1 #ifndef _LINUX_SCHED_H 2 #define _LINUX_SCHED_H 3 4 /* 5 * cloning flags: 6 */ 7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */ 9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 15 #define CLONE_THREAD 0x00010000 /* Same thread group? */ 16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */ 17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */ 25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */ 26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */ 27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */ 28 #define CLONE_NEWNET 0x40000000 /* New network namespace */ 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 /* SCHED_ISO: reserved but not implemented yet */ 38 #define SCHED_IDLE 5 39 40 #ifdef __KERNEL__ 41 42 struct sched_param { 43 int sched_priority; 44 }; 45 46 #include <asm/param.h> /* for HZ */ 47 48 #include <linux/capability.h> 49 #include <linux/threads.h> 50 #include <linux/kernel.h> 51 #include <linux/types.h> 52 #include <linux/timex.h> 53 #include <linux/jiffies.h> 54 #include <linux/rbtree.h> 55 #include <linux/thread_info.h> 56 #include <linux/cpumask.h> 57 #include <linux/errno.h> 58 #include <linux/nodemask.h> 59 #include <linux/mm_types.h> 60 61 #include <asm/system.h> 62 #include <asm/semaphore.h> 63 #include <asm/page.h> 64 #include <asm/ptrace.h> 65 #include <asm/cputime.h> 66 67 #include <linux/smp.h> 68 #include <linux/sem.h> 69 #include <linux/signal.h> 70 #include <linux/securebits.h> 71 #include <linux/fs_struct.h> 72 #include <linux/compiler.h> 73 #include <linux/completion.h> 74 #include <linux/pid.h> 75 #include <linux/percpu.h> 76 #include <linux/topology.h> 77 #include <linux/seccomp.h> 78 #include <linux/rcupdate.h> 79 #include <linux/futex.h> 80 #include <linux/rtmutex.h> 81 82 #include <linux/time.h> 83 #include <linux/param.h> 84 #include <linux/resource.h> 85 #include <linux/timer.h> 86 #include <linux/hrtimer.h> 87 #include <linux/task_io_accounting.h> 88 #include <linux/kobject.h> 89 90 #include <asm/processor.h> 91 92 struct exec_domain; 93 struct futex_pi_state; 94 struct bio; 95 96 /* 97 * List of flags we want to share for kernel threads, 98 * if only because they are not used by them anyway. 99 */ 100 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 101 102 /* 103 * These are the constant used to fake the fixed-point load-average 104 * counting. Some notes: 105 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 106 * a load-average precision of 10 bits integer + 11 bits fractional 107 * - if you want to count load-averages more often, you need more 108 * precision, or rounding will get you. With 2-second counting freq, 109 * the EXP_n values would be 1981, 2034 and 2043 if still using only 110 * 11 bit fractions. 111 */ 112 extern unsigned long avenrun[]; /* Load averages */ 113 114 #define FSHIFT 11 /* nr of bits of precision */ 115 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 116 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 117 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 118 #define EXP_5 2014 /* 1/exp(5sec/5min) */ 119 #define EXP_15 2037 /* 1/exp(5sec/15min) */ 120 121 #define CALC_LOAD(load,exp,n) \ 122 load *= exp; \ 123 load += n*(FIXED_1-exp); \ 124 load >>= FSHIFT; 125 126 extern unsigned long total_forks; 127 extern int nr_threads; 128 DECLARE_PER_CPU(unsigned long, process_counts); 129 extern int nr_processes(void); 130 extern unsigned long nr_running(void); 131 extern unsigned long nr_uninterruptible(void); 132 extern unsigned long nr_active(void); 133 extern unsigned long nr_iowait(void); 134 extern unsigned long weighted_cpuload(const int cpu); 135 136 struct seq_file; 137 struct cfs_rq; 138 struct task_group; 139 #ifdef CONFIG_SCHED_DEBUG 140 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 141 extern void proc_sched_set_task(struct task_struct *p); 142 extern void 143 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 144 #else 145 static inline void 146 proc_sched_show_task(struct task_struct *p, struct seq_file *m) 147 { 148 } 149 static inline void proc_sched_set_task(struct task_struct *p) 150 { 151 } 152 static inline void 153 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 154 { 155 } 156 #endif 157 158 /* 159 * Task state bitmask. NOTE! These bits are also 160 * encoded in fs/proc/array.c: get_task_state(). 161 * 162 * We have two separate sets of flags: task->state 163 * is about runnability, while task->exit_state are 164 * about the task exiting. Confusing, but this way 165 * modifying one set can't modify the other one by 166 * mistake. 167 */ 168 #define TASK_RUNNING 0 169 #define TASK_INTERRUPTIBLE 1 170 #define TASK_UNINTERRUPTIBLE 2 171 #define TASK_STOPPED 4 172 #define TASK_TRACED 8 173 /* in tsk->exit_state */ 174 #define EXIT_ZOMBIE 16 175 #define EXIT_DEAD 32 176 /* in tsk->state again */ 177 #define TASK_DEAD 64 178 179 #define __set_task_state(tsk, state_value) \ 180 do { (tsk)->state = (state_value); } while (0) 181 #define set_task_state(tsk, state_value) \ 182 set_mb((tsk)->state, (state_value)) 183 184 /* 185 * set_current_state() includes a barrier so that the write of current->state 186 * is correctly serialised wrt the caller's subsequent test of whether to 187 * actually sleep: 188 * 189 * set_current_state(TASK_UNINTERRUPTIBLE); 190 * if (do_i_need_to_sleep()) 191 * schedule(); 192 * 193 * If the caller does not need such serialisation then use __set_current_state() 194 */ 195 #define __set_current_state(state_value) \ 196 do { current->state = (state_value); } while (0) 197 #define set_current_state(state_value) \ 198 set_mb(current->state, (state_value)) 199 200 /* Task command name length */ 201 #define TASK_COMM_LEN 16 202 203 #include <linux/spinlock.h> 204 205 /* 206 * This serializes "schedule()" and also protects 207 * the run-queue from deletions/modifications (but 208 * _adding_ to the beginning of the run-queue has 209 * a separate lock). 210 */ 211 extern rwlock_t tasklist_lock; 212 extern spinlock_t mmlist_lock; 213 214 struct task_struct; 215 216 extern void sched_init(void); 217 extern void sched_init_smp(void); 218 extern void init_idle(struct task_struct *idle, int cpu); 219 extern void init_idle_bootup_task(struct task_struct *idle); 220 221 extern cpumask_t nohz_cpu_mask; 222 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 223 extern int select_nohz_load_balancer(int cpu); 224 #else 225 static inline int select_nohz_load_balancer(int cpu) 226 { 227 return 0; 228 } 229 #endif 230 231 /* 232 * Only dump TASK_* tasks. (0 for all tasks) 233 */ 234 extern void show_state_filter(unsigned long state_filter); 235 236 static inline void show_state(void) 237 { 238 show_state_filter(0); 239 } 240 241 extern void show_regs(struct pt_regs *); 242 243 /* 244 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 245 * task), SP is the stack pointer of the first frame that should be shown in the back 246 * trace (or NULL if the entire call-chain of the task should be shown). 