1 #ifndef _LINUX_SCHED_H 2 #define _LINUX_SCHED_H 3 4 /* 5 * cloning flags: 6 */ 7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */ 9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 15 #define CLONE_THREAD 0x00010000 /* Same thread group? */ 16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */ 17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */ 25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */ 26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */ 27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */ 28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */ 29 #define CLONE_NEWNET 0x40000000 /* New network namespace */ 30 #define CLONE_IO 0x80000000 /* Clone io context */ 31 32 /* 33 * Scheduling policies 34 */ 35 #define SCHED_NORMAL 0 36 #define SCHED_FIFO 1 37 #define SCHED_RR 2 38 #define SCHED_BATCH 3 39 /* SCHED_ISO: reserved but not implemented yet */ 40 #define SCHED_IDLE 5 41 42 #ifdef __KERNEL__ 43 44 struct sched_param { 45 int sched_priority; 46 }; 47 48 #include <asm/param.h> /* for HZ */ 49 50 #include <linux/capability.h> 51 #include <linux/threads.h> 52 #include <linux/kernel.h> 53 #include <linux/types.h> 54 #include <linux/timex.h> 55 #include <linux/jiffies.h> 56 #include <linux/rbtree.h> 57 #include <linux/thread_info.h> 58 #include <linux/cpumask.h> 59 #include <linux/errno.h> 60 #include <linux/nodemask.h> 61 #include <linux/mm_types.h> 62 63 #include <asm/system.h> 64 #include <asm/page.h> 65 #include <asm/ptrace.h> 66 #include <asm/cputime.h> 67 68 #include <linux/smp.h> 69 #include <linux/sem.h> 70 #include <linux/signal.h> 71 #include <linux/fs_struct.h> 72 #include <linux/compiler.h> 73 #include <linux/completion.h> 74 #include <linux/pid.h> 75 #include <linux/percpu.h> 76 #include <linux/topology.h> 77 #include <linux/proportions.h> 78 #include <linux/seccomp.h> 79 #include <linux/rcupdate.h> 80 #include <linux/rtmutex.h> 81 82 #include <linux/time.h> 83 #include <linux/param.h> 84 #include <linux/resource.h> 85 #include <linux/timer.h> 86 #include <linux/hrtimer.h> 87 #include <linux/task_io_accounting.h> 88 #include <linux/kobject.h> 89 #include <linux/latencytop.h> 90 #include <linux/cred.h> 91 92 #include <asm/processor.h> 93 94 struct mem_cgroup; 95 struct exec_domain; 96 struct futex_pi_state; 97 struct robust_list_head; 98 struct bio; 99 100 /* 101 * List of flags we want to share for kernel threads, 102 * if only because they are not used by them anyway. 103 */ 104 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 105 106 /* 107 * These are the constant used to fake the fixed-point load-average 108 * counting. Some notes: 109 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 110 * a load-average precision of 10 bits integer + 11 bits fractional 111 * - if you want to count load-averages more often, you need more 112 * precision, or rounding will get you. With 2-second counting freq, 113 * the EXP_n values would be 1981, 2034 and 2043 if still using only 114 * 11 bit fractions. 115 */ 116 extern unsigned long avenrun[]; /* Load averages */ 117 118 #define FSHIFT 11 /* nr of bits of precision */ 119 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 120 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 121 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 122 #define EXP_5 2014 /* 1/exp(5sec/5min) */ 123 #define EXP_15 2037 /* 1/exp(5sec/15min) */ 124 125 #define CALC_LOAD(load,exp,n) \ 126 load *= exp; \ 127 load += n*(FIXED_1-exp); \ 128 load >>= FSHIFT; 129 130 extern unsigned long total_forks; 131 extern int nr_threads; 132 DECLARE_PER_CPU(unsigned long, process_counts); 133 extern int nr_processes(void); 134 extern unsigned long nr_running(void); 135 extern unsigned long nr_uninterruptible(void); 136 extern unsigned long nr_active(void); 137 extern unsigned long nr_iowait(void); 138 139 struct seq_file; 140 struct cfs_rq; 141 struct task_group; 142 #ifdef CONFIG_SCHED_DEBUG 143 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 144 extern void proc_sched_set_task(struct task_struct *p); 145 extern void 146 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 147 #else 148 static inline void 149 proc_sched_show_task(struct task_struct *p, struct seq_file *m) 150 { 151 } 152 static inline void proc_sched_set_task(struct task_struct *p) 153 { 154 } 155 static inline void 156 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 157 { 158 } 159 #endif 160 161 extern unsigned long long time_sync_thresh; 162 163 /* 164 * Task state bitmask. NOTE! These bits are also 165 * encoded in fs/proc/array.c: get_task_state(). 166 * 167 * We have two separate sets of flags: task->state 168 * is about runnability, while task->exit_state are 169 * about the task exiting. Confusing, but this way 170 * modifying one set can't modify the other one by 171 * mistake. 172 */ 173 #define TASK_RUNNING 0 174 #define TASK_INTERRUPTIBLE 1 175 #define TASK_UNINTERRUPTIBLE 2 176 #define __TASK_STOPPED 4 177 #define __TASK_TRACED 8 178 /* in tsk->exit_state */ 179 #define EXIT_ZOMBIE 16 180 #define EXIT_DEAD 32 181 /* in tsk->state again */ 182 #define TASK_DEAD 64 183 #define TASK_WAKEKILL 128 184 185 /* Convenience macros for the sake of set_task_state */ 186 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 187 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 188 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 189 190 /* Convenience macros for the sake of wake_up */ 191 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 192 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 193 194 /* get_task_state() */ 195 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 196 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 197 __TASK_TRACED) 198 199 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 200 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 201 #define task_is_stopped_or_traced(task) \ 202 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 203 #define task_contributes_to_load(task) \ 204 ((task->state & TASK_UNINTERRUPTIBLE) != 0) 205 206 #define __set_task_state(tsk, state_value) \ 207 do { (tsk)->state = (state_value); } while (0) 208 #define set_task_state(tsk, state_value) \ 209 set_mb((tsk)->state, (state_value)) 210 211 /* 212 * set_current_state() includes a barrier so that the write of current->state 213 * is correctly serialised wrt the caller's subsequent test of whether to 214 * actually sleep: 215 * 216 * set_current_state(TASK_UNINTERRUPTIBLE); 217 * if (do_i_need_to_sleep()) 218 * schedule(); 219 * 220 * If the caller does not need such serialisation then use __set_current_state() 221 */ 222 #define __set_current_state(state_value) \ 223 do { current->state = (state_value); } while (0) 224 #define set_current_state(state_value) \ 225 set_mb(current->state, (state_value)) 226 227 /* Task command name length */ 228 #define TASK_COMM_LEN 16 229 230 #include <linux/spinlock.h> 231 232 /* 233 * This serializes "schedule()" and also protects 234 * the run-queue from deletions/modifications (but 235 * _adding_ to the beginning of the run-queue has 236 * a separate lock). 237 */ 238 extern rwlock_t tasklist_lock; 239 extern spinlock_t mmlist_lock; 240 241 struct task_struct; 242 243 extern void sched_init(void); 244 extern void sched_init_smp(void); 245 extern asmlinkage void schedule_tail(struct task_struct *prev); 246 extern void init_idle(struct task_struct *idle, int cpu); 247 extern void init_idle_bootup_task(struct task_struct *idle); 248 249 extern int runqueue_is_locked(void); 250 extern void task_rq_unlock_wait(struct task_struct *p); 251 252 extern cpumask_t nohz_cpu_mask; 253 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 254 extern int select_nohz_load_balancer(int cpu); 255 #else 256 static inline int select_nohz_load_balancer(int cpu) 257 { 258 return 0; 259 } 260 #endif 261 262 extern unsigned long rt_needs_cpu(int cpu); 263 264 /* 265 * Only dump TASK_* tasks. (0 for all tasks) 266 */ 267 extern void show_state_filter(unsigned long state_filter); 268 269 static inline void show_state(void) 270 { 271 show_state_filter(0); 272 } 273 274 extern void show_regs(struct pt_regs *); 275 276 /* 277 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 278 * task), SP is the stack pointer of the first frame that should be shown in the back 279 * trace (or NULL if the entire call-chain of the task should be shown). 