1 #ifndef _LINUX_SCHED_H 2 #define _LINUX_SCHED_H 3 4 /* 5 * cloning flags: 6 */ 7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */ 9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 15 #define CLONE_THREAD 0x00010000 /* Same thread group? */ 16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */ 17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */ 25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */ 26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */ 27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */ 28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */ 29 #define CLONE_NEWNET 0x40000000 /* New network namespace */ 30 #define CLONE_IO 0x80000000 /* Clone io context */ 31 32 /* 33 * Scheduling policies 34 */ 35 #define SCHED_NORMAL 0 36 #define SCHED_FIFO 1 37 #define SCHED_RR 2 38 #define SCHED_BATCH 3 39 /* SCHED_ISO: reserved but not implemented yet */ 40 #define SCHED_IDLE 5 41 42 #ifdef __KERNEL__ 43 44 struct sched_param { 45 int sched_priority; 46 }; 47 48 #include <asm/param.h> /* for HZ */ 49 50 #include <linux/capability.h> 51 #include <linux/threads.h> 52 #include <linux/kernel.h> 53 #include <linux/types.h> 54 #include <linux/timex.h> 55 #include <linux/jiffies.h> 56 #include <linux/rbtree.h> 57 #include <linux/thread_info.h> 58 #include <linux/cpumask.h> 59 #include <linux/errno.h> 60 #include <linux/nodemask.h> 61 #include <linux/mm_types.h> 62 63 #include <asm/system.h> 64 #include <asm/page.h> 65 #include <asm/ptrace.h> 66 #include <asm/cputime.h> 67 68 #include <linux/smp.h> 69 #include <linux/sem.h> 70 #include <linux/signal.h> 71 #include <linux/fs_struct.h> 72 #include <linux/compiler.h> 73 #include <linux/completion.h> 74 #include <linux/pid.h> 75 #include <linux/percpu.h> 76 #include <linux/topology.h> 77 #include <linux/proportions.h> 78 #include <linux/seccomp.h> 79 #include <linux/rcupdate.h> 80 #include <linux/rtmutex.h> 81 82 #include <linux/time.h> 83 #include <linux/param.h> 84 #include <linux/resource.h> 85 #include <linux/timer.h> 86 #include <linux/hrtimer.h> 87 #include <linux/task_io_accounting.h> 88 #include <linux/kobject.h> 89 #include <linux/latencytop.h> 90 91 #include <asm/processor.h> 92 93 struct mem_cgroup; 94 struct exec_domain; 95 struct futex_pi_state; 96 struct robust_list_head; 97 struct bio; 98 99 /* 100 * List of flags we want to share for kernel threads, 101 * if only because they are not used by them anyway. 102 */ 103 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 104 105 /* 106 * These are the constant used to fake the fixed-point load-average 107 * counting. Some notes: 108 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 109 * a load-average precision of 10 bits integer + 11 bits fractional 110 * - if you want to count load-averages more often, you need more 111 * precision, or rounding will get you. With 2-second counting freq, 112 * the EXP_n values would be 1981, 2034 and 2043 if still using only 113 * 11 bit fractions. 114 */ 115 extern unsigned long avenrun[]; /* Load averages */ 116 117 #define FSHIFT 11 /* nr of bits of precision */ 118 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 119 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 120 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 121 #define EXP_5 2014 /* 1/exp(5sec/5min) */ 122 #define EXP_15 2037 /* 1/exp(5sec/15min) */ 123 124 #define CALC_LOAD(load,exp,n) \ 125 load *= exp; \ 126 load += n*(FIXED_1-exp); \ 127 load >>= FSHIFT; 128 129 extern unsigned long total_forks; 130 extern int nr_threads; 131 DECLARE_PER_CPU(unsigned long, process_counts); 132 extern int nr_processes(void); 133 extern unsigned long nr_running(void); 134 extern unsigned long nr_uninterruptible(void); 135 extern unsigned long nr_active(void); 136 extern unsigned long nr_iowait(void); 137 138 struct seq_file; 139 struct cfs_rq; 140 struct task_group; 141 #ifdef CONFIG_SCHED_DEBUG 142 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 143 extern void proc_sched_set_task(struct task_struct *p); 144 extern void 145 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 146 #else 147 static inline void 148 proc_sched_show_task(struct task_struct *p, struct seq_file *m) 149 { 150 } 151 static inline void proc_sched_set_task(struct task_struct *p) 152 { 153 } 154 static inline void 155 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 156 { 157 } 158 #endif 159 160 extern unsigned long long time_sync_thresh; 161 162 /* 163 * Task state bitmask. NOTE! These bits are also 164 * encoded in fs/proc/array.c: get_task_state(). 165 * 166 * We have two separate sets of flags: task->state 167 * is about runnability, while task->exit_state are 168 * about the task exiting. Confusing, but this way 169 * modifying one set can't modify the other one by 170 * mistake. 171 */ 172 #define TASK_RUNNING 0 173 #define TASK_INTERRUPTIBLE 1 174 #define TASK_UNINTERRUPTIBLE 2 175 #define __TASK_STOPPED 4 176 #define __TASK_TRACED 8 177 /* in tsk->exit_state */ 178 #define EXIT_ZOMBIE 16 179 #define EXIT_DEAD 32 180 /* in tsk->state again */ 181 #define TASK_DEAD 64 182 #define TASK_WAKEKILL 128 183 184 /* Convenience macros for the sake of set_task_state */ 185 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 186 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 187 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 188 189 /* Convenience macros for the sake of wake_up */ 190 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 191 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 192 193 /* get_task_state() */ 194 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 195 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 196 __TASK_TRACED) 197 198 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 199 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 200 #define task_is_stopped_or_traced(task) \ 201 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 202 #define task_contributes_to_load(task) \ 203 ((task->state & TASK_UNINTERRUPTIBLE) != 0) 204 205 #define __set_task_state(tsk, state_value) \ 206 do { (tsk)->state = (state_value); } while (0) 207 #define set_task_state(tsk, state_value) \ 208 set_mb((tsk)->state, (state_value)) 209 210 /* 211 * set_current_state() includes a barrier so that the write of current->state 212 * is correctly serialised wrt the caller's subsequent test of whether to 213 * actually sleep: 214 * 215 * set_current_state(TASK_UNINTERRUPTIBLE); 216 * if (do_i_need_to_sleep()) 217 * schedule(); 218 * 219 * If the caller does not need such serialisation then use __set_current_state() 220 */ 221 #define __set_current_state(state_value) \ 222 do { current->state = (state_value); } while (0) 223 #define set_current_state(state_value) \ 224 set_mb(current->state, (state_value)) 225 226 /* Task command name length */ 227 #define TASK_COMM_LEN 16 228 229 #include <linux/spinlock.h> 230 231 /* 232 * This serializes "schedule()" and also protects 233 * the run-queue from deletions/modifications (but 234 * _adding_ to the beginning of the run-queue has 235 * a separate lock). 236 */ 237 extern rwlock_t tasklist_lock; 238 extern spinlock_t mmlist_lock; 239 240 struct task_struct; 241 242 extern void sched_init(void); 243 extern void sched_init_smp(void); 244 extern asmlinkage void schedule_tail(struct task_struct *prev); 245 extern void init_idle(struct task_struct *idle, int cpu); 246 extern void init_idle_bootup_task(struct task_struct *idle); 247 248 extern int runqueue_is_locked(void); 249 250 extern cpumask_t nohz_cpu_mask; 251 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 252 extern int select_nohz_load_balancer(int cpu); 253 #else 254 static inline int select_nohz_load_balancer(int cpu) 255 { 256 return 0; 257 } 258 #endif 259 260 extern unsigned long rt_needs_cpu(int cpu); 261 262 /* 263 * Only dump TASK_* tasks. (0 for all tasks) 264 */ 265 extern void show_state_filter(unsigned long state_filter); 266 267 static inline void show_state(void) 268 { 269 show_state_filter(0); 270 } 271 272 extern void show_regs(struct pt_regs *); 273 274 /* 275 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 276 * task), SP is the stack pointer of the first frame that should be shown in the back 277 * trace (or NULL if the entire call-chain of the task should be shown). 