first user space task the kernel created. 1662 */ 1663static inline int is_global_init(struct task_struct *tsk) 1664{ 1665 return tsk->pid == 1; 1666} 1667 1668/* 1669 * is_container_init: 1670 * check whether in the task is init in its own pid namespace. 1671 */ 1672extern int is_container_init(struct task_struct *tsk); 1673 1674extern struct pid *cad_pid; 1675 1676extern void free_task(struct task_struct *tsk); 1677#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1678 1679extern void __put_task_struct(struct task_struct *t); 1680 1681static inline void put_task_struct(struct task_struct *t) 1682{ 1683 if (atomic_dec_and_test(&t->usage)) 1684 __put_task_struct(t); 1685} 1686 1687extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st); 1688extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st); 1689 1690/* 1691 * Per process flags 1692 */ 1693#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1694 /* Not implemented yet, only for 486*/ 1695#define PF_STARTING 0x00000002 /* being created */ 1696#define PF_EXITING 0x00000004 /* getting shut down */ 1697#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1698#define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1699#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1700#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */ 1701#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1702#define PF_DUMPCORE 0x00000200 /* dumped core */ 1703#define PF_SIGNALED 0x00000400 /* killed by a signal */ 1704#define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1705#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1706#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1707#define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */ 1708#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1709#define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1710#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1711#define PF_KSWAPD 0x00040000 /* I am kswapd */ 1712#define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */ 1713#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1714#define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1715#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1716#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1717#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1718#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1719#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1720#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ 1721#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1722#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1723#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1724#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1725 1726/* 1727 * Only the _current_ task can read/write to tsk->flags, but other 1728 * tasks can access tsk->flags in readonly mode for example 1729 * with tsk_used_math (like during threaded core dumping). 1730 * There is however an exception to this rule during ptrace 1731 * or during fork: the ptracer task is allowed to write to the 1732 * child->flags of its traced child (same goes for fork, the parent 1733 * can write to the child->flags), because we're guaranteed the 1734 * child is not running and in turn not changing child->flags 1735 * at the same time the parent does it. 1736 */ 1737#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1738#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1739#define clear_used_math() clear_stopped_child_used_math(current) 1740#define set_used_math() set_stopped_child_used_math(current) 1741#define conditional_stopped_child_used_math(condition, child) \ 1742 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1743#define conditional_used_math(condition) \ 1744 conditional_stopped_child_used_math(condition, current) 1745#define copy_to_stopped_child_used_math(child) \ 1746 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1747/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1748#define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1749#define used_math() tsk_used_math(current) 1750 1751#ifdef CONFIG_TREE_PREEMPT_RCU 1752 