280 */ 281 extern void show_stack(struct task_struct *task, unsigned long *sp); 282 283 void io_schedule(void); 284 long io_schedule_timeout(long timeout); 285 286 extern void cpu_init (void); 287 extern void trap_init(void); 288 extern void account_process_tick(struct task_struct *task, int user); 289 extern void update_process_times(int user); 290 extern void scheduler_tick(void); 291 292 extern void sched_show_task(struct task_struct *p); 293 294 #ifdef CONFIG_DETECT_SOFTLOCKUP 295 extern void softlockup_tick(void); 296 extern void touch_softlockup_watchdog(void); 297 extern void touch_all_softlockup_watchdogs(void); 298 extern unsigned int softlockup_panic; 299 extern unsigned long sysctl_hung_task_check_count; 300 extern unsigned long sysctl_hung_task_timeout_secs; 301 extern unsigned long sysctl_hung_task_warnings; 302 extern int softlockup_thresh; 303 #else 304 static inline void softlockup_tick(void) 305 { 306 } 307 static inline void spawn_softlockup_task(void) 308 { 309 } 310 static inline void touch_softlockup_watchdog(void) 311 { 312 } 313 static inline void touch_all_softlockup_watchdogs(void) 314 { 315 } 316 #endif 317 318 319 /* Attach to any functions which should be ignored in wchan output. */ 320 #define __sched __attribute__((__section__(".sched.text"))) 321 322 /* Linker adds these: start and end of __sched functions */ 323 extern char __sched_text_start[], __sched_text_end[]; 324 325 /* Is this address in the __sched functions? */ 326 extern int in_sched_functions(unsigned long addr); 327 328 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 329 extern signed long schedule_timeout(signed long timeout); 330 extern signed long schedule_timeout_interruptible(signed long timeout); 331 extern signed long schedule_timeout_killable(signed long timeout); 332 extern signed long schedule_timeout_uninterruptible(signed long timeout); 333 asmlinkage void schedule(void); 334 335 struct nsproxy; 336 struct user_namespace; 337 338 /* Maximum number of active map areas.. This is a random (large) number */ 339 #define DEFAULT_MAX_MAP_COUNT 65536 340 341 extern int sysctl_max_map_count; 342 343 #include <linux/aio.h> 344 345 extern unsigned long 346 arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 347 unsigned long, unsigned long); 348 extern unsigned long 349 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 350 unsigned long len, unsigned long pgoff, 351 unsigned long flags); 352 extern void arch_unmap_area(struct mm_struct *, unsigned long); 353 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 354 355 #if USE_SPLIT_PTLOCKS 356 /* 357 * The mm counters are not protected by its page_table_lock, 358 * so must be incremented atomically. 359 */ 360 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 361 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 362 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 363 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 364 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 365 366 #else /* !USE_SPLIT_PTLOCKS */ 367 /* 368 * The mm counters are protected by its page_table_lock, 369 * so can be incremented directly. 370 */ 371 #define set_mm_counter(mm, member, value) (mm)->_##member = (value) 372 #define get_mm_counter(mm, member) ((mm)->_##member) 373 #define add_mm_counter(mm, member, value) (mm)->_##member += (value) 374 #define inc_mm_counter(mm, member) (mm)->_##member++ 375 #define dec_mm_counter(mm, member) (mm)->_##member-- 376 377 #endif /* !USE_SPLIT_PTLOCKS */ 378 379 #define get_mm_rss(mm) \ 380 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 381 #define update_hiwater_rss(mm) do { \ 382 unsigned long _rss = get_mm_rss(mm); \ 383 if ((mm)->hiwater_rss < _rss) \ 384 (mm)->hiwater_rss = _rss; \ 385 } while (0) 386 #define update_hiwater_vm(mm) do { \ 387 if ((mm)->hiwater_vm < (mm)->total_vm) \ 388 (mm)->hiwater_vm = (mm)->total_vm; \ 389 } while (0) 390 391 extern void set_dumpable(struct mm_struct *mm, int value); 392 extern int get_dumpable(struct mm_struct *mm); 393 394 /* mm flags */ 395 /* dumpable bits */ 396 #define MMF_DUMPABLE 0 /* core dump is permitted */ 397 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 398 #define MMF_DUMPABLE_BITS 2 399 400 /* coredump filter bits */ 401 #define MMF_DUMP_ANON_PRIVATE 2 402 #define MMF_DUMP_ANON_SHARED 3 403 #define MMF_DUMP_MAPPED_PRIVATE 4 404 #define MMF_DUMP_MAPPED_SHARED 5 405 #define MMF_DUMP_ELF_HEADERS 6 406 #define MMF_DUMP_HUGETLB_PRIVATE 7 407 #define MMF_DUMP_HUGETLB_SHARED 8 408 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 409 #define MMF_DUMP_FILTER_BITS 7 410 #define MMF_DUMP_FILTER_MASK \ 411 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 412 #define MMF_DUMP_FILTER_DEFAULT \ 413 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ 414 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) 415 416 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS 417 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) 418 #else 419 # define MMF_DUMP_MASK_DEFAULT_ELF 0 420 #endif 421 422 struct sighand_struct { 423 atomic_t count; 424 struct k_sigaction action[_NSIG]; 425 spinlock_t siglock; 426 wait_queue_head_t signalfd_wqh; 427 }; 428 429 struct pacct_struct { 430 int ac_flag; 431 long ac_exitcode; 432 unsigned long ac_mem; 433 cputime_t ac_utime, ac_stime; 434 unsigned long ac_minflt, ac_majflt; 435 }; 436 437 /** 438 * struct task_cputime - collected CPU time counts 439 * @utime: time spent in user mode, in &cputime_t units 440 * @stime: time spent in kernel mode, in &cputime_t units 441 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds 442 * 443 * This structure groups together three kinds of CPU time that are 444 * tracked for threads and thread groups. Most things considering 445 * CPU time want to group these counts together and treat all three 446 * of them in parallel. 447 */ 448 struct task_cputime { 449 cputime_t utime; 450 cputime_t stime; 451 unsigned long long sum_exec_runtime; 452 }; 453 /* Alternate field names when used to cache expirations. */ 454 #define prof_exp stime 455 #define virt_exp utime 456 #define sched_exp sum_exec_runtime 457 458 /** 459 * struct thread_group_cputime - thread group interval timer counts 460 * @totals: thread group interval timers; substructure for 461 * uniprocessor kernel, per-cpu for SMP kernel. 462 * 463 * This structure contains the version of task_cputime, above, that is 464 * used for thread group CPU clock calculations. 465 */ 466 struct thread_group_cputime { 467 struct task_cputime *totals; 468 }; 469 470 /* 471 * NOTE! "signal_struct" does not have it's own 472 * locking, because a shared signal_struct always 473 * implies a shared sighand_struct, so locking 474 * sighand_struct is always a proper superset of 475 * the locking of signal_struct. 476 */ 477 struct signal_struct { 478 atomic_t count; 479 atomic_t live; 480 481 wait_queue_head_t wait_chldexit; /* for wait4() */ 482 483 /* current thread group signal load-balancing target: */ 484 struct task_struct *curr_target; 485 486 /* shared signal handling: */ 487 struct sigpending shared_pending; 488 489 /* thread group exit support */ 490 int group_exit_code; 491 /* overloaded: 492 * - notify group_exit_task when ->count is equal to notify_count 493 * - everyone except group_exit_task is stopped during signal delivery 494 * of fatal signals, group_exit_task processes the signal. 495 */ 496 int notify_count; 497 struct task_struct *group_exit_task; 498 499 /* thread group stop support, overloads group_exit_code too */ 500 int group_stop_count; 501 unsigned int flags; /* see SIGNAL_* flags below */ 502 503 /* POSIX.1b Interval Timers */ 504 struct list_head posix_timers; 505 506 /* ITIMER_REAL timer for the process */ 507 struct hrtimer real_timer; 508 struct pid *leader_pid; 509 ktime_t it_real_incr; 510 511 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 512 cputime_t it_prof_expires, it_virt_expires; 513 cputime_t it_prof_incr, it_virt_incr; 514 515 /* 516 * Thread group totals for process CPU clocks. 