1 #ifndef _LINUX_SCHED_H 2 #define _LINUX_SCHED_H 3 4 #include <asm/param.h> /* for HZ */ 5 6 #include <linux/config.h> 7 #include <linux/capability.h> 8 #include <linux/threads.h> 9 #include <linux/kernel.h> 10 #include <linux/types.h> 11 #include <linux/timex.h> 12 #include <linux/jiffies.h> 13 #include <linux/rbtree.h> 14 #include <linux/thread_info.h> 15 #include <linux/cpumask.h> 16 #include <linux/errno.h> 17 #include <linux/nodemask.h> 18 19 #include <asm/system.h> 20 #include <asm/semaphore.h> 21 #include <asm/page.h> 22 #include <asm/ptrace.h> 23 #include <asm/mmu.h> 24 #include <asm/cputime.h> 25 26 #include <linux/smp.h> 27 #include <linux/sem.h> 28 #include <linux/signal.h> 29 #include <linux/securebits.h> 30 #include <linux/fs_struct.h> 31 #include <linux/compiler.h> 32 #include <linux/completion.h> 33 #include <linux/pid.h> 34 #include <linux/percpu.h> 35 #include <linux/topology.h> 36 #include <linux/seccomp.h> 37 38 #include <linux/auxvec.h> /* For AT_VECTOR_SIZE */ 39 40 struct exec_domain; 41 42 /* 43 * cloning flags: 44 */ 45 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 46 #define CLONE_VM 0x00000100 /* set if VM shared between processes */ 47 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 48 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 49 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 50 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 51 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 52 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 53 #define CLONE_THREAD 0x00010000 /* Same thread group? */ 54 #define CLONE_NEWNS 0x00020000 /* New namespace group? */ 55 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 56 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 57 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 58 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 59 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 60 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 61 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 62 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */ 63 64 /* 65 * List of flags we want to share for kernel threads, 66 * if only because they are not used by them anyway. 67 */ 68 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 69 70 /* 71 * These are the constant used to fake the fixed-point load-average 72 * counting. Some notes: 73 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 74 * a load-average precision of 10 bits integer + 11 bits fractional 75 * - if you want to count load-averages more often, you need more 76 * precision, or rounding will get you. With 2-second counting freq, 77 * the EXP_n values would be 1981, 2034 and 2043 if still using only 78 * 11 bit fractions. 79 */ 80 extern unsigned long avenrun[]; /* Load averages */ 81 82 #define FSHIFT 11 /* nr of bits of precision */ 83 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 84 #define LOAD_FREQ (5*HZ) /* 5 sec intervals */ 85 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 86 #define EXP_5 2014 /* 1/exp(5sec/5min) */ 87 #define EXP_15 2037 /* 1/exp(5sec/15min) */ 88 89 #define CALC_LOAD(load,exp,n) \ 90 load *= exp; \ 91 load += n*(FIXED_1-exp); \ 92 load >>= FSHIFT; 93 94 extern unsigned long total_forks; 95 extern int nr_threads; 96 extern int last_pid; 97 DECLARE_PER_CPU(unsigned long, process_counts); 98 extern int nr_processes(void); 99 extern unsigned long nr_running(void); 100 extern unsigned long nr_uninterruptible(void); 101 extern unsigned long nr_iowait(void); 102 103 #include <linux/time.h> 104 #include <linux/param.h> 105 #include <linux/resource.h> 106 #include <linux/timer.h> 107 108 #include <asm/processor.h> 109 110 /* 111 * Task state bitmask. NOTE! These bits are also 112 * encoded in fs/proc/array.c: get_task_state(). 113 * 114 * We have two separate sets of flags: task->state 115 * is about runnability, while task->exit_state are 116 * about the task exiting. Confusing, but this way 117 * modifying one set can't modify the other one by 118 * mistake. 119 */ 120 #define TASK_RUNNING 0 121 #define TASK_INTERRUPTIBLE 1 122 #define TASK_UNINTERRUPTIBLE 2 123 #define TASK_STOPPED 4 124 #define TASK_TRACED 8 125 /* in tsk->exit_state */ 126 #define EXIT_ZOMBIE 16 127 #define EXIT_DEAD 32 128 /* in tsk->state again */ 129 #define TASK_NONINTERACTIVE 64 130 131 #define __set_task_state(tsk, state_value) \ 132 do { (tsk)->state = (state_value); } while (0) 133 #define set_task_state(tsk, state_value) \ 134 set_mb((tsk)->state, (state_value)) 135 136 /* 137 * set_current_state() includes a barrier so that the write of current->state 138 * is correctly serialised wrt the caller's subsequent test of whether to 139 * actually sleep: 140 * 141 * set_current_state(TASK_UNINTERRUPTIBLE); 142 * if (do_i_need_to_sleep()) 143 * schedule(); 144 * 145 * If the caller does not need such serialisation then use __set_current_state() 146 */ 147 #define __set_current_state(state_value) \ 148 do { current->state = (state_value); } while (0) 149 #define set_current_state(state_value) \ 150 set_mb(current->state, (state_value)) 151 152 /* Task command name length */ 153 #define TASK_COMM_LEN 16 154 155 /* 156 * Scheduling policies 157 */ 158 #define SCHED_NORMAL 0 159 #define SCHED_FIFO 1 160 #define SCHED_RR 2 161 162 struct sched_param { 163 int sched_priority; 164 }; 165 166 #ifdef __KERNEL__ 167 168 #include <linux/spinlock.h> 169 170 /* 171 * This serializes "schedule()" and also protects 172 * the run-queue from deletions/modifications (but 173 * _adding_ to the beginning of the run-queue has 174 * a separate lock). 175 */ 176 extern rwlock_t tasklist_lock; 177 extern spinlock_t mmlist_lock; 178 179 typedef struct task_struct task_t; 180 181 extern void sched_init(void); 182 extern void sched_init_smp(void); 183 extern void init_idle(task_t *idle, int cpu); 184 185 extern cpumask_t nohz_cpu_mask; 186 187 extern void show_state(void); 188 extern void show_regs(struct pt_regs *); 189 190 /* 191 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 192 * task), SP is the stack pointer of the first frame that should be shown in the back 193 * trace (or NULL if the entire call-chain of the task should be shown). 