1 #ifndef _LINUX_SCHED_H 2 #define _LINUX_SCHED_H 3 4 #include <linux/auxvec.h> /* For AT_VECTOR_SIZE */ 5 6 /* 7 * cloning flags: 8 */ 9 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 10 #define CLONE_VM 0x00000100 /* set if VM shared between processes */ 11 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 12 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 13 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 14 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 15 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 16 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 17 #define CLONE_THREAD 0x00010000 /* Same thread group? */ 18 #define CLONE_NEWNS 0x00020000 /* New namespace group? */ 19 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 20 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 21 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 22 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 23 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 24 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 25 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 26 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */ 27 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */ 28 #define CLONE_NEWIPC 0x08000000 /* New ipcs */ 29 30 /* 31 * Scheduling policies 32 */ 33 #define SCHED_NORMAL 0 34 #define SCHED_FIFO 1 35 #define SCHED_RR 2 36 #define SCHED_BATCH 3 37 38 #ifdef __KERNEL__ 39 40 struct sched_param { 41 int sched_priority; 42 }; 43 44 #include <asm/param.h> /* for HZ */ 45 46 #include <linux/capability.h> 47 #include <linux/threads.h> 48 #include <linux/kernel.h> 49 #include <linux/types.h> 50 #include <linux/timex.h> 51 #include <linux/jiffies.h> 52 #include <linux/rbtree.h> 53 #include <linux/thread_info.h> 54 #include <linux/cpumask.h> 55 #include <linux/errno.h> 56 #include <linux/nodemask.h> 57 58 #include <asm/system.h> 59 #include <asm/semaphore.h> 60 #include <asm/page.h> 61 #include <asm/ptrace.h> 62 #include <asm/mmu.h> 63 #include <asm/cputime.h> 64 65 #include <linux/smp.h> 66 #include <linux/sem.h> 67 #include <linux/signal.h> 68 #include <linux/securebits.h> 69 #include <linux/fs_struct.h> 70 #include <linux/compiler.h> 71 #include <linux/completion.h> 72 #include <linux/pid.h> 73 #include <linux/percpu.h> 74 #include <linux/topology.h> 75 #include <linux/seccomp.h> 76 #include <linux/rcupdate.h> 77 #include <linux/futex.h> 78 #include <linux/rtmutex.h> 79 80 #include <linux/time.h> 81 #include <linux/param.h> 82 #include <linux/resource.h> 83 #include <linux/timer.h> 84 #include <linux/hrtimer.h> 85 86 #include <asm/processor.h> 87 88 struct exec_domain; 89 struct futex_pi_state; 90 91 /* 92 * List of flags we want to share for kernel threads, 93 * if only because they are not used by them anyway. 94 */ 95 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 96 97 /* 98 * These are the constant used to fake the fixed-point load-average 99 * counting. Some notes: 100 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 101 * a load-average precision of 10 bits integer + 11 bits fractional 102 * - if you want to count load-averages more often, you need more 103 * precision, or rounding will get you. With 2-second counting freq, 104 * the EXP_n values would be 1981, 2034 and 2043 if still using only 105 * 11 bit fractions. 106 */ 107 extern unsigned long avenrun[]; /* Load averages */ 108 109 #define FSHIFT 11 /* nr of bits of precision */ 110 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 111 #define LOAD_FREQ (5*HZ) /* 5 sec intervals */ 112 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 113 #define EXP_5 2014 /* 1/exp(5sec/5min) */ 114 #define EXP_15 2037 /* 1/exp(5sec/15min) */ 115 116 #define CALC_LOAD(load,exp,n) \ 117 load *= exp; \ 118 load += n*(FIXED_1-exp); \ 119 load >>= FSHIFT; 120 121 extern unsigned long total_forks; 122 extern int nr_threads; 123 DECLARE_PER_CPU(unsigned long, process_counts); 124 extern int nr_processes(void); 125 extern unsigned long nr_running(void); 126 extern unsigned long nr_uninterruptible(void); 127 extern unsigned long nr_active(void); 128 extern unsigned long nr_iowait(void); 129 extern unsigned long weighted_cpuload(const int cpu); 130 131 132 /* 133 * Task state bitmask. NOTE! These bits are also 134 * encoded in fs/proc/array.c: get_task_state(). 135 * 136 * We have two separate sets of flags: task->state 137 * is about runnability, while task->exit_state are 138 * about the task exiting. Confusing, but this way 139 * modifying one set can't modify the other one by 140 * mistake. 141 */ 142 #define TASK_RUNNING 0 143 #define TASK_INTERRUPTIBLE 1 144 #define TASK_UNINTERRUPTIBLE 2 145 #define TASK_STOPPED 4 146 #define TASK_TRACED 8 147 /* in tsk->exit_state */ 148 #define EXIT_ZOMBIE 16 149 #define EXIT_DEAD 32 150 /* in tsk->state again */ 151 #define TASK_NONINTERACTIVE 64 152 #define TASK_DEAD 128 153 154 #define __set_task_state(tsk, state_value) \ 155 do { (tsk)->state = (state_value); } while (0) 156 #define set_task_state(tsk, state_value) \ 157 set_mb((tsk)->state, (state_value)) 158 159 /* 160 * set_current_state() includes a barrier so that the write of current->state 161 * is correctly serialised wrt the caller's subsequent test of whether to 162 * actually sleep: 163 * 164 * set_current_state(TASK_UNINTERRUPTIBLE); 165 * if (do_i_need_to_sleep()) 166 * schedule(); 167 * 168 * If the caller does not need such serialisation then use __set_current_state() 169 */ 170 #define __set_current_state(state_value) \ 171 do { current->state = (state_value); } while (0) 172 #define set_current_state(state_value) \ 173 set_mb(current->state, (state_value)) 174 175 /* Task command name length */ 176 #define TASK_COMM_LEN 16 177 178 #include <linux/spinlock.h> 179 180 /* 181 * This serializes "schedule()" and also protects 182 * the run-queue from deletions/modifications (but 183 * _adding_ to the beginning of the run-queue has 184 * a separate lock). 185 */ 186 extern rwlock_t tasklist_lock; 187 extern spinlock_t mmlist_lock; 188 189 struct task_struct; 190 191 extern void sched_init(void); 192 extern void sched_init_smp(void); 193 extern void init_idle(struct task_struct *idle, int cpu); 194 195 extern cpumask_t nohz_cpu_mask; 196 197 /* 198 * Only dump TASK_* tasks. (-1 for all tasks) 199 */ 200 extern void show_state_filter(unsigned long state_filter); 201 202 static inline void show_state(void) 203 { 204 show_state_filter(-1); 205 } 206 207 extern void show_regs(struct pt_regs *); 208 209 /* 210 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 211 * task), SP is the stack pointer of the first frame that should be shown in the back 212 * trace (or NULL if the entire call-chain of the task should be shown). 