517 * See thread_group_cputime(), et al, for details. 518 */ 519 struct thread_group_cputime cputime; 520 521 /* Earliest-expiration cache. */ 522 struct task_cputime cputime_expires; 523 524 struct list_head cpu_timers[3]; 525 526 /* job control IDs */ 527 528 /* 529 * pgrp and session fields are deprecated. 530 * use the task_session_Xnr and task_pgrp_Xnr routines below 531 */ 532 533 union { 534 pid_t pgrp __deprecated; 535 pid_t __pgrp; 536 }; 537 538 struct pid *tty_old_pgrp; 539 540 union { 541 pid_t session __deprecated; 542 pid_t __session; 543 }; 544 545 /* boolean value for session group leader */ 546 int leader; 547 548 struct tty_struct *tty; /* NULL if no tty */ 549 550 /* 551 * Cumulative resource counters for dead threads in the group, 552 * and for reaped dead child processes forked by this group. 553 * Live threads maintain their own counters and add to these 554 * in __exit_signal, except for the group leader. 555 */ 556 cputime_t cutime, cstime; 557 cputime_t gtime; 558 cputime_t cgtime; 559 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 560 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 561 unsigned long inblock, oublock, cinblock, coublock; 562 struct task_io_accounting ioac; 563 564 /* 565 * We don't bother to synchronize most readers of this at all, 566 * because there is no reader checking a limit that actually needs 567 * to get both rlim_cur and rlim_max atomically, and either one 568 * alone is a single word that can safely be read normally. 569 * getrlimit/setrlimit use task_lock(current->group_leader) to 570 * protect this instead of the siglock, because they really 571 * have no need to disable irqs. 572 */ 573 struct rlimit rlim[RLIM_NLIMITS]; 574 575 /* keep the process-shared keyrings here so that they do the right 576 * thing in threads created with CLONE_THREAD */ 577 #ifdef CONFIG_KEYS 578 struct key *session_keyring; /* keyring inherited over fork */ 579 struct key *process_keyring; /* keyring private to this process */ 580 #endif 581 #ifdef CONFIG_BSD_PROCESS_ACCT 582 struct pacct_struct pacct; /* per-process accounting information */ 583 #endif 584 #ifdef CONFIG_TASKSTATS 585 struct taskstats *stats; 586 #endif 587 #ifdef CONFIG_AUDIT 588 unsigned audit_tty; 589 struct tty_audit_buf *tty_audit_buf; 590 #endif 591 }; 592 593 /* Context switch must be unlocked if interrupts are to be enabled */ 594 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 595 # define __ARCH_WANT_UNLOCKED_CTXSW 596 #endif 597 598 /* 599 * Bits in flags field of signal_struct. 600 */ 601 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 602 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 603 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 604 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 605 /* 606 * Pending notifications to parent. 607 */ 608 #define SIGNAL_CLD_STOPPED 0x00000010 609 #define SIGNAL_CLD_CONTINUED 0x00000020 610 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 611 612 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 613 614 /* If true, all threads except ->group_exit_task have pending SIGKILL */ 615 static inline int signal_group_exit(const struct signal_struct *sig) 616 { 617 return (sig->flags & SIGNAL_GROUP_EXIT) || 618 (sig->group_exit_task != NULL); 619 } 620 621 /* 622 * Some day this will be a full-fledged user tracking system.. 623 */ 624 struct user_struct { 625 atomic_t __count; /* reference count */ 626 atomic_t processes; /* How many processes does this user have? */ 627 atomic_t files; /* How many open files does this user have? */ 628 atomic_t sigpending; /* How many pending signals does this user have? */ 629 #ifdef CONFIG_INOTIFY_USER 630 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 631 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 632 #endif 633 #ifdef CONFIG_EPOLL 634 atomic_t epoll_devs; /* The number of epoll descriptors currently open */ 635 atomic_t epoll_watches; /* The number of file descriptors currently watched */ 636 #endif 637 #ifdef CONFIG_POSIX_MQUEUE 638 /* protected by mq_lock */ 639 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 640 #endif 641 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 642 643 #ifdef CONFIG_KEYS 644 struct key *uid_keyring; /* UID specific keyring */ 645 struct key *session_keyring; /* UID's default session keyring */ 646 #endif 647 648 /* Hash table maintenance information */ 649 struct hlist_node uidhash_node; 650 uid_t uid; 651 652 #ifdef CONFIG_USER_SCHED 653 struct task_group *tg; 654 #ifdef CONFIG_SYSFS 655 struct kobject kobj; 656 struct work_struct work; 657 #endif 658 #endif 659 }; 660 661 extern int uids_sysfs_init(void); 662 663 extern struct user_struct *find_user(uid_t); 664 665 extern struct user_struct root_user; 666 #define INIT_USER (&root_user) 667 668 struct backing_dev_info; 669 struct reclaim_state; 670 671 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 672 struct sched_info { 673 /* cumulative counters */ 674 unsigned long pcount; /* # of times run on this cpu */ 675 unsigned long long cpu_time, /* time spent on the cpu */ 676 run_delay; /* time spent waiting on a runqueue */ 677 678 /* timestamps */ 679 unsigned long long last_arrival,/* when we last ran on a cpu */ 680 last_queued; /* when we were last queued to run */ 681 #ifdef CONFIG_SCHEDSTATS 682 /* BKL stats */ 683 unsigned int bkl_count; 684 #endif 685 }; 686 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 687 688 #ifdef CONFIG_TASK_DELAY_ACCT 689 struct task_delay_info { 690 spinlock_t lock; 691 unsigned int flags; /* Private per-task flags */ 692 693 /* For each stat XXX, add following, aligned appropriately 694 * 695 * struct timespec XXX_start, XXX_end; 696 * u64 XXX_delay; 697 * u32 XXX_count; 698 * 699 * Atomicity of updates to XXX_delay, XXX_count protected by 700 * single lock above (split into XXX_lock if contention is an issue). 701 */ 702 703 /* 704 * XXX_count is incremented on every XXX operation, the delay 705 * associated with the operation is added to XXX_delay. 706 * XXX_delay contains the accumulated delay time in nanoseconds. 707 */ 708 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 709 u64 blkio_delay; /* wait for sync block io completion */ 710 u64 swapin_delay; /* wait for swapin block io completion */ 711 u32 blkio_count; /* total count of the number of sync block */ 712 /* io operations performed */ 713 u32 swapin_count; /* total count of the number of swapin block */ 714 /* io operations performed */ 715 716 struct timespec freepages_start, freepages_end; 717 u64 freepages_delay; /* wait for memory reclaim */ 718 u32 freepages_count; /* total count of memory reclaim */ 719 }; 720 #endif /* CONFIG_TASK_DELAY_ACCT */ 721 722 static inline int sched_info_on(void) 723 { 724 #ifdef CONFIG_SCHEDSTATS 725 return 1; 726 #elif defined(CONFIG_TASK_DELAY_ACCT) 727 extern int delayacct_on; 728 return delayacct_on; 729 #else 730 return 0; 731 #endif 732 } 733 734 enum cpu_idle_type { 735 CPU_IDLE, 736 CPU_NOT_IDLE, 737 CPU_NEWLY_IDLE, 738 CPU_MAX_IDLE_TYPES 739 }; 740 741 /* 742 * sched-domains (multiprocessor balancing) declarations: 743 */ 744 745 /* 746 * Increase resolution of nice-level calculations: 747 */ 748 #define SCHED_LOAD_SHIFT 10 749 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 750 751 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 752 753 #ifdef CONFIG_SMP 754 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 755 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 756 #define SD_BALANCE_EXEC 4 /* Balance on exec */ 757 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 758 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 759 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 760 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 761 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 762 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 763 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 764 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 765 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 766 767 #define BALANCE_FOR_MC_POWER \ 768 (sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0) 769 770 #define BALANCE_FOR_PKG_POWER \ 771 ((sched_mc_power_savings || sched_smt_power_savings) ? \ 772 SD_POWERSAVINGS_BALANCE : 0) 773 774 #define test_sd_parent(sd, flag) ((sd->parent && \ 775 (sd->parent->flags & flag)) ? 