194 */ 195 extern void show_stack(struct task_struct *task, unsigned long *sp); 196 197 void io_schedule(void); 198 long io_schedule_timeout(long timeout); 199 200 extern void cpu_init (void); 201 extern void trap_init(void); 202 extern void update_process_times(int user); 203 extern void scheduler_tick(void); 204 205 #ifdef CONFIG_DETECT_SOFTLOCKUP 206 extern void softlockup_tick(struct pt_regs *regs); 207 extern void spawn_softlockup_task(void); 208 extern void touch_softlockup_watchdog(void); 209 #else 210 static inline void softlockup_tick(struct pt_regs *regs) 211 { 212 } 213 static inline void spawn_softlockup_task(void) 214 { 215 } 216 static inline void touch_softlockup_watchdog(void) 217 { 218 } 219 #endif 220 221 222 /* Attach to any functions which should be ignored in wchan output. */ 223 #define __sched __attribute__((__section__(".sched.text"))) 224 /* Is this address in the __sched functions? */ 225 extern int in_sched_functions(unsigned long addr); 226 227 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 228 extern signed long FASTCALL(schedule_timeout(signed long timeout)); 229 extern signed long schedule_timeout_interruptible(signed long timeout); 230 extern signed long schedule_timeout_uninterruptible(signed long timeout); 231 asmlinkage void schedule(void); 232 233 struct namespace; 234 235 /* Maximum number of active map areas.. This is a random (large) number */ 236 #define DEFAULT_MAX_MAP_COUNT 65536 237 238 extern int sysctl_max_map_count; 239 240 #include <linux/aio.h> 241 242 extern unsigned long 243 arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 244 unsigned long, unsigned long); 245 extern unsigned long 246 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 247 unsigned long len, unsigned long pgoff, 248 unsigned long flags); 249 extern void arch_unmap_area(struct mm_struct *, unsigned long); 250 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 251 252 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS 253 /* 254 * The mm counters are not protected by its page_table_lock, 255 * so must be incremented atomically. 256 */ 257 #ifdef ATOMIC64_INIT 258 #define set_mm_counter(mm, member, value) atomic64_set(&(mm)->_##member, value) 259 #define get_mm_counter(mm, member) ((unsigned long)atomic64_read(&(mm)->_##member)) 260 #define add_mm_counter(mm, member, value) atomic64_add(value, &(mm)->_##member) 261 #define inc_mm_counter(mm, member) atomic64_inc(&(mm)->_##member) 262 #define dec_mm_counter(mm, member) atomic64_dec(&(mm)->_##member) 263 typedef atomic64_t mm_counter_t; 264 #else /* !ATOMIC64_INIT */ 265 /* 266 * The counters wrap back to 0 at 2^32 * PAGE_SIZE, 267 * that is, at 16TB if using 4kB page size. 268 */ 269 #define set_mm_counter(mm, member, value) atomic_set(&(mm)->_##member, value) 270 #define get_mm_counter(mm, member) ((unsigned long)atomic_read(&(mm)->_##member)) 271 #define add_mm_counter(mm, member, value) atomic_add(value, &(mm)->_##member) 272 #define inc_mm_counter(mm, member) atomic_inc(&(mm)->_##member) 273 #define dec_mm_counter(mm, member) atomic_dec(&(mm)->_##member) 274 typedef atomic_t mm_counter_t; 275 #endif /* !ATOMIC64_INIT */ 276 277 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 278 /* 279 * The mm counters are protected by its page_table_lock, 280 * so can be incremented directly. 281 */ 282 #define set_mm_counter(mm, member, value) (mm)->_##member = (value) 283 #define get_mm_counter(mm, member) ((mm)->_##member) 284 #define add_mm_counter(mm, member, value) (mm)->_##member += (value) 285 #define inc_mm_counter(mm, member) (mm)->_##member++ 286 #define dec_mm_counter(mm, member) (mm)->_##member-- 287 typedef unsigned long mm_counter_t; 288 289 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 290 291 #define get_mm_rss(mm) \ 292 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 293 #define update_hiwater_rss(mm) do { \ 294 unsigned long _rss = get_mm_rss(mm); \ 295 if ((mm)->hiwater_rss < _rss) \ 296 (mm)->hiwater_rss = _rss; \ 297 } while (0) 298 #define update_hiwater_vm(mm) do { \ 299 if ((mm)->hiwater_vm < (mm)->total_vm) \ 300 (mm)->hiwater_vm = (mm)->total_vm; \ 301 } while (0) 302 303 struct mm_struct { 304 struct vm_area_struct * mmap; /* list of VMAs */ 305 struct rb_root mm_rb; 306 struct vm_area_struct * mmap_cache; /* last find_vma result */ 307 unsigned long (*get_unmapped_area) (struct file *filp, 308 unsigned long addr, unsigned long len, 309 unsigned long pgoff, unsigned long flags); 310 void (*unmap_area) (struct mm_struct *mm, unsigned long addr); 311 unsigned long mmap_base; /* base of mmap area */ 312 unsigned long cached_hole_size; /* if non-zero, the largest hole below free_area_cache */ 313 unsigned long free_area_cache; /* first hole of size cached_hole_size or larger */ 314 pgd_t * pgd; 315 atomic_t mm_users; /* How many users with user space? */ 316 atomic_t mm_count; /* How many references to "struct mm_struct" (users count as 1) */ 317 int map_count; /* number of VMAs */ 318 struct rw_semaphore mmap_sem; 319 spinlock_t page_table_lock; /* Protects page tables and some counters */ 320 321 struct list_head mmlist; /* List of maybe swapped mm's. These are globally strung 322 * together off init_mm.mmlist, and are protected 323 * by mmlist_lock 324 */ 325 326 /* Special counters, in some configurations protected by the 327 * page_table_lock, in other configurations by being atomic. 