247 */ 248 extern void show_stack(struct task_struct *task, unsigned long *sp); 249 250 void io_schedule(void); 251 long io_schedule_timeout(long timeout); 252 253 extern void cpu_init (void); 254 extern void trap_init(void); 255 extern void update_process_times(int user); 256 extern void scheduler_tick(void); 257 258 #ifdef CONFIG_DETECT_SOFTLOCKUP 259 extern void softlockup_tick(void); 260 extern void spawn_softlockup_task(void); 261 extern void touch_softlockup_watchdog(void); 262 extern void touch_all_softlockup_watchdogs(void); 263 #else 264 static inline void softlockup_tick(void) 265 { 266 } 267 static inline void spawn_softlockup_task(void) 268 { 269 } 270 static inline void touch_softlockup_watchdog(void) 271 { 272 } 273 static inline void touch_all_softlockup_watchdogs(void) 274 { 275 } 276 #endif 277 278 279 /* Attach to any functions which should be ignored in wchan output. */ 280 #define __sched __attribute__((__section__(".sched.text"))) 281 /* Is this address in the __sched functions? */ 282 extern int in_sched_functions(unsigned long addr); 283 284 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 285 extern signed long FASTCALL(schedule_timeout(signed long timeout)); 286 extern signed long schedule_timeout_interruptible(signed long timeout); 287 extern signed long schedule_timeout_uninterruptible(signed long timeout); 288 asmlinkage void schedule(void); 289 290 struct nsproxy; 291 struct user_namespace; 292 293 /* Maximum number of active map areas.. This is a random (large) number */ 294 #define DEFAULT_MAX_MAP_COUNT 65536 295 296 extern int sysctl_max_map_count; 297 298 #include <linux/aio.h> 299 300 extern unsigned long 301 arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 302 unsigned long, unsigned long); 303 extern unsigned long 304 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 305 unsigned long len, unsigned long pgoff, 306 unsigned long flags); 307 extern void arch_unmap_area(struct mm_struct *, unsigned long); 308 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 309 310 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS 311 /* 312 * The mm counters are not protected by its page_table_lock, 313 * so must be incremented atomically. 314 */ 315 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 316 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 317 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 318 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 319 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 320 321 #else /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 322 /* 323 * The mm counters are protected by its page_table_lock, 324 * so can be incremented directly. 325 */ 326 #define set_mm_counter(mm, member, value) (mm)->_##member = (value) 327 #define get_mm_counter(mm, member) ((mm)->_##member) 328 #define add_mm_counter(mm, member, value) (mm)->_##member += (value) 329 #define inc_mm_counter(mm, member) (mm)->_##member++ 330 #define dec_mm_counter(mm, member) (mm)->_##member-- 331 332 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ 333 334 #define get_mm_rss(mm) \ 335 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 336 #define update_hiwater_rss(mm) do { \ 337 unsigned long _rss = get_mm_rss(mm); \ 338 if ((mm)->hiwater_rss < _rss) \ 339 (mm)->hiwater_rss = _rss; \ 340 } while (0) 341 #define update_hiwater_vm(mm) do { \ 342 if ((mm)->hiwater_vm < (mm)->total_vm) \ 343 (mm)->hiwater_vm = (mm)->total_vm; \ 344 } while (0) 345 346 extern void set_dumpable(struct mm_struct *mm, int value); 347 extern int get_dumpable(struct mm_struct *mm); 348 349 /* mm flags */ 350 /* dumpable bits */ 351 #define MMF_DUMPABLE 0 /* core dump is permitted */ 352 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 353 #define MMF_DUMPABLE_BITS 2 354 355 /* coredump filter bits */ 356 #define MMF_DUMP_ANON_PRIVATE 2 357 #define MMF_DUMP_ANON_SHARED 3 358 #define MMF_DUMP_MAPPED_PRIVATE 4 359 #define MMF_DUMP_MAPPED_SHARED 5 360 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 361 #define MMF_DUMP_FILTER_BITS 4 362 #define MMF_DUMP_FILTER_MASK \ 363 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 364 #define MMF_DUMP_FILTER_DEFAULT \ 365 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED)) 366 367 struct sighand_struct { 368 atomic_t count; 369 struct k_sigaction action[_NSIG]; 370 spinlock_t siglock; 371 wait_queue_head_t signalfd_wqh; 372 }; 373 374 struct pacct_struct { 375 int ac_flag; 376 long ac_exitcode; 377 unsigned long ac_mem; 378 cputime_t ac_utime, ac_stime; 379 unsigned long ac_minflt, ac_majflt; 380 }; 381 382 /* 383 * NOTE! "signal_struct" does not have it's own 384 * locking, because a shared signal_struct always 385 * implies a shared sighand_struct, so locking 386 * sighand_struct is always a proper superset of 387 * the locking of signal_struct. 388 */ 389 struct signal_struct { 390 atomic_t count; 391 atomic_t live; 392 393 wait_queue_head_t wait_chldexit; /* for wait4() */ 394 395 /* current thread group signal load-balancing target: */ 396 struct task_struct *curr_target; 397 398 /* shared signal handling: */ 399 struct sigpending shared_pending; 400 401 /* thread group exit support */ 402 int group_exit_code; 403 /* overloaded: 404 * - notify group_exit_task when ->count is equal to notify_count 405 * - everyone except group_exit_task is stopped during signal delivery 406 * of fatal signals, group_exit_task processes the signal. 407 */ 408 struct task_struct *group_exit_task; 409 int notify_count; 410 411 /* thread group stop support, overloads group_exit_code too */ 412 int group_stop_count; 413 unsigned int flags; /* see SIGNAL_* flags below */ 414 415 /* POSIX.1b Interval Timers */ 416 struct list_head posix_timers; 417 418 /* ITIMER_REAL timer for the process */ 419 struct hrtimer real_timer; 420 struct task_struct *tsk; 421 ktime_t it_real_incr; 422 423 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 424 cputime_t it_prof_expires, it_virt_expires; 425 cputime_t it_prof_incr, it_virt_incr; 426 427 /* job control IDs */ 428 pid_t pgrp; 429 struct pid *tty_old_pgrp; 430 431 union { 432 pid_t session __deprecated; 433 pid_t __session; 434 }; 435 436 /* boolean value for session group leader */ 437 int leader; 438 439 struct tty_struct *tty; /* NULL if no tty */ 440 441 /* 442 * Cumulative resource counters for dead threads in the group, 443 * and for reaped dead child processes forked by this group. 444 * Live threads maintain their own counters and add to these 445 * in __exit_signal, except for the group leader. 446 */ 447 cputime_t utime, stime, cutime, cstime; 448 cputime_t gtime; 449 cputime_t cgtime; 450 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 451 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 452 unsigned long inblock, oublock, cinblock, coublock; 453 454 /* 455 * Cumulative ns of scheduled CPU time for dead threads in the 456 * group, not including a zombie group leader. (This only differs 457 * from jiffies_to_ns(utime + stime) if sched_clock uses something 458 * other than jiffies.) 459 */ 460 unsigned long long sum_sched_runtime; 461 462 /* 463 * We don't bother to synchronize most readers of this at all, 464 * because there is no reader checking a limit that actually needs 465 * to get both rlim_cur and rlim_max atomically, and either one 466 * alone is a single word that can safely be read normally. 467 * getrlimit/setrlimit use task_lock(current->group_leader) to 468 * protect this instead of the siglock, because they really 469 * have no need to disable irqs. 470 */ 471 struct rlimit rlim[RLIM_NLIMITS]; 472 473 struct list_head cpu_timers[3]; 474 475 /* keep the process-shared keyrings here so that they do the right 476 * thing in threads created with CLONE_THREAD */ 477 #ifdef CONFIG_KEYS 478 struct key *session_keyring; /* keyring inherited over fork */ 479 struct key *process_keyring; /* keyring private to this process */ 480 #endif 481 #ifdef CONFIG_BSD_PROCESS_ACCT 482 struct pacct_struct pacct; /* per-process accounting information */ 483 #endif 484 #ifdef CONFIG_TASKSTATS 485 struct taskstats *stats; 486 #endif 487 #ifdef CONFIG_AUDIT 488 unsigned audit_tty; 489 struct tty_audit_buf *tty_audit_buf; 490 #endif 491 }; 492 493 /* Context switch must be unlocked if interrupts are to be enabled */ 494 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 495 # define __ARCH_WANT_UNLOCKED_CTXSW 496 #endif 497 498 /* 499 * Bits in flags field of signal_struct. 