1753#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */ 1754#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */ 1755 1756static inline void rcu_copy_process(struct task_struct *p) 1757{ 1758 p->rcu_read_lock_nesting = 0; 1759 p->rcu_read_unlock_special = 0; 1760 p->rcu_blocked_node = NULL; 1761 INIT_LIST_HEAD(&p->rcu_node_entry); 1762} 1763 1764#else 1765 1766static inline void rcu_copy_process(struct task_struct *p) 1767{ 1768} 1769 1770#endif 1771 1772#ifdef CONFIG_SMP 1773extern int set_cpus_allowed_ptr(struct task_struct *p, 1774 const struct cpumask *new_mask); 1775#else 1776static inline int set_cpus_allowed_ptr(struct task_struct *p, 1777 const struct cpumask *new_mask) 1778{ 1779 if (!cpumask_test_cpu(0, new_mask)) 1780 return -EINVAL; 1781 return 0; 1782} 1783#endif 1784 1785#ifndef CONFIG_CPUMASK_OFFSTACK 1786static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1787{ 1788 return set_cpus_allowed_ptr(p, &new_mask); 1789} 1790#endif 1791 1792/* 1793 * Architectures can set this to 1 if they have specified 1794 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, 1795 * but then during bootup it turns out that sched_clock() 1796 * is reliable after all: 1797 */ 1798#ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1799extern int sched_clock_stable; 1800#endif 1801 1802/* ftrace calls sched_clock() directly */ 1803extern unsigned long long notrace sched_clock(void); 1804 1805extern void sched_clock_init(void); 1806extern u64 sched_clock_cpu(int cpu); 1807 1808#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1809static inline void sched_clock_tick(void) 1810{ 1811} 1812 1813static inline void sched_clock_idle_sleep_event(void) 1814{ 1815} 1816 1817static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1818{ 1819} 1820#else 1821extern void sched_clock_tick(void); 1822extern void sched_clock_idle_sleep_event(void); 1823extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1824#endif 1825 1826/* 1827 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1828 * clock constructed from sched_clock(): 1829 */ 1830extern unsigned long long cpu_clock(int cpu); 1831 1832extern unsigned long long 1833task_sched_runtime(struct task_struct *task); 1834extern unsigned long long thread_group_sched_runtime(struct task_struct *task); 1835 1836/* sched_exec is called by processes performing an exec */ 1837#ifdef CONFIG_SMP 1838extern void sched_exec(void); 1839#else 1840#define sched_exec() {} 1841#endif 1842 1843extern void sched_clock_idle_sleep_event(void); 1844extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1845 1846#ifdef CONFIG_HOTPLUG_CPU 1847extern void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p); 1848extern void idle_task_exit(void); 1849#else 1850static inline void idle_task_exit(void) {} 1851#endif 1852 1853extern void sched_idle_next(void); 1854 1855#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1856extern void wake_up_idle_cpu(int cpu); 1857#else 1858static inline void wake_up_idle_cpu(int cpu) { } 1859#endif 1860 1861extern unsigned int sysctl_sched_latency; 1862extern unsigned int sysctl_sched_min_granularity; 1863extern unsigned int sysctl_sched_wakeup_granularity; 1864extern unsigned int sysctl_sched_shares_ratelimit; 1865extern unsigned int sysctl_sched_shares_thresh; 1866extern unsigned int sysctl_sched_child_runs_first; 1867 1868enum sched_tunable_scaling { 1869 SCHED_TUNABLESCALING_NONE, 1870 SCHED_TUNABLESCALING_LOG, 1871 SCHED_TUNABLESCALING_LINEAR, 1872 SCHED_TUNABLESCALING_END, 1873}; 1874extern enum sched_tunable_scaling sysctl_sched_tunable_scaling; 1875 1876#ifdef CONFIG_SCHED_DEBUG 1877extern unsigned int sysctl_sched_migration_cost; 1878extern unsigned int sysctl_sched_nr_migrate; 1879extern unsigned int sysctl_sched_time_avg; 1880extern unsigned int sysctl_timer_migration; 1881 1882int sched_proc_update_handler(struct ctl_table *table, int write, 1883 void __user *buffer, size_t *length, 1884 loff_t *ppos); 1885#endif 1886#ifdef CONFIG_SCHED_DEBUG 1887static inline unsigned int get_sysctl_timer_migration(void) 1888{ 1889 return sysctl_timer_migration; 1890} 1891#else 1892static inline unsigned int