1 : 0) 776 777 778 struct sched_group { 779 struct sched_group *next; /* Must be a circular list */ 780 cpumask_t cpumask; 781 782 /* 783 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 784 * single CPU. This is read only (except for setup, hotplug CPU). 785 * Note : Never change cpu_power without recompute its reciprocal 786 */ 787 unsigned int __cpu_power; 788 /* 789 * reciprocal value of cpu_power to avoid expensive divides 790 * (see include/linux/reciprocal_div.h) 791 */ 792 u32 reciprocal_cpu_power; 793 }; 794 795 enum sched_domain_level { 796 SD_LV_NONE = 0, 797 SD_LV_SIBLING, 798 SD_LV_MC, 799 SD_LV_CPU, 800 SD_LV_NODE, 801 SD_LV_ALLNODES, 802 SD_LV_MAX 803 }; 804 805 struct sched_domain_attr { 806 int relax_domain_level; 807 }; 808 809 #define SD_ATTR_INIT (struct sched_domain_attr) { \ 810 .relax_domain_level = -1, \ 811 } 812 813 struct sched_domain { 814 /* These fields must be setup */ 815 struct sched_domain *parent; /* top domain must be null terminated */ 816 struct sched_domain *child; /* bottom domain must be null terminated */ 817 struct sched_group *groups; /* the balancing groups of the domain */ 818 cpumask_t span; /* span of all CPUs in this domain */ 819 unsigned long min_interval; /* Minimum balance interval ms */ 820 unsigned long max_interval; /* Maximum balance interval ms */ 821 unsigned int busy_factor; /* less balancing by factor if busy */ 822 unsigned int imbalance_pct; /* No balance until over watermark */ 823 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 824 unsigned int busy_idx; 825 unsigned int idle_idx; 826 unsigned int newidle_idx; 827 unsigned int wake_idx; 828 unsigned int forkexec_idx; 829 int flags; /* See SD_* */ 830 enum sched_domain_level level; 831 832 /* Runtime fields. */ 833 unsigned long last_balance; /* init to jiffies. units in jiffies */ 834 unsigned int balance_interval; /* initialise to 1. units in ms. */ 835 unsigned int nr_balance_failed; /* initialise to 0 */ 836 837 u64 last_update; 838 839 #ifdef CONFIG_SCHEDSTATS 840 /* load_balance() stats */ 841 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 842 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 843 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 844 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 845 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 846 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 847 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 848 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 849 850 /* Active load balancing */ 851 unsigned int alb_count; 852 unsigned int alb_failed; 853 unsigned int alb_pushed; 854 855 /* SD_BALANCE_EXEC stats */ 856 unsigned int sbe_count; 857 unsigned int sbe_balanced; 858 unsigned int sbe_pushed; 859 860 /* SD_BALANCE_FORK stats */ 861 unsigned int sbf_count; 862 unsigned int sbf_balanced; 863 unsigned int sbf_pushed; 864 865 /* try_to_wake_up() stats */ 866 unsigned int ttwu_wake_remote; 867 unsigned int ttwu_move_affine; 868 unsigned int ttwu_move_balance; 869 #endif 870 #ifdef CONFIG_SCHED_DEBUG 871 char *name; 872 #endif 873 }; 874 875 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 876 struct sched_domain_attr *dattr_new); 877 extern int arch_reinit_sched_domains(void); 878 879 #else /* CONFIG_SMP */ 880 881 struct sched_domain_attr; 882 883 static inline void 884 partition_sched_domains(int ndoms_new, cpumask_t *doms_new, 885 struct sched_domain_attr *dattr_new) 886 { 887 } 888 #endif /* !CONFIG_SMP */ 889 890 struct io_context; /* See blkdev.h */ 891 #define NGROUPS_SMALL 32 892 #define NGROUPS_PER_BLOCK ((unsigned int)(PAGE_SIZE / sizeof(gid_t))) 893 struct group_info { 894 int ngroups; 895 atomic_t usage; 896 gid_t small_block[NGROUPS_SMALL]; 897 int nblocks; 898 gid_t *blocks[0]; 899 }; 900 901 /* 902 * get_group_info() must be called with the owning task locked (via task_lock()) 903 * when task != current. The reason being that the vast majority of callers are 904 * looking at current->group_info, which can not be changed except by the 905 * current task. Changing current->group_info requires the task lock, too. 906 */ 907 #define get_group_info(group_info) do { \ 908 atomic_inc(&(group_info)->usage); \ 909 } while (0) 910 911 #define put_group_info(group_info) do { \ 912 if (atomic_dec_and_test(&(group_info)->usage)) \ 913 groups_free(group_info); \ 914 } while (0) 915 916 extern struct group_info *groups_alloc(int gidsetsize); 917 extern void groups_free(struct group_info *group_info); 918 extern int set_current_groups(struct group_info *group_info); 919 extern int groups_search(struct group_info *group_info, gid_t grp); 920 /* access the groups "array" with this macro */ 921 #define GROUP_AT(gi, i) \ 922 ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK]) 923 924 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 925 extern void prefetch_stack(struct task_struct *t); 926 #else 927 static inline void prefetch_stack(struct task_struct *t) { } 928 #endif 929 930 struct audit_context; /* See audit.c */ 931 struct mempolicy; 932 struct pipe_inode_info; 933 struct uts_namespace; 934 935 struct rq; 936 struct sched_domain; 937 938 struct sched_class { 939 const struct sched_class *next; 940 941 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 942 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 943 void (*yield_task) (struct rq *rq); 944 945 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync); 946 947 struct task_struct * (*pick_next_task) (struct rq *rq); 948 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 949 950 #ifdef CONFIG_SMP 951 int (*select_task_rq)(struct task_struct *p, int sync); 952 953 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 954 struct rq *busiest, unsigned long max_load_move, 955 struct sched_domain *sd, enum cpu_idle_type idle, 956 int *all_pinned, int *this_best_prio); 957 958 int (*move_one_task) (struct rq *this_rq, int this_cpu, 959 struct rq *busiest, struct sched_domain *sd, 960 enum cpu_idle_type idle); 961 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 962 void (*post_schedule) (struct rq *this_rq); 963 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 964 965 void (*set_cpus_allowed)(struct task_struct *p, 966 const cpumask_t *newmask); 967 968 void (*rq_online)(struct rq *rq); 969 void (*rq_offline)(struct rq *rq); 970 #endif 971 972 void (*set_curr_task) (struct rq *rq); 973 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 974 void (*task_new) (struct rq *rq, struct task_struct *p); 975 976 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 977 int running); 978 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 979 int running); 980 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 981 int oldprio, int running); 982 983 #ifdef CONFIG_FAIR_GROUP_SCHED 984 void (*moved_group) (struct task_struct *p); 985 #endif 986 }; 987 988 struct load_weight { 989 unsigned long weight, inv_weight; 990 }; 991 992 /* 993 * CFS stats for a schedulable entity (task, task-group etc) 994 * 995 * Current field usage histogram: 996 * 997 * 4 se->block_start 998 * 4 se->run_node 999 * 4 se->sleep_start 1000 * 6 se->load.weight 1001 */ 1002 struct sched_entity { 1003 struct load_weight load; /* for load-balancing */ 1004 struct rb_node run_node; 1005 struct list_head group_node; 1006 unsigned int on_rq; 1007 1008 u64 exec_start; 1009 u64 sum_exec_runtime; 1010 u64 vruntime; 1011 u64 prev_sum_exec_runtime; 1012 1013 u64 last_wakeup; 1014 u64 avg_overlap; 1015 1016 #ifdef CONFIG_SCHEDSTATS 1017 u64 wait_start; 1018 u64 wait_max; 1019 u64 wait_count; 1020 u64 wait_sum; 1021 1022 u64 