213 */ 214 extern void show_stack(struct task_struct *task, unsigned long *sp); 215 216 void io_schedule(void); 217 long io_schedule_timeout(long timeout); 218 219 extern void cpu_init (void); 220 extern void trap_init(void); 221 extern void update_process_times(int user); 222 extern void scheduler_tick(void); 223 224 #ifdef CONFIG_DETECT_SOFTLOCKUP 225 extern void softlockup_tick(void); 226 extern void spawn_softlockup_task(void); 227 extern void touch_softlockup_watchdog(void); 228 #else 229 static inline void softlockup_tick(void) 230 { 231 } 232 static inline void spawn_softlockup_task(void) 233 { 234 } 235 static inline void touch_softlockup_watchdog(void) 236 { 237 } 238 #endif 239 240 241 /* Attach to any functions which should be ignored in wchan output. */ 242 #define __sched __attribute__((__section__(".sched.text"))) 243 /* Is this address in the __sched functions? */ 244 extern int in_sched_functions(unsigned long addr); 245 246 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 247 extern signed long FASTCALL(schedule_timeout(signed long timeout)); 248 extern signed long schedule_timeout_interruptible(signed long timeout); 249 extern signed long schedule_timeout_uninterruptible(signed long timeout); 250 asmlinkage void schedule(void); 251 252 struct nsproxy; 253 254 /* Maximum number of active map areas.. This is a random (large) number */ 255 #define DEFAULT_MAX_MAP_COUNT 65536 256 257 extern int sysctl_max_map_count; 258 259 #include <linux/aio.h> 260 261 extern unsigned long 262 arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 263 unsigned long, unsigned long); 264 extern unsigned long 265 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 266 unsigned long len, unsigned long pgoff, 267 unsigned long flags); 268 extern void arch_unmap_area(struct mm_struct *, unsigned long); 269 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 270 271 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS 272 /* 273 * The mm counters are not protected by its page_table_lock, 274 * so must be incremented atomically. 275 */ 276 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 277 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 278 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 279 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 280 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 281 typedef atomic_long_t mm_counter_t; 282 283 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 284 /* 285 * The mm counters are protected by its page_table_lock, 286 * so can be incremented directly. 287 */ 288 #define set_mm_counter(mm, member, value) (mm)->_##member = (value) 289 #define get_mm_counter(mm, member) ((mm)->_##member) 290 #define add_mm_counter(mm, member, value) (mm)->_##member += (value) 291 #define inc_mm_counter(mm, member) (mm)->_##member++ 292 #define dec_mm_counter(mm, member) (mm)->_##member-- 293 typedef unsigned long mm_counter_t; 294 295 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 296 297 #define get_mm_rss(mm) \ 298 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 299 #define update_hiwater_rss(mm) do { \ 300 unsigned long _rss = get_mm_rss(mm); \ 301 if ((mm)->hiwater_rss < _rss) \ 302 (mm)->hiwater_rss = _rss; \ 303 } while (0) 304 #define update_hiwater_vm(mm) do { \ 305 if ((mm)->hiwater_vm < (mm)->total_vm) \ 306 (mm)->hiwater_vm = (mm)->total_vm; \ 307 } while (0) 308 309 struct mm_struct { 310 struct vm_area_struct * mmap; /* list of VMAs */ 311 struct rb_root mm_rb; 312 struct vm_area_struct * mmap_cache; /* last find_vma result */ 313 unsigned long (*get_unmapped_area) (struct file *filp, 314 unsigned long addr, unsigned long len, 315 unsigned long pgoff, unsigned long flags); 316 void (*unmap_area) (struct mm_struct *mm, unsigned long addr); 317 unsigned long mmap_base; /* base of mmap area */ 318 unsigned long task_size; /* size of task vm space */ 319 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */ 320 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */ 321 pgd_t * pgd; 322 atomic_t mm_users; /* How many users with user space? */ 323 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */ 324 int map_count; /* number of VMAs */ 325 struct rw_semaphore mmap_sem; 326 spinlock_t page_table_lock; /* Protects page tables and some counters */ 327 328 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung 329 * together off init_mm.mmlist, and are protected 330 * by mmlist_lock 331 */ 332 333 /* Special counters, in some configurations protected by the 334 * page_table_lock, in other configurations by being atomic. 335 */ 336 mm_counter_t _file_rss; 337 mm_counter_t _anon_rss; 338 339 unsigned long hiwater_rss; /* High-watermark of RSS usage */ 340 unsigned long hiwater_vm; /* High-water virtual memory usage */ 341 342 unsigned long total_vm, locked_vm, shared_vm, exec_vm; 343 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes; 344 unsigned long start_code, end_code, start_data, end_data; 345 unsigned long start_brk, brk, start_stack; 346 unsigned long arg_start, arg_end, env_start, env_end; 347 348 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */ 349 350 cpumask_t cpu_vm_mask; 351 352 /* Architecture-specific MM context */ 353 mm_context_t context; 354 355 /* Swap token stuff */ 356 /* 357 * Last value of global fault stamp as seen by this process. 358 * In other words, this value gives an indication of how long 359 * it has been since this task got the token. 360 * Look at mm/thrash.c 361 */ 362 unsigned int faultstamp; 363 unsigned int token_priority; 364 unsigned int last_interval; 365 366 unsigned char dumpable:2; 367 368 /* coredumping support */ 369 int core_waiters; 370 struct completion *core_startup_done, core_done; 371 372 /* aio bits */ 373 rwlock_t ioctx_list_lock; 374 struct kioctx *ioctx_list; 375 }; 376 377 struct sighand_struct { 378 atomic_t count; 379 struct k_sigaction action[_NSIG]; 380 spinlock_t siglock; 381 }; 382 383 struct pacct_struct { 384 int ac_flag; 385 long ac_exitcode; 386 unsigned long ac_mem; 387 cputime_t ac_utime, ac_stime; 388 unsigned long ac_minflt, ac_majflt; 389 }; 390 391 /* 392 * NOTE! "signal_struct" does not have it's own 393 * locking, because a shared signal_struct always 394 * implies a shared sighand_struct, so locking 395 * sighand_struct is always a proper superset of 396 * the locking of signal_struct. 397 */ 398 struct signal_struct { 399 atomic_t count; 400 atomic_t live; 401 402 wait_queue_head_t wait_chldexit; /* for wait4() */ 403 404 /* current thread group signal load-balancing target: */ 405 struct task_struct *curr_target; 406 407 /* shared signal handling: */ 408 struct sigpending shared_pending; 409 410 /* thread group exit support */ 411 int group_exit_code; 412 /* overloaded: 413 * - notify group_exit_task when ->count is equal to notify_count 414 * - everyone except group_exit_task is stopped during signal delivery 415 * of fatal signals, group_exit_task processes the signal. 