500 */ 501 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 502 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 503 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 504 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 505 506 /* 507 * Some day this will be a full-fledged user tracking system.. 508 */ 509 struct user_struct { 510 atomic_t __count; /* reference count */ 511 atomic_t processes; /* How many processes does this user have? */ 512 atomic_t files; /* How many open files does this user have? */ 513 atomic_t sigpending; /* How many pending signals does this user have? */ 514 #ifdef CONFIG_INOTIFY_USER 515 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 516 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 517 #endif 518 /* protected by mq_lock */ 519 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 520 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 521 522 #ifdef CONFIG_KEYS 523 struct key *uid_keyring; /* UID specific keyring */ 524 struct key *session_keyring; /* UID's default session keyring */ 525 #endif 526 527 /* Hash table maintenance information */ 528 struct hlist_node uidhash_node; 529 uid_t uid; 530 531 #ifdef CONFIG_FAIR_USER_SCHED 532 struct task_group *tg; 533 struct kset kset; 534 struct subsys_attribute user_attr; 535 struct work_struct work; 536 #endif 537 }; 538 539 #ifdef CONFIG_FAIR_USER_SCHED 540 extern int uids_kobject_init(void); 541 #else 542 static inline int uids_kobject_init(void) { return 0; } 543 #endif 544 545 extern struct user_struct *find_user(uid_t); 546 547 extern struct user_struct root_user; 548 #define INIT_USER (&root_user) 549 550 struct backing_dev_info; 551 struct reclaim_state; 552 553 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 554 struct sched_info { 555 /* cumulative counters */ 556 unsigned long pcount; /* # of times run on this cpu */ 557 unsigned long long cpu_time, /* time spent on the cpu */ 558 run_delay; /* time spent waiting on a runqueue */ 559 560 /* timestamps */ 561 unsigned long long last_arrival,/* when we last ran on a cpu */ 562 last_queued; /* when we were last queued to run */ 563 #ifdef CONFIG_SCHEDSTATS 564 /* BKL stats */ 565 unsigned long bkl_count; 566 #endif 567 }; 568 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 569 570 #ifdef CONFIG_SCHEDSTATS 571 extern const struct file_operations proc_schedstat_operations; 572 #endif /* CONFIG_SCHEDSTATS */ 573 574 #ifdef CONFIG_TASK_DELAY_ACCT 575 struct task_delay_info { 576 spinlock_t lock; 577 unsigned int flags; /* Private per-task flags */ 578 579 /* For each stat XXX, add following, aligned appropriately 580 * 581 * struct timespec XXX_start, XXX_end; 582 * u64 XXX_delay; 583 * u32 XXX_count; 584 * 585 * Atomicity of updates to XXX_delay, XXX_count protected by 586 * single lock above (split into XXX_lock if contention is an issue). 587 */ 588 589 /* 590 * XXX_count is incremented on every XXX operation, the delay 591 * associated with the operation is added to XXX_delay. 592 * XXX_delay contains the accumulated delay time in nanoseconds. 593 */ 594 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 595 u64 blkio_delay; /* wait for sync block io completion */ 596 u64 swapin_delay; /* wait for swapin block io completion */ 597 u32 blkio_count; /* total count of the number of sync block */ 598 /* io operations performed */ 599 u32 swapin_count; /* total count of the number of swapin block */ 600 /* io operations performed */ 601 }; 602 #endif /* CONFIG_TASK_DELAY_ACCT */ 603 604 static inline int sched_info_on(void) 605 { 606 #ifdef CONFIG_SCHEDSTATS 607 return 1; 608 #elif defined(CONFIG_TASK_DELAY_ACCT) 609 extern int delayacct_on; 610 return delayacct_on; 611 #else 612 return 0; 613 #endif 614 } 615 616 enum cpu_idle_type { 617 CPU_IDLE, 618 CPU_NOT_IDLE, 619 CPU_NEWLY_IDLE, 620 CPU_MAX_IDLE_TYPES 621 }; 622 623 /* 624 * sched-domains (multiprocessor balancing) declarations: 625 */ 626 627 /* 628 * Increase resolution of nice-level calculations: 629 */ 630 #define SCHED_LOAD_SHIFT 10 631 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 632 633 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 634 635 #ifdef CONFIG_SMP 636 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 637 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 638 #define SD_BALANCE_EXEC 4 /* Balance on exec */ 639 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 640 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 641 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 642 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 643 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 644 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 645 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 646 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 647 648 #define BALANCE_FOR_MC_POWER \ 649 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0) 650 651 #define BALANCE_FOR_PKG_POWER \ 652 ((sched_mc_power_savings || sched_smt_power_savings) ? \ 653 SD_POWERSAVINGS_BALANCE : 0) 654 655 #define test_sd_parent(sd, flag) ((sd->parent && \ 656 (sd->parent->flags & flag)) ? 1 : 0) 657 658 659 struct sched_group { 660 struct sched_group *next; /* Must be a circular list */ 661 cpumask_t cpumask; 662 663 /* 664 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 665 * single CPU. This is read only (except for setup, hotplug CPU). 666 * Note : Never change cpu_power without recompute its reciprocal 667 */ 668 unsigned int __cpu_power; 669 /* 670 * reciprocal value of cpu_power to avoid expensive divides 671 * (see include/linux/reciprocal_div.h) 672 */ 673 u32 reciprocal_cpu_power; 674 }; 675 676 struct sched_domain { 677 /* These fields must be setup */ 678 struct sched_domain *parent; /* top domain must be null terminated */ 679 struct sched_domain *child; /* bottom domain must be null terminated */ 680 struct sched_group *groups; /* the balancing groups of the domain */ 681 cpumask_t span; /* span of all CPUs in this domain */ 682 unsigned long min_interval; /* Minimum balance interval ms */ 683 unsigned long max_interval; /* Maximum balance interval ms */ 684 unsigned int busy_factor; /* less balancing by factor if busy */ 685 unsigned int imbalance_pct; /* No balance until over watermark */ 686 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 687 unsigned int busy_idx; 688 unsigned int idle_idx; 689 unsigned int newidle_idx; 690 unsigned int wake_idx; 691 unsigned int forkexec_idx; 692 int flags; /* See SD_* */ 693 694 /* Runtime fields. */ 695 unsigned long last_balance; /* init to jiffies. units in jiffies */ 696 unsigned int balance_interval; /* initialise to 1. units in ms. */ 697 unsigned int nr_balance_failed; /* initialise to 0 */ 698 699 #ifdef CONFIG_SCHEDSTATS 700 /* load_balance() stats */ 701 unsigned long lb_count[CPU_MAX_IDLE_TYPES]; 702 unsigned long lb_failed[CPU_MAX_IDLE_TYPES]; 703 unsigned long lb_balanced[CPU_MAX_IDLE_TYPES]; 704 