get_sysctl_timer_migration(void) 1893{ 1894 return 1; 1895} 1896#endif 1897extern unsigned int sysctl_sched_rt_period; 1898extern int sysctl_sched_rt_runtime; 1899 1900int sched_rt_handler(struct ctl_table *table, int write, 1901 void __user *buffer, size_t *lenp, 1902 loff_t *ppos); 1903 1904extern unsigned int sysctl_sched_compat_yield; 1905 1906#ifdef CONFIG_RT_MUTEXES 1907extern int rt_mutex_getprio(struct task_struct *p); 1908extern void rt_mutex_setprio(struct task_struct *p, int prio); 1909extern void rt_mutex_adjust_pi(struct task_struct *p); 1910#else 1911static inline int rt_mutex_getprio(struct task_struct *p) 1912{ 1913 return p->normal_prio; 1914} 1915# define rt_mutex_adjust_pi(p) do { } while (0) 1916#endif 1917 1918extern void set_user_nice(struct task_struct *p, long nice); 1919extern int task_prio(const struct task_struct *p); 1920extern int task_nice(const struct task_struct *p); 1921extern int can_nice(const struct task_struct *p, const int nice); 1922extern int task_curr(const struct task_struct *p); 1923extern int idle_cpu(int cpu); 1924extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1925extern int sched_setscheduler_nocheck(struct task_struct *, int, 1926 struct sched_param *); 1927extern struct task_struct *idle_task(int cpu); 1928extern struct task_struct *curr_task(int cpu); 1929extern void set_curr_task(int cpu, struct task_struct *p); 1930 1931void yield(void); 1932 1933/* 1934 * The default (Linux) execution domain. 1935 */ 1936extern struct exec_domain default_exec_domain; 1937 1938union thread_union { 1939 struct thread_info thread_info; 1940 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1941}; 1942 1943#ifndef __HAVE_ARCH_KSTACK_END 1944static inline int kstack_end(void *addr) 1945{ 1946 /* Reliable end of stack detection: 1947 * Some APM bios versions misalign the stack 1948 */ 1949 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1950} 1951#endif 1952 1953extern union thread_union init_thread_union; 1954extern struct task_struct init_task; 1955 1956extern struct mm_struct init_mm; 1957 1958extern struct pid_namespace init_pid_ns; 1959 1960/* 1961 * find a task by one of its numerical ids 1962 * 1963 * find_task_by_pid_ns(): 1964 * finds a task by its pid in the specified namespace 1965 * find_task_by_vpid(): 1966 * finds a task by its virtual pid 1967 * 1968 * see also find_vpid() etc in include/linux/pid.h 1969 */ 1970 1971extern struct task_struct *find_task_by_vpid(pid_t nr); 1972extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1973 struct pid_namespace *ns); 1974 1975extern void __set_special_pids(struct pid *pid); 1976 1977/* per-UID process charging. */ 1978extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1979static inline struct user_struct *get_uid(struct user_struct *u) 1980{ 1981 atomic_inc(&u->__count); 1982 return u; 1983} 1984extern void free_uid(struct user_struct *); 1985extern void release_uids(struct user_namespace *ns); 1986 1987#include <asm/current.h> 1988 1989extern void do_timer(unsigned long ticks); 1990 1991extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1992extern int wake_up_process(struct task_struct *tsk); 1993extern void wake_up_new_task(struct task_struct *tsk, 1994 unsigned long clone_flags); 1995#ifdef CONFIG_SMP 1996 extern void kick_process(struct task_struct *tsk); 1997#else 1998 static inline void kick_process(struct task_struct *tsk) { } 1999#endif 2000extern void sched_fork(struct task_struct *p, int clone_flags); 2001extern void sched_dead(struct task_struct *p); 2002 2003extern void proc_caches_init(void); 2004extern void flush_signals(struct task_struct *); 2005extern void __flush_signals(struct task_struct *); 2006extern void ignore_signals(struct task_struct *); 2007extern void flush_signal_handlers(struct task_struct *, int force_default); 2008extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 2009 2010static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 2011{ 2012 unsigned long flags; 2013 int ret; 2014 2015 spin_lock_irqsave(&tsk->sighand->siglock, flags); 2016 ret = dequeue_signal(tsk, mask, info); 2017 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 