416 */ 417 struct task_struct *group_exit_task; 418 int notify_count; 419 420 /* thread group stop support, overloads group_exit_code too */ 421 int group_stop_count; 422 unsigned int flags; /* see SIGNAL_* flags below */ 423 424 /* POSIX.1b Interval Timers */ 425 struct list_head posix_timers; 426 427 /* ITIMER_REAL timer for the process */ 428 struct hrtimer real_timer; 429 struct task_struct *tsk; 430 ktime_t it_real_incr; 431 432 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 433 cputime_t it_prof_expires, it_virt_expires; 434 cputime_t it_prof_incr, it_virt_incr; 435 436 /* job control IDs */ 437 pid_t pgrp; 438 pid_t tty_old_pgrp; 439 440 union { 441 pid_t session __deprecated; 442 pid_t __session; 443 }; 444 445 /* boolean value for session group leader */ 446 int leader; 447 448 struct tty_struct *tty; /* NULL if no tty */ 449 450 /* 451 * Cumulative resource counters for dead threads in the group, 452 * and for reaped dead child processes forked by this group. 453 * Live threads maintain their own counters and add to these 454 * in __exit_signal, except for the group leader. 455 */ 456 cputime_t utime, stime, cutime, cstime; 457 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 458 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 459 460 /* 461 * Cumulative ns of scheduled CPU time for dead threads in the 462 * group, not including a zombie group leader. (This only differs 463 * from jiffies_to_ns(utime + stime) if sched_clock uses something 464 * other than jiffies.) 465 */ 466 unsigned long long sched_time; 467 468 /* 469 * We don't bother to synchronize most readers of this at all, 470 * because there is no reader checking a limit that actually needs 471 * to get both rlim_cur and rlim_max atomically, and either one 472 * alone is a single word that can safely be read normally. 473 * getrlimit/setrlimit use task_lock(current->group_leader) to 474 * protect this instead of the siglock, because they really 475 * have no need to disable irqs. 476 */ 477 struct rlimit rlim[RLIM_NLIMITS]; 478 479 struct list_head cpu_timers[3]; 480 481 /* keep the process-shared keyrings here so that they do the right 482 * thing in threads created with CLONE_THREAD */ 483 #ifdef CONFIG_KEYS 484 struct key *session_keyring; /* keyring inherited over fork */ 485 struct key *process_keyring; /* keyring private to this process */ 486 #endif 487 #ifdef CONFIG_BSD_PROCESS_ACCT 488 struct pacct_struct pacct; /* per-process accounting information */ 489 #endif 490 #ifdef CONFIG_TASKSTATS 491 struct taskstats *stats; 492 #endif 493 }; 494 495 /* Context switch must be unlocked if interrupts are to be enabled */ 496 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 497 # define __ARCH_WANT_UNLOCKED_CTXSW 498 #endif 499 500 /* 501 * Bits in flags field of signal_struct. 502 */ 503 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 504 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 505 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 506 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 507 508 509 /* 510 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 511 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 512 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 513 * values are inverted: lower p->prio value means higher priority. 514 * 515 * The MAX_USER_RT_PRIO value allows the actual maximum 516 * RT priority to be separate from the value exported to 517 * user-space. This allows kernel threads to set their 518 * priority to a value higher than any user task. Note: 519 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 520 */ 521 522 #define MAX_USER_RT_PRIO 100 523 #define MAX_RT_PRIO MAX_USER_RT_PRIO 524 525 #define MAX_PRIO (MAX_RT_PRIO + 40) 526 527 #define rt_prio(prio) unlikely((prio) < MAX_RT_PRIO) 528 #define rt_task(p) rt_prio((p)->prio) 529 #define batch_task(p) (unlikely((p)->policy == SCHED_BATCH)) 530 #define is_rt_policy(p) ((p) != SCHED_NORMAL && (p) != SCHED_BATCH) 531 #define has_rt_policy(p) unlikely(is_rt_policy((p)->policy)) 532 533 /* 534 * Some day this will be a full-fledged user tracking system.. 535 */ 536 struct user_struct { 537 atomic_t __count; /* reference count */ 538 atomic_t processes; /* How many processes does this user have? */ 539 atomic_t files; /* How many open files does this user have? */ 540 atomic_t sigpending; /* How many pending signals does this user have? */ 541 #ifdef CONFIG_INOTIFY_USER 542 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 543 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 544 #endif 545 /* protected by mq_lock */ 546 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 547 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 548 549 #ifdef CONFIG_KEYS 550 struct key *uid_keyring; /* UID specific keyring */ 551 struct key *session_keyring; /* UID's default session keyring */ 552 #endif 553 554 /* Hash table maintenance information */ 555 struct list_head uidhash_list; 556 uid_t uid; 557 }; 558 559 extern struct user_struct *find_user(uid_t); 560 561 extern struct user_struct root_user; 562 #define INIT_USER (&root_user) 563 564 struct backing_dev_info; 565 struct reclaim_state; 566 567 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 568 struct sched_info { 569 /* cumulative counters */ 570 unsigned long cpu_time, /* time spent on the cpu */ 571 run_delay, /* time spent waiting on a runqueue */ 572 pcnt; /* # of timeslices run on this cpu */ 573 574 /* timestamps */ 575 unsigned long last_arrival, /* when we last ran on a cpu */ 576 last_queued; /* when we were last queued to run */ 577 }; 578 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 579 580 #ifdef CONFIG_SCHEDSTATS 581 extern const struct file_operations proc_schedstat_operations; 582 #endif /* CONFIG_SCHEDSTATS */ 583 584 #ifdef CONFIG_TASK_DELAY_ACCT 585 struct task_delay_info { 586 spinlock_t lock; 587 unsigned int flags; /* Private per-task flags */ 588 589 /* For each stat XXX, add following, aligned appropriately 590 * 591 * struct timespec XXX_start, XXX_end; 592 * u64 XXX_delay; 593 * u32 XXX_count; 594 * 595 * Atomicity of updates to XXX_delay, XXX_count protected by 596 * single lock above (split into XXX_lock if contention is an issue). 597 */ 598 599 /* 600 * XXX_count is incremented on every XXX operation, the delay 601 * associated with the operation is added to XXX_delay. 602 * XXX_delay contains the accumulated delay time in nanoseconds. 