328 */ 329 mm_counter_t _file_rss; 330 mm_counter_t _anon_rss; 331 332 unsigned long hiwater_rss; /* High-watermark of RSS usage */ 333 unsigned long hiwater_vm; /* High-water virtual memory usage */ 334 335 unsigned long total_vm, locked_vm, shared_vm, exec_vm; 336 unsigned long stack_vm, reserved_vm, def_flags, nr_ptes; 337 unsigned long start_code, end_code, start_data, end_data; 338 unsigned long start_brk, brk, start_stack; 339 unsigned long arg_start, arg_end, env_start, env_end; 340 341 unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */ 342 343 unsigned dumpable:2; 344 cpumask_t cpu_vm_mask; 345 346 /* Architecture-specific MM context */ 347 mm_context_t context; 348 349 /* Token based thrashing protection. */ 350 unsigned long swap_token_time; 351 char recent_pagein; 352 353 /* coredumping support */ 354 int core_waiters; 355 struct completion *core_startup_done, core_done; 356 357 /* aio bits */ 358 rwlock_t ioctx_list_lock; 359 struct kioctx *ioctx_list; 360 struct kioctx default_kioctx; 361 }; 362 363 struct sighand_struct { 364 atomic_t count; 365 struct k_sigaction action[_NSIG]; 366 spinlock_t siglock; 367 }; 368 369 /* 370 * NOTE! "signal_struct" does not have it's own 371 * locking, because a shared signal_struct always 372 * implies a shared sighand_struct, so locking 373 * sighand_struct is always a proper superset of 374 * the locking of signal_struct. 375 */ 376 struct signal_struct { 377 atomic_t count; 378 atomic_t live; 379 380 wait_queue_head_t wait_chldexit; /* for wait4() */ 381 382 /* current thread group signal load-balancing target: */ 383 task_t *curr_target; 384 385 /* shared signal handling: */ 386 struct sigpending shared_pending; 387 388 /* thread group exit support */ 389 int group_exit_code; 390 /* overloaded: 391 * - notify group_exit_task when ->count is equal to notify_count 392 * - everyone except group_exit_task is stopped during signal delivery 393 * of fatal signals, group_exit_task processes the signal. 394 */ 395 struct task_struct *group_exit_task; 396 int notify_count; 397 398 /* thread group stop support, overloads group_exit_code too */ 399 int group_stop_count; 400 unsigned int flags; /* see SIGNAL_* flags below */ 401 402 /* POSIX.1b Interval Timers */ 403 struct list_head posix_timers; 404 405 /* ITIMER_REAL timer for the process */ 406 struct timer_list real_timer; 407 unsigned long it_real_value, it_real_incr; 408 409 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 410 cputime_t it_prof_expires, it_virt_expires; 411 cputime_t it_prof_incr, it_virt_incr; 412 413 /* job control IDs */ 414 pid_t pgrp; 415 pid_t tty_old_pgrp; 416 pid_t session; 417 /* boolean value for session group leader */ 418 int leader; 419 420 struct tty_struct *tty; /* NULL if no tty */ 421 422 /* 423 * Cumulative resource counters for dead threads in the group, 424 * and for reaped dead child processes forked by this group. 425 * Live threads maintain their own counters and add to these 426 * in __exit_signal, except for the group leader. 427 */ 428 cputime_t utime, stime, cutime, cstime; 429 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 430 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 431 432 /* 433 * Cumulative ns of scheduled CPU time for dead threads in the 434 * group, not including a zombie group leader. (This only differs 435 * from jiffies_to_ns(utime + stime) if sched_clock uses something 436 * other than jiffies.) 437 */ 438 unsigned long long sched_time; 439 440 /* 441 * We don't bother to synchronize most readers of this at all, 442 * because there is no reader checking a limit that actually needs 443 * to get both rlim_cur and rlim_max atomically, and either one 444 * alone is a single word that can safely be read normally. 445 * getrlimit/setrlimit use task_lock(current->group_leader) to 446 * protect this instead of the siglock, because they really 447 * have no need to disable irqs. 448 */ 449 struct rlimit rlim[RLIM_NLIMITS]; 450 451 struct list_head cpu_timers[3]; 452 453 /* keep the process-shared keyrings here so that they do the right 454 * thing in threads created with CLONE_THREAD */ 455 #ifdef CONFIG_KEYS 456 struct key *session_keyring; /* keyring inherited over fork */ 457 struct key *process_keyring; /* keyring private to this process */ 458 #endif 459 }; 460 461 /* Context switch must be unlocked if interrupts are to be enabled */ 462 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 463 # define __ARCH_WANT_UNLOCKED_CTXSW 464 #endif 465 466 /* 467 * Bits in flags field of signal_struct. 468 */ 469 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 470 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 471 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 472 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 473 474 475 /* 476 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 477 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL tasks are 478 * in the range MAX_RT_PRIO..MAX_PRIO-1. Priority values 479 * are inverted: lower p->prio value means higher priority. 480 * 481 * The MAX_USER_RT_PRIO value allows the actual maximum 482 * RT priority to be separate from the value exported to 483 * user-space. This allows kernel threads to set their 484 * priority to a value higher than any user task. Note: 485 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 486 */ 487 488 #define MAX_USER_RT_PRIO 100 489 #define MAX_RT_PRIO MAX_USER_RT_PRIO 490 491 #define MAX_PRIO (MAX_RT_PRIO + 40) 492 493 #define rt_task(p) (unlikely((p)->prio < MAX_RT_PRIO)) 494 495 /* 496 * Some day this will be a full-fledged user tracking system.. 497 */ 498 struct user_struct { 499 atomic_t __count; /* reference count */ 500 atomic_t processes; /* How many processes does this user have? */ 501 atomic_t files; /* How many open files does this user have? */ 502 atomic_t sigpending; /* How many pending signals does this user have? */ 503 #ifdef CONFIG_INOTIFY 504 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 505 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 506 #endif 507 /* protected by mq_lock */ 508 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 509 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 510 511 #ifdef CONFIG_KEYS 512 struct key *uid_keyring; /* UID specific keyring */ 513 struct key *session_keyring; /* UID's default session keyring */ 514 #endif 515 516 /* Hash table maintenance information */ 517 struct list_head uidhash_list; 518 uid_t uid; 519 }; 520 521 extern struct user_struct *find_user(uid_t); 522 523 extern struct user_struct root_user; 524 #define INIT_USER (&root_user) 525 526 typedef struct prio_array prio_array_t; 527 