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(int cpu); 143extern unsigned long this_cpu_load(void); 144 145 146extern void calc_global_load(unsigned long ticks); 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_dead(task) ((task)->exit_state != 0) 218#define task_is_stopped_or_traced(task) \ 219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 220#define task_contributes_to_load(task) \ 221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ 222 (task->flags & PF_FREEZING) == 0) 223 224#define __set_task_state(tsk, state_value) \ 225 do { (tsk)->state = (state_value); } while (0) 226#define set_task_state(tsk, state_value) \ 227 set_mb((tsk)->state, (state_value)) 228 229/* 230 * set_current_state() includes a barrier so that the write of current->state 231 * is correctly serialised wrt the caller's subsequent test of whether to 232 * actually sleep: 233 * 234 * set_current_state(TASK_UNINTERRUPTIBLE); 235 * if (do_i_need_to_sleep()) 236 * schedule(); 237 * 238 * If the caller does not need such serialisation then use __set_current_state() 239 */ 240#define __set_current_state(state_value) \ 241 do { current->state = (state_value); } while (0) 242#define set_current_state(state_value) \ 243 set_mb(current->state, (state_value)) 244 245/* Task command name length */ 246#define TASK_COMM_LEN 16 247 248#include <linux/spinlock.h> 249 250/* 251 * This serializes "schedule()" and also protects 252 * the run-queue from deletions/modifications (but 253 * _adding_ to the beginning of the run-queue has 254 * a separate lock). 255 */ 256extern rwlock_t tasklist_lock; 257extern spinlock_t mmlist_lock; 258 259struct task_struct; 260 261#ifdef CONFIG_PROVE_RCU 262extern int lockdep_tasklist_lock_is_held(void); 263#endif /* #ifdef CONFIG_PROVE_RCU */ 264 265extern void sched_init(void); 266extern void sched_init_smp(void); 267extern asmlinkage void schedule_tail(struct task_struct *prev); 268extern void init_idle(struct task_struct *idle, int cpu); 269extern void init_idle_bootup_task(struct task_struct *idle); 270 271extern int runqueue_is_locked(int cpu); 272 273extern cpumask_var_t nohz_cpu_mask; 274#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 275extern void select_nohz_load_balancer(int stop_tick); 276extern int get_nohz_timer_target(void); 277#else 278static inline void select_nohz_load_balancer(int stop_tick) { } 279#endif 280 281/* 282 * Only dump TASK_* tasks. (0 for all tasks) 283 */ 284extern void show_state_filter(unsigned long state_filter); 285 286static inline void show_state(void) 287{ 288 show_state_filter(0); 289} 290 291extern void show_regs(struct pt_regs *); 292 293/* 294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 295 * task), SP is the stack pointer of the first frame that should be shown in the back 296 * trace (or NULL if the entire call-chain of the task should be shown). 297 */ 298extern void show_stack(struct task_struct *task, unsigned long *sp); 299 300void io_schedule(void); 301long io_schedule_timeout(long timeout); 302 303extern void cpu_init (void); 304extern void trap_init(void); 305extern void update_process_times(int user); 306extern void scheduler_tick(void); 307 308extern void sched_show_task(struct task_struct *p); 309 310#ifdef CONFIG_LOCKUP_DETECTOR 311extern void touch_softlockup_watchdog(void); 312extern void touch_softlockup_watchdog_sync(void); 313extern void touch_all_softlockup_watchdogs(void); 314extern int proc_dowatchdog_thresh(struct ctl_table *table, int write, 315 void __user *buffer, 316 size_t *lenp, loff_t *ppos); 317extern unsigned int softlockup_panic; 318extern int softlockup_thresh; 319#else 320static inline void touch_softlockup_watchdog(void) 321{ 322} 323static inline void touch_softlockup_watchdog_sync(void) 324{ 325} 326static inline void touch_all_softlockup_watchdogs(void) 327{ 328} 329#endif 330 331#ifdef CONFIG_DETECT_HUNG_TASK 332extern unsigned int sysctl_hung_task_panic; 333extern unsigned long sysctl_hung_task_check_count; 334extern unsigned long sysctl_hung_task_timeout_secs; 335extern unsigned long sysctl_hung_task_warnings; 336extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write, 337 void __user *buffer, 338 size_t *lenp, loff_t *ppos); 339#endif 340 341/* Attach to any functions which should be ignored in wchan output. */ 342#define __sched __attribute__((__section__(".sched.text"))) 343 344/* Linker adds these: start and end of __sched functions */ 345extern char __sched_text_start[], __sched_text_end[]; 346 347/* Is this address in the __sched functions? */ 348extern int in_sched_functions(unsigned long addr); 349 350#define MAX_SCHEDULE_TIMEOUT LONG_MAX 351extern signed long schedule_timeout(signed long timeout); 352extern signed long schedule_timeout_interruptible(signed long timeout); 353extern signed long schedule_timeout_killable(signed long timeout); 354extern signed long schedule_timeout_uninterruptible(signed long timeout); 355asmlinkage void schedule(void); 356extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner); 357 358struct nsproxy; 359struct user_namespace; 360 361/* 362 * Default maximum number of active map areas, this limits the number of vmas 363 * per mm struct. Users can overwrite this number by sysctl but there is a 364 * problem. 365 * 366 * When a program's coredump is generated as ELF format, a section is created 367 * per a vma. In ELF, the number of sections is represented in unsigned short. 368 * This means the number of sections should be smaller than 65535 at coredump. 369 * Because the kernel adds some informative sections to a image of program at 370 * generating coredump, we need some margin. The number of extra sections is 371 * 1-3 now and depends on arch. We use "5" as safe margin, here. 372 */ 373#define MAPCOUNT_ELF_CORE_MARGIN (5) 374#define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN) 375 376extern int sysctl_max_map_count; 377 378#include <linux/aio.h> 379 380#ifdef CONFIG_MMU 381extern void arch_pick_mmap_layout(struct mm_struct *mm); 382extern unsigned long 383arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 384 unsigned long, unsigned long); 385extern unsigned long 386arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 387 unsigned long len, unsigned long pgoff, 388 unsigned long flags); 389extern void arch_unmap_area(struct mm_struct *, unsigned long); 390extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 391#else 392static inline void arch_pick_mmap_layout(struct mm_struct *mm) {} 393#endif 394 395 396extern void set_dumpable(struct mm_struct *mm, int value); 397extern int get_dumpable(struct mm_struct *mm); 398 399/* mm flags */ 400/* dumpable bits */ 401#define MMF_DUMPABLE 0 /* core dump is permitted */ 402#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 403 404#define MMF_DUMPABLE_BITS 2 405#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1) 406 407/* coredump filter bits */ 408#define MMF_DUMP_ANON_PRIVATE 2 409#define MMF_DUMP_ANON_SHARED 3 410#define MMF_DUMP_MAPPED_PRIVATE 4 411#define MMF_DUMP_MAPPED_SHARED 5 412#define MMF_DUMP_ELF_HEADERS 6 413#define MMF_DUMP_HUGETLB_PRIVATE 7 414#define MMF_DUMP_HUGETLB_SHARED 8 415 416#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 417#define MMF_DUMP_FILTER_BITS 7 418#define MMF_DUMP_FILTER_MASK \ 419 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 420#define MMF_DUMP_FILTER_DEFAULT \ 421 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ 422 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) 423 424#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS 425# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) 426#else 427# define MMF_DUMP_MASK_DEFAULT_ELF 0 428#endif 429 /* leave room for more dump flags */ 430#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */ 431 432#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK) 433 434struct sighand_struct { 435 atomic_t count; 436 struct k_sigaction action[_NSIG]; 437 spinlock_t siglock; 438 wait_queue_head_t signalfd_wqh; 439}; 440 441struct pacct_struct { 442 int ac_flag; 443 long ac_exitcode; 444 unsigned long ac_mem; 445 cputime_t ac_utime, ac_stime; 446 unsigned long ac_minflt, ac_majflt; 447}; 448 449struct cpu_itimer { 450 cputime_t expires; 451 cputime_t incr; 452 u32 error; 453 u32 incr_error; 454}; 455 456/** 457 * struct task_cputime - collected CPU time counts 458 * @utime: time spent in user mode, in &cputime_t units 459 * @stime: time spent in kernel mode, in &cputime_t units 460 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds 461 * 462 * This structure groups together three kinds of CPU time that are 463 * tracked for threads and thread groups. Most things considering 464 * CPU time want to group these counts together and treat all three 465 * of them in parallel. 466 */ 467struct task_cputime { 468 cputime_t utime; 469 cputime_t stime; 470 unsigned long long sum_exec_runtime; 471}; 472/* Alternate field names when used to cache expirations. */ 473#define prof_exp stime 474#define virt_exp utime 475#define sched_exp sum_exec_runtime 476 477#define INIT_CPUTIME \ 478 (struct task_cputime) { \ 479 .utime = cputime_zero, \ 480 .stime = cputime_zero, \ 481 .sum_exec_runtime = 0, \ 482 } 483 484/* 485 * Disable preemption until the scheduler is running. 