278 */ 279 extern void show_stack(struct task_struct *task, unsigned long *sp); 280 281 void io_schedule(void); 282 long io_schedule_timeout(long timeout); 283 284 extern void cpu_init (void); 285 extern void trap_init(void); 286 extern void account_process_tick(struct task_struct *task, int user); 287 extern void update_process_times(int user); 288 extern void scheduler_tick(void); 289 extern void hrtick_resched(void); 290 291 extern void sched_show_task(struct task_struct *p); 292 293 #ifdef CONFIG_DETECT_SOFTLOCKUP 294 extern void softlockup_tick(void); 295 extern void touch_softlockup_watchdog(void); 296 extern void touch_all_softlockup_watchdogs(void); 297 extern unsigned int softlockup_panic; 298 extern unsigned long sysctl_hung_task_check_count; 299 extern unsigned long sysctl_hung_task_timeout_secs; 300 extern unsigned long sysctl_hung_task_warnings; 301 extern int softlockup_thresh; 302 #else 303 static inline void softlockup_tick(void) 304 { 305 } 306 static inline void spawn_softlockup_task(void) 307 { 308 } 309 static inline void touch_softlockup_watchdog(void) 310 { 311 } 312 static inline void touch_all_softlockup_watchdogs(void) 313 { 314 } 315 #endif 316 317 318 /* Attach to any functions which should be ignored in wchan output. */ 319 #define __sched __attribute__((__section__(".sched.text"))) 320 321 /* Linker adds these: start and end of __sched functions */ 322 extern char __sched_text_start[], __sched_text_end[]; 323 324 /* Is this address in the __sched functions? */ 325 extern int in_sched_functions(unsigned long addr); 326 327 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 328 extern signed long schedule_timeout(signed long timeout); 329 extern signed long schedule_timeout_interruptible(signed long timeout); 330 extern signed long schedule_timeout_killable(signed long timeout); 331 extern signed long schedule_timeout_uninterruptible(signed long timeout); 332 asmlinkage void schedule(void); 333 334 struct nsproxy; 335 struct user_namespace; 336 337 /* Maximum number of active map areas.. This is a random (large) number */ 338 #define DEFAULT_MAX_MAP_COUNT 65536 339 340 extern int sysctl_max_map_count; 341 342 #include <linux/aio.h> 343 344 extern unsigned long 345 arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 346 unsigned long, unsigned long); 347 extern unsigned long 348 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 349 unsigned long len, unsigned long pgoff, 350 unsigned long flags); 351 extern void arch_unmap_area(struct mm_struct *, unsigned long); 352 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 353 354 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS 355 /* 356 * The mm counters are not protected by its page_table_lock, 357 * so must be incremented atomically. 358 */ 359 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 360 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 361 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 362 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 363 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 364 365 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 366 /* 367 * The mm counters are protected by its page_table_lock, 368 * so can be incremented directly. 369 */ 370 #define set_mm_counter(mm, member, value) (mm)->_##member = (value) 371 #define get_mm_counter(mm, member) ((mm)->_##member) 372 #define add_mm_counter(mm, member, value) (mm)->_##member += (value) 373 #define inc_mm_counter(mm, member) (mm)->_##member++ 374 #define dec_mm_counter(mm, member) (mm)->_##member-- 375 376 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 377 378 #define get_mm_rss(mm) \ 379 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 380 #define update_hiwater_rss(mm) do { \ 381 unsigned long _rss = get_mm_rss(mm); \ 382 if ((mm)->hiwater_rss < _rss) \ 383 (mm)->hiwater_rss = _rss; \ 384 } while (0) 385 #define update_hiwater_vm(mm) do { \ 386 if ((mm)->hiwater_vm < (mm)->total_vm) \ 387 (mm)->hiwater_vm = (mm)->total_vm; \ 388 } while (0) 389 390 extern void set_dumpable(struct mm_struct *mm, int value); 391 extern int get_dumpable(struct mm_struct *mm); 392 393 /* mm flags */ 394 /* dumpable bits */ 395 #define MMF_DUMPABLE 0 /* core dump is permitted */ 396 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 397 #define MMF_DUMPABLE_BITS 2 398 399 /* coredump filter bits */ 400 #define MMF_DUMP_ANON_PRIVATE 2 401 #define MMF_DUMP_ANON_SHARED 3 402 #define MMF_DUMP_MAPPED_PRIVATE 4 403 #define MMF_DUMP_MAPPED_SHARED 5 404 #define MMF_DUMP_ELF_HEADERS 6 405 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 406 #define MMF_DUMP_FILTER_BITS 5 407 #define MMF_DUMP_FILTER_MASK \ 408 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 409 #define MMF_DUMP_FILTER_DEFAULT \ 410 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED)) 411 412 struct sighand_struct { 413 atomic_t count; 414 struct k_sigaction action[_NSIG]; 415 spinlock_t siglock; 416 wait_queue_head_t signalfd_wqh; 417 }; 418 419 struct pacct_struct { 420 int ac_flag; 421 long ac_exitcode; 422 unsigned long ac_mem; 423 cputime_t ac_utime, ac_stime; 424 unsigned long ac_minflt, ac_majflt; 425 }; 426 427 /* 428 * NOTE! "signal_struct" does not have it's own 429 * locking, because a shared signal_struct always 430 * implies a shared sighand_struct, so locking 431 * sighand_struct is always a proper superset of 432 * the locking of signal_struct. 433 */ 434 struct signal_struct { 435 atomic_t count; 436 atomic_t live; 437 438 wait_queue_head_t wait_chldexit; /* for wait4() */ 439 440 /* current thread group signal load-balancing target: */ 441 struct task_struct *curr_target; 442 443 /* shared signal handling: */ 444 struct sigpending shared_pending; 445 446 /* thread group exit support */ 447 int group_exit_code; 448 /* overloaded: 449 * - notify group_exit_task when ->count is equal to notify_count 450 * - everyone except group_exit_task is stopped during signal delivery 451 * of fatal signals, group_exit_task processes the signal. 452 */ 453 struct task_struct *group_exit_task; 454 int notify_count; 455 456 /* thread group stop support, overloads group_exit_code too */ 457 int group_stop_count; 458 unsigned int flags; /* see SIGNAL_* flags below */ 459 460 /* POSIX.1b Interval Timers */ 461 struct list_head posix_timers; 462 463 /* ITIMER_REAL timer for the process */ 464 struct hrtimer real_timer; 465 struct pid *leader_pid; 466 ktime_t it_real_incr; 467 468 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 469 cputime_t it_prof_expires, it_virt_expires; 470 cputime_t it_prof_incr, it_virt_incr; 471 472 /* job control IDs */ 473 474 /* 475 * pgrp and session fields are deprecated. 476 * use the task_session_Xnr and task_pgrp_Xnr routines below 477 */ 478 479 union { 480 pid_t pgrp __deprecated; 481 pid_t __pgrp; 482 }; 483 484 struct pid *tty_old_pgrp; 485 486 union { 487 pid_t session __deprecated; 488 pid_t __session; 489 }; 490 491 /* boolean value for session group leader */ 492 int leader; 493 494 struct tty_struct *tty; /* NULL if no tty */ 495 496 /* 497 * Cumulative resource counters for dead threads in the group, 498 * and for reaped dead child processes forked by this group. 499 * Live threads maintain their own counters and add to these 500 * in __exit_signal, except for the group leader. 501 */ 502 cputime_t utime, stime, cutime, cstime; 503 cputime_t gtime; 504 cputime_t cgtime; 505 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 506 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 507 unsigned long inblock, oublock, cinblock, coublock; 508 struct task_io_accounting ioac; 509 510 /* 511 * Cumulative ns of scheduled CPU time for dead threads in the 512 * group, not including a zombie group leader. (This only differs 513 * from jiffies_to_ns(utime + stime) if sched_clock uses something 514 * other than jiffies.) 515 */ 516 unsigned long long sum_sched_runtime; 517 518 /* 519 * We don't bother to synchronize most readers of this at all, 520 * because there is no reader checking a limit that actually needs 521 * to get both rlim_cur and rlim_max atomically, and either one 522 * alone is a single word that can safely be read normally. 523 * getrlimit/setrlimit use task_lock(current->group_leader) to 524 * protect this instead of the siglock, because they really 525 * have no need to disable irqs. 526 */ 527 struct rlimit rlim[RLIM_NLIMITS]; 528 529 struct list_head cpu_timers[3]; 530 531 /* keep the process-shared keyrings here so that they do the right 532 * thing in threads created with CLONE_THREAD */ 533 #ifdef CONFIG_KEYS 534 struct key *session_keyring; /* keyring inherited over fork */ 535 struct key *process_keyring; /* keyring private to this process */ 536 #endif 537 #ifdef CONFIG_BSD_PROCESS_ACCT 538 struct pacct_struct pacct; /* per-process accounting information */ 539 #endif 540 #ifdef CONFIG_TASKSTATS 541 struct taskstats *stats; 542 #endif 543 #ifdef CONFIG_AUDIT 544 unsigned audit_tty; 545 struct tty_audit_buf *tty_audit_buf; 546 #endif 547 }; 548 549 /* Context switch must be unlocked if interrupts are to be enabled */ 550 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 551 # define __ARCH_WANT_UNLOCKED_CTXSW 552 #endif 553 554 /* 555 * Bits in flags field of signal_struct. 556 */ 557 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 558 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 559 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 560 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 561 /* 562 * Pending notifications to parent. 