struct backing_dev_info; 528 struct reclaim_state; 529 530 #ifdef CONFIG_SCHEDSTATS 531 struct sched_info { 532 /* cumulative counters */ 533 unsigned long cpu_time, /* time spent on the cpu */ 534 run_delay, /* time spent waiting on a runqueue */ 535 pcnt; /* # of timeslices run on this cpu */ 536 537 /* timestamps */ 538 unsigned long last_arrival, /* when we last ran on a cpu */ 539 last_queued; /* when we were last queued to run */ 540 }; 541 542 extern struct file_operations proc_schedstat_operations; 543 #endif 544 545 enum idle_type 546 { 547 SCHED_IDLE, 548 NOT_IDLE, 549 NEWLY_IDLE, 550 MAX_IDLE_TYPES 551 }; 552 553 /* 554 * sched-domains (multiprocessor balancing) declarations: 555 */ 556 #ifdef CONFIG_SMP 557 #define SCHED_LOAD_SCALE 128UL /* increase resolution of load */ 558 559 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 560 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 561 #define SD_BALANCE_EXEC 4 /* Balance on exec */ 562 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 563 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 564 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 565 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 566 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 567 568 struct sched_group { 569 struct sched_group *next; /* Must be a circular list */ 570 cpumask_t cpumask; 571 572 /* 573 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 574 * single CPU. This is read only (except for setup, hotplug CPU). 575 */ 576 unsigned long cpu_power; 577 }; 578 579 struct sched_domain { 580 /* These fields must be setup */ 581 struct sched_domain *parent; /* top domain must be null terminated */ 582 struct sched_group *groups; /* the balancing groups of the domain */ 583 cpumask_t span; /* span of all CPUs in this domain */ 584 unsigned long min_interval; /* Minimum balance interval ms */ 585 unsigned long max_interval; /* Maximum balance interval ms */ 586 unsigned int busy_factor; /* less balancing by factor if busy */ 587 unsigned int imbalance_pct; /* No balance until over watermark */ 588 unsigned long long cache_hot_time; /* Task considered cache hot (ns) */ 589 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 590 unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */ 591 unsigned int busy_idx; 592 unsigned int idle_idx; 593 unsigned int newidle_idx; 594 unsigned int wake_idx; 595 unsigned int forkexec_idx; 596 int flags; /* See SD_* */ 597 598 /* Runtime fields. */ 599 unsigned long last_balance; /* init to jiffies. units in jiffies */ 600 unsigned int balance_interval; /* initialise to 1. units in ms. */ 601 unsigned int nr_balance_failed; /* initialise to 0 */ 602 603 #ifdef CONFIG_SCHEDSTATS 604 /* load_balance() stats */ 605 unsigned long lb_cnt[MAX_IDLE_TYPES]; 606 unsigned long lb_failed[MAX_IDLE_TYPES]; 607 unsigned long lb_balanced[MAX_IDLE_TYPES]; 608 unsigned long lb_imbalance[MAX_IDLE_TYPES]; 609 unsigned long lb_gained[MAX_IDLE_TYPES]; 610 unsigned long lb_hot_gained[MAX_IDLE_TYPES]; 611 unsigned long lb_nobusyg[MAX_IDLE_TYPES]; 612 unsigned long lb_nobusyq[MAX_IDLE_TYPES]; 613 614 /* Active load balancing */ 615 unsigned long alb_cnt; 616 unsigned long alb_failed; 617 unsigned long alb_pushed; 618 619 /* SD_BALANCE_EXEC stats */ 620 unsigned long sbe_cnt; 621 unsigned long sbe_balanced; 622 unsigned long sbe_pushed; 623 624 /* SD_BALANCE_FORK stats */ 625 unsigned long sbf_cnt; 626 unsigned long sbf_balanced; 627 unsigned long sbf_pushed; 628 629 /* try_to_wake_up() stats */ 630 unsigned long ttwu_wake_remote; 631 unsigned long ttwu_move_affine; 632 unsigned long ttwu_move_balance; 633 #endif 634 }; 635 636 extern void partition_sched_domains(cpumask_t *partition1, 637 cpumask_t *partition2); 638 #endif /* CONFIG_SMP */ 639 640 641 struct io_context; /* See blkdev.h */ 642 void exit_io_context(void); 643 struct cpuset; 644 645 #define NGROUPS_SMALL 32 646 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t))) 647 struct group_info { 648 int ngroups; 649 atomic_t usage; 650 gid_t small_block[NGROUPS_SMALL]; 651 int nblocks; 652 gid_t *blocks[0]; 653 }; 654 655 /* 656 * get_group_info() must be called with the owning task locked (via task_lock()) 657 * when task != current. The reason being that the vast majority of callers are 658 * looking at current->group_info, which can not be changed except by the 659 * current task. Changing current->group_info requires the task lock, too. 660 */ 661 #define get_group_info(group_info) do { \ 662 atomic_inc(&(group_info)->usage); \ 663 } while (0) 664 665 #define put_group_info(group_info) do { \ 666 if (atomic_dec_and_test(&(group_info)->usage)) \ 667 groups_free(group_info); \ 668 } while (0) 669 670 extern struct group_info *groups_alloc(int gidsetsize); 671 extern void groups_free(struct group_info *group_info); 672 extern int set_current_groups(struct group_info *group_info); 673 extern int groups_search(struct group_info *group_info, gid_t grp); 674 /* access the groups "array" with this macro */ 675 #define GROUP_AT(gi, i) \ 676 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 677 678 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 679 extern void prefetch_stack(struct task_struct*); 680 #else 681 static inline void prefetch_stack(struct task_struct *t) { } 682 #endif 683 684 struct audit_context; /* See audit.c */ 685 struct mempolicy; 686 687 struct task_struct { 688 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 689 struct thread_info *thread_info; 690 atomic_t usage; 691 unsigned long flags; /* per process flags, defined below */ 692 unsigned long ptrace; 693 694 int lock_depth; /* BKL lock depth */ 695 696 #if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW) 697 int oncpu; 698 #endif 699 int prio, static_prio; 700 struct list_head run_list; 701 prio_array_t *array; 702 703 unsigned short ioprio; 704 705 unsigned long sleep_avg; 706 unsigned long long timestamp, last_ran; 707 unsigned long long sched_time; /* sched_clock time spent running */ 708 int activated; 709 710 unsigned long policy; 711 cpumask_t cpus_allowed; 712 unsigned int time_slice, first_time_slice; 713 714 #ifdef CONFIG_SCHEDSTATS 715 struct sched_info sched_info; 716 #endif 717 718 struct list_head tasks; 719 /* 720 * ptrace_list/ptrace_children forms the list of my children 721 * that were stolen by a ptracer. 