sleep_start; 1023 u64 sleep_max; 1024 s64 sum_sleep_runtime; 1025 1026 u64 block_start; 1027 u64 block_max; 1028 u64 exec_max; 1029 u64 slice_max; 1030 1031 u64 nr_migrations; 1032 u64 nr_migrations_cold; 1033 u64 nr_failed_migrations_affine; 1034 u64 nr_failed_migrations_running; 1035 u64 nr_failed_migrations_hot; 1036 u64 nr_forced_migrations; 1037 u64 nr_forced2_migrations; 1038 1039 u64 nr_wakeups; 1040 u64 nr_wakeups_sync; 1041 u64 nr_wakeups_migrate; 1042 u64 nr_wakeups_local; 1043 u64 nr_wakeups_remote; 1044 u64 nr_wakeups_affine; 1045 u64 nr_wakeups_affine_attempts; 1046 u64 nr_wakeups_passive; 1047 u64 nr_wakeups_idle; 1048 #endif 1049 1050 #ifdef CONFIG_FAIR_GROUP_SCHED 1051 struct sched_entity *parent; 1052 /* rq on which this entity is (to be) queued: */ 1053 struct cfs_rq *cfs_rq; 1054 /* rq "owned" by this entity/group: */ 1055 struct cfs_rq *my_q; 1056 #endif 1057 }; 1058 1059 struct sched_rt_entity { 1060 struct list_head run_list; 1061 unsigned long timeout; 1062 unsigned int time_slice; 1063 int nr_cpus_allowed; 1064 1065 struct sched_rt_entity *back; 1066 #ifdef CONFIG_RT_GROUP_SCHED 1067 struct sched_rt_entity *parent; 1068 /* rq on which this entity is (to be) queued: */ 1069 struct rt_rq *rt_rq; 1070 /* rq "owned" by this entity/group: */ 1071 struct rt_rq *my_q; 1072 #endif 1073 }; 1074 1075 struct task_struct { 1076 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1077 void *stack; 1078 atomic_t usage; 1079 unsigned int flags; /* per process flags, defined below */ 1080 unsigned int ptrace; 1081 1082 int lock_depth; /* BKL lock depth */ 1083 1084 #ifdef CONFIG_SMP 1085 #ifdef __ARCH_WANT_UNLOCKED_CTXSW 1086 int oncpu; 1087 #endif 1088 #endif 1089 1090 int prio, static_prio, normal_prio; 1091 unsigned int rt_priority; 1092 const struct sched_class *sched_class; 1093 struct sched_entity se; 1094 struct sched_rt_entity rt; 1095 1096 #ifdef CONFIG_PREEMPT_NOTIFIERS 1097 /* list of struct preempt_notifier: */ 1098 struct hlist_head preempt_notifiers; 1099 #endif 1100 1101 /* 1102 * fpu_counter contains the number of consecutive context switches 1103 * that the FPU is used. If this is over a threshold, the lazy fpu 1104 * saving becomes unlazy to save the trap. This is an unsigned char 1105 * so that after 256 times the counter wraps and the behavior turns 1106 * lazy again; this to deal with bursty apps that only use FPU for 1107 * a short time 1108 */ 1109 unsigned char fpu_counter; 1110 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */ 1111 #ifdef CONFIG_BLK_DEV_IO_TRACE 1112 unsigned int btrace_seq; 1113 #endif 1114 1115 unsigned int policy; 1116 cpumask_t cpus_allowed; 1117 1118 #ifdef CONFIG_PREEMPT_RCU 1119 int rcu_read_lock_nesting; 1120 int rcu_flipctr_idx; 1121 #endif /* #ifdef CONFIG_PREEMPT_RCU */ 1122 1123 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1124 struct sched_info sched_info; 1125 #endif 1126 1127 struct list_head tasks; 1128 1129 struct mm_struct *mm, *active_mm; 1130 1131 /* task state */ 1132 struct linux_binfmt *binfmt; 1133 int exit_state; 1134 int exit_code, exit_signal; 1135 int pdeath_signal; /* The signal sent when the parent dies */ 1136 /* ??? */ 1137 unsigned int personality; 1138 unsigned did_exec:1; 1139 pid_t pid; 1140 pid_t tgid; 1141 1142 #ifdef CONFIG_CC_STACKPROTECTOR 1143 /* Canary value for the -fstack-protector gcc feature */ 1144 unsigned long stack_canary; 1145 #endif 1146 /* 1147 * pointers to (original) parent process, youngest child, younger sibling, 1148 * older sibling, respectively. (p->father can be replaced with 1149 * p->real_parent->pid) 1150 */ 1151 struct task_struct *real_parent; /* real parent process */ 1152 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1153 /* 1154 * children/sibling forms the list of my natural children 1155 */ 1156 struct list_head children; /* list of my children */ 1157 struct list_head sibling; /* linkage in my parent's children list */ 1158 struct task_struct *group_leader; /* threadgroup leader */ 1159 1160 /* 1161 * ptraced is the list of tasks this task is using ptrace on. 1162 * This includes both natural children and PTRACE_ATTACH targets. 1163 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1164 */ 1165 struct list_head ptraced; 1166 struct list_head ptrace_entry; 1167 1168 /* PID/PID hash table linkage. */ 1169 struct pid_link pids[PIDTYPE_MAX]; 1170 struct list_head thread_group; 1171 1172 struct completion *vfork_done; /* for vfork() */ 1173 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1174 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1175 1176 cputime_t utime, stime, utimescaled, stimescaled; 1177 cputime_t gtime; 1178 cputime_t prev_utime, prev_stime; 1179 unsigned long nvcsw, nivcsw; /* context switch counts */ 1180 struct timespec start_time; /* monotonic time */ 1181 struct timespec real_start_time; /* boot based time */ 1182 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1183 unsigned long min_flt, maj_flt; 1184 1185 struct task_cputime cputime_expires; 1186 struct list_head cpu_timers[3]; 1187 1188 /* process credentials */ 1189 uid_t uid,euid,suid,fsuid; 1190 gid_t gid,egid,sgid,fsgid; 1191 struct group_info *group_info; 1192 kernel_cap_t cap_effective, cap_inheritable, cap_permitted, cap_bset; 1193 struct user_struct *user; 1194 unsigned securebits; 1195 #ifdef CONFIG_KEYS 1196 unsigned char jit_keyring; /* default keyring to attach requested keys to */ 1197 struct key *request_key_auth; /* assumed request_key authority */ 1198 struct key *thread_keyring; /* keyring private to this thread */ 1199 #endif 1200 char comm[TASK_COMM_LEN]; /* executable name excluding path 1201 - access with [gs]et_task_comm (which lock 1202 it with task_lock()) 1203 - initialized normally by flush_old_exec */ 1204 /* file system info */ 1205 int link_count, total_link_count; 1206 #ifdef CONFIG_SYSVIPC 1207 /* ipc stuff */ 1208 struct sysv_sem sysvsem; 1209 #endif 1210 #ifdef CONFIG_DETECT_SOFTLOCKUP 1211 /* hung task detection */ 1212 unsigned long last_switch_timestamp; 1213 unsigned long last_switch_count; 1214 #endif 1215 /* CPU-specific state of this task */ 1216 struct thread_struct thread; 1217 /* filesystem information */ 1218 struct fs_struct *fs; 1219 /* open file information */ 1220 struct files_struct *files; 1221 /* namespaces */ 1222 struct nsproxy *nsproxy; 1223 /* signal handlers */ 1224 struct signal_struct *signal; 1225 struct sighand_struct *sighand; 1226 1227 sigset_t blocked, real_blocked; 1228 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1229 struct sigpending pending; 1230 1231 unsigned long sas_ss_sp; 1232 size_t sas_ss_size; 1233 int (*notifier)(void *priv); 1234 void *notifier_data; 1235 sigset_t *notifier_mask; 1236 #ifdef CONFIG_SECURITY 1237 void *security; 1238 #endif 1239 struct audit_context *audit_context; 1240 #ifdef CONFIG_AUDITSYSCALL 1241 uid_t loginuid; 1242 unsigned int sessionid; 1243 #endif 1244 seccomp_t seccomp; 1245 1246 /* Thread group tracking */ 1247 u32 parent_exec_id; 1248 u32 self_exec_id; 1249 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */ 1250 spinlock_t alloc_lock; 1251 1252 /* Protection of the PI data structures: */ 1253 spinlock_t pi_lock; 1254 1255 #ifdef CONFIG_RT_MUTEXES 1256 /* PI waiters blocked on a rt_mutex held by this task */ 1257 struct plist_head pi_waiters; 1258 /* Deadlock detection and priority inheritance handling */ 1259 struct rt_mutex_waiter *pi_blocked_on; 1260 #endif 1261 1262 #ifdef CONFIG_DEBUG_MUTEXES 1263 /* mutex deadlock detection */ 1264 struct mutex_waiter *blocked_on; 1265 #endif 1266 #ifdef CONFIG_TRACE_IRQFLAGS 1267 unsigned int irq_events; 1268 int hardirqs_enabled; 1269 unsigned long hardirq_enable_ip; 1270 unsigned int hardirq_enable_event; 1271 unsigned long hardirq_disable_ip; 1272 unsigned int hardirq_disable_event; 1273 int softirqs_enabled; 1274 unsigned long softirq_disable_ip; 1275 unsigned int softirq_disable_event; 1276 unsigned long softirq_enable_ip; 1277 unsigned int softirq_enable_event; 1278 int hardirq_context; 1279 int softirq_context; 1280 #endif 1281 #ifdef CONFIG_LOCKDEP 1282 # define MAX_LOCK_DEPTH 