486 * Reset by start_kernel()->sched_init()->init_idle(). 487 * 488 * We include PREEMPT_ACTIVE to avoid cond_resched() from working 489 * before the scheduler is active -- see should_resched(). 490 */ 491#define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE) 492 493/** 494 * struct thread_group_cputimer - thread group interval timer counts 495 * @cputime: thread group interval timers. 496 * @running: non-zero when there are timers running and 497 * @cputime receives updates. 498 * @lock: lock for fields in this struct. 499 * 500 * This structure contains the version of task_cputime, above, that is 501 * used for thread group CPU timer calculations. 502 */ 503struct thread_group_cputimer { 504 struct task_cputime cputime; 505 int running; 506 spinlock_t lock; 507}; 508 509/* 510 * NOTE! "signal_struct" does not have it's own 511 * locking, because a shared signal_struct always 512 * implies a shared sighand_struct, so locking 513 * sighand_struct is always a proper superset of 514 * the locking of signal_struct. 515 */ 516struct signal_struct { 517 atomic_t sigcnt; 518 atomic_t live; 519 int nr_threads; 520 521 wait_queue_head_t wait_chldexit; /* for wait4() */ 522 523 /* current thread group signal load-balancing target: */ 524 struct task_struct *curr_target; 525 526 /* shared signal handling: */ 527 struct sigpending shared_pending; 528 529 /* thread group exit support */ 530 int group_exit_code; 531 /* overloaded: 532 * - notify group_exit_task when ->count is equal to notify_count 533 * - everyone except group_exit_task is stopped during signal delivery 534 * of fatal signals, group_exit_task processes the signal. 535 */ 536 int notify_count; 537 struct task_struct *group_exit_task; 538 539 /* thread group stop support, overloads group_exit_code too */ 540 int group_stop_count; 541 unsigned int flags; /* see SIGNAL_* flags below */ 542 543 /* POSIX.1b Interval Timers */ 544 struct list_head posix_timers; 545 546 /* ITIMER_REAL timer for the process */ 547 struct hrtimer real_timer; 548 struct pid *leader_pid; 549 ktime_t it_real_incr; 550 551 /* 552 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use 553 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these 554 * values are defined to 0 and 1 respectively 555 */ 556 struct cpu_itimer it[2]; 557 558 /* 559 * Thread group totals for process CPU timers. 560 * See thread_group_cputimer(), et al, for details. 561 */ 562 struct thread_group_cputimer cputimer; 563 564 /* Earliest-expiration cache. */ 565 struct task_cputime cputime_expires; 566 567 struct list_head cpu_timers[3]; 568 569 struct pid *tty_old_pgrp; 570 571 /* boolean value for session group leader */ 572 int leader; 573 574 struct tty_struct *tty; /* NULL if no tty */ 575 576 /* 577 * Cumulative resource counters for dead threads in the group, 578 * and for reaped dead child processes forked by this group. 579 * Live threads maintain their own counters and add to these 580 * in __exit_signal, except for the group leader. 581 */ 582 cputime_t utime, stime, cutime, cstime; 583 cputime_t gtime; 584 cputime_t cgtime; 585#ifndef CONFIG_VIRT_CPU_ACCOUNTING 586 cputime_t prev_utime, prev_stime; 587#endif 588 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 589 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 590 unsigned long inblock, oublock, cinblock, coublock; 591 unsigned long maxrss, cmaxrss; 592 struct task_io_accounting ioac; 593 594 /* 595 * Cumulative ns of schedule CPU time fo dead threads in the 596 * group, not including a zombie group leader, (This only differs 597 * from jiffies_to_ns(utime + stime) if sched_clock uses something 598 * other than jiffies.) 599 */ 600 unsigned long long sum_sched_runtime; 601 602 /* 603 * We don't bother to synchronize most readers of this at all, 604 * because there is no reader checking a limit that actually needs 605 * to get both rlim_cur and rlim_max atomically, and either one 606 * alone is a single word that can safely be read normally. 607 * getrlimit/setrlimit use task_lock(current->group_leader) to 608 * protect this instead of the siglock, because they really 609 * have no need to disable irqs. 610 */ 611 struct rlimit rlim[RLIM_NLIMITS]; 612 613#ifdef CONFIG_BSD_PROCESS_ACCT 614 struct pacct_struct pacct; /* per-process accounting information */ 615#endif 616#ifdef CONFIG_TASKSTATS 617 struct taskstats *stats; 618#endif 619#ifdef CONFIG_AUDIT 620 unsigned audit_tty; 621 struct tty_audit_buf *tty_audit_buf; 622#endif 623 624 int oom_adj; /* OOM kill score adjustment (bit shift) */ 625 int oom_score_adj; /* OOM kill score adjustment */ 626}; 627 628/* Context switch must be unlocked if interrupts are to be enabled */ 629#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 630# define __ARCH_WANT_UNLOCKED_CTXSW 631#endif 632 633/* 634 * Bits in flags field of signal_struct. 635 */ 636#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 637#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 638#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 639#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 640/* 641 * Pending notifications to parent. 642 */ 643#define SIGNAL_CLD_STOPPED 0x00000010 644#define SIGNAL_CLD_CONTINUED 0x00000020 645#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 646 647#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 648 649/* If true, all threads except ->group_exit_task have pending SIGKILL */ 650static inline int signal_group_exit(const struct signal_struct *sig) 651{ 652 return (sig->flags & SIGNAL_GROUP_EXIT) || 653 (sig->group_exit_task != NULL); 654} 655 656/* 657 * Some day this will be a full-fledged user tracking system.. 658 */ 659struct user_struct { 660 atomic_t __count; /* reference count */ 661 atomic_t processes; /* How many processes does this user have? */ 662 atomic_t files; /* How many open files does this user have? */ 663 atomic_t sigpending; /* How many pending signals does this user have? */ 664#ifdef CONFIG_INOTIFY_USER 665 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 666 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 667#endif 668#ifdef CONFIG_EPOLL 669 atomic_t epoll_watches; /* The number of file descriptors currently watched */ 670#endif 671#ifdef CONFIG_POSIX_MQUEUE 672 /* protected by mq_lock */ 673 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 674#endif 675 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 676 677#ifdef CONFIG_KEYS 678 struct key *uid_keyring; /* UID specific keyring */ 679 struct key *session_keyring; /* UID's default session keyring */ 680#endif 681 682 /* Hash table maintenance information */ 683 struct hlist_node uidhash_node; 684 uid_t uid; 685 struct user_namespace *user_ns; 686 687#ifdef CONFIG_PERF_EVENTS 688 atomic_long_t locked_vm; 689#endif 690}; 691 692extern int uids_sysfs_init(void); 693 694extern struct user_struct *find_user(uid_t); 695 696extern struct user_struct root_user; 697#define INIT_USER (&root_user) 698 699 700struct backing_dev_info; 701struct reclaim_state; 702 703#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 704struct sched_info { 705 /* cumulative counters */ 706 unsigned long pcount; /* # of times run on this cpu */ 707 unsigned long long run_delay; /* time spent waiting on a runqueue */ 708 709 /* timestamps */ 710 unsigned long long last_arrival,/* when we last ran on a cpu */ 711 last_queued; /* when we were last queued to run */ 712#ifdef CONFIG_SCHEDSTATS 713 /* BKL stats */ 714 unsigned int bkl_count; 715#endif 716}; 717#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 718 719#ifdef CONFIG_TASK_DELAY_ACCT 720struct task_delay_info { 721 spinlock_t lock; 722 unsigned int flags; /* Private per-task flags */ 723 724 725 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 726 u64 blkio_delay; /* wait for sync block io completion */ 727 u64 swapin_delay; /* wait for swapin block io completion */ 728 u32 blkio_count; /* total count of the number of sync block */ 729 /* io operations performed */ 730 u32 swapin_count; /* total count of the number of swapin block */ 731 /* io operations performed */ 732 733 struct timespec freepages_start, freepages_end; 734 u64 freepages_delay; /* wait for memory reclaim */ 735 u32 freepages_count; /* total count of memory reclaim */ 736}; 737#endif /* CONFIG_TASK_DELAY_ACCT */ 738 739static inline int sched_info_on(void) 740{ 741#ifdef CONFIG_SCHEDSTATS 742 return 1; 743#elif defined(CONFIG_TASK_DELAY_ACCT) 744 extern int delayacct_on; 745 return delayacct_on; 746#else 747 return 0; 748#endif 749} 750 751enum cpu_idle_type { 752 CPU_IDLE, 753 CPU_NOT_IDLE, 754 CPU_NEWLY_IDLE, 755 CPU_MAX_IDLE_TYPES 756}; 757 758/* 759 * sched-domains (multiprocessor balancing) declarations: 760 */ 761 762/* 763 * Increase resolution of nice-level calculations: 764 */ 765#define SCHED_LOAD_SHIFT 10 766#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 767 768#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 769 770#ifdef CONFIG_SMP 771#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */ 772#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */ 773#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */ 774#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */ 775#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */ 776#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */ 777#define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */ 778#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */ 779#define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */ 780#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */ 781#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */ 782#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */ 783#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */ 784 785enum powersavings_balance_level { 786 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */ 787 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package 788 * first for long running threads 789 */ 790 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle 791 * cpu package for power savings 792 */ 793 MAX_POWERSAVINGS_BALANCE_LEVELS 794}; 795 796extern int sched_mc_power_savings, sched_smt_power_savings; 797 798static inline int sd_balance_for_mc_power(void) 799{ 800 if (sched_smt_power_savings) 801 return SD_POWERSAVINGS_BALANCE; 802 803 if (!sched_mc_power_savings) 804 return SD_PREFER_SIBLING; 805 806 return 0; 807} 808 809static inline int sd_balance_for_package_power(void) 810{ 811 if (sched_mc_power_savings | sched_smt_power_savings) 812 return SD_POWERSAVINGS_BALANCE; 813 814 return SD_PREFER_SIBLING; 815} 816 817extern int __weak arch_sd_sibiling_asym_packing(void); 818 819/* 820 * Optimise SD flags for power savings: 821 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings. 822 * Keep default SD flags if sched_{smt,mc}_power_saving=0 823 */ 824 825static inline int sd_power_saving_flags(void) 826{ 827 if (sched_mc_power_savings | sched_smt_power_savings) 828 return SD_BALANCE_NEWIDLE; 829 830 return 0; 831} 832 833struct sched_group { 834 struct sched_group *next; /* Must be a circular list */ 835 836 /* 837 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 838 * single CPU. 839 */ 840 unsigned int cpu_power, cpu_power_orig; 841 842 /* 843 * The CPUs this group covers. 844 * 845 * NOTE: this field is variable length. (Allocated dynamically 846 * by attaching extra space to the end of the structure, 847 * depending on how many CPUs the kernel has booted up with) 848 * 849 * It is also be embedded into static data structures at build 850 * time. (See 'struct static_sched_group' in kernel/sched.c) 851 */ 852 unsigned long cpumask[0]; 853}; 854 855static inline struct cpumask *sched_group_cpus(struct sched_group *sg) 856{ 857 return to_cpumask(sg->cpumask); 858} 859 860enum sched_domain_level { 861 SD_LV_NONE = 0, 862 SD_LV_SIBLING, 863 SD_LV_MC, 864 SD_LV_CPU, 865 SD_LV_NODE, 866 SD_LV_ALLNODES, 867 SD_LV_MAX 868}; 869 870struct sched_domain_attr { 871 int relax_domain_level; 872}; 873 874#define SD_ATTR_INIT (struct sched_domain_attr) { \ 875 .relax_domain_level = -1, \ 876} 877 878struct sched_domain { 879 /* These fields must be setup */ 880 struct sched_domain *parent; /* top domain must be null terminated */ 881 struct sched_domain *child; /* bottom domain must be null terminated */ 882 struct sched_group *groups; /* the balancing groups of the domain */ 883 unsigned long min_interval; /* Minimum balance interval ms */ 884 unsigned long max_interval; /* Maximum balance interval ms */ 885 unsigned int busy_factor; /* less balancing by factor if busy */ 886 unsigned int imbalance_pct; /* No balance until over watermark */ 887 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 888 unsigned int busy_idx; 889 unsigned int idle_idx; 890 unsigned int newidle_idx; 891 unsigned int wake_idx; 892 unsigned int forkexec_idx; 893 unsigned int smt_gain; 894 int flags; /* See SD_* */ 895 enum sched_domain_level level; 896 897 /* Runtime fields. */ 898 unsigned long last_balance; /* init to jiffies. units in jiffies */ 899 unsigned int balance_interval; /* initialise to 1. units in ms. */ 900 unsigned int nr_balance_failed; /* initialise to 0 */ 901 902 u64 last_update; 903 904#ifdef CONFIG_SCHEDSTATS 905 /* load_balance() stats */ 906 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 907 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 908 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 909 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 910 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 911 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 912 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 913 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 914 915 /* Active load balancing */ 916 unsigned int alb_count; 917 unsigned int alb_failed; 918 unsigned int alb_pushed; 919 920 /* SD_BALANCE_EXEC stats */ 921 unsigned int sbe_count; 922 unsigned int sbe_balanced; 923 unsigned int sbe_pushed; 924 925 /* SD_BALANCE_FORK stats */ 926 unsigned int sbf_count; 927 unsigned int sbf_balanced; 928 unsigned int sbf_pushed; 929 930 /* try_to_wake_up() stats */ 931 unsigned int ttwu_wake_remote; 932 unsigned int ttwu_move_affine; 933 unsigned int ttwu_move_balance; 934#endif 935#ifdef CONFIG_SCHED_DEBUG 936 char *name; 937#endif 938 939 unsigned int span_weight; 940 /* 941 * Span of all CPUs in this domain. 942 * 943 * NOTE: this field is variable length. (Allocated dynamically 944 * by attaching extra space to the end of the structure, 945 * depending on how many CPUs the kernel has booted up with) 946 * 947 * It is also be embedded into static data structures at build 948 * time. (See 'struct static_sched_domain' in kernel/sched.c) 949 */ 950 unsigned long span[0]; 951}; 952 953static inline struct cpumask *sched_domain_span(struct sched_domain *sd) 954{ 955 return to_cpumask(sd->span); 956} 957 958extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], 959 struct sched_domain_attr *dattr_new); 960 961/* Allocate an array of sched domains, for partition_sched_domains(). */ 962cpumask_var_t *alloc_sched_domains(unsigned int ndoms); 963void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms); 964 965/* Test a flag in parent sched domain */ 966static inline int test_sd_parent(struct sched_domain *sd, int flag) 967{ 968 if (sd->parent && (sd->parent->flags & flag)) 969 return 1; 970 971 return 0; 972} 973 974unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu); 975unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu); 976 977#else /* CONFIG_SMP */ 978 979struct sched_domain_attr; 980 981static inline void 982partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], 983 struct sched_domain_attr *dattr_new) 984{ 985} 986#endif /* !CONFIG_SMP */ 987 988 989struct io_context; /* See blkdev.h */ 990 991 992#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 993extern void prefetch_stack(struct task_struct *t); 994#else 995static inline void prefetch_stack(struct task_struct *t) { } 996#endif 997 998struct audit_context; /* See audit.c */ 999struct mempolicy; 1000struct pipe_inode_info; 1001struct uts_namespace; 1002 1003struct rq; 1004struct sched_domain; 1005 1006/* 1007 * wake flags 1008 */ 1009#define WF_SYNC 0x01 /* waker goes to sleep after wakup */ 1010#define WF_FORK 0x02 /* child wakeup after fork */ 1011 1012#define ENQUEUE_WAKEUP 1 1013#define ENQUEUE_WAKING 2 1014#define ENQUEUE_HEAD 4 1015 1016#define DEQUEUE_SLEEP 1 1017 1018struct sched_class { 1019 const struct sched_class *next; 1020 1021 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags); 1022 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags); 1023 void (*yield_task) (struct rq *rq); 1024 1025 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags); 1026 1027 struct task_struct * (*pick_next_task) (struct rq *rq); 1028 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 1029 1030#ifdef CONFIG_SMP 1031 int (*select_task_rq)(struct rq *rq, struct task_struct *p, 1032 int sd_flag, int flags); 1033 1034 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 1035 void (*post_schedule) (struct rq *this_rq); 1036 void (*task_waking) (struct rq *this_rq, struct task_struct *task); 1037 void (*task_woken) (struct rq *this_rq, struct task_struct *task); 1038 1039 void (*set_cpus_allowed)(struct task_struct *p, 1040 const struct cpumask *newmask); 1041 1042 void (*rq_online)(struct rq *rq); 1043 void (*rq_offline)(struct rq *rq); 1044#endif 1045 1046 void (*set_curr_task) (struct rq *rq); 1047 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 1048 void (*task_fork) (struct task_struct *p); 1049 1050 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 1051 int running); 1052 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 1053 int running); 1054 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 1055 int oldprio, int running); 1056 1057 unsigned int (*get_rr_interval) (struct rq *rq, 1058 struct task_struct *task); 1059 1060#ifdef CONFIG_FAIR_GROUP_SCHED 1061 void (*moved_group) (struct task_struct *p, int on_rq); 1062#endif 1063}; 1064 1065struct load_weight { 1066 unsigned long weight, inv_weight; 1067}; 1068 1069#ifdef CONFIG_SCHEDSTATS 1070struct sched_statistics { 1071 u64 wait_start; 1072 u64 wait_max; 1073 u64 wait_count; 1074 u64 wait_sum; 1075 u64 iowait_count; 