563 */ 564 #define SIGNAL_CLD_STOPPED 0x00000010 565 #define SIGNAL_CLD_CONTINUED 0x00000020 566 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 567 568 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 569 570 /* If true, all threads except ->group_exit_task have pending SIGKILL */ 571 static inline int signal_group_exit(const struct signal_struct *sig) 572 { 573 return (sig->flags & SIGNAL_GROUP_EXIT) || 574 (sig->group_exit_task != NULL); 575 } 576 577 /* 578 * Some day this will be a full-fledged user tracking system.. 579 */ 580 struct user_struct { 581 atomic_t __count; /* reference count */ 582 atomic_t processes; /* How many processes does this user have? */ 583 atomic_t files; /* How many open files does this user have? */ 584 atomic_t sigpending; /* How many pending signals does this user have? */ 585 #ifdef CONFIG_INOTIFY_USER 586 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 587 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 588 #endif 589 #ifdef CONFIG_POSIX_MQUEUE 590 /* protected by mq_lock */ 591 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 592 #endif 593 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 594 595 #ifdef CONFIG_KEYS 596 struct key *uid_keyring; /* UID specific keyring */ 597 struct key *session_keyring; /* UID's default session keyring */ 598 #endif 599 600 /* Hash table maintenance information */ 601 struct hlist_node uidhash_node; 602 uid_t uid; 603 604 #ifdef CONFIG_USER_SCHED 605 struct task_group *tg; 606 #ifdef CONFIG_SYSFS 607 struct kobject kobj; 608 struct work_struct work; 609 #endif 610 #endif 611 }; 612 613 extern int uids_sysfs_init(void); 614 615 extern struct user_struct *find_user(uid_t); 616 617 extern struct user_struct root_user; 618 #define INIT_USER (&root_user) 619 620 struct backing_dev_info; 621 struct reclaim_state; 622 623 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 624 struct sched_info { 625 /* cumulative counters */ 626 unsigned long pcount; /* # of times run on this cpu */ 627 unsigned long long cpu_time, /* time spent on the cpu */ 628 run_delay; /* time spent waiting on a runqueue */ 629 630 /* timestamps */ 631 unsigned long long last_arrival,/* when we last ran on a cpu */ 632 last_queued; /* when we were last queued to run */ 633 #ifdef CONFIG_SCHEDSTATS 634 /* BKL stats */ 635 unsigned int bkl_count; 636 #endif 637 }; 638 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 639 640 #ifdef CONFIG_SCHEDSTATS 641 extern const struct file_operations proc_schedstat_operations; 642 #endif /* CONFIG_SCHEDSTATS */ 643 644 #ifdef CONFIG_TASK_DELAY_ACCT 645 struct task_delay_info { 646 spinlock_t lock; 647 unsigned int flags; /* Private per-task flags */ 648 649 /* For each stat XXX, add following, aligned appropriately 650 * 651 * struct timespec XXX_start, XXX_end; 652 * u64 XXX_delay; 653 * u32 XXX_count; 654 * 655 * Atomicity of updates to XXX_delay, XXX_count protected by 656 * single lock above (split into XXX_lock if contention is an issue). 657 */ 658 659 /* 660 * XXX_count is incremented on every XXX operation, the delay 661 * associated with the operation is added to XXX_delay. 662 * XXX_delay contains the accumulated delay time in nanoseconds. 663 */ 664 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 665 u64 blkio_delay; /* wait for sync block io completion */ 666 u64 swapin_delay; /* wait for swapin block io completion */ 667 u32 blkio_count; /* total count of the number of sync block */ 668 /* io operations performed */ 669 u32 swapin_count; /* total count of the number of swapin block */ 670 /* io operations performed */ 671 672 struct timespec freepages_start, freepages_end; 673 u64 freepages_delay; /* wait for memory reclaim */ 674 u32 freepages_count; /* total count of memory reclaim */ 675 }; 676 #endif /* CONFIG_TASK_DELAY_ACCT */ 677 678 static inline int sched_info_on(void) 679 { 680 #ifdef CONFIG_SCHEDSTATS 681 return 1; 682 #elif defined(CONFIG_TASK_DELAY_ACCT) 683 extern int delayacct_on; 684 return delayacct_on; 685 #else 686 return 0; 687 #endif 688 } 689 690 enum cpu_idle_type { 691 CPU_IDLE, 692 CPU_NOT_IDLE, 693 CPU_NEWLY_IDLE, 694 CPU_MAX_IDLE_TYPES 695 }; 696 697 /* 698 * sched-domains (multiprocessor balancing) declarations: 699 */ 700 701 /* 702 * Increase resolution of nice-level calculations: 703 */ 704 #define SCHED_LOAD_SHIFT 10 705 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 706 707 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 708 709 #ifdef CONFIG_SMP 710 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 711 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 712 #define SD_BALANCE_EXEC 4 /* Balance on exec */ 713 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 714 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 715 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 716 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 717 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 718 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 719 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 720 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 721 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 722 723 #define BALANCE_FOR_MC_POWER \ 724 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0) 725 726 #define BALANCE_FOR_PKG_POWER \ 727 ((sched_mc_power_savings || sched_smt_power_savings) ? \ 728 SD_POWERSAVINGS_BALANCE : 0) 729 730 #define test_sd_parent(sd, flag) ((sd->parent && \ 731 (sd->parent->flags & flag)) ? 1 : 0) 732 733 734 struct sched_group { 735 struct sched_group *next; /* Must be a circular list */ 736 cpumask_t cpumask; 737 738 /* 739 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 740 * single CPU. This is read only (except for setup, hotplug CPU). 741 * Note : Never change cpu_power without recompute its reciprocal 742 */ 743 unsigned int __cpu_power; 744 /* 745 * reciprocal value of cpu_power to avoid expensive divides 746 * (see include/linux/reciprocal_div.h) 747 */ 748 u32 reciprocal_cpu_power; 749 }; 750 751 enum sched_domain_level { 752 SD_LV_NONE = 0, 753 SD_LV_SIBLING, 754 SD_LV_MC, 755 SD_LV_CPU, 756 SD_LV_NODE, 757 SD_LV_ALLNODES, 758 SD_LV_MAX 759 }; 760 761 struct sched_domain_attr { 762 int relax_domain_level; 763 }; 764 765 #define SD_ATTR_INIT (struct sched_domain_attr) { \ 766 .relax_domain_level = -1, \ 767 } 768 769 struct sched_domain { 770 /* These fields must be setup */ 771 struct sched_domain *parent; /* top domain must be null terminated */ 772 struct sched_domain *child; /* bottom domain must be null terminated */ 773 struct sched_group *groups; /* the balancing groups of the domain */ 774 cpumask_t span; /* span of all CPUs in this domain */ 775 unsigned long min_interval; /* Minimum balance interval ms */ 776 unsigned long max_interval; /* Maximum balance interval ms */ 777 unsigned int busy_factor; /* less balancing by factor if busy */ 778 unsigned int imbalance_pct; /* No balance until over watermark */ 779 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 780 unsigned int busy_idx; 781 unsigned int idle_idx; 782 unsigned int newidle_idx; 783 unsigned int wake_idx; 784 unsigned int forkexec_idx; 785 int flags; /* See SD_* */ 786 enum sched_domain_level level; 787 788 /* Runtime fields. */ 789 unsigned long last_balance; /* init to jiffies. units in jiffies */ 790 unsigned int balance_interval; /* initialise to 1. units in ms. */ 791 unsigned int nr_balance_failed; /* initialise to 0 */ 792 793 u64 last_update; 794 795 #ifdef CONFIG_SCHEDSTATS 796 /* load_balance() stats */ 797 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 798 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 799 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 800 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 801 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 802 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 803 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 804 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 805 806 /* Active load balancing */ 807 unsigned int alb_count; 808 unsigned int alb_failed; 809 unsigned int alb_pushed; 810 811 /* SD_BALANCE_EXEC stats */ 812 unsigned int sbe_count; 813 unsigned int sbe_balanced; 814 unsigned int sbe_pushed; 815 816 /* SD_BALANCE_FORK stats */ 817 unsigned int sbf_count; 818 unsigned int sbf_balanced; 819 unsigned int sbf_pushed; 820 821 /* try_to_wake_up() stats */ 822 unsigned int ttwu_wake_remote; 823 unsigned int ttwu_move_affine; 824 unsigned int ttwu_move_balance; 825 #endif 826 }; 827 828 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 829 struct sched_domain_attr *dattr_new); 830 extern int arch_reinit_sched_domains(void); 831 832 #else /* CONFIG_SMP */ 833 834 