722 */ 723 struct list_head ptrace_children; 724 struct list_head ptrace_list; 725 726 struct mm_struct *mm, *active_mm; 727 728 /* task state */ 729 struct linux_binfmt *binfmt; 730 long exit_state; 731 int exit_code, exit_signal; 732 int pdeath_signal; /* The signal sent when the parent dies */ 733 /* ??? */ 734 unsigned long personality; 735 unsigned did_exec:1; 736 pid_t pid; 737 pid_t tgid; 738 /* 739 * pointers to (original) parent process, youngest child, younger sibling, 740 * older sibling, respectively. (p->father can be replaced with 741 * p->parent->pid) 742 */ 743 struct task_struct *real_parent; /* real parent process (when being debugged) */ 744 struct task_struct *parent; /* parent process */ 745 /* 746 * children/sibling forms the list of my children plus the 747 * tasks I'm ptracing. 748 */ 749 struct list_head children; /* list of my children */ 750 struct list_head sibling; /* linkage in my parent's children list */ 751 struct task_struct *group_leader; /* threadgroup leader */ 752 753 /* PID/PID hash table linkage. */ 754 struct pid pids[PIDTYPE_MAX]; 755 756 struct completion *vfork_done; /* for vfork() */ 757 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 758 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 759 760 unsigned long rt_priority; 761 cputime_t utime, stime; 762 unsigned long nvcsw, nivcsw; /* context switch counts */ 763 struct timespec start_time; 764 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 765 unsigned long min_flt, maj_flt; 766 767 cputime_t it_prof_expires, it_virt_expires; 768 unsigned long long it_sched_expires; 769 struct list_head cpu_timers[3]; 770 771 /* process credentials */ 772 uid_t uid,euid,suid,fsuid; 773 gid_t gid,egid,sgid,fsgid; 774 struct group_info *group_info; 775 kernel_cap_t cap_effective, cap_inheritable, cap_permitted; 776 unsigned keep_capabilities:1; 777 struct user_struct *user; 778 #ifdef CONFIG_KEYS 779 struct key *thread_keyring; /* keyring private to this thread */ 780 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 781 #endif 782 int oomkilladj; /* OOM kill score adjustment (bit shift). */ 783 char comm[TASK_COMM_LEN]; /* executable name excluding path 784 - access with [gs]et_task_comm (which lock 785 it with task_lock()) 786 - initialized normally by flush_old_exec */ 787 /* file system info */ 788 int link_count, total_link_count; 789 /* ipc stuff */ 790 struct sysv_sem sysvsem; 791 /* CPU-specific state of this task */ 792 struct thread_struct thread; 793 /* filesystem information */ 794 struct fs_struct *fs; 795 /* open file information */ 796 struct files_struct *files; 797 /* namespace */ 798 struct namespace *namespace; 799 /* signal handlers */ 800 struct signal_struct *signal; 801 struct sighand_struct *sighand; 802 803 sigset_t blocked, real_blocked; 804 struct sigpending pending; 805 806 unsigned long sas_ss_sp; 807 size_t sas_ss_size; 808 int (*notifier)(void *priv); 809 void *notifier_data; 810 sigset_t *notifier_mask; 811 812 void *security; 813 struct audit_context *audit_context; 814 seccomp_t seccomp; 815 816 /* Thread group tracking */ 817 u32 parent_exec_id; 818 u32 self_exec_id; 819 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 820 spinlock_t alloc_lock; 821 /* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */ 822 spinlock_t proc_lock; 823 824 /* journalling filesystem info */ 825 void *journal_info; 826 827 /* VM state */ 828 struct reclaim_state *reclaim_state; 829 830 struct dentry *proc_dentry; 831 struct backing_dev_info *backing_dev_info; 832 833 struct io_context *io_context; 834 835 unsigned long ptrace_message; 836 siginfo_t *last_siginfo; /* For ptrace use. */ 837 /* 838 * current io wait handle: wait queue entry to use for io waits 839 * If this thread is processing aio, this points at the waitqueue 840 * inside the currently handled kiocb. It may be NULL (i.e. default 841 * to a stack based synchronous wait) if its doing sync IO. 842 */ 843 wait_queue_t *io_wait; 844 /* i/o counters(bytes read/written, #syscalls */ 845 u64 rchar, wchar, syscr, syscw; 846 #if defined(CONFIG_BSD_PROCESS_ACCT) 847 u64 acct_rss_mem1; /* accumulated rss usage */ 848 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 849 clock_t acct_stimexpd; /* clock_t-converted stime since last update */ 850 #endif 851 #ifdef CONFIG_NUMA 852 struct mempolicy *mempolicy; 853 short il_next; 854 #endif 855 #ifdef CONFIG_CPUSETS 856 struct cpuset *cpuset; 857 nodemask_t mems_allowed; 858 int cpuset_mems_generation; 859 #endif 860 atomic_t fs_excl; /* holding fs exclusive resources */ 861 }; 862 863 static inline pid_t process_group(struct task_struct *tsk) 864 { 865 return tsk->signal->pgrp; 866 } 867 868 /** 869 * pid_alive - check that a task structure is not stale 870 * @p: Task structure to be checked. 871 * 872 * Test if a process is not yet dead (at most zombie state) 873 * If pid_alive fails, then pointers within the task structure 874 * can be stale and must not be dereferenced. 