unsigned long lb_imbalance[CPU_MAX_IDLE_TYPES]; 705 unsigned long lb_gained[CPU_MAX_IDLE_TYPES]; 706 unsigned long lb_hot_gained[CPU_MAX_IDLE_TYPES]; 707 unsigned long lb_nobusyg[CPU_MAX_IDLE_TYPES]; 708 unsigned long lb_nobusyq[CPU_MAX_IDLE_TYPES]; 709 710 /* Active load balancing */ 711 unsigned long alb_count; 712 unsigned long alb_failed; 713 unsigned long alb_pushed; 714 715 /* SD_BALANCE_EXEC stats */ 716 unsigned long sbe_count; 717 unsigned long sbe_balanced; 718 unsigned long sbe_pushed; 719 720 /* SD_BALANCE_FORK stats */ 721 unsigned long sbf_count; 722 unsigned long sbf_balanced; 723 unsigned long sbf_pushed; 724 725 /* try_to_wake_up() stats */ 726 unsigned long ttwu_wake_remote; 727 unsigned long ttwu_move_affine; 728 unsigned long ttwu_move_balance; 729 #endif 730 }; 731 732 extern int partition_sched_domains(cpumask_t *partition1, 733 cpumask_t *partition2); 734 735 #endif /* CONFIG_SMP */ 736 737 /* 738 * A runqueue laden with a single nice 0 task scores a weighted_cpuload of 739 * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a 740 * task of nice 0 or enough lower priority tasks to bring up the 741 * weighted_cpuload 742 */ 743 static inline int above_background_load(void) 744 { 745 unsigned long cpu; 746 747 for_each_online_cpu(cpu) { 748 if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE) 749 return 1; 750 } 751 return 0; 752 } 753 754 struct io_context; /* See blkdev.h */ 755 struct cpuset; 756 757 #define NGROUPS_SMALL 32 758 #define NGROUPS_PER_BLOCK ((int)(PAGE_SIZE / sizeof(gid_t))) 759 struct group_info { 760 int ngroups; 761 atomic_t usage; 762 gid_t small_block[NGROUPS_SMALL]; 763 int nblocks; 764 gid_t *blocks[0]; 765 }; 766 767 /* 768 * get_group_info() must be called with the owning task locked (via task_lock()) 769 * when task != current. The reason being that the vast majority of callers are 770 * looking at current->group_info, which can not be changed except by the 771 * current task. Changing current->group_info requires the task lock, too. 772 */ 773 #define get_group_info(group_info) do { \ 774 atomic_inc(&(group_info)->usage); \ 775 } while (0) 776 777 #define put_group_info(group_info) do { \ 778 if (atomic_dec_and_test(&(group_info)->usage)) \ 779 groups_free(group_info); \ 780 } while (0) 781 782 extern struct group_info *groups_alloc(int gidsetsize); 783 extern void groups_free(struct group_info *group_info); 784 extern int set_current_groups(struct group_info *group_info); 785 extern int groups_search(struct group_info *group_info, gid_t grp); 786 /* access the groups "array" with this macro */ 787 #define GROUP_AT(gi, i) \ 788 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 789 790 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 791 extern void prefetch_stack(struct task_struct *t); 792 #else 793 static inline void prefetch_stack(struct task_struct *t) { } 794 #endif 795 796 struct audit_context; /* See audit.c */ 797 struct mempolicy; 798 struct pipe_inode_info; 799 struct uts_namespace; 800 801 struct rq; 802 struct sched_domain; 803 804 struct sched_class { 805 const struct sched_class *next; 806 807 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 808 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 809 void (*yield_task) (struct rq *rq); 810 811 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p); 812 813 struct task_struct * (*pick_next_task) (struct rq *rq); 814 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 815 816 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 817 struct rq *busiest, 818 unsigned long max_nr_move, unsigned long max_load_move, 819 struct sched_domain *sd, enum cpu_idle_type idle, 820 int *all_pinned, int *this_best_prio); 821 822 void (*set_curr_task) (struct rq *rq); 823 void (*task_tick) (struct rq *rq, struct task_struct *p); 824 void (*task_new) (struct rq *rq, struct task_struct *p); 825 }; 826 827 struct load_weight { 828 unsigned long weight, inv_weight; 829 }; 830 831 /* 832 * CFS stats for a schedulable entity (task, task-group etc) 833 * 834 * Current field usage histogram: 835 * 836 * 4 se->block_start 837 * 4 se->run_node 838 * 4 se->sleep_start 839 * 6 se->load.weight 840 */ 841 struct sched_entity { 842 struct load_weight load; /* for load-balancing */ 843 struct rb_node run_node; 844 unsigned int on_rq; 845 int peer_preempt; 846 847 u64 exec_start; 848 u64 sum_exec_runtime; 849 u64 vruntime; 850 u64 prev_sum_exec_runtime; 851 852 #ifdef CONFIG_SCHEDSTATS 853 u64 wait_start; 854 u64 wait_max; 855 856 u64 sleep_start; 857 u64 sleep_max; 858 s64 sum_sleep_runtime; 859 860 u64 block_start; 861 u64 block_max; 862 u64 exec_max; 863 u64 slice_max; 864 865 u64 nr_migrations; 866 u64 nr_migrations_cold; 867 u64 nr_failed_migrations_affine; 868 u64 nr_failed_migrations_running; 869 u64 nr_failed_migrations_hot; 870 u64 nr_forced_migrations; 871 u64 nr_forced2_migrations; 872 873 u64 nr_wakeups; 874 u64 nr_wakeups_sync; 875 u64 nr_wakeups_migrate; 876 u64 nr_wakeups_local; 877 u64 nr_wakeups_remote; 878 u64 nr_wakeups_affine; 879 u64 nr_wakeups_affine_attempts; 880 u64 nr_wakeups_passive; 881 u64 nr_wakeups_idle; 882 #endif 883 884 #ifdef CONFIG_FAIR_GROUP_SCHED 885 struct sched_entity *parent; 886 /* rq on which this entity is (to be) queued: */ 887 struct cfs_rq *cfs_rq; 888 /* rq "owned" by this entity/group: */ 889 struct cfs_rq *my_q; 890 #endif 891 }; 892 893 struct task_struct { 894 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 895 void *stack; 896 atomic_t usage; 897 unsigned int flags; /* per process flags, defined below */ 898 unsigned int ptrace; 899 900 int lock_depth; /* BKL lock depth */ 901 902 #ifdef CONFIG_SMP 903 #ifdef __ARCH_WANT_UNLOCKED_CTXSW 904 int oncpu; 905 #endif 906 #endif 907 908 int prio, static_prio, normal_prio; 909 struct list_head run_list; 910 const struct sched_class *sched_class; 911 struct sched_entity se; 912 913 #ifdef CONFIG_PREEMPT_NOTIFIERS 914 /* list of struct preempt_notifier: */ 915 struct hlist_head preempt_notifiers; 916 #endif 917 918 unsigned short ioprio; 919 #ifdef CONFIG_BLK_DEV_IO_TRACE 920 unsigned int btrace_seq; 921 #endif 922 923 unsigned int policy; 924 cpumask_t cpus_allowed; 925 unsigned int time_slice; 926 927 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 928 struct sched_info sched_info; 929 #endif 930 931 struct list_head tasks; 932 /* 933 * ptrace_list/ptrace_children forms the list of my children 934 * that were stolen by a ptracer. 935 */ 936 struct list_head ptrace_children; 937 struct list_head ptrace_list; 938 939 struct mm_struct *mm, *active_mm; 940 941 /* task state */ 942 struct linux_binfmt *binfmt; 943 int exit_state; 944 int exit_code, exit_signal; 945 int pdeath_signal; /* The signal sent when the parent dies */ 946 /* ??? */ 947 unsigned int personality; 948 unsigned did_exec:1; 949 pid_t pid; 950 pid_t tgid; 951 952 #ifdef CONFIG_CC_STACKPROTECTOR 953 /* Canary value for the -fstack-protector gcc feature */ 954 unsigned long stack_canary; 955 #endif 956 /* 957 * pointers to (original) parent process, youngest child, younger sibling, 958 * older sibling, respectively. (p->father can be replaced with 959 * p->parent->pid) 960 */ 961 struct task_struct *real_parent; /* real parent process (when being debugged) */ 962 struct task_struct *parent; /* parent process */ 963 /* 964 * children/sibling forms the list of my children plus the 965 * tasks I'm ptracing. 