603 */ 604 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 605 u64 blkio_delay; /* wait for sync block io completion */ 606 u64 swapin_delay; /* wait for swapin block io completion */ 607 u32 blkio_count; /* total count of the number of sync block */ 608 /* io operations performed */ 609 u32 swapin_count; /* total count of the number of swapin block */ 610 /* io operations performed */ 611 }; 612 #endif /* CONFIG_TASK_DELAY_ACCT */ 613 614 static inline int sched_info_on(void) 615 { 616 #ifdef CONFIG_SCHEDSTATS 617 return 1; 618 #elif defined(CONFIG_TASK_DELAY_ACCT) 619 extern int delayacct_on; 620 return delayacct_on; 621 #else 622 return 0; 623 #endif 624 } 625 626 enum idle_type 627 { 628 SCHED_IDLE, 629 NOT_IDLE, 630 NEWLY_IDLE, 631 MAX_IDLE_TYPES 632 }; 633 634 /* 635 * sched-domains (multiprocessor balancing) declarations: 636 */ 637 #define SCHED_LOAD_SCALE 128UL /* increase resolution of load */ 638 639 #ifdef CONFIG_SMP 640 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 641 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 642 #define SD_BALANCE_EXEC 4 /* Balance on exec */ 643 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 644 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 645 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 646 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 647 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 648 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 649 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 650 651 #define BALANCE_FOR_MC_POWER \ 652 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0) 653 654 #define BALANCE_FOR_PKG_POWER \ 655 ((sched_mc_power_savings || sched_smt_power_savings) ? \ 656 SD_POWERSAVINGS_BALANCE : 0) 657 658 #define test_sd_parent(sd, flag) ((sd->parent && \ 659 (sd->parent->flags & flag)) ? 1 : 0) 660 661 662 struct sched_group { 663 struct sched_group *next; /* Must be a circular list */ 664 cpumask_t cpumask; 665 666 /* 667 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 668 * single CPU. This is read only (except for setup, hotplug CPU). 669 */ 670 unsigned long cpu_power; 671 }; 672 673 struct sched_domain { 674 /* These fields must be setup */ 675 struct sched_domain *parent; /* top domain must be null terminated */ 676 struct sched_domain *child; /* bottom domain must be null terminated */ 677 struct sched_group *groups; /* the balancing groups of the domain */ 678 cpumask_t span; /* span of all CPUs in this domain */ 679 unsigned long min_interval; /* Minimum balance interval ms */ 680 unsigned long max_interval; /* Maximum balance interval ms */ 681 unsigned int busy_factor; /* less balancing by factor if busy */ 682 unsigned int imbalance_pct; /* No balance until over watermark */ 683 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */ 684 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 685 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */ 686 unsigned int busy_idx; 687 unsigned int idle_idx; 688 unsigned int newidle_idx; 689 unsigned int wake_idx; 690 unsigned int forkexec_idx; 691 int flags; /* See SD_* */ 692 693 /* Runtime fields. */ 694 unsigned long last_balance; /* init to jiffies. units in jiffies */ 695 unsigned int balance_interval; /* initialise to 1. units in ms. */ 696 unsigned int nr_balance_failed; /* initialise to 0 */ 697 698 #ifdef CONFIG_SCHEDSTATS 699 /* load_balance() stats */ 700 unsigned long lb_cnt[MAX_IDLE_TYPES]; 701 unsigned long lb_failed[MAX_IDLE_TYPES]; 702 unsigned long lb_balanced[MAX_IDLE_TYPES]; 703 unsigned long lb_imbalance[MAX_IDLE_TYPES]; 704 unsigned long lb_gained[MAX_IDLE_TYPES]; 705 unsigned long lb_hot_gained[MAX_IDLE_TYPES]; 706 unsigned long lb_nobusyg[MAX_IDLE_TYPES]; 707 unsigned long lb_nobusyq[MAX_IDLE_TYPES]; 708 709 /* Active load balancing */ 710 unsigned long alb_cnt; 711 unsigned long alb_failed; 712 unsigned long alb_pushed; 713 714 /* SD_BALANCE_EXEC stats */ 715 unsigned long sbe_cnt; 716 unsigned long sbe_balanced; 717 unsigned long sbe_pushed; 718 719 /* SD_BALANCE_FORK stats */ 720 unsigned long sbf_cnt; 721 unsigned long sbf_balanced; 722 unsigned long sbf_pushed; 723 724 /* try_to_wake_up() stats */ 725 unsigned long ttwu_wake_remote; 726 unsigned long ttwu_move_affine; 727 unsigned long ttwu_move_balance; 728 #endif 729 }; 730 731 extern int partition_sched_domains(cpumask_t *partition1, 732 cpumask_t *partition2); 733 734 /* 735 * Maximum cache size the migration-costs auto-tuning code will 736 * search from: 737 */ 738 extern unsigned int max_cache_size; 739 740 #endif /* CONFIG_SMP */ 741 742 743 struct io_context; /* See blkdev.h */ 744 struct cpuset; 745 746 #define NGROUPS_SMALL 32 747 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t))) 748 struct group_info { 749 int ngroups; 750 atomic_t usage; 751 gid_t small_block[NGROUPS_SMALL]; 752 int nblocks; 753 gid_t *blocks[0]; 754 }; 755 756 /* 757 * get_group_info() must be called with the owning task locked (via task_lock()) 758 * when task != current. The reason being that the vast majority of callers are 759 * looking at current->group_info, which can not be changed except by the 760 * current task. Changing current->group_info requires the task lock, too. 761 */ 762 #define get_group_info(group_info) do { \ 763 atomic_inc(&(group_info)->usage); \ 764 } while (0) 765 766 #define put_group_info(group_info) do { \ 767 if (atomic_dec_and_test(&(group_info)->usage)) \ 768 groups_free(group_info); \ 769 } while (0) 770 771 extern struct group_info *groups_alloc(int gidsetsize); 772 extern void groups_free(struct group_info *group_info); 773 extern int set_current_groups(struct group_info *group_info); 774 extern int groups_search(struct group_info *group_info, gid_t grp); 775 /* access the groups "array" with this macro */ 776 #define GROUP_AT(gi, i) \ 777 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 778 779 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 780 extern void prefetch_stack(struct task_struct *t); 781 #else 782 static inline void prefetch_stack(struct task_struct *t) { } 783 #endif 784 785 struct audit_context; /* See audit.c */ 786 struct mempolicy; 787 struct pipe_inode_info; 788 struct uts_namespace; 789 790 enum sleep_type { 791 SLEEP_NORMAL, 792 SLEEP_NONINTERACTIVE, 793 SLEEP_INTERACTIVE, 794 SLEEP_INTERRUPTED, 795 }; 796 797 struct prio_array; 798 799 struct task_struct { 800 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 801 struct thread_info *thread_info; 802 atomic_t usage; 803 unsigned long flags; /* per process flags, defined below */ 804 unsigned long ptrace; 805 806 int lock_depth; /* BKL lock depth */ 807 808 #ifdef CONFIG_SMP 809 #ifdef __ARCH_WANT_UNLOCKED_CTXSW 810 int oncpu; 811 #endif 812 #endif 813 int load_weight; /* for niceness load balancing purposes */ 814 int prio, static_prio, normal_prio; 815 struct list_head run_list; 816 struct prio_array *array; 817 818 unsigned short ioprio; 819 #ifdef CONFIG_BLK_DEV_IO_TRACE 820 unsigned int btrace_seq; 821 #endif 822 unsigned long sleep_avg; 823 unsigned long long timestamp, last_ran; 824 unsigned long long sched_time; /* sched_clock time spent running */ 825 enum sleep_type sleep_type; 826 827 unsigned long policy; 828 cpumask_t cpus_allowed; 829 unsigned int time_slice, first_time_slice; 830 831 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 832 struct sched_info sched_info; 833 #endif 834 835 struct list_head tasks; 836 /* 837 * ptrace_list/ptrace_children forms the list of my children 838 * that were stolen by a ptracer. 