2018 2019 return ret; 2020} 2021 2022extern void block_all_signals(int (*notifier)(void *priv), void *priv, 2023 sigset_t *mask); 2024extern void unblock_all_signals(void); 2025extern void release_task(struct task_struct * p); 2026extern int send_sig_info(int, struct siginfo *, struct task_struct *); 2027extern int force_sigsegv(int, struct task_struct *); 2028extern int force_sig_info(int, struct siginfo *, struct task_struct *); 2029extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 2030extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 2031extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 2032extern int kill_pgrp(struct pid *pid, int sig, int priv); 2033extern int kill_pid(struct pid *pid, int sig, int priv); 2034extern int kill_proc_info(int, struct siginfo *, pid_t); 2035extern int do_notify_parent(struct task_struct *, int); 2036extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent); 2037extern void force_sig(int, struct task_struct *); 2038extern int send_sig(int, struct task_struct *, int); 2039extern int zap_other_threads(struct task_struct *p); 2040extern struct sigqueue *sigqueue_alloc(void); 2041extern void sigqueue_free(struct sigqueue *); 2042extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 2043extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 2044extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 2045 2046static inline int kill_cad_pid(int sig, int priv) 2047{ 2048 return kill_pid(cad_pid, sig, priv); 2049} 2050 2051/* These can be the second arg to send_sig_info/send_group_sig_info. */ 2052#define SEND_SIG_NOINFO ((struct siginfo *) 0) 2053#define SEND_SIG_PRIV ((struct siginfo *) 1) 2054#define SEND_SIG_FORCED ((struct siginfo *) 2) 2055 2056/* 2057 * True if we are on the alternate signal stack. 2058 */ 2059static inline int on_sig_stack(unsigned long sp) 2060{ 2061#ifdef CONFIG_STACK_GROWSUP 2062 return sp >= current->sas_ss_sp && 2063 sp - current->sas_ss_sp < current->sas_ss_size; 2064#else 2065 return sp > current->sas_ss_sp && 2066 sp - current->sas_ss_sp <= current->sas_ss_size; 2067#endif 2068} 2069 2070static inline int sas_ss_flags(unsigned long sp) 2071{ 2072 return (current->sas_ss_size == 0 ? SS_DISABLE 2073 : on_sig_stack(sp) ? SS_ONSTACK : 0); 2074} 2075 2076/* 2077 * Routines for handling mm_structs 2078 */ 2079extern struct mm_struct * mm_alloc(void); 2080 2081/* mmdrop drops the mm and the page tables */ 2082extern void __mmdrop(struct mm_struct *); 2083static inline void mmdrop(struct mm_struct * mm) 2084{ 2085 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 2086 __mmdrop(mm); 2087} 2088 2089/* mmput gets rid of the mappings and all user-space */ 2090extern void mmput(struct mm_struct *); 2091/* Grab a reference to a task's mm, if it is not already going away */ 2092extern struct mm_struct *get_task_mm(struct task_struct *task); 2093/* Remove the current tasks stale references to the old mm_struct */ 2094extern void mm_release(struct task_struct *, struct mm_struct *); 2095/* Allocate a new mm structure and copy contents from tsk->mm */ 2096extern struct mm_struct *dup_mm(struct task_struct *tsk); 2097 2098extern int copy_thread(unsigned long, unsigned long, unsigned long, 2099 struct task_struct *, struct pt_regs *); 2100extern void flush_thread(void); 2101extern void exit_thread(void); 2102 2103extern void exit_files(struct task_struct *); 2104extern void __cleanup_signal(struct signal_struct *); 2105extern void __cleanup_sighand(struct sighand_struct *); 2106 2107extern void exit_itimers(struct signal_struct *); 2108extern void flush_itimer_signals(void); 2109 2110extern NORET_TYPE void do_group_exit(int); 2111 2112extern void daemonize(const char *, ...); 2113extern int allow_signal(int); 2114extern int disallow_signal(int); 2115 2116extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 2117extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 2118struct task_struct *fork_idle(int); 2119 2120extern void set_task_comm(struct task_struct *tsk, char *from); 2121extern char *get_task_comm(char *to, struct task_struct *tsk); 2122 2123#ifdef CONFIG_SMP 2124extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 2125#else 2126static inline unsigned long wait_task_inactive(struct task_struct *p, 2127 long match_state) 2128{ 2129 return 1; 2130} 2131#endif 2132 2133#define next_task(p) \ 2134 list_entry_rcu((p)->tasks.next, struct task_struct, tasks) 2135 2136#define for_each_process(p) \ 2137 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 2138 2139extern bool current_is_single_threaded(void); 2140 2141/* 2142 * Careful: do_each_thread/while_each_thread is a double loop so 2143 * 'break' will not work as expected - use goto instead. 