966 */ 967 struct list_head children; /* list of my children */ 968 struct list_head sibling; /* linkage in my parent's children list */ 969 struct task_struct *group_leader; /* threadgroup leader */ 970 971 /* PID/PID hash table linkage. */ 972 struct pid_link pids[PIDTYPE_MAX]; 973 struct list_head thread_group; 974 975 struct completion *vfork_done; /* for vfork() */ 976 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 977 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 978 979 unsigned int rt_priority; 980 cputime_t utime, stime; 981 cputime_t gtime; 982 unsigned long nvcsw, nivcsw; /* context switch counts */ 983 struct timespec start_time; /* monotonic time */ 984 struct timespec real_start_time; /* boot based time */ 985 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 986 unsigned long min_flt, maj_flt; 987 988 cputime_t it_prof_expires, it_virt_expires; 989 unsigned long long it_sched_expires; 990 struct list_head cpu_timers[3]; 991 992 /* process credentials */ 993 uid_t uid,euid,suid,fsuid; 994 gid_t gid,egid,sgid,fsgid; 995 struct group_info *group_info; 996 kernel_cap_t cap_effective, cap_inheritable, cap_permitted; 997 unsigned keep_capabilities:1; 998 struct user_struct *user; 999 #ifdef CONFIG_KEYS 1000 struct key *request_key_auth; /* assumed request_key authority */ 1001 struct key *thread_keyring; /* keyring private to this thread */ 1002 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 1003 #endif 1004 /* 1005 * fpu_counter contains the number of consecutive context switches 1006 * that the FPU is used. If this is over a threshold, the lazy fpu 1007 * saving becomes unlazy to save the trap. This is an unsigned char 1008 * so that after 256 times the counter wraps and the behavior turns 1009 * lazy again; this to deal with bursty apps that only use FPU for 1010 * a short time 1011 */ 1012 unsigned char fpu_counter; 1013 int oomkilladj; /* OOM kill score adjustment (bit shift). */ 1014 char comm[TASK_COMM_LEN]; /* executable name excluding path 1015 - access with [gs]et_task_comm (which lock 1016 it with task_lock()) 1017 - initialized normally by flush_old_exec */ 1018 /* file system info */ 1019 int link_count, total_link_count; 1020 #ifdef CONFIG_SYSVIPC 1021 /* ipc stuff */ 1022 struct sysv_sem sysvsem; 1023 #endif 1024 /* CPU-specific state of this task */ 1025 struct thread_struct thread; 1026 /* filesystem information */ 1027 struct fs_struct *fs; 1028 /* open file information */ 1029 struct files_struct *files; 1030 /* namespaces */ 1031 struct nsproxy *nsproxy; 1032 /* signal handlers */ 1033 struct signal_struct *signal; 1034 struct sighand_struct *sighand; 1035 1036 sigset_t blocked, real_blocked; 1037 sigset_t saved_sigmask; /* To be restored with TIF_RESTORE_SIGMASK */ 1038 struct sigpending pending; 1039 1040 unsigned long sas_ss_sp; 1041 size_t sas_ss_size; 1042 int (*notifier)(void *priv); 1043 void *notifier_data; 1044 sigset_t *notifier_mask; 1045 1046 void *security; 1047 struct audit_context *audit_context; 1048 seccomp_t seccomp; 1049 1050 /* Thread group tracking */ 1051 u32 parent_exec_id; 1052 u32 self_exec_id; 1053 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 1054 spinlock_t alloc_lock; 1055 1056 /* Protection of the PI data structures: */ 1057 spinlock_t pi_lock; 1058 1059 #ifdef CONFIG_RT_MUTEXES 1060 /* PI waiters blocked on a rt_mutex held by this task */ 1061 struct plist_head pi_waiters; 1062 /* Deadlock detection and priority inheritance handling */ 1063 struct rt_mutex_waiter *pi_blocked_on; 1064 #endif 1065 1066 #ifdef CONFIG_DEBUG_MUTEXES 1067 /* mutex deadlock detection */ 1068 struct mutex_waiter *blocked_on; 1069 #endif 1070 #ifdef CONFIG_TRACE_IRQFLAGS 1071 unsigned int irq_events; 1072 int hardirqs_enabled; 1073 unsigned long hardirq_enable_ip; 1074 unsigned int hardirq_enable_event; 1075 unsigned long hardirq_disable_ip; 1076 unsigned int hardirq_disable_event; 1077 int softirqs_enabled; 1078 unsigned long softirq_disable_ip; 1079 unsigned int softirq_disable_event; 1080 unsigned long softirq_enable_ip; 1081 unsigned int softirq_enable_event; 1082 int hardirq_context; 1083 int softirq_context; 1084 #endif 1085 #ifdef CONFIG_LOCKDEP 1086 # define MAX_LOCK_DEPTH 30UL 1087 u64 curr_chain_key; 1088 int lockdep_depth; 1089 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1090 unsigned int lockdep_recursion; 1091 #endif 1092 1093 /* journalling filesystem info */ 1094 void *journal_info; 1095 1096 /* stacked block device info */ 1097 struct bio *bio_list, **bio_tail; 1098 1099 /* VM state */ 1100 struct reclaim_state *reclaim_state; 1101 1102 struct backing_dev_info *backing_dev_info; 1103 1104 struct io_context *io_context; 1105 1106 unsigned long ptrace_message; 1107 siginfo_t *last_siginfo; /* For ptrace use. */ 1108 /* 1109 * current io wait handle: wait queue entry to use for io waits 1110 * If this thread is processing aio, this points at the waitqueue 1111 * inside the currently handled kiocb. It may be NULL (i.e. default 1112 * to a stack based synchronous wait) if its doing sync IO. 1113 */ 1114 wait_queue_t *io_wait; 1115 #ifdef CONFIG_TASK_XACCT 1116 /* i/o counters(bytes read/written, #syscalls */ 1117 u64 rchar, wchar, syscr, syscw; 1118 #endif 1119 struct task_io_accounting ioac; 1120 #if defined(CONFIG_TASK_XACCT) 1121 u64 acct_rss_mem1; /* accumulated rss usage */ 1122 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1123 cputime_t acct_stimexpd;/* stime since last update */ 1124 #endif 1125 #ifdef CONFIG_NUMA 1126 struct mempolicy *mempolicy; 1127 short il_next; 1128 #endif 1129 #ifdef CONFIG_CPUSETS 1130 struct cpuset *cpuset; 1131 nodemask_t mems_allowed; 1132 int cpuset_mems_generation; 1133 int cpuset_mem_spread_rotor; 1134 #endif 1135 struct robust_list_head __user *robust_list; 1136 #ifdef CONFIG_COMPAT 1137 struct compat_robust_list_head __user *compat_robust_list; 1138 #endif 1139 struct list_head pi_state_list; 1140 struct futex_pi_state *pi_state_cache; 1141 1142 atomic_t fs_excl; /* holding fs exclusive resources */ 1143 struct rcu_head rcu; 1144 1145 /* 1146 * cache last used pipe for splice 1147 */ 1148 struct pipe_inode_info *splice_pipe; 1149 #ifdef CONFIG_TASK_DELAY_ACCT 1150 struct task_delay_info *delays; 1151 #endif 1152 #ifdef CONFIG_FAULT_INJECTION 1153 int make_it_fail; 1154 #endif 1155 }; 1156 1157 /* 1158 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1159 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1160 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1161 * values are inverted: lower p->prio value means higher priority. 1162 * 1163 * The MAX_USER_RT_PRIO value allows the actual maximum 1164 * RT priority to be separate from the value exported to 1165 * user-space. This allows kernel threads to set their 1166 * priority to a value higher than any user task. Note: 1167 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1168 */ 1169 1170 #define MAX_USER_RT_PRIO 100 1171 #define MAX_RT_PRIO MAX_USER_RT_PRIO 1172 1173 #define MAX_PRIO (MAX_RT_PRIO + 40) 1174 #define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1175 1176 static inline int rt_prio(int prio) 1177 { 1178 if (unlikely(prio < MAX_RT_PRIO)) 1179 return 1; 1180 return 0; 1181 } 1182 1183 static inline int rt_task(struct task_struct *p) 1184 { 1185 return rt_prio(p->prio); 1186 } 1187 1188 static inline pid_t process_group(struct task_struct *tsk) 1189 { 1190 return tsk->signal->pgrp; 1191 } 1192 1193 static inline pid_t signal_session(struct signal_struct *sig) 1194 { 1195 return sig->__session; 1196 } 1197 1198 static inline pid_t process_session(struct task_struct *tsk) 1199 { 1200 return signal_session(tsk->signal); 1201 } 1202 1203 static inline void set_signal_session(struct signal_struct *sig, pid_t session) 1204 { 1205 sig->__session = session; 1206 } 1207 1208 static inline struct pid *task_pid(struct task_struct *task) 1209 { 1210 return task->pids[PIDTYPE_PID].pid; 1211 } 1212 1213 static inline struct pid *task_tgid(struct task_struct *task) 1214 { 1215 return task->group_leader->pids[PIDTYPE_PID].pid; 1216 } 1217 1218 static inline struct pid *task_pgrp(struct task_struct *task) 1219 { 1220 return task->group_leader->pids[PIDTYPE_PGID].pid; 1221 } 1222 1223 static inline struct pid *task_session(struct task_struct *task) 1224 { 1225 return task->group_leader->pids[PIDTYPE_SID].pid; 1226 } 1227 1228 /** 1229 * pid_alive - check that a task structure is not stale 1230 * @p: Task structure to be checked. 