1076 u64 iowait_sum; 1077 1078 u64 sleep_start; 1079 u64 sleep_max; 1080 s64 sum_sleep_runtime; 1081 1082 u64 block_start; 1083 u64 block_max; 1084 u64 exec_max; 1085 u64 slice_max; 1086 1087 u64 nr_migrations_cold; 1088 u64 nr_failed_migrations_affine; 1089 u64 nr_failed_migrations_running; 1090 u64 nr_failed_migrations_hot; 1091 u64 nr_forced_migrations; 1092 1093 u64 nr_wakeups; 1094 u64 nr_wakeups_sync; 1095 u64 nr_wakeups_migrate; 1096 u64 nr_wakeups_local; 1097 u64 nr_wakeups_remote; 1098 u64 nr_wakeups_affine; 1099 u64 nr_wakeups_affine_attempts; 1100 u64 nr_wakeups_passive; 1101 u64 nr_wakeups_idle; 1102}; 1103#endif 1104 1105struct sched_entity { 1106 struct load_weight load; /* for load-balancing */ 1107 struct rb_node run_node; 1108 struct list_head group_node; 1109 unsigned int on_rq; 1110 1111 u64 exec_start; 1112 u64 sum_exec_runtime; 1113 u64 vruntime; 1114 u64 prev_sum_exec_runtime; 1115 1116 u64 nr_migrations; 1117 1118#ifdef CONFIG_SCHEDSTATS 1119 struct sched_statistics statistics; 1120#endif 1121 1122#ifdef CONFIG_FAIR_GROUP_SCHED 1123 struct sched_entity *parent; 1124 /* rq on which this entity is (to be) queued: */ 1125 struct cfs_rq *cfs_rq; 1126 /* rq "owned" by this entity/group: */ 1127 struct cfs_rq *my_q; 1128#endif 1129}; 1130 1131struct sched_rt_entity { 1132 struct list_head run_list; 1133 unsigned long timeout; 1134 unsigned int time_slice; 1135 int nr_cpus_allowed; 1136 1137 struct sched_rt_entity *back; 1138#ifdef CONFIG_RT_GROUP_SCHED 1139 struct sched_rt_entity *parent; 1140 /* rq on which this entity is (to be) queued: */ 1141 struct rt_rq *rt_rq; 1142 /* rq "owned" by this entity/group: */ 1143 struct rt_rq *my_q; 1144#endif 1145}; 1146 1147struct rcu_node; 1148 1149struct task_struct { 1150 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1151 void *stack; 1152 atomic_t usage; 1153 unsigned int flags; /* per process flags, defined below */ 1154 unsigned int ptrace; 1155 1156 int lock_depth; /* BKL lock depth */ 1157 1158#ifdef CONFIG_SMP 1159#ifdef __ARCH_WANT_UNLOCKED_CTXSW 1160 int oncpu; 1161#endif 1162#endif 1163 1164 int prio, static_prio, normal_prio; 1165 unsigned int rt_priority; 1166 const struct sched_class *sched_class; 1167 struct sched_entity se; 1168 struct sched_rt_entity rt; 1169 1170#ifdef CONFIG_PREEMPT_NOTIFIERS 1171 /* list of struct preempt_notifier: */ 1172 struct hlist_head preempt_notifiers; 1173#endif 1174 1175 /* 1176 * fpu_counter contains the number of consecutive context switches 1177 * that the FPU is used. If this is over a threshold, the lazy fpu 1178 * saving becomes unlazy to save the trap. This is an unsigned char 1179 * so that after 256 times the counter wraps and the behavior turns 1180 * lazy again; this to deal with bursty apps that only use FPU for 1181 * a short time 1182 */ 1183 unsigned char fpu_counter; 1184#ifdef CONFIG_BLK_DEV_IO_TRACE 1185 unsigned int btrace_seq; 1186#endif 1187 1188 unsigned int policy; 1189 cpumask_t cpus_allowed; 1190 1191#ifdef CONFIG_TREE_PREEMPT_RCU 1192 int rcu_read_lock_nesting; 1193 char rcu_read_unlock_special; 1194 struct rcu_node *rcu_blocked_node; 1195 struct list_head rcu_node_entry; 1196#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */ 1197 1198#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1199 struct sched_info sched_info; 1200#endif 1201 1202 struct list_head tasks; 1203 struct plist_node pushable_tasks; 1204 1205 struct mm_struct *mm, *active_mm; 1206#if defined(SPLIT_RSS_COUNTING) 1207 struct task_rss_stat rss_stat; 1208#endif 1209/* task state */ 1210 int exit_state; 1211 int exit_code, exit_signal; 1212 int pdeath_signal; /* The signal sent when the parent dies */ 1213 /* ??? */ 1214 unsigned int personality; 1215 unsigned did_exec:1; 1216 unsigned in_execve:1; /* Tell the LSMs that the process is doing an 1217 * execve */ 1218 unsigned in_iowait:1; 1219 1220 1221 /* Revert to default priority/policy when forking */ 1222 unsigned sched_reset_on_fork:1; 1223 1224 pid_t pid; 1225 pid_t tgid; 1226 1227#ifdef CONFIG_CC_STACKPROTECTOR 1228 /* Canary value for the -fstack-protector gcc feature */ 1229 unsigned long stack_canary; 1230#endif 1231 1232 /* 1233 * pointers to (original) parent process, youngest child, younger sibling, 1234 * older sibling, respectively. (p->father can be replaced with 1235 * p->real_parent->pid) 1236 */ 1237 struct task_struct *real_parent; /* real parent process */ 1238 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1239 /* 1240 * children/sibling forms the list of my natural children 1241 */ 1242 struct list_head children; /* list of my children */ 1243 struct list_head sibling; /* linkage in my parent's children list */ 1244 struct task_struct *group_leader; /* threadgroup leader */ 1245 1246 /* 1247 * ptraced is the list of tasks this task is using ptrace on. 1248 * This includes both natural children and PTRACE_ATTACH targets. 1249 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1250 */ 1251 struct list_head ptraced; 1252 struct list_head ptrace_entry; 1253 1254 /* PID/PID hash table linkage. */ 1255 struct pid_link pids[PIDTYPE_MAX]; 1256 struct list_head thread_group; 1257 1258 struct completion *vfork_done; /* for vfork() */ 1259 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1260 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1261 1262 cputime_t utime, stime, utimescaled, stimescaled; 1263 cputime_t gtime; 1264#ifndef CONFIG_VIRT_CPU_ACCOUNTING 1265 cputime_t prev_utime, prev_stime; 1266#endif 1267 unsigned long nvcsw, nivcsw; /* context switch counts */ 1268 struct timespec start_time; /* monotonic time */ 1269 struct timespec real_start_time; /* boot based time */ 1270/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1271 unsigned long min_flt, maj_flt; 1272 1273 struct task_cputime cputime_expires; 1274 struct list_head cpu_timers[3]; 1275 1276/* process credentials */ 1277 const struct cred *real_cred; /* objective and real subjective task 1278 * credentials (COW) */ 1279 const struct cred *cred; /* effective (overridable) subjective task 1280 * credentials (COW) */ 1281 struct mutex cred_guard_mutex; /* guard against foreign influences on 1282 * credential calculations 1283 * (notably. ptrace) */ 1284 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */ 1285 1286 char comm[TASK_COMM_LEN]; /* executable name excluding path 1287 - access with [gs]et_task_comm (which lock 1288 it with task_lock()) 1289 - initialized normally by setup_new_exec */ 1290/* file system info */ 1291 int link_count, total_link_count; 1292#ifdef CONFIG_SYSVIPC 1293/* ipc stuff */ 1294 struct sysv_sem sysvsem; 1295#endif 1296#ifdef CONFIG_DETECT_HUNG_TASK 1297/* hung task detection */ 1298 unsigned long last_switch_count; 1299#endif 1300/* CPU-specific state of this task */ 1301 struct thread_struct thread; 1302/* filesystem information */ 1303 struct fs_struct *fs; 1304/* open file information */ 1305 struct files_struct *files; 1306/* namespaces */ 1307 struct nsproxy *nsproxy; 1308/* signal handlers */ 1309 struct signal_struct *signal; 1310 struct sighand_struct *sighand; 1311 1312 sigset_t blocked, real_blocked; 1313 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1314 struct sigpending pending; 1315 1316 unsigned long sas_ss_sp; 1317 size_t sas_ss_size; 1318 int (*notifier)(void *priv); 1319 void *notifier_data; 1320 sigset_t *notifier_mask; 1321 struct audit_context *audit_context; 1322#ifdef CONFIG_AUDITSYSCALL 1323 uid_t loginuid; 1324 unsigned int sessionid; 1325#endif 1326 seccomp_t seccomp; 1327 1328/* Thread group tracking */ 1329 u32 parent_exec_id; 1330 u32 self_exec_id; 1331/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, 1332 * mempolicy */ 1333 spinlock_t alloc_lock; 1334 1335#ifdef CONFIG_GENERIC_HARDIRQS 1336 /* IRQ handler threads */ 1337 struct irqaction *irqaction; 1338#endif 1339 1340 /* Protection of the PI data structures: */ 1341 raw_spinlock_t pi_lock; 1342 1343#ifdef CONFIG_RT_MUTEXES 1344 /* PI waiters blocked on a rt_mutex held by this task */ 1345 struct plist_head pi_waiters; 1346 /* Deadlock detection and priority inheritance handling */ 1347 struct rt_mutex_waiter *pi_blocked_on; 1348#endif 1349 1350#ifdef CONFIG_DEBUG_MUTEXES 1351 /* mutex deadlock detection */ 1352 struct mutex_waiter *blocked_on; 1353#endif 1354#ifdef CONFIG_TRACE_IRQFLAGS 1355 unsigned int irq_events; 1356 unsigned long hardirq_enable_ip; 1357 unsigned long hardirq_disable_ip; 1358 unsigned int hardirq_enable_event; 1359 unsigned int hardirq_disable_event; 1360 int hardirqs_enabled; 1361 int hardirq_context; 1362 unsigned long softirq_disable_ip; 1363 unsigned long softirq_enable_ip; 1364 unsigned int softirq_disable_event; 1365 unsigned int softirq_enable_event; 1366 int softirqs_enabled; 1367 int softirq_context; 1368#endif 1369#ifdef CONFIG_LOCKDEP 1370# define MAX_LOCK_DEPTH 48UL 1371 u64 curr_chain_key; 1372 int lockdep_depth; 1373 unsigned int lockdep_recursion; 1374 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1375 gfp_t lockdep_reclaim_gfp; 1376#endif 1377 1378/* journalling filesystem info */ 1379 void *journal_info; 1380 1381/* stacked block device info */ 1382 struct bio_list *bio_list; 1383 1384/* VM state */ 1385 