2144 */ 2145#define do_each_thread(g, t) \ 2146 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 2147 2148#define while_each_thread(g, t) \ 2149 while ((t = next_thread(t)) != g) 2150 2151/* de_thread depends on thread_group_leader not being a pid based check */ 2152#define thread_group_leader(p) (p == p->group_leader) 2153 2154/* Do to the insanities of de_thread it is possible for a process 2155 * to have the pid of the thread group leader without actually being 2156 * the thread group leader. For iteration through the pids in proc 2157 * all we care about is that we have a task with the appropriate 2158 * pid, we don't actually care if we have the right task. 2159 */ 2160static inline int has_group_leader_pid(struct task_struct *p) 2161{ 2162 return p->pid == p->tgid; 2163} 2164 2165static inline 2166int same_thread_group(struct task_struct *p1, struct task_struct *p2) 2167{ 2168 return p1->tgid == p2->tgid; 2169} 2170 2171static inline struct task_struct *next_thread(const struct task_struct *p) 2172{ 2173 return list_entry_rcu(p->thread_group.next, 2174 struct task_struct, thread_group); 2175} 2176 2177static inline int thread_group_empty(struct task_struct *p) 2178{ 2179 return list_empty(&p->thread_group); 2180} 2181 2182#define delay_group_leader(p) \ 2183 (thread_group_leader(p) && !thread_group_empty(p)) 2184 2185static inline int task_detached(struct task_struct *p) 2186{ 2187 return p->exit_signal == -1; 2188} 2189 2190/* 2191 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 2192 * subscriptions and synchronises with wait4(). Also used in procfs. Also 2193 * pins the final release of task.io_context. Also protects ->cpuset and 2194 * ->cgroup.subsys[]. 2195 * 2196 * Nests both inside and outside of read_lock(&tasklist_lock). 2197 * It must not be nested with write_lock_irq(&tasklist_lock), 2198 * neither inside nor outside. 2199 */ 2200static inline void task_lock(struct task_struct *p) 2201{ 2202 spin_lock(&p->alloc_lock); 2203} 2204 2205static inline void task_unlock(struct task_struct *p) 2206{ 2207 spin_unlock(&p->alloc_lock); 2208} 2209 2210extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 2211 unsigned long *flags); 2212 2213static inline void unlock_task_sighand(struct task_struct *tsk, 2214 unsigned long *flags) 2215{ 2216 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 2217} 2218 2219#ifndef __HAVE_THREAD_FUNCTIONS 2220 2221#define task_thread_info(task) ((struct thread_info *)(task)->stack) 2222#define task_stack_page(task) ((task)->stack) 2223 2224static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 2225{ 2226 *task_thread_info(p) = *task_thread_info(org); 2227 task_thread_info(p)->task = p; 2228} 2229 2230static inline unsigned long *end_of_stack(struct task_struct *p) 2231{ 2232 return (unsigned long *)(task_thread_info(p) + 1); 2233} 2234 2235#endif 2236 2237static inline int object_is_on_stack(void *obj) 2238{ 2239 void *stack = task_stack_page(current); 2240 2241 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 2242} 2243 2244extern void thread_info_cache_init(void); 2245 2246#ifdef CONFIG_DEBUG_STACK_USAGE 2247static inline unsigned long stack_not_used(struct task_struct *p) 2248{ 2249 unsigned long *n = end_of_stack(p); 2250 2251 do { /* Skip over canary */ 2252 n++; 2253 } while (!