1231 * 1232 * Test if a process is not yet dead (at most zombie state) 1233 * If pid_alive fails, then pointers within the task structure 1234 * can be stale and must not be dereferenced. 1235 */ 1236 static inline int pid_alive(struct task_struct *p) 1237 { 1238 return p->pids[PIDTYPE_PID].pid != NULL; 1239 } 1240 1241 /** 1242 * is_init - check if a task structure is init 1243 * @tsk: Task structure to be checked. 1244 * 1245 * Check if a task structure is the first user space task the kernel created. 1246 */ 1247 static inline int is_init(struct task_struct *tsk) 1248 { 1249 return tsk->pid == 1; 1250 } 1251 1252 extern struct pid *cad_pid; 1253 1254 extern void free_task(struct task_struct *tsk); 1255 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1256 1257 extern void __put_task_struct(struct task_struct *t); 1258 1259 static inline void put_task_struct(struct task_struct *t) 1260 { 1261 if (atomic_dec_and_test(&t->usage)) 1262 __put_task_struct(t); 1263 } 1264 1265 /* 1266 * Per process flags 1267 */ 1268 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1269 /* Not implemented yet, only for 486*/ 1270 #define PF_STARTING 0x00000002 /* being created */ 1271 #define PF_EXITING 0x00000004 /* getting shut down */ 1272 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1273 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1274 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1275 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1276 #define PF_DUMPCORE 0x00000200 /* dumped core */ 1277 #define PF_SIGNALED 0x00000400 /* killed by a signal */ 1278 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1279 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1280 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1281 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1282 #define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1283 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1284 #define PF_KSWAPD 0x00040000 /* I am kswapd */ 1285 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1286 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1287 #define PF_BORROWED_MM 0x00200000 /* I am a kthread doing use_mm */ 1288 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1289 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1290 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1291 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1292 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1293 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1294 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1295 1296 /* 1297 * Only the _current_ task can read/write to tsk->flags, but other 1298 * tasks can access tsk->flags in readonly mode for example 1299 * with tsk_used_math (like during threaded core dumping). 1300 * There is however an exception to this rule during ptrace 1301 * or during fork: the ptracer task is allowed to write to the 1302 * child->flags of its traced child (same goes for fork, the parent 1303 * can write to the child->flags), because we're guaranteed the 1304 * child is not running and in turn not changing child->flags 1305 * at the same time the parent does it. 1306 */ 1307 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1308 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1309 #define clear_used_math() clear_stopped_child_used_math(current) 1310 #define set_used_math() set_stopped_child_used_math(current) 1311 #define conditional_stopped_child_used_math(condition, child) \ 1312 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1313 #define conditional_used_math(condition) \ 1314 conditional_stopped_child_used_math(condition, current) 1315 #define copy_to_stopped_child_used_math(child) \ 1316 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1317 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1318 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1319 #define used_math() tsk_used_math(current) 1320 1321 #ifdef CONFIG_SMP 1322 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask); 1323 #else 1324 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1325 { 1326 if (!cpu_isset(0, new_mask)) 1327 return -EINVAL; 1328 return 0; 1329 } 1330 #endif 1331 1332 extern unsigned long long sched_clock(void); 1333 1334 /* 1335 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1336 * clock constructed from sched_clock(): 1337 */ 1338 extern unsigned long long cpu_clock(int cpu); 1339 1340 extern unsigned long long 1341 task_sched_runtime(struct task_struct *task); 1342 1343 /* sched_exec is called by processes performing an exec */ 1344 #ifdef CONFIG_SMP 1345 extern void sched_exec(void); 1346 #else 1347 #define sched_exec() {} 1348 #endif 1349 1350 extern void sched_clock_idle_sleep_event(void); 1351 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1352 1353 #ifdef CONFIG_HOTPLUG_CPU 1354 extern void idle_task_exit(void); 1355 #else 1356 static inline void idle_task_exit(void) {} 1357 #endif 1358 1359 extern void sched_idle_next(void); 1360 1361 #ifdef CONFIG_SCHED_DEBUG 1362 extern unsigned int sysctl_sched_latency; 1363 extern unsigned int sysctl_sched_nr_latency; 1364 extern unsigned int sysctl_sched_wakeup_granularity; 1365 extern unsigned int sysctl_sched_batch_wakeup_granularity; 1366 extern unsigned int sysctl_sched_child_runs_first; 1367 extern unsigned int sysctl_sched_features; 1368 extern unsigned int sysctl_sched_migration_cost; 1369 #endif 1370 1371 extern unsigned int sysctl_sched_compat_yield; 1372 1373 #ifdef CONFIG_RT_MUTEXES 1374 extern int rt_mutex_getprio(struct task_struct *p); 1375 extern void rt_mutex_setprio(struct task_struct *p, int prio); 1376 extern void rt_mutex_adjust_pi(struct task_struct *p); 1377 #else 1378 static inline int rt_mutex_getprio(struct task_struct *p) 1379 { 1380 return p->normal_prio; 1381 } 1382 # define rt_mutex_adjust_pi(p) do { } while (0) 1383 #endif 1384 1385 extern void set_user_nice(struct task_struct *p, long nice); 1386 extern int task_prio(const struct task_struct *p); 1387 extern int task_nice(const struct task_struct *p); 1388 extern int can_nice(const struct task_struct *p, const int nice); 1389 extern int task_curr(const struct task_struct *p); 1390 extern int idle_cpu(int cpu); 1391 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1392 extern struct task_struct *idle_task(int cpu); 1393 extern struct task_struct *curr_task(int cpu); 1394 extern void set_curr_task(int cpu, struct task_struct *p); 1395 1396 void yield(void); 1397 1398 /* 1399 * The default (Linux) execution domain. 