struct sched_domain_attr; 835 836 static inline void 837 partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 838 struct sched_domain_attr *dattr_new) 839 { 840 } 841 #endif /* !CONFIG_SMP */ 842 843 struct io_context; /* See blkdev.h */ 844 #define NGROUPS_SMALL 32 845 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t))) 846 struct group_info { 847 int ngroups; 848 atomic_t usage; 849 gid_t small_block[NGROUPS_SMALL]; 850 int nblocks; 851 gid_t *blocks[0]; 852 }; 853 854 /* 855 * get_group_info() must be called with the owning task locked (via task_lock()) 856 * when task != current. The reason being that the vast majority of callers are 857 * looking at current->group_info, which can not be changed except by the 858 * current task. Changing current->group_info requires the task lock, too. 859 */ 860 #define get_group_info(group_info) do { \ 861 atomic_inc(&(group_info)->usage); \ 862 } while (0) 863 864 #define put_group_info(group_info) do { \ 865 if (atomic_dec_and_test(&(group_info)->usage)) \ 866 groups_free(group_info); \ 867 } while (0) 868 869 extern struct group_info *groups_alloc(int gidsetsize); 870 extern void groups_free(struct group_info *group_info); 871 extern int set_current_groups(struct group_info *group_info); 872 extern int groups_search(struct group_info *group_info, gid_t grp); 873 /* access the groups "array" with this macro */ 874 #define GROUP_AT(gi, i) \ 875 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 876 877 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 878 extern void prefetch_stack(struct task_struct *t); 879 #else 880 static inline void prefetch_stack(struct task_struct *t) { } 881 #endif 882 883 struct audit_context; /* See audit.c */ 884 struct mempolicy; 885 struct pipe_inode_info; 886 struct uts_namespace; 887 888 struct rq; 889 struct sched_domain; 890 891 struct sched_class { 892 const struct sched_class *next; 893 894 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 895 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 896 void (*yield_task) (struct rq *rq); 897 int (*select_task_rq)(struct task_struct *p, int sync); 898 899 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p); 900 901 struct task_struct * (*pick_next_task) (struct rq *rq); 902 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 903 904 #ifdef CONFIG_SMP 905 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 906 struct rq *busiest, unsigned long max_load_move, 907 struct sched_domain *sd, enum cpu_idle_type idle, 908 int *all_pinned, int *this_best_prio); 909 910 int (*move_one_task) (struct rq *this_rq, int this_cpu, 911 struct rq *busiest, struct sched_domain *sd, 912 enum cpu_idle_type idle); 913 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 914 void (*post_schedule) (struct rq *this_rq); 915 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 916 #endif 917 918 void (*set_curr_task) (struct rq *rq); 919 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 920 void (*task_new) (struct rq *rq, struct task_struct *p); 921 void (*set_cpus_allowed)(struct task_struct *p, 922 const cpumask_t *newmask); 923 924 void (*rq_online)(struct rq *rq); 925 void (*rq_offline)(struct rq *rq); 926 927 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 928 int running); 929 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 930 int running); 931 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 932 int oldprio, int running); 933 934 #ifdef CONFIG_FAIR_GROUP_SCHED 935 void (*moved_group) (struct task_struct *p); 936 #endif 937 }; 938 939 struct load_weight { 940 unsigned long weight, inv_weight; 941 }; 942 943 /* 944 * CFS stats for a schedulable entity (task, task-group etc) 945 * 946 * Current field usage histogram: 947 * 948 * 4 se->block_start 949 * 4 se->run_node 950 * 4 se->sleep_start 951 * 6 se->load.weight 952 */ 953 struct sched_entity { 954 struct load_weight load; /* for load-balancing */ 955 struct rb_node run_node; 956 struct list_head group_node; 957 unsigned int on_rq; 958 959 u64 exec_start; 960 u64 sum_exec_runtime; 961 u64 vruntime; 962 u64 prev_sum_exec_runtime; 963 964 u64 last_wakeup; 965 u64 avg_overlap; 966 967 #ifdef CONFIG_SCHEDSTATS 968 u64 wait_start; 969 u64 wait_max; 970 u64 wait_count; 971 u64 wait_sum; 972 973 u64 sleep_start; 974 u64 sleep_max; 975 s64 sum_sleep_runtime; 976 977 u64 block_start; 978 u64 block_max; 979 u64 exec_max; 980 u64 slice_max; 981 982 u64 nr_migrations; 983 u64 nr_migrations_cold; 984 u64 nr_failed_migrations_affine; 985 u64 nr_failed_migrations_running; 986 u64 nr_failed_migrations_hot; 987 u64 nr_forced_migrations; 988 u64 nr_forced2_migrations; 989 990 u64 nr_wakeups; 991 u64 nr_wakeups_sync; 992 u64 nr_wakeups_migrate; 993 u64 nr_wakeups_local; 994 u64 nr_wakeups_remote; 995 u64 nr_wakeups_affine; 996 u64 nr_wakeups_affine_attempts; 997 u64 nr_wakeups_passive; 998 u64 nr_wakeups_idle; 999 #endif 1000 1001 #ifdef CONFIG_FAIR_GROUP_SCHED 1002 struct sched_entity *parent; 1003 /* rq on which this entity is (to be) queued: */ 1004 struct cfs_rq *cfs_rq; 1005 /* rq "owned" by this entity/group: */ 1006 struct cfs_rq *my_q; 1007 #endif 1008 }; 1009 1010 struct sched_rt_entity { 1011 struct list_head run_list; 1012 unsigned int time_slice; 1013 unsigned long timeout; 1014 int nr_cpus_allowed; 1015 1016 struct sched_rt_entity *back; 1017 #ifdef CONFIG_RT_GROUP_SCHED 1018 struct sched_rt_entity *parent; 1019 /* rq on which this entity is (to be) queued: */ 1020 struct rt_rq *rt_rq; 1021 /* rq "owned" by this entity/group: */ 1022 struct rt_rq *my_q; 1023 #endif 1024 }; 1025 1026 struct task_struct { 1027 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1028 void *stack; 1029 atomic_t usage; 1030 unsigned int flags; /* per process flags, defined below */ 1031 unsigned int ptrace; 1032 1033 int lock_depth; /* BKL lock depth */ 1034 1035 #ifdef CONFIG_SMP 1036 #ifdef __ARCH_WANT_UNLOCKED_CTXSW 1037 int oncpu; 1038 #endif 1039 #endif 1040 1041 int prio, static_prio, normal_prio; 1042 unsigned int rt_priority; 1043 const struct sched_class *sched_class; 1044 struct sched_entity se; 1045 struct sched_rt_entity rt; 1046 1047 #ifdef CONFIG_PREEMPT_NOTIFIERS 1048 /* list of struct preempt_notifier: */ 1049 struct hlist_head preempt_notifiers; 1050 #endif 1051 1052 /* 1053 * fpu_counter contains the number of consecutive context switches 1054 * that the FPU is used. If this is over a threshold, the lazy fpu 1055 * saving becomes unlazy to save the trap. This is an unsigned char 1056 * so that after 256 times the counter wraps and the behavior turns 1057 * lazy again; this to deal with bursty apps that only use FPU for 1058 * a short time 1059 */ 1060 unsigned char fpu_counter; 1061 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */ 1062 #ifdef CONFIG_BLK_DEV_IO_TRACE 1063 unsigned int btrace_seq; 1064 #endif 1065 1066 unsigned int policy; 1067 cpumask_t cpus_allowed; 1068 1069 #ifdef CONFIG_PREEMPT_RCU 1070 int rcu_read_lock_nesting; 1071 int rcu_flipctr_idx; 1072 #endif /* #ifdef CONFIG_PREEMPT_RCU */ 1073 1074 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1075 struct sched_info sched_info; 1076 #endif 1077 1078 struct list_head tasks; 1079 1080 struct mm_struct *mm, *active_mm; 1081 1082 /* task state */ 1083 struct linux_binfmt *binfmt; 1084 int exit_state; 1085 int exit_code, exit_signal; 1086 int pdeath_signal; /* The signal sent when the parent dies */ 1087 /* ??? */ 1088 unsigned int personality; 1089 unsigned did_exec:1; 1090 pid_t pid; 1091 pid_t tgid; 1092 1093 #ifdef CONFIG_CC_STACKPROTECTOR 1094 /* Canary value for the -fstack-protector gcc feature */ 1095 unsigned long stack_canary; 1096 #endif 1097 /* 1098 * pointers to (original) parent process, youngest child, younger sibling, 1099 * older sibling, respectively. (p->father can be replaced with 1100 * p->real_parent->pid) 1101 */ 1102 struct task_struct *real_parent; /* real parent process */ 1103 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1104 /* 1105 * children/sibling forms the list of my natural children 1106 */ 1107 struct list_head children; /* list of my children */ 1108 struct list_head sibling; /* linkage in my parent's children list */ 1109 struct task_struct *group_leader; /* threadgroup leader */ 1110 1111 /* 1112 * ptraced is the list of tasks this task is using ptrace on. 1113 * This includes both natural children and PTRACE_ATTACH targets. 