*n); 2254 2255 return (unsigned long)n - (unsigned long)end_of_stack(p); 2256} 2257#endif 2258 2259/* set thread flags in other task's structures 2260 * - see asm/thread_info.h for TIF_xxxx flags available 2261 */ 2262static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 2263{ 2264 set_ti_thread_flag(task_thread_info(tsk), flag); 2265} 2266 2267static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2268{ 2269 clear_ti_thread_flag(task_thread_info(tsk), flag); 2270} 2271 2272static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2273{ 2274 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2275} 2276 2277static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2278{ 2279 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2280} 2281 2282static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2283{ 2284 return test_ti_thread_flag(task_thread_info(tsk), flag); 2285} 2286 2287static inline void set_tsk_need_resched(struct task_struct *tsk) 2288{ 2289 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2290} 2291 2292static inline void clear_tsk_need_resched(struct task_struct *tsk) 2293{ 2294 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2295} 2296 2297static inline int test_tsk_need_resched(struct task_struct *tsk) 2298{ 2299 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2300} 2301 2302static inline int restart_syscall(void) 2303{ 2304 set_tsk_thread_flag(current, TIF_SIGPENDING); 2305 return -ERESTARTNOINTR; 2306} 2307 2308static inline int signal_pending(struct task_struct *p) 2309{ 2310 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2311} 2312 2313static inline int __fatal_signal_pending(struct task_struct *p) 2314{ 2315 return unlikely(sigismember(&p->pending.signal, SIGKILL)); 2316} 2317 2318static inline int fatal_signal_pending(struct task_struct *p) 2319{ 2320 return signal_pending(p) && __fatal_signal_pending(p); 2321} 2322 2323static inline int signal_pending_state(long state, struct task_struct *p) 2324{ 2325 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2326 return 0; 2327 if (!signal_pending(p)) 2328 return 0; 2329 2330 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2331} 2332 2333static inline int need_resched(void) 2334{ 2335 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2336} 2337 2338/* 2339 * cond_resched() and cond_resched_lock(): latency reduction via 2340 * explicit rescheduling in places that are safe. The return 2341 * value indicates whether a reschedule was done in fact. 2342 * cond_resched_lock() will drop the spinlock before scheduling, 2343 * cond_resched_softirq() will enable bhs before scheduling. 2344 */ 2345extern int _cond_resched(void); 2346 2347#define cond_resched() ({ \ 2348 __might_sleep(__FILE__, __LINE__, 0); \ 2349 _cond_resched(); \ 2350}) 2351 2352extern int __cond_resched_lock(spinlock_t *lock); 2353 2354#ifdef CONFIG_PREEMPT 2355#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET 2356#else 2357#define PREEMPT_LOCK_OFFSET 0 2358#endif 2359 2360#define cond_resched_lock(lock) ({ \ 2361 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \ 2362 __cond_resched_lock(lock); \ 2363}) 2364 2365extern int __cond_resched_softirq(void); 2366 2367#define cond_resched_softirq() ({ \ 2368 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \ 2369 __cond_resched_softirq(); \ 2370}) 2371 2372/* 2373 * Does a critical section need to be broken due to another 2374 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2375 * but a general need for low latency) 2376 */ 2377static inline int spin_needbreak(spinlock_t *lock) 2378{ 2379#ifdef CONFIG_PREEMPT 2380 return spin_is_contended(lock); 2381#else 2382 return 0; 2383#endif 2384} 2385 2386/* 2387 * Thread group CPU time accounting. 2388 */ 2389void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); 2390void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); 2391 2392static inline void thread_group_cputime_init(struct signal_struct *sig) 2393{ 2394 spin_lock_init(&sig->cputimer.lock); 2395} 2396 2397static inline void thread_group_cputime_free(struct signal_struct *sig) 2398{ 2399} 2400 2401/* 2402 * Reevaluate whether the task has signals pending delivery. 2403 * Wake the task if so. 2404 * This is required every time the blocked sigset_t changes. 2405 * callers must hold sighand->siglock. 