839 */ 840 struct list_head ptrace_children; 841 struct list_head ptrace_list; 842 843 struct mm_struct *mm, *active_mm; 844 845 /* task state */ 846 struct linux_binfmt *binfmt; 847 long exit_state; 848 int exit_code, exit_signal; 849 int pdeath_signal; /* The signal sent when the parent dies */ 850 /* ??? */ 851 unsigned long personality; 852 unsigned did_exec:1; 853 pid_t pid; 854 pid_t tgid; 855 856 #ifdef CONFIG_CC_STACKPROTECTOR 857 /* Canary value for the -fstack-protector gcc feature */ 858 unsigned long stack_canary; 859 #endif 860 /* 861 * pointers to (original) parent process, youngest child, younger sibling, 862 * older sibling, respectively. (p->father can be replaced with 863 * p->parent->pid) 864 */ 865 struct task_struct *real_parent; /* real parent process (when being debugged) */ 866 struct task_struct *parent; /* parent process */ 867 /* 868 * children/sibling forms the list of my children plus the 869 * tasks I'm ptracing. 870 */ 871 struct list_head children; /* list of my children */ 872 struct list_head sibling; /* linkage in my parent's children list */ 873 struct task_struct *group_leader; /* threadgroup leader */ 874 875 /* PID/PID hash table linkage. */ 876 struct pid_link pids[PIDTYPE_MAX]; 877 struct list_head thread_group; 878 879 struct completion *vfork_done; /* for vfork() */ 880 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 881 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 882 883 unsigned long rt_priority; 884 cputime_t utime, stime; 885 unsigned long nvcsw, nivcsw; /* context switch counts */ 886 struct timespec start_time; 887 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 888 unsigned long min_flt, maj_flt; 889 890 cputime_t it_prof_expires, it_virt_expires; 891 unsigned long long it_sched_expires; 892 struct list_head cpu_timers[3]; 893 894 /* process credentials */ 895 uid_t uid,euid,suid,fsuid; 896 gid_t gid,egid,sgid,fsgid; 897 struct group_info *group_info; 898 kernel_cap_t cap_effective, cap_inheritable, cap_permitted; 899 unsigned keep_capabilities:1; 900 struct user_struct *user; 901 #ifdef CONFIG_KEYS 902 struct key *request_key_auth; /* assumed request_key authority */ 903 struct key *thread_keyring; /* keyring private to this thread */ 904 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 905 #endif 906 /* 907 * fpu_counter contains the number of consecutive context switches 908 * that the FPU is used. If this is over a threshold, the lazy fpu 909 * saving becomes unlazy to save the trap. This is an unsigned char 910 * so that after 256 times the counter wraps and the behavior turns 911 * lazy again; this to deal with bursty apps that only use FPU for 912 * a short time 913 */ 914 unsigned char fpu_counter; 915 int oomkilladj; /* OOM kill score adjustment (bit shift). */ 916 char comm[TASK_COMM_LEN]; /* executable name excluding path 917 - access with [gs]et_task_comm (which lock 918 it with task_lock()) 919 - initialized normally by flush_old_exec */ 920 /* file system info */ 921 int link_count, total_link_count; 922 #ifdef CONFIG_SYSVIPC 923 /* ipc stuff */ 924 struct sysv_sem sysvsem; 925 #endif 926 /* CPU-specific state of this task */ 927 struct thread_struct thread; 928 /* filesystem information */ 929 struct fs_struct *fs; 930 /* open file information */ 931 struct files_struct *files; 932 /* namespaces */ 933 struct nsproxy *nsproxy; 934 /* signal handlers */ 935 struct signal_struct *signal; 936 struct sighand_struct *sighand; 937 938 sigset_t blocked, real_blocked; 939 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */ 940 struct sigpending pending; 941 942 unsigned long sas_ss_sp; 943 size_t sas_ss_size; 944 int (*notifier)(void *priv); 945 void *notifier_data; 946 sigset_t *notifier_mask; 947 948 void *security; 949 struct audit_context *audit_context; 950 seccomp_t seccomp; 951 952 /* Thread group tracking */ 953 u32 parent_exec_id; 954 u32 self_exec_id; 955 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 956 spinlock_t alloc_lock; 957 958 /* Protection of the PI data structures: */ 959 spinlock_t pi_lock; 960 961 #ifdef CONFIG_RT_MUTEXES 962 /* PI waiters blocked on a rt_mutex held by this task */ 963 struct plist_head pi_waiters; 964 /* Deadlock detection and priority inheritance handling */ 965 struct rt_mutex_waiter *pi_blocked_on; 966 #endif 967 968 #ifdef CONFIG_DEBUG_MUTEXES 969 /* mutex deadlock detection */ 970 struct mutex_waiter *blocked_on; 971 #endif 972 #ifdef CONFIG_TRACE_IRQFLAGS 973 unsigned int irq_events; 974 int hardirqs_enabled; 975 unsigned long hardirq_enable_ip; 976 unsigned int hardirq_enable_event; 977 unsigned long hardirq_disable_ip; 978 unsigned int hardirq_disable_event; 979 int softirqs_enabled; 980 unsigned long softirq_disable_ip; 981 unsigned int softirq_disable_event; 982 unsigned long softirq_enable_ip; 983 unsigned int softirq_enable_event; 984 int hardirq_context; 985 int softirq_context; 986 #endif 987 #ifdef CONFIG_LOCKDEP 988 # define MAX_LOCK_DEPTH 30UL 989 u64 curr_chain_key; 990 int lockdep_depth; 991 struct held_lock held_locks[MAX_LOCK_DEPTH]; 992 unsigned int lockdep_recursion; 993 #endif 994 995 /* journalling filesystem info */ 996 void *journal_info; 997 998 /* VM state */ 999 struct reclaim_state *reclaim_state; 1000 1001 struct backing_dev_info *backing_dev_info; 1002 1003 struct io_context *io_context; 1004 1005 unsigned long ptrace_message; 1006 siginfo_t *last_siginfo; /* For ptrace use. */ 1007 /* 1008 * current io wait handle: wait queue entry to use for io waits 1009 * If this thread is processing aio, this points at the waitqueue 1010 * inside the currently handled kiocb. It may be NULL (i.e. default 1011 * to a stack based synchronous wait) if its doing sync IO. 1012 */ 1013 wait_queue_t *io_wait; 1014 /* i/o counters(bytes read/written, #syscalls */ 1015 u64 rchar, wchar, syscr, syscw; 1016 #if defined(CONFIG_TASK_XACCT) 1017 u64 acct_rss_mem1; /* accumulated rss usage */ 1018 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1019 cputime_t acct_stimexpd;/* stime since last update */ 1020 #endif 1021 #ifdef CONFIG_NUMA 1022 struct mempolicy *mempolicy; 1023 short il_next; 1024 #endif 1025 #ifdef CONFIG_CPUSETS 1026 struct cpuset *cpuset; 1027 nodemask_t mems_allowed; 1028 int cpuset_mems_generation; 1029 int cpuset_mem_spread_rotor; 1030 #endif 1031 struct robust_list_head __user *robust_list; 1032 #ifdef CONFIG_COMPAT 1033 struct compat_robust_list_head __user *compat_robust_list; 1034 #endif 1035 struct list_head pi_state_list; 1036 struct futex_pi_state *pi_state_cache; 1037 1038 atomic_t fs_excl; /* holding fs exclusive resources */ 1039 struct rcu_head rcu; 1040 1041 /* 1042 * cache last used pipe for splice 1043 */ 1044 struct pipe_inode_info *splice_pipe; 1045 #ifdef CONFIG_TASK_DELAY_ACCT 1046 struct task_delay_info *delays; 1047 #endif 1048 #ifdef CONFIG_FAULT_INJECTION 