48UL 1283 u64 curr_chain_key; 1284 int lockdep_depth; 1285 unsigned int lockdep_recursion; 1286 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1287 #endif 1288 1289 /* journalling filesystem info */ 1290 void *journal_info; 1291 1292 /* stacked block device info */ 1293 struct bio *bio_list, **bio_tail; 1294 1295 /* VM state */ 1296 struct reclaim_state *reclaim_state; 1297 1298 struct backing_dev_info *backing_dev_info; 1299 1300 struct io_context *io_context; 1301 1302 unsigned long ptrace_message; 1303 siginfo_t *last_siginfo; /* For ptrace use. */ 1304 struct task_io_accounting ioac; 1305 #if defined(CONFIG_TASK_XACCT) 1306 u64 acct_rss_mem1; /* accumulated rss usage */ 1307 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1308 cputime_t acct_timexpd; /* stime + utime since last update */ 1309 #endif 1310 #ifdef CONFIG_CPUSETS 1311 nodemask_t mems_allowed; 1312 int cpuset_mems_generation; 1313 int cpuset_mem_spread_rotor; 1314 #endif 1315 #ifdef CONFIG_CGROUPS 1316 /* Control Group info protected by css_set_lock */ 1317 struct css_set *cgroups; 1318 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1319 struct list_head cg_list; 1320 #endif 1321 #ifdef CONFIG_FUTEX 1322 struct robust_list_head __user *robust_list; 1323 #ifdef CONFIG_COMPAT 1324 struct compat_robust_list_head __user *compat_robust_list; 1325 #endif 1326 struct list_head pi_state_list; 1327 struct futex_pi_state *pi_state_cache; 1328 #endif 1329 #ifdef CONFIG_NUMA 1330 struct mempolicy *mempolicy; 1331 short il_next; 1332 #endif 1333 atomic_t fs_excl; /* holding fs exclusive resources */ 1334 struct rcu_head rcu; 1335 1336 /* 1337 * cache last used pipe for splice 1338 */ 1339 struct pipe_inode_info *splice_pipe; 1340 #ifdef CONFIG_TASK_DELAY_ACCT 1341 struct task_delay_info *delays; 1342 #endif 1343 #ifdef CONFIG_FAULT_INJECTION 1344 int make_it_fail; 1345 #endif 1346 struct prop_local_single dirties; 1347 #ifdef CONFIG_LATENCYTOP 1348 int latency_record_count; 1349 struct latency_record latency_record[LT_SAVECOUNT]; 1350 #endif 1351 /* 1352 * time slack values; these are used to round up poll() and 1353 * select() etc timeout values. These are in nanoseconds. 1354 */ 1355 unsigned long timer_slack_ns; 1356 unsigned long default_timer_slack_ns; 1357 1358 struct list_head *scm_work_list; 1359 }; 1360 1361 /* 1362 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1363 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1364 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1365 * values are inverted: lower p->prio value means higher priority. 1366 * 1367 * The MAX_USER_RT_PRIO value allows the actual maximum 1368 * RT priority to be separate from the value exported to 1369 * user-space. This allows kernel threads to set their 1370 * priority to a value higher than any user task. Note: 1371 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1372 */ 1373 1374 #define MAX_USER_RT_PRIO 100 1375 #define MAX_RT_PRIO MAX_USER_RT_PRIO 1376 1377 #define MAX_PRIO (MAX_RT_PRIO + 40) 1378 #define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1379 1380 static inline int rt_prio(int prio) 1381 { 1382 if (unlikely(prio < MAX_RT_PRIO)) 1383 return 1; 1384 return 0; 1385 } 1386 1387 static inline int rt_task(struct task_struct *p) 1388 { 1389 return rt_prio(p->prio); 1390 } 1391 1392 static inline void set_task_session(struct task_struct *tsk, pid_t session) 1393 { 1394 tsk->signal->__session = session; 1395 } 1396 1397 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp) 1398 { 1399 tsk->signal->__pgrp = pgrp; 1400 } 1401 1402 static inline struct pid *task_pid(struct task_struct *task) 1403 { 1404 return task->pids[PIDTYPE_PID].pid; 1405 } 1406 1407 static inline struct pid *task_tgid(struct task_struct *task) 1408 { 1409 return task->group_leader->pids[PIDTYPE_PID].pid; 1410 } 1411 1412 static inline struct pid *task_pgrp(struct task_struct *task) 1413 { 1414 return task->group_leader->pids[PIDTYPE_PGID].pid; 1415 } 1416 1417 static inline struct pid *task_session(struct task_struct *task) 1418 { 1419 return task->group_leader->pids[PIDTYPE_SID].pid; 1420 } 1421 1422 struct pid_namespace; 1423 1424 /* 1425 * the helpers to get the task's different pids as they are seen 1426 * from various namespaces 1427 * 1428 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1429 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1430 * current. 1431 * task_xid_nr_ns() : id seen from the ns specified; 1432 * 1433 * set_task_vxid() : assigns a virtual id to a task; 1434 * 1435 * see also pid_nr() etc in include/linux/pid.h 1436 */ 1437 1438 static inline pid_t task_pid_nr(struct task_struct *tsk) 1439 { 1440 return tsk->pid; 1441 } 1442 1443 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1444 1445 static inline pid_t task_pid_vnr(struct task_struct *tsk) 1446 { 1447 return pid_vnr(task_pid(tsk)); 1448 } 1449 1450 1451 static inline pid_t task_tgid_nr(struct task_struct *tsk) 1452 { 1453 return tsk->tgid; 1454 } 1455 1456 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1457 1458 static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1459 { 1460 return pid_vnr(task_tgid(tsk)); 1461 } 1462 1463 1464 static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1465 { 1466 return tsk->signal->__pgrp; 1467 } 1468 1469 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1470 1471 static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1472 { 1473 return pid_vnr(task_pgrp(tsk)); 1474 } 1475 1476 1477 static inline pid_t task_session_nr(struct task_struct *tsk) 1478 { 1479 return tsk->signal->__session; 1480 } 1481 1482 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1483 1484 static inline pid_t task_session_vnr(struct task_struct *tsk) 1485 { 1486 return pid_vnr(task_session(tsk)); 1487 } 1488 1489 1490 /** 1491 * pid_alive - check that a task structure is not stale 1492 * @p: Task structure to be checked. 1493 * 1494 * Test if a process is not yet dead (at most zombie state) 1495 * If pid_alive fails, then pointers within the task structure 1496 * can be stale and must not be dereferenced. 1497 */ 1498 static inline int pid_alive(struct task_struct *p) 1499 { 1500 return p->pids[PIDTYPE_PID].pid != NULL; 1501 } 1502 1503 /** 1504 * is_global_init - check if a task structure is init 1505 * @tsk: Task structure to be checked. 1506 * 1507 * Check if a task structure is the first user space task the kernel created. 1508 */ 1509 static inline int is_global_init(struct task_struct *tsk) 1510 { 1511 return tsk->pid == 1; 1512 } 1513 1514 /* 1515 * is_container_init: 1516 * check whether in the task is init in its own pid namespace. 1517 */ 1518 extern int is_container_init(struct task_struct *tsk); 1519 1520 extern struct pid *cad_pid; 1521 1522 extern void free_task(struct task_struct *tsk); 1523 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1524 1525 extern void __put_task_struct(struct task_struct *t); 1526 1527 static inline void put_task_struct(struct task_struct *t) 1528 { 1529 if (atomic_dec_and_test(&t->usage)) 1530 __put_task_struct(t); 1531 } 1532 1533 extern cputime_t task_utime(struct task_struct *p); 1534 extern cputime_t task_stime(struct task_struct *p); 1535 extern cputime_t task_gtime(struct task_struct *p); 1536 1537 /* 1538 * Per process flags 1539 */ 1540 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1541 /* Not implemented yet, only for 486*/ 1542 #define PF_STARTING 0x00000002 /* being created */ 1543 #define PF_EXITING 0x00000004 /* getting shut down */ 1544 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1545 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1546 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1547 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1548 #define PF_DUMPCORE 0x00000200 /* dumped core */ 1549 #define PF_SIGNALED 0x00000400 /* killed by a signal */ 1550 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1551 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1552 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1553 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1554 #define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1555 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1556 #define PF_KSWAPD 0x00040000 /* I am kswapd */ 1557 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1558 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1559 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1560 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1561 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1562 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1563 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1564 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1565 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1566 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1567 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1568 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1569 1570 /* 1571 * Only the _current_ task can read/write to tsk->flags, but other 1572 * tasks can access tsk->flags in readonly mode for example 1573 * with tsk_used_math (like during threaded core dumping). 