2406 */ 2407extern void recalc_sigpending_and_wake(struct task_struct *t); 2408extern void recalc_sigpending(void); 2409 2410extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2411 2412/* 2413 * Wrappers for p->thread_info->cpu access. No-op on UP. 2414 */ 2415#ifdef CONFIG_SMP 2416 2417static inline unsigned int task_cpu(const struct task_struct *p) 2418{ 2419 return task_thread_info(p)->cpu; 2420} 2421 2422extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2423 2424#else 2425 2426static inline unsigned int task_cpu(const struct task_struct *p) 2427{ 2428 return 0; 2429} 2430 2431static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2432{ 2433} 2434 2435#endif /* CONFIG_SMP */ 2436 2437#ifdef CONFIG_TRACING 2438extern void 2439__trace_special(void *__tr, void *__data, 2440 unsigned long arg1, unsigned long arg2, unsigned long arg3); 2441#else 2442static inline void 2443__trace_special(void *__tr, void *__data, 2444 unsigned long arg1, unsigned long arg2, unsigned long arg3) 2445{ 2446} 2447#endif 2448 2449extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); 2450extern long sched_getaffinity(pid_t pid, struct cpumask *mask); 2451 2452extern void normalize_rt_tasks(void); 2453 2454#ifdef CONFIG_CGROUP_SCHED 2455 2456extern struct task_group init_task_group; 2457 2458extern struct task_group *sched_create_group(struct task_group *parent); 2459extern void sched_destroy_group(struct task_group *tg); 2460extern void sched_move_task(struct task_struct *tsk); 2461#ifdef CONFIG_FAIR_GROUP_SCHED 2462extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2463extern unsigned long sched_group_shares(struct task_group *tg); 2464#endif 2465#ifdef CONFIG_RT_GROUP_SCHED 2466extern int sched_group_set_rt_runtime(struct task_group *tg, 2467 long rt_runtime_us); 2468extern long sched_group_rt_runtime(struct task_group *tg); 2469extern int sched_group_set_rt_period(struct task_group *tg, 2470 long rt_period_us); 2471extern long sched_group_rt_period(struct task_group *tg); 2472extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk); 2473#endif 2474#endif 2475 2476extern int task_can_switch_user(struct user_struct *up, 2477 struct task_struct *tsk); 2478 2479#ifdef CONFIG_TASK_XACCT 2480static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2481{ 2482 tsk->ioac.rchar += amt; 2483} 2484 2485static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2486{ 2487 tsk->ioac.wchar += amt; 2488} 2489 2490static inline void inc_syscr(struct task_struct *tsk) 2491{ 2492 tsk->ioac.syscr++; 2493} 2494 2495static inline void inc_syscw(struct task_struct *tsk) 2496{ 2497 tsk->ioac.syscw++; 2498} 2499#else 2500static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2501{ 2502} 2503 2504static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2505{ 2506} 2507 2508static inline void inc_syscr(struct task_struct *tsk) 2509{ 2510} 2511 2512static inline void inc_syscw(struct task_struct *tsk) 2513{ 2514} 2515#endif 2516 2517#ifndef TASK_SIZE_OF 2518#define TASK_SIZE_OF(tsk) TASK_SIZE 2519#endif 2520 2521/* 2522 * Call the function if the target task is executing on a CPU right now: 2523 */ 2524extern void task_oncpu_function_call(struct task_struct *p, 2525 void (*func) (void *info), void *info); 2526 2527 2528#ifdef CONFIG_MM_OWNER 2529extern void mm_update_next_owner(struct mm_struct *mm); 2530extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2531#else 2532static inline void mm_update_next_owner(struct mm_struct *mm) 2533{ 2534} 2535 2536static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2537{ 2538} 2539#endif /* CONFIG_MM_OWNER */ 2540 2541static inline unsigned long task_rlimit(const struct task_struct *tsk, 2542 unsigned int limit) 2543{ 2544 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur); 2545} 2546 2547static inline unsigned long task_rlimit_max(const struct task_struct *tsk, 2548 unsigned int limit) 2549{ 2550 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max); 2551} 2552 2553static inline unsigned long rlimit(unsigned int limit) 2554{ 2555 return task_rlimit(current, limit); 2556} 2557 2558static inline unsigned long rlimit_max(unsigned int limit) 2559{ 2560 return task_rlimit_max(current, limit); 2561} 2562 2563#endif /* __KERNEL__ */ 2564 2565#endif 2566