1049 int make_it_fail; 1050 #endif 1051 }; 1052 1053 static inline pid_t process_group(struct task_struct *tsk) 1054 { 1055 return tsk->signal->pgrp; 1056 } 1057 1058 static inline pid_t signal_session(struct signal_struct *sig) 1059 { 1060 return sig->__session; 1061 } 1062 1063 static inline pid_t process_session(struct task_struct *tsk) 1064 { 1065 return signal_session(tsk->signal); 1066 } 1067 1068 static inline void set_signal_session(struct signal_struct *sig, pid_t session) 1069 { 1070 sig->__session = session; 1071 } 1072 1073 static inline struct pid *task_pid(struct task_struct *task) 1074 { 1075 return task->pids[PIDTYPE_PID].pid; 1076 } 1077 1078 static inline struct pid *task_tgid(struct task_struct *task) 1079 { 1080 return task->group_leader->pids[PIDTYPE_PID].pid; 1081 } 1082 1083 static inline struct pid *task_pgrp(struct task_struct *task) 1084 { 1085 return task->group_leader->pids[PIDTYPE_PGID].pid; 1086 } 1087 1088 static inline struct pid *task_session(struct task_struct *task) 1089 { 1090 return task->group_leader->pids[PIDTYPE_SID].pid; 1091 } 1092 1093 /** 1094 * pid_alive - check that a task structure is not stale 1095 * @p: Task structure to be checked. 1096 * 1097 * Test if a process is not yet dead (at most zombie state) 1098 * If pid_alive fails, then pointers within the task structure 1099 * can be stale and must not be dereferenced. 1100 */ 1101 static inline int pid_alive(struct task_struct *p) 1102 { 1103 return p->pids[PIDTYPE_PID].pid != NULL; 1104 } 1105 1106 /** 1107 * is_init - check if a task structure is init 1108 * @tsk: Task structure to be checked. 1109 * 1110 * Check if a task structure is the first user space task the kernel created. 1111 */ 1112 static inline int is_init(struct task_struct *tsk) 1113 { 1114 return tsk->pid == 1; 1115 } 1116 1117 extern struct pid *cad_pid; 1118 1119 extern void free_task(struct task_struct *tsk); 1120 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1121 1122 extern void __put_task_struct(struct task_struct *t); 1123 1124 static inline void put_task_struct(struct task_struct *t) 1125 { 1126 if (atomic_dec_and_test(&t->usage)) 1127 __put_task_struct(t); 1128 } 1129 1130 /* 1131 * Per process flags 1132 */ 1133 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1134 /* Not implemented yet, only for 486*/ 1135 #define PF_STARTING 0x00000002 /* being created */ 1136 #define PF_EXITING 0x00000004 /* getting shut down */ 1137 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1138 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1139 #define PF_DUMPCORE 0x00000200 /* dumped core */ 1140 #define PF_SIGNALED 0x00000400 /* killed by a signal */ 1141 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1142 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1143 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1144 #define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */ 1145 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1146 #define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1147 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1148 #define PF_KSWAPD 0x00040000 /* I am kswapd */ 1149 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1150 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1151 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */ 1152 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1153 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1154 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1155 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1156 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1157 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1158 1159 /* 1160 * Only the _current_ task can read/write to tsk->flags, but other 1161 * tasks can access tsk->flags in readonly mode for example 1162 * with tsk_used_math (like during threaded core dumping). 1163 * There is however an exception to this rule during ptrace 1164 * or during fork: the ptracer task is allowed to write to the 1165 * child->flags of its traced child (same goes for fork, the parent 1166 * can write to the child->flags), because we're guaranteed the 1167 * child is not running and in turn not changing child->flags 1168 * at the same time the parent does it. 1169 */ 1170 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1171 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1172 #define clear_used_math() clear_stopped_child_used_math(current) 1173 #define set_used_math() set_stopped_child_used_math(current) 1174 #define conditional_stopped_child_used_math(condition, child) \ 1175 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1176 #define conditional_used_math(condition) \ 1177 conditional_stopped_child_used_math(condition, current) 1178 #define copy_to_stopped_child_used_math(child) \ 1179 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1180 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1181 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1182 #define used_math() tsk_used_math(current) 1183 1184 #ifdef CONFIG_SMP 1185 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask); 1186 #else 1187 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1188 { 1189 if (!cpu_isset(0, new_mask)) 1190 return -EINVAL; 1191 return 0; 1192 } 1193 #endif 1194 1195 extern unsigned long long sched_clock(void); 1196 extern unsigned long long 1197 current_sched_time(const struct task_struct *current_task); 1198 1199 /* sched_exec is called by processes performing an exec */ 1200 #ifdef CONFIG_SMP 1201 extern void sched_exec(void); 1202 #else 1203 #define sched_exec() {} 1204 #endif 1205 1206 #ifdef CONFIG_HOTPLUG_CPU 1207 extern void idle_task_exit(void); 1208 #else 1209 static inline void idle_task_exit(void) {} 1210 #endif 1211 1212 extern void sched_idle_next(void); 1213 1214 #ifdef CONFIG_RT_MUTEXES 1215 extern int rt_mutex_getprio(struct task_struct *p); 1216 extern void rt_mutex_setprio(struct task_struct *p, int prio); 1217 extern void rt_mutex_adjust_pi(struct task_struct *p); 1218 #else 1219 static inline int rt_mutex_getprio(struct task_struct *p) 1220 { 1221 return p->normal_prio; 1222 } 1223 # define rt_mutex_adjust_pi(p) do { } while (0) 1224 #endif 1225 1226 extern void set_user_nice(struct task_struct *p, long nice); 1227 extern int task_prio(const struct task_struct *p); 1228 extern int task_nice(const struct task_struct *p); 1229 extern int can_nice(const struct task_struct *p, const int nice); 1230 extern int task_curr(const struct task_struct *p); 1231 extern int idle_cpu(int cpu); 1232 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1233 extern struct task_struct *idle_task(int cpu); 1234 extern struct task_struct *curr_task(int cpu); 1235 extern void set_curr_task(int cpu, struct task_struct *p); 1236 1237 void yield(void); 1238 1239 /* 1240 * The default (Linux) execution domain. 