875 */ 876 static inline int pid_alive(struct task_struct *p) 877 { 878 return p->pids[PIDTYPE_PID].nr != 0; 879 } 880 881 extern void free_task(struct task_struct *tsk); 882 extern void __put_task_struct(struct task_struct *tsk); 883 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 884 #define put_task_struct(tsk) \ 885 do { if (atomic_dec_and_test(&(tsk)->usage)) __put_task_struct(tsk); } while(0) 886 887 /* 888 * Per process flags 889 */ 890 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 891 /* Not implemented yet, only for 486*/ 892 #define PF_STARTING 0x00000002 /* being created */ 893 #define PF_EXITING 0x00000004 /* getting shut down */ 894 #define PF_DEAD 0x00000008 /* Dead */ 895 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 896 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 897 #define PF_DUMPCORE 0x00000200 /* dumped core */ 898 #define PF_SIGNALED 0x00000400 /* killed by a signal */ 899 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 900 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 901 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 902 #define PF_FREEZE 0x00004000 /* this task is being frozen for suspend now */ 903 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 904 #define PF_FROZEN 0x00010000 /* frozen for system suspend */ 905 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 906 #define PF_KSWAPD 0x00040000 /* I am kswapd */ 907 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 908 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 909 #define PF_SYNCWRITE 0x00200000 /* I am doing a sync write */ 910 #define PF_BORROWED_MM 0x00400000 /* I am a kthread doing use_mm */ 911 #define PF_RANDOMIZE 0x00800000 /* randomize virtual address space */ 912 913 /* 914 * Only the _current_ task can read/write to tsk->flags, but other 915 * tasks can access tsk->flags in readonly mode for example 916 * with tsk_used_math (like during threaded core dumping). 917 * There is however an exception to this rule during ptrace 918 * or during fork: the ptracer task is allowed to write to the 919 * child->flags of its traced child (same goes for fork, the parent 920 * can write to the child->flags), because we're guaranteed the 921 * child is not running and in turn not changing child->flags 922 * at the same time the parent does it. 923 */ 924 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 925 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 926 #define clear_used_math() clear_stopped_child_used_math(current) 927 #define set_used_math() set_stopped_child_used_math(current) 928 #define conditional_stopped_child_used_math(condition, child) \ 929 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 930 #define conditional_used_math(condition) \ 931 conditional_stopped_child_used_math(condition, current) 932 #define copy_to_stopped_child_used_math(child) \ 933 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 934 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 935 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 936 #define used_math() tsk_used_math(current) 937 938 #ifdef CONFIG_SMP 939 extern int set_cpus_allowed(task_t *p, cpumask_t new_mask); 940 #else 941 static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask) 942 { 943 if (!cpu_isset(0, new_mask)) 944 return -EINVAL; 945 return 0; 946 } 947 #endif 948 949 extern unsigned long long sched_clock(void); 950 extern unsigned long long current_sched_time(const task_t *current_task); 951 952 /* sched_exec is called by processes performing an exec */ 953 #ifdef CONFIG_SMP 954 extern void sched_exec(void); 955 #else 956 #define sched_exec() {} 957 #endif 958 959 #ifdef CONFIG_HOTPLUG_CPU 960 extern void idle_task_exit(void); 961 #else 962 static inline void idle_task_exit(void) {} 963 #endif 964 965 extern void sched_idle_next(void); 966 extern void set_user_nice(task_t *p, long nice); 967 extern int task_prio(const task_t *p); 968 extern int task_nice(const task_t *p); 969 extern int can_nice(const task_t *p, const int nice); 970 extern int task_curr(const task_t *p); 971 extern int idle_cpu(int cpu); 972 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 973 extern task_t *idle_task(int cpu); 974 extern task_t *curr_task(int cpu); 975 extern void set_curr_task(int cpu, task_t *p); 976 977 void yield(void); 978 979 /* 980 * The default (Linux) execution domain. 981 */ 982 extern struct exec_domain default_exec_domain; 983 984 union thread_union { 985 struct thread_info thread_info; 986 unsigned long stack[THREAD_SIZE/sizeof(long)]; 987 }; 988 989 #ifndef __HAVE_ARCH_KSTACK_END 990 static inline int kstack_end(void *addr) 991 { 992 /* Reliable end of stack detection: 993 * Some APM bios versions misalign the stack 994 */ 995 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 996 } 997 #endif 998 999 extern union thread_union init_thread_union; 1000 extern struct task_struct init_task; 1001 1002 extern struct mm_struct init_mm; 1003 1004 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr) 1005 extern struct task_struct *find_task_by_pid_type(int type, int pid); 1006 extern void set_special_pids(pid_t session, pid_t pgrp); 1007 extern void __set_special_pids(pid_t session, pid_t pgrp); 1008 1009 /* per-UID process charging. */ 1010 extern struct user_struct * alloc_uid(uid_t); 1011 static inline struct user_struct *get_uid(struct user_struct *u) 1012 { 1013 atomic_inc(&u->__count); 1014 return u; 1015 } 1016 extern void free_uid(struct user_struct *); 1017 extern void switch_uid(struct user_struct *); 1018 1019 #include <asm/current.h> 1020 1021 extern void do_timer(struct pt_regs *); 1022 1023 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state)); 1024 extern int FASTCALL(wake_up_process(struct task_struct * tsk)); 1025 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk, 1026 unsigned long clone_flags)); 1027 #ifdef CONFIG_SMP 1028 extern void kick_process(struct task_struct *tsk); 1029 #else 1030 static inline void kick_process(struct task_struct *tsk) { } 1031 #endif 1032 extern void FASTCALL(sched_fork(task_t * p, int clone_flags)); 1033 extern void FASTCALL(sched_exit(task_t * p)); 1034 1035 extern int in_group_p(gid_t); 1036 extern int in_egroup_p(gid_t); 1037 1038 extern void proc_caches_init(void); 1039 extern void flush_signals(struct task_struct *); 1040 extern void flush_signal_handlers(struct task_struct *, int force_default); 1041 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1042 1043 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1044 { 1045 unsigned long flags; 1046 int ret; 1047 1048 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1049 ret = dequeue_signal(tsk, mask, info); 1050 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1051 1052 return ret; 1053 } 1054 1055 extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1056 sigset_t *mask); 1057 extern void unblock_all_signals(void); 1058 extern void release_task(struct task_struct * p); 1059 extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1060 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *); 1061 extern int force_sigsegv(int, struct task_struct *); 1062 extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1063 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp); 1064 extern int kill_pg_info(int, struct siginfo *, pid_t); 1065 extern int kill_proc_info(int, struct siginfo *, pid_t); 1066 extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t); 1067 extern void do_notify_parent(struct task_struct *, int); 1068 extern void force_sig(int, struct task_struct *); 1069 extern void force_sig_specific(int, struct task_struct *); 1070 extern int send_sig(int, struct task_struct *, int); 1071 extern void zap_other_threads(struct task_struct *p); 1072 extern int kill_pg(pid_t, int, int); 1073 extern int kill_sl(pid_t, int, int); 1074 extern int kill_proc(pid_t, int, int); 1075 extern struct sigqueue *sigqueue_alloc(void); 1076 extern void sigqueue_free(struct sigqueue *); 1077 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *); 1078 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *); 1079 extern int do_sigaction(int, const struct k_sigaction *, struct k_sigaction *); 1080 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1081 1082 /* These can be the second arg to send_sig_info/send_group_sig_info. */ 1083 #define SEND_SIG_NOINFO ((struct siginfo *) 0) 1084 #define SEND_SIG_PRIV ((struct siginfo *) 1) 1085 #define SEND_SIG_FORCED ((struct siginfo *) 2) 1086 1087 static inline int is_si_special(const struct siginfo *info) 1088 { 1089 return info <= SEND_SIG_FORCED; 1090 } 1091 1092 /* True if we are on the alternate signal stack. */ 1093 1094 static inline int on_sig_stack(unsigned long sp) 1095 { 1096 return (sp - current->sas_ss_sp < current->sas_ss_size); 1097 } 1098 1099 static inline int sas_ss_flags(unsigned long sp) 1100 { 1101 return (current->sas_ss_size == 0 ? SS_DISABLE 1102 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1103 } 1104 1105 1106 #ifdef CONFIG_SECURITY 1107 /* code is in security.c */ 1108 extern int capable(int cap); 1109 #else 1110 static inline int capable(int cap) 1111 { 1112 if (cap_raised(current->cap_effective, cap)) { 1113 current->flags |= PF_SUPERPRIV; 1114 return 1; 1115 } 1116 return 0; 1117 } 1118 #endif 1119 1120 /* 1121 * Routines for handling mm_structs 1122 */ 1123 extern struct mm_struct * mm_alloc(void); 1124 1125 /* mmdrop drops the mm and the page tables */ 1126 extern void FASTCALL(__mmdrop(struct mm_struct *)); 1127 static inline void mmdrop(struct mm_struct * mm) 1128 { 1129 if (atomic_dec_and_test(&mm->mm_count)) 1130 __mmdrop(mm); 1131 } 1132 1133 /* mmput gets rid of the mappings and all user-space */ 1134 extern void mmput(struct mm_struct *); 1135 /* Grab a reference to a task's mm, if it is not already going away */ 1136 extern struct mm_struct *get_task_mm(struct task_struct *task); 1137 /* Remove the current tasks stale references to the old mm_struct */ 1138 extern void mm_release(struct task_struct *, struct mm_struct *); 1139 1140 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1141 extern void flush_thread(void); 1142 extern void exit_thread(void); 1143 1144 extern void exit_files(struct task_struct *); 1145 extern void exit_signal(struct task_struct *); 1146 extern void __exit_signal(struct task_struct *); 1147 extern void exit_sighand(struct task_struct *); 1148 extern void __exit_sighand(struct task_struct *); 1149 extern void exit_itimers(struct signal_struct *); 1150 1151 extern NORET_TYPE void do_group_exit(int); 1152 1153 extern void daemonize(const char *, ...); 1154 extern int allow_signal(int); 1155 extern int disallow_signal(int); 1156 extern task_t *child_reaper; 1157 1158 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1159 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1160 task_t *fork_idle(int); 1161 1162 extern void set_task_comm(struct task_struct *tsk, char *from); 1163 extern void get_task_comm(char *to, struct task_struct *tsk); 1164 1165 #ifdef CONFIG_SMP 1166 extern void wait_task_inactive(task_t * p); 1167 #else 1168 #define wait_task_inactive(p) do { } while (0) 1169 #endif 1170 1171 #define remove_parent(p) list_del_init(&(p)->sibling) 1172 #define add_parent(p, parent) list_add_tail(&(p)->sibling,&(parent)->children) 1173 1174 #define REMOVE_LINKS(p) do { \ 1175 if (thread_group_leader(p)) \ 1176 list_del_init(&(p)->tasks); \ 1177 remove_parent(p); \ 1178 } while (0) 1179 1180 #define SET_LINKS(p) do { \ 1181 if (thread_group_leader(p)) \ 1182 list_add_tail(&(p)->tasks,&init_task.tasks); \ 1183 add_parent(p, (p)->parent); \ 1184 } while (0) 1185 1186 #define next_task(p) list_entry((p)->tasks.next, struct task_struct, tasks) 1187 #define prev_task(p) list_entry((p)->tasks.prev, struct task_struct, tasks) 1188 1189 #define for_each_process(p) \ 1190 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1191 1192 /* 1193 * Careful: do_each_thread/while_each_thread is a double loop so 1194 * 'break' will not work as expected - use goto instead. 1195 */ 1196 #define do_each_thread(g, t) \ 1197 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1198 1199 #define while_each_thread(g, t) \ 1200 while ((t = next_thread(t)) != g) 1201 1202 extern task_t * FASTCALL(next_thread(const task_t *p)); 1203 1204 #define thread_group_leader(p) (p->pid == p->tgid) 1205 1206 static inline int thread_group_empty(task_t *p) 1207 { 1208 return list_empty(&p->pids[PIDTYPE_TGID].pid_list); 1209 } 1210 1211 #define delay_group_leader(p) \ 1212 (thread_group_leader(p) && !thread_group_empty(p)) 1213 1214 extern void unhash_process(struct task_struct *p); 1215 1216 /* 1217 * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring 1218 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1219 * pins the final release of task.io_context. Also protects ->cpuset. 1220 * 1221 * Nests both inside and outside of read_lock(&tasklist_lock). 1222 * It must not be nested with write_lock_irq(&tasklist_lock), 1223 * neither inside nor outside. 