struct reclaim_state *reclaim_state; 1386 1387 struct backing_dev_info *backing_dev_info; 1388 1389 struct io_context *io_context; 1390 1391 unsigned long ptrace_message; 1392 siginfo_t *last_siginfo; /* For ptrace use. */ 1393 struct task_io_accounting ioac; 1394#if defined(CONFIG_TASK_XACCT) 1395 u64 acct_rss_mem1; /* accumulated rss usage */ 1396 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1397 cputime_t acct_timexpd; /* stime + utime since last update */ 1398#endif 1399#ifdef CONFIG_CPUSETS 1400 nodemask_t mems_allowed; /* Protected by alloc_lock */ 1401 int mems_allowed_change_disable; 1402 int cpuset_mem_spread_rotor; 1403 int cpuset_slab_spread_rotor; 1404#endif 1405#ifdef CONFIG_CGROUPS 1406 /* Control Group info protected by css_set_lock */ 1407 struct css_set *cgroups; 1408 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1409 struct list_head cg_list; 1410#endif 1411#ifdef CONFIG_FUTEX 1412 struct robust_list_head __user *robust_list; 1413#ifdef CONFIG_COMPAT 1414 struct compat_robust_list_head __user *compat_robust_list; 1415#endif 1416 struct list_head pi_state_list; 1417 struct futex_pi_state *pi_state_cache; 1418#endif 1419#ifdef CONFIG_PERF_EVENTS 1420 struct perf_event_context *perf_event_ctxp; 1421 struct mutex perf_event_mutex; 1422 struct list_head perf_event_list; 1423#endif 1424#ifdef CONFIG_NUMA 1425 struct mempolicy *mempolicy; /* Protected by alloc_lock */ 1426 short il_next; 1427#endif 1428 atomic_t fs_excl; /* holding fs exclusive resources */ 1429 struct rcu_head rcu; 1430 1431 /* 1432 * cache last used pipe for splice 1433 */ 1434 struct pipe_inode_info *splice_pipe; 1435#ifdef CONFIG_TASK_DELAY_ACCT 1436 struct task_delay_info *delays; 1437#endif 1438#ifdef CONFIG_FAULT_INJECTION 1439 int make_it_fail; 1440#endif 1441 struct prop_local_single dirties; 1442#ifdef CONFIG_LATENCYTOP 1443 int latency_record_count; 1444 struct latency_record latency_record[LT_SAVECOUNT]; 1445#endif 1446 /* 1447 * time slack values; these are used to round up poll() and 1448 * select() etc timeout values. These are in nanoseconds. 1449 */ 1450 unsigned long timer_slack_ns; 1451 unsigned long default_timer_slack_ns; 1452 1453 struct list_head *scm_work_list; 1454#ifdef CONFIG_FUNCTION_GRAPH_TRACER 1455 /* Index of current stored address in ret_stack */ 1456 int curr_ret_stack; 1457 /* Stack of return addresses for return function tracing */ 1458 struct ftrace_ret_stack *ret_stack; 1459 /* time stamp for last schedule */ 1460 unsigned long long ftrace_timestamp; 1461 /* 1462 * Number of functions that haven't been traced 1463 * because of depth overrun. 1464 */ 1465 atomic_t trace_overrun; 1466 /* Pause for the tracing */ 1467 atomic_t tracing_graph_pause; 1468#endif 1469#ifdef CONFIG_TRACING 1470 /* state flags for use by tracers */ 1471 unsigned long trace; 1472 /* bitmask of trace recursion */ 1473 unsigned long trace_recursion; 1474#endif /* CONFIG_TRACING */ 1475#ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */ 1476 struct memcg_batch_info { 1477 int do_batch; /* incremented when batch uncharge started */ 1478 struct mem_cgroup *memcg; /* target memcg of uncharge */ 1479 unsigned long bytes; /* uncharged usage */ 1480 unsigned long memsw_bytes; /* uncharged mem+swap usage */ 1481 } memcg_batch; 1482#endif 1483}; 1484 1485/* Future-safe accessor for struct task_struct's cpus_allowed. */ 1486#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed) 1487 1488/* 1489 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1490 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1491 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1492 * values are inverted: lower p->prio value means higher priority. 1493 * 1494 * The MAX_USER_RT_PRIO value allows the actual maximum 1495 * RT priority to be separate from the value exported to 1496 * user-space. This allows kernel threads to set their 1497 * priority to a value higher than any user task. Note: 1498 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1499 */ 1500 1501#define MAX_USER_RT_PRIO 100 1502#define MAX_RT_PRIO MAX_USER_RT_PRIO 1503 1504#define MAX_PRIO (MAX_RT_PRIO + 40) 1505#define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1506 1507static inline int rt_prio(int prio) 1508{ 1509 if (unlikely(prio < MAX_RT_PRIO)) 1510 return 1; 1511 return 0; 1512} 1513 1514static inline int rt_task(struct task_struct *p) 1515{ 1516 return rt_prio(p->prio); 1517} 1518 1519static inline struct pid *task_pid(struct task_struct *task) 1520{ 1521 return task->pids[PIDTYPE_PID].pid; 1522} 1523 1524static inline struct pid *task_tgid(struct task_struct *task) 1525{ 1526 return task->group_leader->pids[PIDTYPE_PID].pid; 1527} 1528 1529/* 1530 * Without tasklist or rcu lock it is not safe to dereference 1531 * the result of task_pgrp/task_session even if task == current, 1532 * we can race with another thread doing sys_setsid/sys_setpgid. 1533 */ 1534static inline struct pid *task_pgrp(struct task_struct *task) 1535{ 1536 return task->group_leader->pids[PIDTYPE_PGID].pid; 1537} 1538 1539static inline struct pid *task_session(struct task_struct *task) 1540{ 1541 return task->group_leader->pids[PIDTYPE_SID].pid; 1542} 1543 1544struct pid_namespace; 1545 1546/* 1547 * the helpers to get the task's different pids as they are seen 1548 * from various namespaces 1549 * 1550 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1551 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1552 * current. 1553 * task_xid_nr_ns() : id seen from the ns specified; 1554 * 1555 * set_task_vxid() : assigns a virtual id to a task; 1556 * 1557 * see also pid_nr() etc in include/linux/pid.h 1558 */ 1559pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, 1560 struct pid_namespace *ns); 1561 1562static inline pid_t task_pid_nr(struct task_struct *tsk) 1563{ 1564 return tsk->pid; 1565} 1566 1567static inline pid_t task_pid_nr_ns(struct task_struct *tsk, 1568 struct pid_namespace *ns) 1569{ 1570 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); 1571} 1572 1573static inline pid_t task_pid_vnr(struct task_struct *tsk) 1574{ 1575 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); 1576} 1577 1578 1579static inline pid_t task_tgid_nr(struct task_struct *tsk) 1580{ 1581 return tsk->tgid; 1582} 1583 1584pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1585 1586static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1587{ 1588 return pid_vnr(task_tgid(tsk)); 1589} 1590 1591 1592static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, 1593 struct pid_namespace *ns) 1594{ 1595 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); 1596} 1597 1598static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1599{ 1600 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); 1601} 1602 1603 1604static inline pid_t task_session_nr_ns(struct task_struct *tsk, 1605 struct pid_namespace *ns) 1606{ 1607 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); 1608} 1609 1610static inline pid_t task_session_vnr(struct task_struct *tsk) 1611{ 1612 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); 1613} 1614 1615/* obsolete, do not use */ 1616static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1617{ 1618 return task_pgrp_nr_ns(tsk, &init_pid_ns); 1619} 1620 1621/** 1622 * pid_alive - check that a task structure is not stale 1623 * @p: Task structure to be checked. 1624 * 1625 * Test if a process is not yet dead (at most zombie state) 1626 * If pid_alive fails, then pointers within the task structure 1627 * can be stale and must not be dereferenced. 1628 */ 1629static inline int pid_alive(struct task_struct *p) 1630{ 1631 return p->pids[PIDTYPE_PID].pid != NULL; 1632} 1633 1634/** 1635 * is_global_init - check if a task structure is init 1636 * @tsk: Task structure to be checked. 1637 * 1638 * Check if a task structure is the first user space task the kernel created. 1639 */ 1640static inline int is_global_init(struct task_struct *tsk) 1641{ 1642 return tsk->pid == 1; 1643} 1644 1645/* 1646 * is_container_init: 1647 * check whether in the task is init in its own pid namespace. 