1241 */ 1242 extern struct exec_domain default_exec_domain; 1243 1244 union thread_union { 1245 struct thread_info thread_info; 1246 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1247 }; 1248 1249 #ifndef __HAVE_ARCH_KSTACK_END 1250 static inline int kstack_end(void *addr) 1251 { 1252 /* Reliable end of stack detection: 1253 * Some APM bios versions misalign the stack 1254 */ 1255 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1256 } 1257 #endif 1258 1259 extern union thread_union init_thread_union; 1260 extern struct task_struct init_task; 1261 1262 extern struct mm_struct init_mm; 1263 1264 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr) 1265 extern struct task_struct *find_task_by_pid_type(int type, int pid); 1266 extern void __set_special_pids(pid_t session, pid_t pgrp); 1267 1268 /* per-UID process charging. */ 1269 extern struct user_struct * alloc_uid(uid_t); 1270 static inline struct user_struct *get_uid(struct user_struct *u) 1271 { 1272 atomic_inc(&u->__count); 1273 return u; 1274 } 1275 extern void free_uid(struct user_struct *); 1276 extern void switch_uid(struct user_struct *); 1277 1278 #include <asm/current.h> 1279 1280 extern void do_timer(unsigned long ticks); 1281 1282 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state)); 1283 extern int FASTCALL(wake_up_process(struct task_struct * tsk)); 1284 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk, 1285 unsigned long clone_flags)); 1286 #ifdef CONFIG_SMP 1287 extern void kick_process(struct task_struct *tsk); 1288 #else 1289 static inline void kick_process(struct task_struct *tsk) { } 1290 #endif 1291 extern void FASTCALL(sched_fork(struct task_struct * p, int clone_flags)); 1292 extern void FASTCALL(sched_exit(struct task_struct * p)); 1293 1294 extern int in_group_p(gid_t); 1295 extern int in_egroup_p(gid_t); 1296 1297 extern void proc_caches_init(void); 1298 extern void flush_signals(struct task_struct *); 1299 extern void flush_signal_handlers(struct task_struct *, int force_default); 1300 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1301 1302 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1303 { 1304 unsigned long flags; 1305 int ret; 1306 1307 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1308 ret = dequeue_signal(tsk, mask, info); 1309 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1310 1311 return ret; 1312 } 1313 1314 extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1315 sigset_t *mask); 1316 extern void unblock_all_signals(void); 1317 extern void release_task(struct task_struct * p); 1318 extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1319 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *); 1320 extern int force_sigsegv(int, struct task_struct *); 1321 extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1322 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1323 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1324 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1325 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1326 extern int kill_pgrp(struct pid *pid, int sig, int priv); 1327 extern int kill_pid(struct pid *pid, int sig, int priv); 1328 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp); 1329 extern int kill_pg_info(int, struct siginfo *, pid_t); 1330 extern void do_notify_parent(struct task_struct *, int); 1331 extern void force_sig(int, struct task_struct *); 1332 extern void force_sig_specific(int, struct task_struct *); 1333 extern int send_sig(int, struct task_struct *, int); 1334 extern void zap_other_threads(struct task_struct *p); 1335 extern int kill_pg(pid_t, int, int); 1336 extern int kill_proc(pid_t, int, int); 1337 extern struct sigqueue *sigqueue_alloc(void); 1338 extern void sigqueue_free(struct sigqueue *); 1339 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *); 1340 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *); 1341 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1342 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1343 1344 static inline int kill_cad_pid(int sig, int priv) 1345 { 1346 return kill_pid(cad_pid, sig, priv); 1347 } 1348 1349 /* These can be the second arg to send_sig_info/send_group_sig_info. */ 1350 #define SEND_SIG_NOINFO ((struct siginfo *) 0) 1351 #define SEND_SIG_PRIV ((struct siginfo *) 1) 1352 #define SEND_SIG_FORCED ((struct siginfo *) 2) 1353 1354 static inline int is_si_special(const struct siginfo *info) 1355 { 1356 return info <= SEND_SIG_FORCED; 1357 } 1358 1359 /* True if we are on the alternate signal stack. */ 1360 1361 static inline int on_sig_stack(unsigned long sp) 1362 { 1363 return (sp - current->sas_ss_sp < current->sas_ss_size); 1364 } 1365 1366 static inline int sas_ss_flags(unsigned long sp) 1367 { 1368 return (current->sas_ss_size == 0 ? SS_DISABLE 1369 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1370 } 1371 1372 /* 1373 * Routines for handling mm_structs 1374 */ 1375 extern struct mm_struct * mm_alloc(void); 1376 1377 /* mmdrop drops the mm and the page tables */ 1378 extern void FASTCALL(__mmdrop(struct mm_struct *)); 1379 static inline void mmdrop(struct mm_struct * mm) 1380 { 1381 if (atomic_dec_and_test(&mm->mm_count)) 1382 __mmdrop(mm); 1383 } 1384 1385 /* mmput gets rid of the mappings and all user-space */ 1386 extern void mmput(struct mm_struct *); 1387 /* Grab a reference to a task's mm, if it is not already going away */ 1388 extern struct mm_struct *get_task_mm(struct task_struct *task); 1389 /* Remove the current tasks stale references to the old mm_struct */ 1390 extern void mm_release(struct task_struct *, struct mm_struct *); 1391 1392 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1393 extern void flush_thread(void); 1394 extern void exit_thread(void); 1395 1396 extern void exit_files(struct task_struct *); 1397 extern void __cleanup_signal(struct signal_struct *); 1398 extern void __cleanup_sighand(struct sighand_struct *); 1399 extern void exit_itimers(struct signal_struct *); 1400 1401 extern NORET_TYPE void do_group_exit(int); 1402 1403 extern void daemonize(const char *, ...); 1404 extern int allow_signal(int); 1405 extern int disallow_signal(int); 1406 1407 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1408 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1409 struct task_struct *fork_idle(int); 1410 1411 extern void set_task_comm(struct task_struct *tsk, char *from); 1412 extern void get_task_comm(char *to, struct task_struct *tsk); 1413 1414 #ifdef CONFIG_SMP 1415 extern void wait_task_inactive(struct task_struct * p); 1416 #else 1417 #define wait_task_inactive(p) do { } while (0) 1418 #endif 1419 1420 #define remove_parent(p) list_del_init(&(p)->sibling) 1421 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children) 1422 1423 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1424 1425 #define for_each_process(p) \ 1426 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1427 1428 /* 1429 * Careful: do_each_thread/while_each_thread is a double loop so 1430 * 'break' will not work as expected - use goto instead. 