1224 */ 1225 static inline void task_lock(struct task_struct *p) 1226 { 1227 spin_lock(&p->alloc_lock); 1228 } 1229 1230 static inline void task_unlock(struct task_struct *p) 1231 { 1232 spin_unlock(&p->alloc_lock); 1233 } 1234 1235 /* set thread flags in other task's structures 1236 * - see asm/thread_info.h for TIF_xxxx flags available 1237 */ 1238 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1239 { 1240 set_ti_thread_flag(tsk->thread_info,flag); 1241 } 1242 1243 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1244 { 1245 clear_ti_thread_flag(tsk->thread_info,flag); 1246 } 1247 1248 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 1249 { 1250 return test_and_set_ti_thread_flag(tsk->thread_info,flag); 1251 } 1252 1253 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1254 { 1255 return test_and_clear_ti_thread_flag(tsk->thread_info,flag); 1256 } 1257 1258 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 1259 { 1260 return test_ti_thread_flag(tsk->thread_info,flag); 1261 } 1262 1263 static inline void set_tsk_need_resched(struct task_struct *tsk) 1264 { 1265 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1266 } 1267 1268 static inline void clear_tsk_need_resched(struct task_struct *tsk) 1269 { 1270 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1271 } 1272 1273 static inline int signal_pending(struct task_struct *p) 1274 { 1275 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 1276 } 1277 1278 static inline int need_resched(void) 1279 { 1280 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 1281 } 1282 1283 /* 1284 * cond_resched() and cond_resched_lock(): latency reduction via 1285 * explicit rescheduling in places that are safe. The return 1286 * value indicates whether a reschedule was done in fact. 1287 * cond_resched_lock() will drop the spinlock before scheduling, 1288 * cond_resched_softirq() will enable bhs before scheduling. 1289 */ 1290 extern int cond_resched(void); 1291 extern int cond_resched_lock(spinlock_t * lock); 1292 extern int cond_resched_softirq(void); 1293 1294 /* 1295 * Does a critical section need to be broken due to another 1296 * task waiting?: 1297 */ 1298 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP) 1299 # define need_lockbreak(lock) ((lock)->break_lock) 1300 #else 1301 # define need_lockbreak(lock) 0 1302 #endif 1303 1304 /* 1305 * Does a critical section need to be broken due to another 1306 * task waiting or preemption being signalled: 1307 */ 1308 static inline int lock_need_resched(spinlock_t *lock) 1309 { 1310 if (need_lockbreak(lock) || need_resched()) 1311 return 1; 1312 return 0; 1313 } 1314 1315 /* Reevaluate whether the task has signals pending delivery. 1316 This is required every time the blocked sigset_t changes. 1317 callers must hold sighand->siglock. */ 1318 1319 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t)); 1320 extern void recalc_sigpending(void); 1321 1322 extern void signal_wake_up(struct task_struct *t, int resume_stopped); 1323 1324 /* 1325 * Wrappers for p->thread_info->cpu access. No-op on UP. 1326 */ 1327 #ifdef CONFIG_SMP 1328 1329 static inline unsigned int task_cpu(const struct task_struct *p) 1330 { 1331 return p->thread_info->cpu; 1332 } 1333 1334 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 1335 { 1336 p->thread_info->cpu = cpu; 1337 } 1338 1339 #else 1340 1341 static inline unsigned int task_cpu(const struct task_struct *p) 1342 { 1343 return 0; 1344 } 1345 1346 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 1347 { 1348 } 1349 1350 #endif /* CONFIG_SMP */ 1351 1352 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1353 extern void arch_pick_mmap_layout(struct mm_struct *mm); 1354 #else 1355 static inline void arch_pick_mmap_layout(struct mm_struct *mm) 1356 { 1357 mm->mmap_base = TASK_UNMAPPED_BASE; 1358 mm->get_unmapped_area = arch_get_unmapped_area; 1359 mm->unmap_area = arch_unmap_area; 1360 } 1361 #endif 1362 1363 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask); 1364 extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 1365 1366 #ifdef CONFIG_MAGIC_SYSRQ 1367 1368 extern void normalize_rt_tasks(void); 1369 1370 #endif 1371 1372 #ifdef CONFIG_PM 1373 /* 1374 * Check if a process has been frozen 1375 */ 1376 static inline int frozen(struct task_struct *p) 1377 { 1378 return p->flags & PF_FROZEN; 1379 } 1380 1381 /* 1382 * Check if there is a request to freeze a process 1383 */ 1384 static inline int freezing(struct task_struct *p) 1385 { 1386 return p->flags & PF_FREEZE; 1387 } 1388 1389 /* 1390 * Request that a process be frozen 1391 * FIXME: SMP problem. We may not modify other process' flags! 1392 */ 1393 static inline void freeze(struct task_struct *p) 1394 { 1395 p->flags |= PF_FREEZE; 1396 } 1397 1398 /* 1399 * Wake up a frozen process 1400 */ 1401 static inline int thaw_process(struct task_struct *p) 1402 { 1403 if (frozen(p)) { 1404 p->flags &= ~PF_FROZEN; 1405 wake_up_process(p); 1406 return 1; 1407 } 1408 return 0; 1409 } 1410 1411 /* 1412 * freezing is complete, mark process as frozen 1413 */ 1414 static inline void frozen_process(struct task_struct *p) 1415 { 1416 p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN; 1417 } 1418 1419 extern void refrigerator(void); 1420 extern int freeze_processes(void); 1421 extern void thaw_processes(void); 1422 1423 static inline int try_to_freeze(void) 1424 { 1425 if (freezing(current)) { 1426 refrigerator(); 1427 return 1; 1428 } else 1429 return 0; 1430 } 1431 #else 1432 static inline int frozen(struct task_struct *p) { return 0; } 1433 static inline int freezing(struct task_struct *p) { return 0; } 1434 static inline void freeze(struct task_struct *p) { BUG(); } 1435 static inline int thaw_process(struct task_struct *p) { return 1; } 1436 static inline void frozen_process(struct task_struct *p) { BUG(); } 1437 1438 static inline void refrigerator(void) {} 1439 static inline int freeze_processes(void) { BUG(); return 0; } 1440 static inline void thaw_processes(void) {} 1441 1442 static inline int try_to_freeze(void) { return 0; } 1443 1444 #endif /* CONFIG_PM */ 1445 #endif /* __KERNEL__ */ 1446 1447 #endif 1448