1114 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1115 */ 1116 struct list_head ptraced; 1117 struct list_head ptrace_entry; 1118 1119 /* PID/PID hash table linkage. */ 1120 struct pid_link pids[PIDTYPE_MAX]; 1121 struct list_head thread_group; 1122 1123 struct completion *vfork_done; /* for vfork() */ 1124 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1125 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1126 1127 cputime_t utime, stime, utimescaled, stimescaled; 1128 cputime_t gtime; 1129 cputime_t prev_utime, prev_stime; 1130 unsigned long nvcsw, nivcsw; /* context switch counts */ 1131 struct timespec start_time; /* monotonic time */ 1132 struct timespec real_start_time; /* boot based time */ 1133 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1134 unsigned long min_flt, maj_flt; 1135 1136 cputime_t it_prof_expires, it_virt_expires; 1137 unsigned long long it_sched_expires; 1138 struct list_head cpu_timers[3]; 1139 1140 /* process credentials */ 1141 uid_t uid,euid,suid,fsuid; 1142 gid_t gid,egid,sgid,fsgid; 1143 struct group_info *group_info; 1144 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset; 1145 struct user_struct *user; 1146 unsigned securebits; 1147 #ifdef CONFIG_KEYS 1148 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 1149 struct key *request_key_auth; /* assumed request_key authority */ 1150 struct key *thread_keyring; /* keyring private to this thread */ 1151 #endif 1152 char comm[TASK_COMM_LEN]; /* executable name excluding path 1153 - access with [gs]et_task_comm (which lock 1154 it with task_lock()) 1155 - initialized normally by flush_old_exec */ 1156 /* file system info */ 1157 int link_count, total_link_count; 1158 #ifdef CONFIG_SYSVIPC 1159 /* ipc stuff */ 1160 struct sysv_sem sysvsem; 1161 #endif 1162 #ifdef CONFIG_DETECT_SOFTLOCKUP 1163 /* hung task detection */ 1164 unsigned long last_switch_timestamp; 1165 unsigned long last_switch_count; 1166 #endif 1167 /* CPU-specific state of this task */ 1168 struct thread_struct thread; 1169 /* filesystem information */ 1170 struct fs_struct *fs; 1171 /* open file information */ 1172 struct files_struct *files; 1173 /* namespaces */ 1174 struct nsproxy *nsproxy; 1175 /* signal handlers */ 1176 struct signal_struct *signal; 1177 struct sighand_struct *sighand; 1178 1179 sigset_t blocked, real_blocked; 1180 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1181 struct sigpending pending; 1182 1183 unsigned long sas_ss_sp; 1184 size_t sas_ss_size; 1185 int (*notifier)(void *priv); 1186 void *notifier_data; 1187 sigset_t *notifier_mask; 1188 #ifdef CONFIG_SECURITY 1189 void *security; 1190 #endif 1191 struct audit_context *audit_context; 1192 #ifdef CONFIG_AUDITSYSCALL 1193 uid_t loginuid; 1194 unsigned int sessionid; 1195 #endif 1196 seccomp_t seccomp; 1197 1198 /* Thread group tracking */ 1199 u32 parent_exec_id; 1200 u32 self_exec_id; 1201 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 1202 spinlock_t alloc_lock; 1203 1204 /* Protection of the PI data structures: */ 1205 spinlock_t pi_lock; 1206 1207 #ifdef CONFIG_RT_MUTEXES 1208 /* PI waiters blocked on a rt_mutex held by this task */ 1209 struct plist_head pi_waiters; 1210 /* Deadlock detection and priority inheritance handling */ 1211 struct rt_mutex_waiter *pi_blocked_on; 1212 #endif 1213 1214 #ifdef CONFIG_DEBUG_MUTEXES 1215 /* mutex deadlock detection */ 1216 struct mutex_waiter *blocked_on; 1217 #endif 1218 #ifdef CONFIG_TRACE_IRQFLAGS 1219 unsigned int irq_events; 1220 int hardirqs_enabled; 1221 unsigned long hardirq_enable_ip; 1222 unsigned int hardirq_enable_event; 1223 unsigned long hardirq_disable_ip; 1224 unsigned int hardirq_disable_event; 1225 int softirqs_enabled; 1226 unsigned long softirq_disable_ip; 1227 unsigned int softirq_disable_event; 1228 unsigned long softirq_enable_ip; 1229 unsigned int softirq_enable_event; 1230 int hardirq_context; 1231 int softirq_context; 1232 #endif 1233 #ifdef CONFIG_LOCKDEP 1234 # define MAX_LOCK_DEPTH 48UL 1235 u64 curr_chain_key; 1236 int lockdep_depth; 1237 unsigned int lockdep_recursion; 1238 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1239 #endif 1240 1241 /* journalling filesystem info */ 1242 void *journal_info; 1243 1244 /* stacked block device info */ 1245 struct bio *bio_list, **bio_tail; 1246 1247 /* VM state */ 1248 struct reclaim_state *reclaim_state; 1249 1250 struct backing_dev_info *backing_dev_info; 1251 1252 struct io_context *io_context; 1253 1254 unsigned long ptrace_message; 1255 siginfo_t *last_siginfo; /* For ptrace use. */ 1256 struct task_io_accounting ioac; 1257 #if defined(CONFIG_TASK_XACCT) 1258 u64 acct_rss_mem1; /* accumulated rss usage */ 1259 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1260 cputime_t acct_timexpd; /* stime + utime since last update */ 1261 #endif 1262 #ifdef CONFIG_CPUSETS 1263 nodemask_t mems_allowed; 1264 int cpuset_mems_generation; 1265 int cpuset_mem_spread_rotor; 1266 #endif 1267 #ifdef CONFIG_CGROUPS 1268 /* Control Group info protected by css_set_lock */ 1269 struct css_set *cgroups; 1270 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1271 struct list_head cg_list; 1272 #endif 1273 #ifdef CONFIG_FUTEX 1274 struct robust_list_head __user *robust_list; 1275 #ifdef CONFIG_COMPAT 1276 struct compat_robust_list_head __user *compat_robust_list; 1277 #endif 1278 struct list_head pi_state_list; 1279 struct futex_pi_state *pi_state_cache; 1280 #endif 1281 #ifdef CONFIG_NUMA 1282 struct mempolicy *mempolicy; 1283 short il_next; 1284 #endif 1285 atomic_t fs_excl; /* holding fs exclusive resources */ 1286 struct rcu_head rcu; 1287 1288 /* 1289 * cache last used pipe for splice 1290 */ 1291 struct pipe_inode_info *splice_pipe; 1292 #ifdef CONFIG_TASK_DELAY_ACCT 1293 struct task_delay_info *delays; 1294 #endif 1295 #ifdef CONFIG_FAULT_INJECTION 1296 int make_it_fail; 1297 #endif 1298 struct prop_local_single dirties; 1299 #ifdef CONFIG_LATENCYTOP 1300 int latency_record_count; 1301 struct latency_record latency_record[LT_SAVECOUNT]; 1302 #endif 1303 }; 1304 1305 /* 1306 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1307 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1308 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1309 * values are inverted: lower p->prio value means higher priority. 1310 * 1311 * The MAX_USER_RT_PRIO value allows the actual maximum 1312 * RT priority to be separate from the value exported to 1313 * user-space. This allows kernel threads to set their 1314 * priority to a value higher than any user task. Note: 1315 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1316 */ 1317 1318 #define MAX_USER_RT_PRIO 100 1319 #define MAX_RT_PRIO MAX_USER_RT_PRIO 1320 1321 #define MAX_PRIO (MAX_RT_PRIO + 40) 1322 #define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1323 1324 static inline int rt_prio(int prio) 1325 { 1326 if (unlikely(prio < MAX_RT_PRIO)) 1327 return 1; 1328 return 0; 1329 } 1330 1331 static inline int rt_task(struct task_struct *p) 1332 { 1333 return rt_prio(p->prio); 1334 } 1335 1336 static inline void set_task_session(struct task_struct *tsk, pid_t session) 1337 { 1338 tsk->signal->__session = session; 1339 } 1340 1341 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp) 1342 { 1343 tsk->signal->__pgrp = pgrp; 1344 } 1345 1346 static inline struct pid *task_pid(struct task_struct *task) 1347 { 1348 return task->pids[PIDTYPE_PID].pid; 1349 } 1350 1351 static inline struct pid *task_tgid(struct task_struct *task) 1352 { 1353 return task->group_leader->pids[PIDTYPE_PID].pid; 1354 } 1355 1356 static inline struct pid *task_pgrp(struct task_struct *task) 1357 { 1358 return task->group_leader->pids[PIDTYPE_PGID].pid; 1359 } 1360 1361 static inline struct pid *task_session(struct task_struct *task) 1362 { 1363 return task->group_leader->pids[PIDTYPE_SID].pid; 1364 } 1365 1366 struct pid_namespace; 1367 1368 /* 1369 * the helpers to get the task's different pids as they are seen 1370 * from various namespaces 1371 * 1372 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1373 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1374 * current. 1375 * task_xid_nr_ns() : id seen from the ns specified; 1376 * 1377 * set_task_vxid() : assigns a virtual id to a task; 1378 * 1379 * see also pid_nr() etc in include/linux/pid.h 1380 */ 1381 1382 static inline pid_t task_pid_nr(struct task_struct *tsk) 1383 { 1384 return tsk->pid; 1385 } 1386 1387 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1388 1389 static inline pid_t task_pid_vnr(struct task_struct *tsk) 1390 { 1391 return pid_vnr(task_pid(tsk)); 1392 } 1393 1394 1395 static inline pid_t task_tgid_nr(struct task_struct *tsk) 1396 { 1397 return tsk->tgid; 1398 } 1399 1400 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1401 1402 static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1403 { 1404 return pid_vnr(task_tgid(tsk)); 1405 } 1406 1407 1408 static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1409 { 1410 return tsk->signal->__pgrp; 1411 } 1412 1413 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1414 1415 static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1416 { 1417 return pid_vnr(task_pgrp(tsk)); 1418 } 1419 1420 1421 static inline pid_t task_session_nr(struct task_struct *tsk) 1422 { 1423 return tsk->signal->__session; 1424 } 1425 1426 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1427 1428 static inline pid_t task_session_vnr(struct task_struct *tsk) 1429 { 1430 return pid_vnr(task_session(tsk)); 1431 } 1432 1433 1434 /** 1435 * pid_alive - check that a task structure is not stale 1436 * @p: Task structure to be checked. 