1431 */ 1432 #define do_each_thread(g, t) \ 1433 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1434 1435 #define while_each_thread(g, t) \ 1436 while ((t = next_thread(t)) != g) 1437 1438 /* de_thread depends on thread_group_leader not being a pid based check */ 1439 #define thread_group_leader(p) (p == p->group_leader) 1440 1441 /* Do to the insanities of de_thread it is possible for a process 1442 * to have the pid of the thread group leader without actually being 1443 * the thread group leader. For iteration through the pids in proc 1444 * all we care about is that we have a task with the appropriate 1445 * pid, we don't actually care if we have the right task. 1446 */ 1447 static inline int has_group_leader_pid(struct task_struct *p) 1448 { 1449 return p->pid == p->tgid; 1450 } 1451 1452 static inline struct task_struct *next_thread(const struct task_struct *p) 1453 { 1454 return list_entry(rcu_dereference(p->thread_group.next), 1455 struct task_struct, thread_group); 1456 } 1457 1458 static inline int thread_group_empty(struct task_struct *p) 1459 { 1460 return list_empty(&p->thread_group); 1461 } 1462 1463 #define delay_group_leader(p) \ 1464 (thread_group_leader(p) && !thread_group_empty(p)) 1465 1466 /* 1467 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 1468 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1469 * pins the final release of task.io_context. Also protects ->cpuset. 1470 * 1471 * Nests both inside and outside of read_lock(&tasklist_lock). 1472 * It must not be nested with write_lock_irq(&tasklist_lock), 1473 * neither inside nor outside. 1474 */ 1475 static inline void task_lock(struct task_struct *p) 1476 { 1477 spin_lock(&p->alloc_lock); 1478 } 1479 1480 static inline void task_unlock(struct task_struct *p) 1481 { 1482 spin_unlock(&p->alloc_lock); 1483 } 1484 1485 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 1486 unsigned long *flags); 1487 1488 static inline void unlock_task_sighand(struct task_struct *tsk, 1489 unsigned long *flags) 1490 { 1491 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 1492 } 1493 1494 #ifndef __HAVE_THREAD_FUNCTIONS 1495 1496 #define task_thread_info(task) (task)->thread_info 1497 #define task_stack_page(task) ((void*)((task)->thread_info)) 1498 1499 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 1500 { 1501 *task_thread_info(p) = *task_thread_info(org); 1502 task_thread_info(p)->task = p; 1503 } 1504 1505 static inline unsigned long *end_of_stack(struct task_struct *p) 1506 { 1507 return (unsigned long *)(p->thread_info + 1); 1508 } 1509 1510 #endif 1511 1512 /* set thread flags in other task's structures 1513 * - see asm/thread_info.h for TIF_xxxx flags available 1514 */ 1515 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1516 { 1517 set_ti_thread_flag(task_thread_info(tsk), flag); 1518 } 1519 1520 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1521 { 1522 clear_ti_thread_flag(task_thread_info(tsk), flag); 1523 } 1524 1525 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 1526 { 1527 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 1528 } 1529 1530 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1531 { 1532 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 1533 } 1534 1535 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 1536 { 1537 return test_ti_thread_flag(task_thread_info(tsk), flag); 1538 } 1539 1540 static inline void set_tsk_need_resched(struct task_struct *tsk) 1541 { 1542 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1543 } 1544 1545 static inline void clear_tsk_need_resched(struct task_struct *tsk) 1546 { 1547 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1548 } 1549 1550 static inline int signal_pending(struct task_struct *p) 1551 { 1552 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 1553 } 1554 1555 static inline int need_resched(void) 1556 { 1557 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 1558 } 1559 1560 /* 1561 * cond_resched() and cond_resched_lock(): latency reduction via 1562 * explicit rescheduling in places that are safe. The return 1563 * value indicates whether a reschedule was done in fact. 1564 * cond_resched_lock() will drop the spinlock before scheduling, 1565 * cond_resched_softirq() will enable bhs before scheduling. 1566 */ 1567 extern int cond_resched(void); 1568 extern int cond_resched_lock(spinlock_t * lock); 1569 extern int cond_resched_softirq(void); 1570 1571 /* 1572 * Does a critical section need to be broken due to another 1573 * task waiting?: 1574 */ 1575 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP) 1576 # define need_lockbreak(lock) ((lock)->break_lock) 1577 #else 1578 # define need_lockbreak(lock) 0 1579 #endif 1580 1581 /* 1582 * Does a critical section need to be broken due to another 1583 * task waiting or preemption being signalled: 1584 */ 1585 static inline int lock_need_resched(spinlock_t *lock) 1586 { 1587 if (need_lockbreak(lock) || need_resched()) 1588 return 1; 1589 return 0; 1590 } 1591 1592 /* Reevaluate whether the task has signals pending delivery. 1593 This is required every time the blocked sigset_t changes. 1594 callers must hold sighand->siglock. */ 1595 1596 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t)); 1597 extern void recalc_sigpending(void); 1598 1599 extern void signal_wake_up(struct task_struct *t, int resume_stopped); 1600 1601 /* 1602 * Wrappers for p->thread_info->cpu access. No-op on UP. 1603 */ 1604 #ifdef CONFIG_SMP 1605 1606 static inline unsigned int task_cpu(const struct task_struct *p) 1607 { 1608 return task_thread_info(p)->cpu; 1609 } 1610 1611 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 1612 { 1613 task_thread_info(p)->cpu = cpu; 1614 } 1615 1616 #else 1617 1618 static inline unsigned int task_cpu(const struct task_struct *p) 1619 { 1620 return 0; 1621 } 1622 1623 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 1624 { 1625 } 1626 1627 #endif /* CONFIG_SMP */ 1628 1629 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1630 extern void arch_pick_mmap_layout(struct mm_struct *mm); 1631 #else 1632 static inline void arch_pick_mmap_layout(struct mm_struct *mm) 1633 { 1634 mm->mmap_base = TASK_UNMAPPED_BASE; 1635 mm->get_unmapped_area = arch_get_unmapped_area; 1636 mm->unmap_area = arch_unmap_area; 1637 } 1638 #endif 1639 1640 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask); 1641 extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 1642 1643 #include <linux/sysdev.h> 1644 extern int sched_mc_power_savings, sched_smt_power_savings; 1645 extern struct sysdev_attribute attr_sched_mc_power_savings, attr_sched_smt_power_savings; 1646 extern int sched_create_sysfs_power_savings_entries(struct sysdev_class *cls); 1647 1648 extern void normalize_rt_tasks(void); 1649 1650 #endif /* __KERNEL__ */ 1651 1652 #endif 1653