1400 */ 1401 extern struct exec_domain default_exec_domain; 1402 1403 union thread_union { 1404 struct thread_info thread_info; 1405 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1406 }; 1407 1408 #ifndef __HAVE_ARCH_KSTACK_END 1409 static inline int kstack_end(void *addr) 1410 { 1411 /* Reliable end of stack detection: 1412 * Some APM bios versions misalign the stack 1413 */ 1414 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1415 } 1416 #endif 1417 1418 extern union thread_union init_thread_union; 1419 extern struct task_struct init_task; 1420 1421 extern struct mm_struct init_mm; 1422 1423 #define find_task_by_pid(nr) find_task_by_pid_type(PIDTYPE_PID, nr) 1424 extern struct task_struct *find_task_by_pid_type(int type, int pid); 1425 extern void __set_special_pids(pid_t session, pid_t pgrp); 1426 1427 /* per-UID process charging. */ 1428 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1429 static inline struct user_struct *get_uid(struct user_struct *u) 1430 { 1431 atomic_inc(&u->__count); 1432 return u; 1433 } 1434 extern void free_uid(struct user_struct *); 1435 extern void switch_uid(struct user_struct *); 1436 extern void release_uids(struct user_namespace *ns); 1437 1438 #include <asm/current.h> 1439 1440 extern void do_timer(unsigned long ticks); 1441 1442 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state)); 1443 extern int FASTCALL(wake_up_process(struct task_struct * tsk)); 1444 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk, 1445 unsigned long clone_flags)); 1446 #ifdef CONFIG_SMP 1447 extern void kick_process(struct task_struct *tsk); 1448 #else 1449 static inline void kick_process(struct task_struct *tsk) { } 1450 #endif 1451 extern void sched_fork(struct task_struct *p, int clone_flags); 1452 extern void sched_dead(struct task_struct *p); 1453 1454 extern int in_group_p(gid_t); 1455 extern int in_egroup_p(gid_t); 1456 1457 extern void proc_caches_init(void); 1458 extern void flush_signals(struct task_struct *); 1459 extern void ignore_signals(struct task_struct *); 1460 extern void flush_signal_handlers(struct task_struct *, int force_default); 1461 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1462 1463 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1464 { 1465 unsigned long flags; 1466 int ret; 1467 1468 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1469 ret = dequeue_signal(tsk, mask, info); 1470 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1471 1472 return ret; 1473 } 1474 1475 extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1476 sigset_t *mask); 1477 extern void unblock_all_signals(void); 1478 extern void release_task(struct task_struct * p); 1479 extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1480 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *); 1481 extern int force_sigsegv(int, struct task_struct *); 1482 extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1483 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1484 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1485 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1486 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1487 extern int kill_pgrp(struct pid *pid, int sig, int priv); 1488 extern int kill_pid(struct pid *pid, int sig, int priv); 1489 extern int kill_proc_info(int, struct siginfo *, pid_t); 1490 extern void do_notify_parent(struct task_struct *, int); 1491 extern void force_sig(int, struct task_struct *); 1492 extern void force_sig_specific(int, struct task_struct *); 1493 extern int send_sig(int, struct task_struct *, int); 1494 extern void zap_other_threads(struct task_struct *p); 1495 extern int kill_proc(pid_t, int, int); 1496 extern struct sigqueue *sigqueue_alloc(void); 1497 extern void sigqueue_free(struct sigqueue *); 1498 extern int send_sigqueue(int, struct sigqueue *, struct task_struct *); 1499 extern int send_group_sigqueue(int, struct sigqueue *, struct task_struct *); 1500 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1501 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1502 1503 static inline int kill_cad_pid(int sig, int priv) 1504 { 1505 return kill_pid(cad_pid, sig, priv); 1506 } 1507 1508 /* These can be the second arg to send_sig_info/send_group_sig_info. */ 1509 #define SEND_SIG_NOINFO ((struct siginfo *) 0) 1510 #define SEND_SIG_PRIV ((struct siginfo *) 1) 1511 #define SEND_SIG_FORCED ((struct siginfo *) 2) 1512 1513 static inline int is_si_special(const struct siginfo *info) 1514 { 1515 return info <= SEND_SIG_FORCED; 1516 } 1517 1518 /* True if we are on the alternate signal stack. */ 1519 1520 static inline int on_sig_stack(unsigned long sp) 1521 { 1522 return (sp - current->sas_ss_sp < current->sas_ss_size); 1523 } 1524 1525 static inline int sas_ss_flags(unsigned long sp) 1526 { 1527 return (current->sas_ss_size == 0 ? SS_DISABLE 1528 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1529 } 1530 1531 /* 1532 * Routines for handling mm_structs 1533 */ 1534 extern struct mm_struct * mm_alloc(void); 1535 1536 /* mmdrop drops the mm and the page tables */ 1537 extern void FASTCALL(__mmdrop(struct mm_struct *)); 1538 static inline void mmdrop(struct mm_struct * mm) 1539 { 1540 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 1541 __mmdrop(mm); 1542 } 1543 1544 /* mmput gets rid of the mappings and all user-space */ 1545 extern void mmput(struct mm_struct *); 1546 /* Grab a reference to a task's mm, if it is not already going away */ 1547 extern struct mm_struct *get_task_mm(struct task_struct *task); 1548 /* Remove the current tasks stale references to the old mm_struct */ 1549 extern void mm_release(struct task_struct *, struct mm_struct *); 1550 1551 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1552 extern void flush_thread(void); 1553 extern void exit_thread(void); 1554 1555 extern void exit_files(struct task_struct *); 1556 extern void __cleanup_signal(struct signal_struct *); 1557 extern void __cleanup_sighand(struct sighand_struct *); 1558 extern void exit_itimers(struct signal_struct *); 1559 1560 extern NORET_TYPE void do_group_exit(int); 1561 1562 extern void daemonize(const char *, ...); 1563 extern int allow_signal(int); 1564 extern int disallow_signal(int); 1565 1566 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1567 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1568 struct task_struct *fork_idle(int); 1569 1570 extern void set_task_comm(struct task_struct *tsk, char *from); 1571 extern void get_task_comm(char *to, struct task_struct *tsk); 1572 1573 #ifdef CONFIG_SMP 1574 extern void wait_task_inactive(struct task_struct * p); 1575 #else 1576 #define wait_task_inactive(p) do { } while (0) 1577 #endif 1578 1579 #define remove_parent(p) list_del_init(&(p)->sibling) 1580 #define add_parent(p) list_add_tail(&(p)->sibling,&(p)->parent->children) 1581 1582 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1583 1584 #define for_each_process(p) \ 1585 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1586 1587 /* 1588 * Careful: do_each_thread/while_each_thread is a double loop so 1589 * 'break' will not work as expected - use goto instead. 