1437 * 1438 * Test if a process is not yet dead (at most zombie state) 1439 * If pid_alive fails, then pointers within the task structure 1440 * can be stale and must not be dereferenced. 1441 */ 1442 static inline int pid_alive(struct task_struct *p) 1443 { 1444 return p->pids[PIDTYPE_PID].pid != NULL; 1445 } 1446 1447 /** 1448 * is_global_init - check if a task structure is init 1449 * @tsk: Task structure to be checked. 1450 * 1451 * Check if a task structure is the first user space task the kernel created. 1452 */ 1453 static inline int is_global_init(struct task_struct *tsk) 1454 { 1455 return tsk->pid == 1; 1456 } 1457 1458 /* 1459 * is_container_init: 1460 * check whether in the task is init in its own pid namespace. 1461 */ 1462 extern int is_container_init(struct task_struct *tsk); 1463 1464 extern struct pid *cad_pid; 1465 1466 extern void free_task(struct task_struct *tsk); 1467 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1468 1469 extern void __put_task_struct(struct task_struct *t); 1470 1471 static inline void put_task_struct(struct task_struct *t) 1472 { 1473 if (atomic_dec_and_test(&t->usage)) 1474 __put_task_struct(t); 1475 } 1476 1477 /* 1478 * Per process flags 1479 */ 1480 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1481 /* Not implemented yet, only for 486*/ 1482 #define PF_STARTING 0x00000002 /* being created */ 1483 #define PF_EXITING 0x00000004 /* getting shut down */ 1484 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1485 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1486 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1487 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1488 #define PF_DUMPCORE 0x00000200 /* dumped core */ 1489 #define PF_SIGNALED 0x00000400 /* killed by a signal */ 1490 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1491 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1492 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1493 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1494 #define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1495 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1496 #define PF_KSWAPD 0x00040000 /* I am kswapd */ 1497 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1498 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1499 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1500 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1501 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1502 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1503 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1504 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1505 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1506 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1507 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1508 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1509 1510 /* 1511 * Only the _current_ task can read/write to tsk->flags, but other 1512 * tasks can access tsk->flags in readonly mode for example 1513 * with tsk_used_math (like during threaded core dumping). 1514 * There is however an exception to this rule during ptrace 1515 * or during fork: the ptracer task is allowed to write to the 1516 * child->flags of its traced child (same goes for fork, the parent 1517 * can write to the child->flags), because we're guaranteed the 1518 * child is not running and in turn not changing child->flags 1519 * at the same time the parent does it. 1520 */ 1521 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1522 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1523 #define clear_used_math() clear_stopped_child_used_math(current) 1524 #define set_used_math() set_stopped_child_used_math(current) 1525 #define conditional_stopped_child_used_math(condition, child) \ 1526 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1527 #define conditional_used_math(condition) \ 1528 conditional_stopped_child_used_math(condition, current) 1529 #define copy_to_stopped_child_used_math(child) \ 1530 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1531 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1532 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1533 #define used_math() tsk_used_math(current) 1534 1535 #ifdef CONFIG_SMP 1536 extern int set_cpus_allowed_ptr(struct task_struct *p, 1537 const cpumask_t *new_mask); 1538 #else 1539 static inline int set_cpus_allowed_ptr(struct task_struct *p, 1540 const cpumask_t *new_mask) 1541 { 1542 if (!cpu_isset(0, *new_mask)) 1543 return -EINVAL; 1544 return 0; 1545 } 1546 #endif 1547 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1548 { 1549 return set_cpus_allowed_ptr(p, &new_mask); 1550 } 1551 1552 extern unsigned long long sched_clock(void); 1553 1554 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1555 static inline void sched_clock_init(void) 1556 { 1557 } 1558 1559 static inline u64 sched_clock_cpu(int cpu) 1560 { 1561 return sched_clock(); 1562 } 1563 1564 static inline void sched_clock_tick(void) 1565 { 1566 } 1567 1568 static inline void sched_clock_idle_sleep_event(void) 1569 { 1570 } 1571 1572 static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1573 { 1574 } 1575 1576 #ifdef CONFIG_NO_HZ 1577 static inline void sched_clock_tick_stop(int cpu) 1578 { 1579 } 1580 1581 static inline void sched_clock_tick_start(int cpu) 1582 { 1583 } 1584 #endif 1585 1586 #else /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ 1587 extern void sched_clock_init(void); 1588 extern u64 sched_clock_cpu(int cpu); 1589 extern void sched_clock_tick(void); 1590 extern void sched_clock_idle_sleep_event(void); 1591 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1592 #ifdef CONFIG_NO_HZ 1593 extern void sched_clock_tick_stop(int cpu); 1594 extern void sched_clock_tick_start(int cpu); 1595 #endif 1596 #endif /* CONFIG_HAVE_UNSTABLE_SCHED_CLOCK */ 1597 1598 /* 1599 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1600 * clock constructed from sched_clock(): 1601 */ 1602 extern unsigned long long cpu_clock(int cpu); 1603 1604 extern unsigned long long 1605 task_sched_runtime(struct task_struct *task); 1606 1607 /* sched_exec is called by processes performing an exec */ 1608 #ifdef CONFIG_SMP 1609 extern void sched_exec(void); 1610 #else 1611 #define sched_exec() {} 1612 #endif 1613 1614 extern void sched_clock_idle_sleep_event(void); 1615 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1616 1617 #ifdef CONFIG_HOTPLUG_CPU 1618 extern void idle_task_exit(void); 1619 #else 1620 static inline void idle_task_exit(void) {} 1621 #endif 1622 1623 extern void sched_idle_next(void); 1624 1625 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1626 extern void wake_up_idle_cpu(int cpu); 1627 #else 1628 static inline void wake_up_idle_cpu(int cpu) { } 1629 #endif 1630 1631 #ifdef CONFIG_SCHED_DEBUG 1632 extern unsigned int sysctl_sched_latency; 1633 extern unsigned int sysctl_sched_min_granularity; 1634 extern unsigned int sysctl_sched_wakeup_granularity; 1635 extern unsigned int sysctl_sched_child_runs_first; 1636 extern unsigned int sysctl_sched_features; 1637 extern unsigned int sysctl_sched_migration_cost; 1638 extern unsigned int sysctl_sched_nr_migrate; 1639 extern unsigned int sysctl_sched_shares_ratelimit; 1640 1641 int sched_nr_latency_handler(struct ctl_table *table, int write, 1642 struct file *file, void __user *buffer, size_t *length, 1643 loff_t *ppos); 1644 #endif 1645 extern unsigned int sysctl_sched_rt_period; 1646 extern int sysctl_sched_rt_runtime; 1647 1648 int sched_rt_handler(struct ctl_table *table, int write, 1649 struct file *filp, void __user *buffer, size_t *lenp, 1650 loff_t *ppos); 1651 1652 extern unsigned int sysctl_sched_compat_yield; 1653 1654 #ifdef CONFIG_RT_MUTEXES 1655 extern int rt_mutex_getprio(struct task_struct *p); 1656 extern void rt_mutex_setprio(struct task_struct *p, int prio); 1657 extern void rt_mutex_adjust_pi(struct task_struct *p); 1658 #else 1659 static inline int rt_mutex_getprio(struct task_struct *p) 1660 { 1661 return p->normal_prio; 1662 } 1663 # define rt_mutex_adjust_pi(p) do { } while (0) 1664 #endif 1665 1666 extern void set_user_nice(struct task_struct *p, long nice); 1667 extern int task_prio(const struct task_struct *p); 1668 extern int task_nice(const struct task_struct *p); 1669 extern int can_nice(const struct task_struct *p, const int nice); 1670 extern int task_curr(const struct task_struct *p); 1671 extern int idle_cpu(int cpu); 1672 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1673 extern int sched_setscheduler_nocheck(struct task_struct *, int, 1674 struct sched_param *); 1675 extern struct task_struct *idle_task(int cpu); 1676 extern struct task_struct *curr_task(int cpu); 1677 extern void set_curr_task(int cpu, struct task_struct *p); 1678 1679 void yield(void); 1680 1681 /* 1682 * The default (Linux) execution domain. 