1648 */ 1649extern int is_container_init(struct task_struct *tsk); 1650 1651extern struct pid *cad_pid; 1652 1653extern void free_task(struct task_struct *tsk); 1654#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1655 1656extern void __put_task_struct(struct task_struct *t); 1657 1658static inline void put_task_struct(struct task_struct *t) 1659{ 1660 if (atomic_dec_and_test(&t->usage)) 1661 __put_task_struct(t); 1662} 1663 1664extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st); 1665extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st); 1666 1667/* 1668 * Per process flags 1669 */ 1670#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1671 /* Not implemented yet, only for 486*/ 1672#define PF_STARTING 0x00000002 /* being created */ 1673#define PF_EXITING 0x00000004 /* getting shut down */ 1674#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1675#define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1676#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ 1677#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1678#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */ 1679#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1680#define PF_DUMPCORE 0x00000200 /* dumped core */ 1681#define PF_SIGNALED 0x00000400 /* killed by a signal */ 1682#define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1683#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1684#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1685#define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */ 1686#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1687#define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1688#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1689#define PF_KSWAPD 0x00040000 /* I am kswapd */ 1690#define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */ 1691#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1692#define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1693#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1694#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1695#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1696#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1697#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1698#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ 1699#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1700#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1701#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1702#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1703 1704/* 1705 * Only the _current_ task can read/write to tsk->flags, but other 1706 * tasks can access tsk->flags in readonly mode for example 1707 * with tsk_used_math (like during threaded core dumping). 1708 * There is however an exception to this rule during ptrace 1709 * or during fork: the ptracer task is allowed to write to the 1710 * child->flags of its traced child (same goes for fork, the parent 1711 * can write to the child->flags), because we're guaranteed the 1712 * child is not running and in turn not changing child->flags 1713 * at the same time the parent does it. 1714 */ 1715#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1716#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1717#define clear_used_math() clear_stopped_child_used_math(current) 1718#define set_used_math() set_stopped_child_used_math(current) 1719#define conditional_stopped_child_used_math(condition, child) \ 1720 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1721#define conditional_used_math(condition) \ 1722 conditional_stopped_child_used_math(condition, current) 1723#define copy_to_stopped_child_used_math(child) \ 1724 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1725/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1726#define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1727#define used_math() tsk_used_math(current) 1728 1729#ifdef CONFIG_TREE_PREEMPT_RCU 1730 1731#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */ 1732#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */ 1733 1734static inline void rcu_copy_process(struct task_struct *p) 1735{ 1736 p->rcu_read_lock_nesting = 0; 1737 p->rcu_read_unlock_special = 0; 1738 p->rcu_blocked_node = NULL; 1739 INIT_LIST_HEAD(&p->rcu_node_entry); 1740} 1741 1742#else 1743 1744static inline void rcu_copy_process(struct task_struct *p) 1745{ 1746} 1747 1748#endif 1749 1750#ifdef CONFIG_SMP 1751extern int set_cpus_allowed_ptr(struct task_struct *p, 1752 const struct cpumask *new_mask); 1753#else 1754static inline int set_cpus_allowed_ptr(struct task_struct *p, 1755 const struct cpumask *new_mask) 1756{ 1757 if (!cpumask_test_cpu(0, new_mask)) 1758 return -EINVAL; 1759 return 0; 1760} 1761#endif 1762 1763#ifndef CONFIG_CPUMASK_OFFSTACK 1764static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1765{ 1766 return set_cpus_allowed_ptr(p, &new_mask); 1767} 1768#endif 1769 1770/* 1771 * Do not use outside of architecture code which knows its limitations. 1772 * 1773 * sched_clock() has no promise of monotonicity or bounded drift between 1774 * CPUs, use (which you should not) requires disabling IRQs. 1775 * 1776 * Please use one of the three interfaces below. 1777 */ 1778extern unsigned long long notrace sched_clock(void); 1779/* 1780 * See the comment in kernel/sched_clock.c 1781 */ 1782extern u64 cpu_clock(int cpu); 1783extern u64 local_clock(void); 1784extern u64 sched_clock_cpu(int cpu); 1785 1786 1787extern void sched_clock_init(void); 1788 1789#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1790static inline void sched_clock_tick(void) 1791{ 1792} 1793 1794static inline void sched_clock_idle_sleep_event(void) 1795{ 1796} 1797 1798static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1799{ 1800} 1801#else 1802/* 1803 * Architectures can set this to 1 if they have specified 1804 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, 1805 * but then during bootup it turns out that sched_clock() 1806 * is reliable after all: 1807 */ 1808extern int sched_clock_stable; 1809 1810extern void sched_clock_tick(void); 1811extern void sched_clock_idle_sleep_event(void); 1812extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1813#endif 1814 1815extern unsigned long long 1816task_sched_runtime(struct task_struct *task); 1817extern unsigned long long thread_group_sched_runtime(struct task_struct *task); 1818 1819/* sched_exec is called by processes performing an exec */ 1820#ifdef CONFIG_SMP 1821extern void sched_exec(void); 1822#else 1823#define sched_exec() {} 1824#endif 1825 1826extern void sched_clock_idle_sleep_event(void); 1827extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1828 1829#ifdef CONFIG_HOTPLUG_CPU 1830extern void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p); 1831extern void idle_task_exit(void); 1832#else 1833static inline void idle_task_exit(void) {} 1834#endif 1835 1836extern void sched_idle_next(void); 1837 1838#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1839extern void wake_up_idle_cpu(int cpu); 1840#else 1841static inline void wake_up_idle_cpu(int cpu) { } 1842#endif 1843 1844extern unsigned int sysctl_sched_latency; 1845extern unsigned int sysctl_sched_min_granularity; 1846extern unsigned int sysctl_sched_wakeup_granularity; 1847extern unsigned int sysctl_sched_shares_ratelimit; 1848extern unsigned int sysctl_sched_shares_thresh; 1849extern unsigned int sysctl_sched_child_runs_first; 1850 1851enum sched_tunable_scaling { 1852 SCHED_TUNABLESCALING_NONE, 1853 SCHED_TUNABLESCALING_LOG, 1854 SCHED_TUNABLESCALING_LINEAR, 1855 SCHED_TUNABLESCALING_END, 1856}; 1857extern enum sched_tunable_scaling sysctl_sched_tunable_scaling; 1858 1859#ifdef CONFIG_SCHED_DEBUG 1860extern unsigned int sysctl_sched_migration_cost; 1861extern unsigned int sysctl_sched_nr_migrate; 1862extern unsigned int sysctl_sched_time_avg; 1863extern unsigned int sysctl_timer_migration; 1864 1865int sched_proc_update_handler(struct ctl_table *table, int write, 1866 void __user *buffer, size_t *length, 1867 loff_t *ppos); 1868#endif 1869#ifdef CONFIG_SCHED_DEBUG 1870static inline unsigned int get_sysctl_timer_migration(void) 1871{ 1872 return sysctl_timer_migration; 1873} 1874#else 1875static inline unsigned int get_sysctl_timer_migration(void) 1876{ 1877 return 1; 1878} 1879#endif 1880extern unsigned int sysctl_sched_rt_period; 1881extern int sysctl_sched_rt_runtime; 1882 1883int sched_rt_handler(struct ctl_table *table, int write, 1884 void __user *buffer, size_t *lenp, 1885 loff_t *ppos); 1886 1887extern unsigned int sysctl_sched_compat_yield; 1888 1889#ifdef CONFIG_RT_MUTEXES 1890extern int rt_mutex_getprio(struct task_struct *p); 1891extern void rt_mutex_setprio(struct task_struct *p, int prio); 1892extern void rt_mutex_adjust_pi(struct task_struct *p); 1893#else 1894static inline int rt_mutex_getprio(struct task_struct *p) 1895{ 1896 return p->normal_prio; 1897} 1898# define rt_mutex_adjust_pi(p) do { } while (0) 1899#endif 1900 1901extern void set_user_nice(struct task_struct *p, long nice); 1902extern int task_prio(const struct task_struct *p); 1903extern int task_nice(const struct task_struct *p); 1904extern int can_nice(const struct task_struct *p, const int nice); 1905extern int task_curr(const struct task_struct *p); 1906extern int idle_cpu(int cpu); 1907extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1908extern int sched_setscheduler_nocheck(struct task_struct *, int, 1909 struct sched_param *); 1910extern struct task_struct *idle_task(int cpu); 1911extern struct task_struct *curr_task(int cpu); 1912extern void set_curr_task(int cpu, struct task_struct *p); 1913 1914void yield(void); 1915 1916/* 1917 * The default (Linux) execution domain. 