1590 */ 1591 #define do_each_thread(g, t) \ 1592 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1593 1594 #define while_each_thread(g, t) \ 1595 while ((t = next_thread(t)) != g) 1596 1597 /* de_thread depends on thread_group_leader not being a pid based check */ 1598 #define thread_group_leader(p) (p == p->group_leader) 1599 1600 /* Do to the insanities of de_thread it is possible for a process 1601 * to have the pid of the thread group leader without actually being 1602 * the thread group leader. For iteration through the pids in proc 1603 * all we care about is that we have a task with the appropriate 1604 * pid, we don't actually care if we have the right task. 1605 */ 1606 static inline int has_group_leader_pid(struct task_struct *p) 1607 { 1608 return p->pid == p->tgid; 1609 } 1610 1611 static inline struct task_struct *next_thread(const struct task_struct *p) 1612 { 1613 return list_entry(rcu_dereference(p->thread_group.next), 1614 struct task_struct, thread_group); 1615 } 1616 1617 static inline int thread_group_empty(struct task_struct *p) 1618 { 1619 return list_empty(&p->thread_group); 1620 } 1621 1622 #define delay_group_leader(p) \ 1623 (thread_group_leader(p) && !thread_group_empty(p)) 1624 1625 /* 1626 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 1627 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1628 * pins the final release of task.io_context. Also protects ->cpuset. 1629 * 1630 * Nests both inside and outside of read_lock(&tasklist_lock). 1631 * It must not be nested with write_lock_irq(&tasklist_lock), 1632 * neither inside nor outside. 1633 */ 1634 static inline void task_lock(struct task_struct *p) 1635 { 1636 spin_lock(&p->alloc_lock); 1637 } 1638 1639 static inline void task_unlock(struct task_struct *p) 1640 { 1641 spin_unlock(&p->alloc_lock); 1642 } 1643 1644 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 1645 unsigned long *flags); 1646 1647 static inline void unlock_task_sighand(struct task_struct *tsk, 1648 unsigned long *flags) 1649 { 1650 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 1651 } 1652 1653 #ifndef __HAVE_THREAD_FUNCTIONS 1654 1655 #define task_thread_info(task) ((struct thread_info *)(task)->stack) 1656 #define task_stack_page(task) ((task)->stack) 1657 1658 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 1659 { 1660 *task_thread_info(p) = *task_thread_info(org); 1661 task_thread_info(p)->task = p; 1662 } 1663 1664 static inline unsigned long *end_of_stack(struct task_struct *p) 1665 { 1666 return (unsigned long *)(task_thread_info(p) + 1); 1667 } 1668 1669 #endif 1670 1671 /* set thread flags in other task's structures 1672 * - see asm/thread_info.h for TIF_xxxx flags available 1673 */ 1674 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 1675 { 1676 set_ti_thread_flag(task_thread_info(tsk), flag); 1677 } 1678 1679 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1680 { 1681 clear_ti_thread_flag(task_thread_info(tsk), flag); 1682 } 1683 1684 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 1685 { 1686 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 1687 } 1688 1689 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 1690 { 1691 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 1692 } 1693 1694 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 1695 { 1696 return test_ti_thread_flag(task_thread_info(tsk), flag); 1697 } 1698 1699 static inline void set_tsk_need_resched(struct task_struct *tsk) 1700 { 1701 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1702 } 1703 1704 static inline void clear_tsk_need_resched(struct task_struct *tsk) 1705 { 1706 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 1707 } 1708 1709 static inline int signal_pending(struct task_struct *p) 1710 { 1711 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 1712 } 1713 1714 static inline int need_resched(void) 1715 { 1716 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 1717 } 1718 1719 /* 1720 * cond_resched() and cond_resched_lock(): latency reduction via 1721 * explicit rescheduling in places that are safe. The return 1722 * value indicates whether a reschedule was done in fact. 1723 * cond_resched_lock() will drop the spinlock before scheduling, 1724 * cond_resched_softirq() will enable bhs before scheduling. 1725 */ 1726 extern int cond_resched(void); 1727 extern int cond_resched_lock(spinlock_t * lock); 1728 extern int cond_resched_softirq(void); 1729 1730 /* 1731 * Does a critical section need to be broken due to another 1732 * task waiting?: 1733 */ 1734 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP) 1735 # define need_lockbreak(lock) ((lock)->break_lock) 1736 #else 1737 # define need_lockbreak(lock) 0 1738 #endif 1739 1740 /* 1741 * Does a critical section need to be broken due to another 1742 * task waiting or preemption being signalled: 1743 */ 1744 static inline int lock_need_resched(spinlock_t *lock) 1745 { 1746 if (need_lockbreak(lock) || need_resched()) 1747 return 1; 1748 return 0; 1749 } 1750 1751 /* 1752 * Reevaluate whether the task has signals pending delivery. 1753 * Wake the task if so. 1754 * This is required every time the blocked sigset_t changes. 1755 * callers must hold sighand->siglock. 1756 */ 1757 extern void recalc_sigpending_and_wake(struct task_struct *t); 1758 extern void recalc_sigpending(void); 1759 1760 extern void signal_wake_up(struct task_struct *t, int resume_stopped); 1761 1762 /* 1763 * Wrappers for p->thread_info->cpu access. No-op on UP. 1764 */ 1765 #ifdef CONFIG_SMP 1766 1767 static inline unsigned int task_cpu(const struct task_struct *p) 1768 { 1769 return task_thread_info(p)->cpu; 1770 } 1771 1772 extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 1773 1774 #else 1775 1776 static inline unsigned int task_cpu(const struct task_struct *p) 1777 { 1778 return 0; 1779 } 1780 1781 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 1782 { 1783 } 1784 1785 #endif /* CONFIG_SMP */ 1786 1787 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT 1788 extern void arch_pick_mmap_layout(struct mm_struct *mm); 1789 #else 1790 static inline void arch_pick_mmap_layout(struct mm_struct *mm) 1791 { 1792 mm->mmap_base = TASK_UNMAPPED_BASE; 1793 mm->get_unmapped_area = arch_get_unmapped_area; 1794 mm->unmap_area = arch_unmap_area; 1795 } 1796 #endif 1797 1798 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask); 1799 extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 1800 1801 extern int sched_mc_power_savings, sched_smt_power_savings; 1802 1803 extern void normalize_rt_tasks(void); 1804 1805 #ifdef CONFIG_FAIR_GROUP_SCHED 1806 1807 extern struct task_group init_task_group; 1808 1809 extern struct task_group *sched_create_group(void); 1810 extern void sched_destroy_group(struct task_group *tg); 1811 extern void sched_move_task(struct task_struct *tsk); 1812 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 1813 extern unsigned long sched_group_shares(struct task_group *tg); 1814 1815 #endif 1816 1817 #ifdef CONFIG_TASK_XACCT 1818 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 1819 { 1820 tsk->rchar += amt; 1821 } 1822 1823 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 1824 { 1825 tsk->wchar += amt; 1826 } 1827 1828 static inline void inc_syscr(struct task_struct *tsk) 1829 { 1830 tsk->syscr++; 1831 } 1832 1833 static inline void inc_syscw(struct task_struct *tsk) 1834 { 1835 tsk->syscw++; 1836 } 1837 #else 1838 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 1839 { 1840 } 1841 1842 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 1843 { 1844 } 1845 1846 static inline void inc_syscr(struct task_struct *tsk) 1847 { 1848 } 1849 1850 static inline void inc_syscw(struct task_struct *tsk) 1851 { 1852 } 1853 #endif 1854 1855 #endif /* __KERNEL__ */ 1856 1857 #endif 1858