1683 */ 1684 extern struct exec_domain default_exec_domain; 1685 1686 union thread_union { 1687 struct thread_info thread_info; 1688 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1689 }; 1690 1691 #ifndef __HAVE_ARCH_KSTACK_END 1692 static inline int kstack_end(void *addr) 1693 { 1694 /* Reliable end of stack detection: 1695 * Some APM bios versions misalign the stack 1696 */ 1697 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1698 } 1699 #endif 1700 1701 extern union thread_union init_thread_union; 1702 extern struct task_struct init_task; 1703 1704 extern struct mm_struct init_mm; 1705 1706 extern struct pid_namespace init_pid_ns; 1707 1708 /* 1709 * find a task by one of its numerical ids 1710 * 1711 * find_task_by_pid_type_ns(): 1712 * it is the most generic call - it finds a task by all id, 1713 * type and namespace specified 1714 * find_task_by_pid_ns(): 1715 * finds a task by its pid in the specified namespace 1716 * find_task_by_vpid(): 1717 * finds a task by its virtual pid 1718 * 1719 * see also find_vpid() etc in include/linux/pid.h 1720 */ 1721 1722 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid, 1723 struct pid_namespace *ns); 1724 1725 extern struct task_struct *find_task_by_vpid(pid_t nr); 1726 extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1727 struct pid_namespace *ns); 1728 1729 extern void __set_special_pids(struct pid *pid); 1730 1731 /* per-UID process charging. */ 1732 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1733 static inline struct user_struct *get_uid(struct user_struct *u) 1734 { 1735 atomic_inc(&u->__count); 1736 return u; 1737 } 1738 extern void free_uid(struct user_struct *); 1739 extern void switch_uid(struct user_struct *); 1740 extern void release_uids(struct user_namespace *ns); 1741 1742 #include <asm/current.h> 1743 1744 extern void do_timer(unsigned long ticks); 1745 1746 extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1747 extern int wake_up_process(struct task_struct *tsk); 1748 extern void wake_up_new_task(struct task_struct *tsk, 1749 unsigned long clone_flags); 1750 #ifdef CONFIG_SMP 1751 extern void kick_process(struct task_struct *tsk); 1752 #else 1753 static inline void kick_process(struct task_struct *tsk) { } 1754 #endif 1755 extern void sched_fork(struct task_struct *p, int clone_flags); 1756 extern void sched_dead(struct task_struct *p); 1757 1758 extern int in_group_p(gid_t); 1759 extern int in_egroup_p(gid_t); 1760 1761 extern void proc_caches_init(void); 1762 extern void flush_signals(struct task_struct *); 1763 extern void ignore_signals(struct task_struct *); 1764 extern void flush_signal_handlers(struct task_struct *, int force_default); 1765 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1766 1767 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1768 { 1769 unsigned long flags; 1770 int ret; 1771 1772 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1773 ret = dequeue_signal(tsk, mask, info); 1774 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1775 1776 return ret; 1777 } 1778 1779 extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1780 sigset_t *mask); 1781 extern void unblock_all_signals(void); 1782 extern void release_task(struct task_struct * p); 1783 extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1784 extern int force_sigsegv(int, struct task_struct *); 1785 extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1786 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1787 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1788 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1789 extern int kill_pgrp(struct pid *pid, int sig, int priv); 1790 extern int kill_pid(struct pid *pid, int sig, int priv); 1791 extern int kill_proc_info(int, struct siginfo *, pid_t); 1792 extern int do_notify_parent(struct task_struct *, int); 1793 extern void force_sig(int, struct task_struct *); 1794 extern void force_sig_specific(int, struct task_struct *); 1795 extern int send_sig(int, struct task_struct *, int); 1796 extern void zap_other_threads(struct task_struct *p); 1797 extern struct sigqueue *sigqueue_alloc(void); 1798 extern void sigqueue_free(struct sigqueue *); 1799 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 1800 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1801 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1802 1803 static inline int kill_cad_pid(int sig, int priv) 1804 { 1805 return kill_pid(cad_pid, sig, priv); 1806 } 1807 1808 /* These can be the second arg to send_sig_info/send_group_sig_info. */ 1809 #define SEND_SIG_NOINFO ((struct siginfo *) 0) 1810 #define SEND_SIG_PRIV ((struct siginfo *) 1) 1811 #define SEND_SIG_FORCED ((struct siginfo *) 2) 1812 1813 static inline int is_si_special(const struct siginfo *info) 1814 { 1815 return info <= SEND_SIG_FORCED; 1816 } 1817 1818 /* True if we are on the alternate signal stack. */ 1819 1820 static inline int on_sig_stack(unsigned long sp) 1821 { 1822 return (sp - current->sas_ss_sp < current->sas_ss_size); 1823 } 1824 1825 static inline int sas_ss_flags(unsigned long sp) 1826 { 1827 return (current->sas_ss_size == 0 ? SS_DISABLE 1828 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1829 } 1830 1831 /* 1832 * Routines for handling mm_structs 1833 */ 1834 extern struct mm_struct * mm_alloc(void); 1835 1836 /* mmdrop drops the mm and the page tables */ 1837 extern void __mmdrop(struct mm_struct *); 1838 static inline void mmdrop(struct mm_struct * mm) 1839 { 1840 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 1841 __mmdrop(mm); 1842 } 1843 1844 /* mmput gets rid of the mappings and all user-space */ 1845 extern void mmput(struct mm_struct *); 1846 /* Grab a reference to a task's mm, if it is not already going away */ 1847 extern struct mm_struct *get_task_mm(struct task_struct *task); 1848 /* Remove the current tasks stale references to the old mm_struct */ 1849 extern void mm_release(struct task_struct *, struct mm_struct *); 1850 /* Allocate a new mm structure and copy contents from tsk->mm */ 1851 extern struct mm_struct *dup_mm(struct task_struct *tsk); 1852 1853 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1854 extern void flush_thread(void); 1855 extern void exit_thread(void); 1856 1857 extern void exit_files(struct task_struct *); 1858 extern void __cleanup_signal(struct signal_struct *); 1859 extern void __cleanup_sighand(struct sighand_struct *); 1860 1861 extern void exit_itimers(struct signal_struct *); 1862 extern void flush_itimer_signals(void); 1863 1864 extern NORET_TYPE void do_group_exit(int); 1865 1866 extern void daemonize(const char *, ...); 1867 extern int allow_signal(int); 1868 extern int disallow_signal(int); 1869 1870 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1871 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1872 struct task_struct *fork_idle(int); 1873 1874 extern void set_task_comm(struct task_struct *tsk, char *from); 1875 extern char *get_task_comm(char *to, struct task_struct *tsk); 1876 1877 #ifdef CONFIG_SMP 1878 extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 1879 #else 1880 static inline unsigned long wait_task_inactive(struct task_struct *p, 1881 long match_state) 1882 { 1883 return 1; 1884 } 1885 #endif 1886 1887 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1888 1889 #define for_each_process(p) \ 1890 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1891 1892 /* 1893 * Careful: do_each_thread/while_each_thread is a double loop so 1894 * 'break' will not work as expected - use goto instead. 1895 */ 1896 #define do_each_thread(g, t) \ 1897 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1898 1899 #define while_each_thread(g, t) \ 1900 while ((t = next_thread(t)) != g) 1901 1902 /* de_thread depends on thread_group_leader not being a pid based check */ 1903 #define thread_group_leader(p) (p == p->group_leader) 1904 1905 /* Do to the insanities of de_thread it is possible for a process 1906 * to have the pid of the thread group leader without actually being 1907 * the thread group leader. For iteration through the pids in proc 1908 * all we care about is that we have a task with the appropriate 1909 * pid, we don't actually care if we have the right task. 1910 */ 1911 static inline int has_group_leader_pid(struct task_struct *p) 1912 { 1913 return p->pid == p->tgid; 1914 } 1915 1916 static inline 1917 int same_thread_group(struct task_struct *p1, struct task_struct *p2) 1918 { 1919 return p1->tgid == p2->tgid; 1920 } 1921 1922 static inline struct task_struct *next_thread(const struct task_struct *p) 1923 { 1924 return list_entry(rcu_dereference(p->thread_group.next), 1925 struct task_struct, thread_group); 1926 } 1927 1928 static inline int thread_group_empty(struct task_struct *p) 1929 { 1930 return list_empty(&p->thread_group); 1931 } 1932 1933 #define delay_group_leader(p) \ 1934 (thread_group_leader(p) && !thread_group_empty(p)) 1935 1936 /* 1937 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 1938 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1939 * pins the final release of task.io_context. Also protects ->cpuset and 1940 * ->cgroup.subsys[]. 