1918 */ 1919extern struct exec_domain default_exec_domain; 1920 1921union thread_union { 1922 struct thread_info thread_info; 1923 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1924}; 1925 1926#ifndef __HAVE_ARCH_KSTACK_END 1927static inline int kstack_end(void *addr) 1928{ 1929 /* Reliable end of stack detection: 1930 * Some APM bios versions misalign the stack 1931 */ 1932 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1933} 1934#endif 1935 1936extern union thread_union init_thread_union; 1937extern struct task_struct init_task; 1938 1939extern struct mm_struct init_mm; 1940 1941extern struct pid_namespace init_pid_ns; 1942 1943/* 1944 * find a task by one of its numerical ids 1945 * 1946 * find_task_by_pid_ns(): 1947 * finds a task by its pid in the specified namespace 1948 * find_task_by_vpid(): 1949 * finds a task by its virtual pid 1950 * 1951 * see also find_vpid() etc in include/linux/pid.h 1952 */ 1953 1954extern struct task_struct *find_task_by_vpid(pid_t nr); 1955extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1956 struct pid_namespace *ns); 1957 1958extern void __set_special_pids(struct pid *pid); 1959 1960/* per-UID process charging. */ 1961extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1962static inline struct user_struct *get_uid(struct user_struct *u) 1963{ 1964 atomic_inc(&u->__count); 1965 return u; 1966} 1967extern void free_uid(struct user_struct *); 1968extern void release_uids(struct user_namespace *ns); 1969 1970#include <asm/current.h> 1971 1972extern void do_timer(unsigned long ticks); 1973 1974extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1975extern int wake_up_process(struct task_struct *tsk); 1976extern void wake_up_new_task(struct task_struct *tsk, 1977 unsigned long clone_flags); 1978#ifdef CONFIG_SMP 1979 extern void kick_process(struct task_struct *tsk); 1980#else 1981 static inline void kick_process(struct task_struct *tsk) { } 1982#endif 1983extern void sched_fork(struct task_struct *p, int clone_flags); 1984extern void sched_dead(struct task_struct *p); 1985 1986extern void proc_caches_init(void); 1987extern void flush_signals(struct task_struct *); 1988extern void __flush_signals(struct task_struct *); 1989extern void ignore_signals(struct task_struct *); 1990extern void flush_signal_handlers(struct task_struct *, int force_default); 1991extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1992 1993static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1994{ 1995 unsigned long flags; 1996 int ret; 1997 1998 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1999 ret = dequeue_signal(tsk, mask, info); 2000 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 2001 2002 return ret; 2003} 2004 2005extern void block_all_signals(int (*notifier)(void *priv), void *priv, 2006 sigset_t *mask); 2007extern void unblock_all_signals(void); 2008extern void release_task(struct task_struct * p); 2009extern int send_sig_info(int, struct siginfo *, struct task_struct *); 2010extern int force_sigsegv(int, struct task_struct *); 2011extern int force_sig_info(int, struct siginfo *, struct task_struct *); 2012extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 2013extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 2014extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 2015extern int kill_pgrp(struct pid *pid, int sig, int priv); 2016extern int kill_pid(struct pid *pid, int sig, int priv); 2017extern int kill_proc_info(int, struct siginfo *, pid_t); 2018extern int do_notify_parent(struct task_struct *, int); 2019extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent); 2020extern void force_sig(int, struct task_struct *); 2021extern int send_sig(int, struct task_struct *, int); 2022extern int zap_other_threads(struct task_struct *p); 2023extern struct sigqueue *sigqueue_alloc(void); 2024extern void sigqueue_free(struct sigqueue *); 2025extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 2026extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 2027extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 2028 2029static inline int kill_cad_pid(int sig, int priv) 2030{ 2031 return kill_pid(cad_pid, sig, priv); 2032} 2033 2034/* These can be the second arg to send_sig_info/send_group_sig_info. */ 2035#define SEND_SIG_NOINFO ((struct siginfo *) 0) 2036#define SEND_SIG_PRIV ((struct siginfo *) 1) 2037#define SEND_SIG_FORCED ((struct siginfo *) 2) 2038 2039/* 2040 * True if we are on the alternate signal stack. 2041 */ 2042static inline int on_sig_stack(unsigned long sp) 2043{ 2044#ifdef CONFIG_STACK_GROWSUP 2045 return sp >= current->sas_ss_sp && 2046 sp - current->sas_ss_sp < current->sas_ss_size; 2047#else 2048 return sp > current->sas_ss_sp && 2049 sp - current->sas_ss_sp <= current->sas_ss_size; 2050#endif 2051} 2052 2053static inline int sas_ss_flags(unsigned long sp) 2054{ 2055 return (current->sas_ss_size == 0 ? SS_DISABLE 2056 : on_sig_stack(sp) ? SS_ONSTACK : 0); 2057} 2058 2059/* 2060 * Routines for handling mm_structs 2061 */ 2062extern struct mm_struct * mm_alloc(void); 2063 2064/* mmdrop drops the mm and the page tables */ 2065extern void __mmdrop(struct mm_struct *); 2066static inline void mmdrop(struct mm_struct * mm) 2067{ 2068 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 2069 __mmdrop(mm); 2070} 2071 2072/* mmput gets rid of the mappings and all user-space */ 2073extern void mmput(struct mm_struct *); 2074/* Grab a reference to a task's mm, if it is not already going away */ 2075extern struct mm_struct *get_task_mm(struct task_struct *task); 2076/* Remove the current tasks stale references to the old mm_struct */ 2077extern void mm_release(struct task_struct *, struct mm_struct *); 2078/* Allocate a new mm structure and copy contents from tsk->mm */ 2079extern struct mm_struct *dup_mm(struct task_struct *tsk); 2080 2081extern int copy_thread(unsigned long, unsigned long, unsigned long, 2082 struct task_struct *, struct pt_regs *); 2083extern void flush_thread(void); 2084extern void exit_thread(void); 2085 2086extern void exit_files(struct task_struct *); 2087extern void __cleanup_sighand(struct sighand_struct *); 2088 2089extern void exit_itimers(struct signal_struct *); 2090extern void flush_itimer_signals(void); 2091 2092extern NORET_TYPE void do_group_exit(int); 2093 2094extern void daemonize(const char *, ...); 2095extern int allow_signal(int); 2096extern int disallow_signal(int); 2097 2098extern int do_execve(const char *, 2099 const char __user * const __user *, 2100 const char __user * const __user *, struct pt_regs *); 2101extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 2102struct task_struct *fork_idle(int); 2103 2104extern void set_task_comm(struct task_struct *tsk, char *from); 2105extern char *get_task_comm(char *to, struct task_struct *tsk); 2106 2107#ifdef CONFIG_SMP 2108extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 2109#else 2110static inline unsigned long wait_task_inactive(struct task_struct *p, 2111 long match_state) 2112{ 2113 return 1; 2114} 2115#endif 2116 2117#define next_task(p) \ 2118 list_entry_rcu((p)->tasks.next, struct task_struct, tasks) 2119 2120#define for_each_process(p) \ 2121 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 2122 2123extern bool current_is_single_threaded(void); 2124 2125/* 2126 * Careful: do_each_thread/while_each_thread is a double loop so 2127 * 'break' will not work as expected - use goto instead. 2128 */ 2129#define do_each_thread(g, t) \ 2130 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 2131 2132#define while_each_thread(g, t) \ 2133 while ((t = next_thread(t)) != g) 2134 2135static inline int get_nr_threads(struct task_struct *tsk) 2136{ 2137 return tsk->signal->nr_threads; 2138} 2139 2140/* de_thread depends on thread_group_leader not being a pid based check */ 2141#define thread_group_leader(p) (p == p->group_leader) 2142 2143/* Do to the insanities of de_thread it is possible for a process 2144 * to have the pid of the thread group leader without actually being 2145 * the thread group leader. For iteration through the pids in proc 2146 * all we care about is that we have a task with the appropriate 2147 * pid, we don't actually care if we have the right task. 2148 */ 2149static inline int has_group_leader_pid(struct task_struct *p) 2150{ 2151 return p->pid == p->tgid; 2152} 2153 2154static inline 2155int same_thread_group(struct task_struct *p1, struct task_struct *p2) 2156{ 2157 return p1->tgid == p2->tgid; 2158} 2159 2160static inline struct task_struct *next_thread(const struct task_struct *p) 2161{ 2162 return list_entry_rcu(p->thread_group.next, 2163 struct task_struct, thread_group); 2164} 2165 2166static inline int thread_group_empty(struct task_struct *p) 2167{ 2168 return list_empty(&p->thread_group); 2169} 2170 2171#define delay_group_leader(p) \ 2172 (thread_group_leader(p) && !thread_group_empty(p)) 2173 2174static inline int task_detached(struct task_struct *p) 2175{ 2176 return p->exit_signal == -1; 2177} 2178 2179/* 2180 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 2181 * subscriptions and synchronises with wait4(). Also used in procfs. Also 2182 * pins the final release of task.io_context. Also protects ->cpuset and 2183 * ->cgroup.subsys[]. 2184 * 2185 * Nests both inside and outside of read_lock(&tasklist_lock). 2186 * It must not be nested with write_lock_irq(&tasklist_lock), 2187 * neither inside nor outside. 