1941 * 1942 * Nests both inside and outside of read_lock(&tasklist_lock). 1943 * It must not be nested with write_lock_irq(&tasklist_lock), 1944 * neither inside nor outside. 1945 */ 1946 static inline void task_lock(struct task_struct *p) 1947 { 1948 spin_lock(&p->alloc_lock); 1949 } 1950 1951 static inline void task_unlock(struct task_struct *p) 1952 { 1953 spin_unlock(&p->alloc_lock); 1954 } 1955 1956 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 1957 unsigned long *flags); 1958 1959 static inline void unlock_task_sighand(struct task_struct *tsk, 1960 unsigned long *flags) 1961 { 1962 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 1963 } 1964 1965 #ifndef __HAVE_THREAD_FUNCTIONS 1966 1967 #define task_thread_info(task) ((struct thread_info *)(task)->stack) 1968 #define task_stack_page(task) ((task)->stack) 1969 1970 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 1971 { 1972 *task_thread_info(p) = *task_thread_info(org); 1973 task_thread_info(p)->task = p; 1974 } 1975 1976 static inline unsigned long *end_of_stack(struct task_struct *p) 1977 { 1978 return (unsigned long *)(task_thread_info(p) + 1); 1979 } 1980 1981 #endif 1982 1983 static inline int object_is_on_stack(void *obj) 1984 { 1985 void *stack = task_stack_page(current); 1986 1987 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 1988 } 1989 1990 extern void thread_info_cache_init(void); 1991 1992 /* set thread flags in other task's structures 1993 * - see asm/thread_info.h for TIF_xxxx flags available 1994 */ 1995 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1996 { 1997 set_ti_thread_flag(task_thread_info(tsk), flag); 1998 } 1999 2000 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2001 { 2002 clear_ti_thread_flag(task_thread_info(tsk), flag); 2003 } 2004 2005 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2006 { 2007 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2008 } 2009 2010 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2011 { 2012 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2013 } 2014 2015 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2016 { 2017 return test_ti_thread_flag(task_thread_info(tsk), flag); 2018 } 2019 2020 static inline void set_tsk_need_resched(struct task_struct *tsk) 2021 { 2022 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2023 } 2024 2025 static inline void clear_tsk_need_resched(struct task_struct *tsk) 2026 { 2027 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2028 } 2029 2030 static inline int test_tsk_need_resched(struct task_struct *tsk) 2031 { 2032 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2033 } 2034 2035 static inline int signal_pending(struct task_struct *p) 2036 { 2037 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2038 } 2039 2040 extern int __fatal_signal_pending(struct task_struct *p); 2041 2042 static inline int fatal_signal_pending(struct task_struct *p) 2043 { 2044 return signal_pending(p) && __fatal_signal_pending(p); 2045 } 2046 2047 static inline int signal_pending_state(long state, struct task_struct *p) 2048 { 2049 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2050 return 0; 2051 if (!signal_pending(p)) 2052 return 0; 2053 2054 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2055 } 2056 2057 static inline int need_resched(void) 2058 { 2059 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2060 } 2061 2062 /* 2063 * cond_resched() and cond_resched_lock(): latency reduction via 2064 * explicit rescheduling in places that are safe. The return 2065 * value indicates whether a reschedule was done in fact. 2066 * cond_resched_lock() will drop the spinlock before scheduling, 2067 * cond_resched_softirq() will enable bhs before scheduling. 2068 */ 2069 extern int _cond_resched(void); 2070 #ifdef CONFIG_PREEMPT_BKL 2071 static inline int cond_resched(void) 2072 { 2073 return 0; 2074 } 2075 #else 2076 static inline int cond_resched(void) 2077 { 2078 return _cond_resched(); 2079 } 2080 #endif 2081 extern int cond_resched_lock(spinlock_t * lock); 2082 extern int cond_resched_softirq(void); 2083 static inline int cond_resched_bkl(void) 2084 { 2085 return _cond_resched(); 2086 } 2087 2088 /* 2089 * Does a critical section need to be broken due to another 2090 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2091 * but a general need for low latency) 2092 */ 2093 static inline int spin_needbreak(spinlock_t *lock) 2094 { 2095 #ifdef CONFIG_PREEMPT 2096 return spin_is_contended(lock); 2097 #else 2098 return 0; 2099 #endif 2100 } 2101 2102 /* 2103 * Reevaluate whether the task has signals pending delivery. 2104 * Wake the task if so. 2105 * This is required every time the blocked sigset_t changes. 2106 * callers must hold sighand->siglock. 2107 */ 2108 extern void recalc_sigpending_and_wake(struct task_struct *t); 2109 extern void recalc_sigpending(void); 2110 2111 extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2112 2113 /* 2114 * Wrappers for p->thread_info->cpu access. No-op on UP. 2115 */ 2116 #ifdef CONFIG_SMP 2117 2118 static inline unsigned int task_cpu(const struct task_struct *p) 2119 { 2120 return task_thread_info(p)->cpu; 2121 } 2122 2123 extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2124 2125 #else 2126 2127 static inline unsigned int task_cpu(const struct task_struct *p) 2128 { 2129 return 0; 2130 } 2131 2132 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2133 { 2134 } 2135 2136 #endif /* CONFIG_SMP */ 2137 2138 extern void arch_pick_mmap_layout(struct mm_struct *mm); 2139 2140 #ifdef CONFIG_TRACING 2141 extern void 2142 __trace_special(void *__tr, void *__data, 2143 unsigned long arg1, unsigned long arg2, unsigned long arg3); 2144 #else 2145 static inline void 2146 __trace_special(void *__tr, void *__data, 2147 unsigned long arg1, unsigned long arg2, unsigned long arg3) 2148 { 2149 } 2150 #endif 2151 2152 extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); 2153 extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 2154 2155 extern int sched_mc_power_savings, sched_smt_power_savings; 2156 2157 extern void normalize_rt_tasks(void); 2158 2159 #ifdef CONFIG_GROUP_SCHED 2160 2161 extern struct task_group init_task_group; 2162 #ifdef CONFIG_USER_SCHED 2163 extern struct task_group root_task_group; 2164 #endif 2165 2166 extern struct task_group *sched_create_group(struct task_group *parent); 2167 extern void sched_destroy_group(struct task_group *tg); 2168 extern void sched_move_task(struct task_struct *tsk); 2169 #ifdef CONFIG_FAIR_GROUP_SCHED 2170 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2171 extern unsigned long sched_group_shares(struct task_group *tg); 2172 #endif 2173 #ifdef CONFIG_RT_GROUP_SCHED 2174 extern int sched_group_set_rt_runtime(struct task_group *tg, 2175 long rt_runtime_us); 2176 extern long sched_group_rt_runtime(struct task_group *tg); 2177 extern int sched_group_set_rt_period(struct task_group *tg, 2178 long rt_period_us); 2179 extern long sched_group_rt_period(struct task_group *tg); 2180 #endif 2181 #endif 2182 2183 #ifdef CONFIG_TASK_XACCT 2184 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2185 { 2186 tsk->ioac.rchar += amt; 2187 } 2188 2189 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2190 { 2191 tsk->ioac.wchar += amt; 2192 } 2193 2194 static inline void inc_syscr(struct task_struct *tsk) 2195 { 2196 tsk->ioac.syscr++; 2197 } 2198 2199 static inline void inc_syscw(struct task_struct *tsk) 2200 { 2201 tsk->ioac.syscw++; 2202 } 2203 #else 2204 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2205 { 2206 } 2207 2208 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2209 { 2210 } 2211 2212 static inline void inc_syscr(struct task_struct *tsk) 2213 { 2214 } 2215 2216 static inline void inc_syscw(struct task_struct *tsk) 2217 { 2218 } 2219 #endif 2220 2221 #ifndef TASK_SIZE_OF 2222 #define TASK_SIZE_OF(tsk) TASK_SIZE 2223 #endif 2224 2225 #ifdef CONFIG_MM_OWNER 2226 extern void mm_update_next_owner(struct mm_struct *mm); 2227 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2228 #else 2229 static inline void mm_update_next_owner(struct mm_struct *mm) 2230 { 2231 } 2232 2233 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2234 { 2235 } 2236 #endif /* CONFIG_MM_OWNER */ 2237 2238 #define TASK_STATE_TO_CHAR_STR "RSDTtZX" 2239 2240 #endif /* __KERNEL__ */ 2241 2242 #endif 2243