2188 */ 2189static inline void task_lock(struct task_struct *p) 2190{ 2191 spin_lock(&p->alloc_lock); 2192} 2193 2194static inline void task_unlock(struct task_struct *p) 2195{ 2196 spin_unlock(&p->alloc_lock); 2197} 2198 2199extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 2200 unsigned long *flags); 2201 2202static inline void unlock_task_sighand(struct task_struct *tsk, 2203 unsigned long *flags) 2204{ 2205 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 2206} 2207 2208#ifndef __HAVE_THREAD_FUNCTIONS 2209 2210#define task_thread_info(task) ((struct thread_info *)(task)->stack) 2211#define task_stack_page(task) ((task)->stack) 2212 2213static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 2214{ 2215 *task_thread_info(p) = *task_thread_info(org); 2216 task_thread_info(p)->task = p; 2217} 2218 2219static inline unsigned long *end_of_stack(struct task_struct *p) 2220{ 2221 return (unsigned long *)(task_thread_info(p) + 1); 2222} 2223 2224#endif 2225 2226static inline int object_is_on_stack(void *obj) 2227{ 2228 void *stack = task_stack_page(current); 2229 2230 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 2231} 2232 2233extern void thread_info_cache_init(void); 2234 2235#ifdef CONFIG_DEBUG_STACK_USAGE 2236static inline unsigned long stack_not_used(struct task_struct *p) 2237{ 2238 unsigned long *n = end_of_stack(p); 2239 2240 do { /* Skip over canary */ 2241 n++; 2242 } while (!*n); 2243 2244 return (unsigned long)n - (unsigned long)end_of_stack(p); 2245} 2246#endif 2247 2248/* set thread flags in other task's structures 2249 * - see asm/thread_info.h for TIF_xxxx flags available 2250 */ 2251static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 2252{ 2253 set_ti_thread_flag(task_thread_info(tsk), flag); 2254} 2255 2256static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2257{ 2258 clear_ti_thread_flag(task_thread_info(tsk), flag); 2259} 2260 2261static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2262{ 2263 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2264} 2265 2266static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2267{ 2268 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2269} 2270 2271static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2272{ 2273 return test_ti_thread_flag(task_thread_info(tsk), flag); 2274} 2275 2276static inline void set_tsk_need_resched(struct task_struct *tsk) 2277{ 2278 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2279} 2280 2281static inline void clear_tsk_need_resched(struct task_struct *tsk) 2282{ 2283 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2284} 2285 2286static inline int test_tsk_need_resched(struct task_struct *tsk) 2287{ 2288 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2289} 2290 2291static inline int restart_syscall(void) 2292{ 2293 set_tsk_thread_flag(current, TIF_SIGPENDING); 2294 return -ERESTARTNOINTR; 2295} 2296 2297static inline int signal_pending(struct task_struct *p) 2298{ 2299 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2300} 2301 2302static inline int __fatal_signal_pending(struct task_struct *p) 2303{ 2304 return unlikely(sigismember(&p->pending.signal, SIGKILL)); 2305} 2306 2307static inline int fatal_signal_pending(struct task_struct *p) 2308{ 2309 return signal_pending(p) && __fatal_signal_pending(p); 2310} 2311 2312static inline int signal_pending_state(long state, struct task_struct *p) 2313{ 2314 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2315 return 0; 2316 if (!signal_pending(p)) 2317 return 0; 2318 2319 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2320} 2321 2322static inline int need_resched(void) 2323{ 2324 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2325} 2326 2327/* 2328 * cond_resched() and cond_resched_lock(): latency reduction via 2329 * explicit rescheduling in places that are safe. The return 2330 * value indicates whether a reschedule was done in fact. 2331 * cond_resched_lock() will drop the spinlock before scheduling, 2332 * cond_resched_softirq() will enable bhs before scheduling. 2333 */ 2334extern int _cond_resched(void); 2335 2336#define cond_resched() ({ \ 2337 __might_sleep(__FILE__, __LINE__, 0); \ 2338 _cond_resched(); \ 2339}) 2340 2341extern int __cond_resched_lock(spinlock_t *lock); 2342 2343#ifdef CONFIG_PREEMPT 2344#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET 2345#else 2346#define PREEMPT_LOCK_OFFSET 0 2347#endif 2348 2349#define cond_resched_lock(lock) ({ \ 2350 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \ 2351 __cond_resched_lock(lock); \ 2352}) 2353 2354extern int __cond_resched_softirq(void); 2355 2356#define cond_resched_softirq() ({ \ 2357 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \ 2358 __cond_resched_softirq(); \ 2359}) 2360 2361/* 2362 * Does a critical section need to be broken due to another 2363 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2364 * but a general need for low latency) 2365 */ 2366static inline int spin_needbreak(spinlock_t *lock) 2367{ 2368#ifdef CONFIG_PREEMPT 2369 return spin_is_contended(lock); 2370#else 2371 return 0; 2372#endif 2373} 2374 2375/* 2376 * Thread group CPU time accounting. 2377 */ 2378void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); 2379void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); 2380 2381static inline void thread_group_cputime_init(struct signal_struct *sig) 2382{ 2383 spin_lock_init(&sig->cputimer.lock); 2384} 2385 2386/* 2387 * Reevaluate whether the task has signals pending delivery. 2388 * Wake the task if so. 2389 * This is required every time the blocked sigset_t changes. 2390 * callers must hold sighand->siglock. 2391 */ 2392extern void recalc_sigpending_and_wake(struct task_struct *t); 2393extern void recalc_sigpending(void); 2394 2395extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2396 2397/* 2398 * Wrappers for p->thread_info->cpu access. No-op on UP. 2399 */ 2400#ifdef CONFIG_SMP 2401 2402static inline unsigned int task_cpu(const struct task_struct *p) 2403{ 2404 return task_thread_info(p)->cpu; 2405} 2406 2407extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2408 2409#else 2410 2411static inline unsigned int task_cpu(const struct task_struct *p) 2412{ 2413 return 0; 2414} 2415 2416static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2417{ 2418} 2419 2420#endif /* CONFIG_SMP */ 2421 2422extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); 2423extern long sched_getaffinity(pid_t pid, struct cpumask *mask); 2424 2425extern void normalize_rt_tasks(void); 2426 2427#ifdef CONFIG_CGROUP_SCHED 2428 2429extern struct task_group init_task_group; 2430 2431extern struct task_group *sched_create_group(struct task_group *parent); 2432extern void sched_destroy_group(struct task_group *tg); 2433extern void sched_move_task(struct task_struct *tsk); 2434#ifdef CONFIG_FAIR_GROUP_SCHED 2435extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2436extern unsigned long sched_group_shares(struct task_group *tg); 2437#endif 2438#ifdef CONFIG_RT_GROUP_SCHED 2439extern int sched_group_set_rt_runtime(struct task_group *tg, 2440 long rt_runtime_us); 2441extern long sched_group_rt_runtime(struct task_group *tg); 2442extern int sched_group_set_rt_period(struct task_group *tg, 2443 long rt_period_us); 2444extern long sched_group_rt_period(struct task_group *tg); 2445extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk); 2446#endif 2447#endif 2448 2449extern int task_can_switch_user(struct user_struct *up, 2450 struct task_struct *tsk); 2451 2452#ifdef CONFIG_TASK_XACCT 2453static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2454{ 2455 tsk->ioac.rchar += amt; 2456} 2457 2458static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2459{ 2460 tsk->ioac.wchar += amt; 2461} 2462 2463static inline void inc_syscr(struct task_struct *tsk) 2464{ 2465 tsk->ioac.syscr++; 2466} 2467 2468static inline void inc_syscw(struct task_struct *tsk) 2469{ 2470 tsk->ioac.syscw++; 2471} 2472#else 2473static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2474{ 2475} 2476 2477static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2478{ 2479} 2480 2481static inline void inc_syscr(struct task_struct *tsk) 2482{ 2483} 2484 2485static inline void inc_syscw(struct task_struct *tsk) 2486{ 2487} 2488#endif 2489 2490#ifndef TASK_SIZE_OF 2491#define TASK_SIZE_OF(tsk) TASK_SIZE 2492#endif 2493 2494/* 2495 * Call the function if the target task is executing on a CPU right now: 2496 */ 2497extern void task_oncpu_function_call(struct task_struct *p, 2498 void (*func) (void *info), void *info); 2499 2500 2501#ifdef CONFIG_MM_OWNER 2502extern void mm_update_next_owner(struct mm_struct *mm); 2503extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2504#else 2505static inline void mm_update_next_owner(struct mm_struct *mm) 2506{ 2507} 2508 2509static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2510{ 2511} 2512#endif /* CONFIG_MM_OWNER */ 2513 2514static inline unsigned long task_rlimit(const struct task_struct *tsk, 2515 unsigned int limit) 2516{ 2517 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur); 2518} 2519 2520static inline unsigned long task_rlimit_max(const struct task_struct *tsk, 2521 unsigned int limit) 2522{ 2523 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max); 2524} 2525 2526static inline unsigned long rlimit(unsigned int limit) 2527{ 2528 return task_rlimit(current, limit); 2529} 2530 2531static inline unsigned long rlimit_max(unsigned int limit) 2532{ 2533 return task_rlimit_max(current, limit); 2534} 2535 2536#endif /* __KERNEL__ */ 2537 2538#endif 2539