1574 * There is however an exception to this rule during ptrace 1575 * or during fork: the ptracer task is allowed to write to the 1576 * child->flags of its traced child (same goes for fork, the parent 1577 * can write to the child->flags), because we're guaranteed the 1578 * child is not running and in turn not changing child->flags 1579 * at the same time the parent does it. 1580 */ 1581 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1582 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1583 #define clear_used_math() clear_stopped_child_used_math(current) 1584 #define set_used_math() set_stopped_child_used_math(current) 1585 #define conditional_stopped_child_used_math(condition, child) \ 1586 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1587 #define conditional_used_math(condition) \ 1588 conditional_stopped_child_used_math(condition, current) 1589 #define copy_to_stopped_child_used_math(child) \ 1590 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1591 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1592 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1593 #define used_math() tsk_used_math(current) 1594 1595 #ifdef CONFIG_SMP 1596 extern int set_cpus_allowed_ptr(struct task_struct *p, 1597 const cpumask_t *new_mask); 1598 #else 1599 static inline int set_cpus_allowed_ptr(struct task_struct *p, 1600 const cpumask_t *new_mask) 1601 { 1602 if (!cpu_isset(0, *new_mask)) 1603 return -EINVAL; 1604 return 0; 1605 } 1606 #endif 1607 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1608 { 1609 return set_cpus_allowed_ptr(p, &new_mask); 1610 } 1611 1612 extern unsigned long long sched_clock(void); 1613 1614 extern void sched_clock_init(void); 1615 extern u64 sched_clock_cpu(int cpu); 1616 1617 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1618 static inline void sched_clock_tick(void) 1619 { 1620 } 1621 1622 static inline void sched_clock_idle_sleep_event(void) 1623 { 1624 } 1625 1626 static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1627 { 1628 } 1629 #else 1630 extern void sched_clock_tick(void); 1631 extern void sched_clock_idle_sleep_event(void); 1632 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1633 #endif 1634 1635 /* 1636 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1637 * clock constructed from sched_clock(): 1638 */ 1639 extern unsigned long long cpu_clock(int cpu); 1640 1641 extern unsigned long long 1642 task_sched_runtime(struct task_struct *task); 1643 extern unsigned long long thread_group_sched_runtime(struct task_struct *task); 1644 1645 /* sched_exec is called by processes performing an exec */ 1646 #ifdef CONFIG_SMP 1647 extern void sched_exec(void); 1648 #else 1649 #define sched_exec() {} 1650 #endif 1651 1652 extern void sched_clock_idle_sleep_event(void); 1653 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1654 1655 #ifdef CONFIG_HOTPLUG_CPU 1656 extern void idle_task_exit(void); 1657 #else 1658 static inline void idle_task_exit(void) {} 1659 #endif 1660 1661 extern void sched_idle_next(void); 1662 1663 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1664 extern void wake_up_idle_cpu(int cpu); 1665 #else 1666 static inline void wake_up_idle_cpu(int cpu) { } 1667 #endif 1668 1669 #ifdef CONFIG_SCHED_DEBUG 1670 extern unsigned int sysctl_sched_latency; 1671 extern unsigned int sysctl_sched_min_granularity; 1672 extern unsigned int sysctl_sched_wakeup_granularity; 1673 extern unsigned int sysctl_sched_child_runs_first; 1674 extern unsigned int sysctl_sched_features; 1675 extern unsigned int sysctl_sched_migration_cost; 1676 extern unsigned int sysctl_sched_nr_migrate; 1677 extern unsigned int sysctl_sched_shares_ratelimit; 1678 extern unsigned int sysctl_sched_shares_thresh; 1679 1680 int sched_nr_latency_handler(struct ctl_table *table, int write, 1681 struct file *file, void __user *buffer, size_t *length, 1682 loff_t *ppos); 1683 #endif 1684 extern unsigned int sysctl_sched_rt_period; 1685 extern int sysctl_sched_rt_runtime; 1686 1687 int sched_rt_handler(struct ctl_table *table, int write, 1688 struct file *filp, void __user *buffer, size_t *lenp, 1689 loff_t *ppos); 1690 1691 extern unsigned int sysctl_sched_compat_yield; 1692 1693 #ifdef CONFIG_RT_MUTEXES 1694 extern int rt_mutex_getprio(struct task_struct *p); 1695 extern void rt_mutex_setprio(struct task_struct *p, int prio); 1696 extern void rt_mutex_adjust_pi(struct task_struct *p); 1697 #else 1698 static inline int rt_mutex_getprio(struct task_struct *p) 1699 { 1700 return p->normal_prio; 1701 } 1702 # define rt_mutex_adjust_pi(p) do { } while (0) 1703 #endif 1704 1705 extern void set_user_nice(struct task_struct *p, long nice); 1706 extern int task_prio(const struct task_struct *p); 1707 extern int task_nice(const struct task_struct *p); 1708 extern int can_nice(const struct task_struct *p, const int nice); 1709 extern int task_curr(const struct task_struct *p); 1710 extern int idle_cpu(int cpu); 1711 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1712 extern int sched_setscheduler_nocheck(struct task_struct *, int, 1713 struct sched_param *); 1714 extern struct task_struct *idle_task(int cpu); 1715 extern struct task_struct *curr_task(int cpu); 1716 extern void set_curr_task(int cpu, struct task_struct *p); 1717 1718 void yield(void); 1719 1720 /* 1721 * The default (Linux) execution domain. 1722 */ 1723 extern struct exec_domain default_exec_domain; 1724 1725 union thread_union { 1726 struct thread_info thread_info; 1727 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1728 }; 1729 1730 #ifndef __HAVE_ARCH_KSTACK_END 1731 static inline int kstack_end(void *addr) 1732 { 1733 /* Reliable end of stack detection: 1734 * Some APM bios versions misalign the stack 1735 */ 1736 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1737 } 1738 #endif 1739 1740 extern union thread_union init_thread_union; 1741 extern struct task_struct init_task; 1742 1743 extern struct mm_struct init_mm; 1744 1745 extern struct pid_namespace init_pid_ns; 1746 1747 /* 1748 * find a task by one of its numerical ids 1749 * 1750 * find_task_by_pid_type_ns(): 1751 * it is the most generic call - it finds a task by all id, 1752 * type and namespace specified 1753 * find_task_by_pid_ns(): 1754 * finds a task by its pid in the specified namespace 1755 * find_task_by_vpid(): 1756 * finds a task by its virtual pid 1757 * 1758 * see also find_vpid() etc in include/linux/pid.h 1759 */ 1760 1761 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid, 1762 struct pid_namespace *ns); 1763 1764 extern struct task_struct *find_task_by_vpid(pid_t nr); 1765 extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1766 struct pid_namespace *ns); 1767 1768 extern void __set_special_pids(struct pid *pid); 1769 1770 /* per-UID process charging. */ 1771 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1772 static inline struct user_struct *get_uid(struct user_struct *u) 1773 { 1774 atomic_inc(&u->__count); 1775 return u; 1776 } 1777 extern void free_uid(struct user_struct *); 1778 extern void switch_uid(struct user_struct *); 1779 extern void release_uids(struct user_namespace *ns); 1780 1781 #include <asm/current.h> 1782 1783 extern void do_timer(unsigned long ticks); 1784 1785 extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1786 extern int wake_up_process(struct task_struct *tsk); 1787 extern void wake_up_new_task(struct task_struct *tsk, 1788 unsigned long clone_flags); 1789 #ifdef CONFIG_SMP 1790 extern void kick_process(struct task_struct *tsk); 1791 #else 1792 static inline void kick_process(struct task_struct *tsk) { } 1793 #endif 1794 extern void sched_fork(struct task_struct *p, int clone_flags); 1795 extern void sched_dead(struct task_struct *p); 1796 1797 extern int in_group_p(gid_t); 1798 extern int in_egroup_p(gid_t); 1799 1800 extern void proc_caches_init(void); 1801 extern void flush_signals(struct task_struct *); 1802 extern void ignore_signals(struct task_struct *); 1803 extern void flush_signal_handlers(struct task_struct *, int force_default); 1804 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1805 1806 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1807 { 1808 unsigned long flags; 1809 int ret; 1810 1811 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1812 ret = dequeue_signal(tsk, mask, info); 1813 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1814 1815 return ret; 1816 } 1817 1818 extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1819 sigset_t *mask); 1820 extern void unblock_all_signals(void); 1821 extern void release_task(struct task_struct * p); 1822 extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1823 extern int force_sigsegv(int, struct task_struct *); 1824 extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1825 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1826 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1827 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1828 extern int kill_pgrp(struct pid *pid, int sig, int priv); 1829 extern int kill_pid(struct pid *pid, int sig, int priv); 1830 extern int kill_proc_info(int, struct siginfo *, pid_t); 1831 extern int do_notify_parent(struct task_struct *, int); 1832 extern void force_sig(int, struct task_struct *); 1833 extern void force_sig_specific(int, struct task_struct *); 1834 extern int send_sig(int, struct task_struct *, int); 1835 extern void zap_other_threads(struct task_struct *p); 1836 extern struct sigqueue *sigqueue_alloc(void); 1837 extern void sigqueue_free(struct sigqueue *); 1838 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 1839 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1840 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1841 1842 static inline int kill_cad_pid(int sig, int priv) 1843 { 1844 return kill_pid(cad_pid, sig, priv); 1845 } 1846 1847 /* These can be the second arg to send_sig_info/send_group_sig_info. */ 1848 #define SEND_SIG_NOINFO ((struct siginfo *) 0) 1849 #define SEND_SIG_PRIV ((struct siginfo *) 1) 1850 #define SEND_SIG_FORCED ((struct siginfo *) 2) 1851 1852 static inline int is_si_special(const struct siginfo *info) 1853 { 1854 return info <= SEND_SIG_FORCED; 1855 } 1856 1857 /* True if we are on the alternate signal stack. */ 1858 1859 static inline int on_sig_stack(unsigned long sp) 1860 { 1861 return (sp - current->sas_ss_sp < current->sas_ss_size); 1862 } 1863 1864 static inline int sas_ss_flags(unsigned long sp) 1865 { 1866 return (current->sas_ss_size == 0 ? SS_DISABLE 1867 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1868 } 1869 1870 /* 1871 * Routines for handling mm_structs 1872 */ 1873 extern struct mm_struct * mm_alloc(void); 1874 1875 /* mmdrop drops the mm and the page tables */ 1876 extern void __mmdrop(struct mm_struct *); 1877 static inline void mmdrop(struct mm_struct * mm) 1878 { 1879 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 1880 __mmdrop(mm); 1881 } 1882 1883 /* mmput gets rid of the mappings and all user-space */ 1884 extern void mmput(struct mm_struct *); 1885 /* Grab a reference to a task's mm, if it is not already going away */ 1886 extern struct mm_struct *get_task_mm(struct task_struct *task); 1887 /* Remove the current tasks stale references to the old mm_struct */ 1888 extern void mm_release(struct task_struct *, struct mm_struct *); 1889 /* Allocate a new mm structure and copy contents from tsk->mm */ 1890 extern struct mm_struct *dup_mm(struct task_struct *tsk); 1891 1892 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *); 1893 extern void flush_thread(void); 1894 extern void exit_thread(void); 1895 1896 extern void exit_files(struct task_struct *); 1897 extern void __cleanup_signal(struct signal_struct *); 1898 extern void __cleanup_sighand(struct sighand_struct *); 1899 1900 extern void exit_itimers(struct signal_struct *); 1901 extern void flush_itimer_signals(void); 1902 1903 extern NORET_TYPE void do_group_exit(int); 1904 1905 extern void daemonize(const char *, ...); 1906 extern int allow_signal(int); 1907 extern int disallow_signal(int); 1908 1909 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 1910 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 1911 struct task_struct *fork_idle(int); 1912 1913 extern void set_task_comm(struct task_struct *tsk, char *from); 1914 extern char *get_task_comm(char *to, struct task_struct *tsk); 1915 1916 #ifdef CONFIG_SMP 1917 extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 1918 #else 1919 static inline unsigned long wait_task_inactive(struct task_struct *p, 1920 long match_state) 1921 { 1922 return 1; 1923 } 1924 #endif 1925 1926 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks) 1927 1928 #define for_each_process(p) \ 1929 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 1930 1931 /* 1932 * Careful: do_each_thread/while_each_thread is a double loop so 1933 * 'break' will not work as expected - use goto instead. 1934 */ 1935 #define do_each_thread(g, t) \ 1936 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 1937 1938 #define while_each_thread(g, t) \ 1939 while ((t = next_thread(t)) != g) 1940 1941 /* de_thread depends on thread_group_leader not being a pid based check */ 1942 #define thread_group_leader(p) (p == p->group_leader) 1943 1944 /* Do to the insanities of de_thread it is possible for a process 1945 * to have the pid of the thread group leader without actually being 1946 * the thread group leader. For iteration through the pids in proc 1947 * all we care about is that we have a task with the appropriate 1948 * pid, we don't actually care if we have the right task. 1949 */ 1950 static inline int has_group_leader_pid(struct task_struct *p) 1951 { 1952 return p->pid == p->tgid; 1953 } 1954 1955 static inline 1956 int same_thread_group(struct task_struct *p1, struct task_struct *p2) 1957 { 1958 return p1->tgid == p2->tgid; 1959 } 1960 1961 static inline struct task_struct *next_thread(const struct task_struct *p) 1962 { 1963 return list_entry(rcu_dereference(p->thread_group.next), 1964 struct task_struct, thread_group); 1965 } 1966 1967 static inline int thread_group_empty(struct task_struct *p) 1968 { 1969 return list_empty(&p->thread_group); 1970 } 1971 1972 #define delay_group_leader(p) \ 1973 (thread_group_leader(p) && !thread_group_empty(p)) 1974 1975 /* 1976 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 1977 * subscriptions and synchronises with wait4(). Also used in procfs. Also 1978 * pins the final release of task.io_context. Also protects ->cpuset and 1979 * ->cgroup.subsys[]. 1980 * 1981 * Nests both inside and outside of read_lock(&tasklist_lock). 1982 * It must not be nested with write_lock_irq(&tasklist_lock), 1983 * neither inside nor outside. 1984 */ 1985 static inline void task_lock(struct task_struct *p) 1986 { 1987 spin_lock(&p->alloc_lock); 1988 } 1989 1990 static inline void task_unlock(struct task_struct *p) 1991 { 1992 spin_unlock(&p->alloc_lock); 1993 } 1994 1995 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 1996 unsigned long *flags); 1997 1998 static inline void unlock_task_sighand(struct task_struct *tsk, 1999 unsigned long *flags) 2000 { 2001 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 2002 } 2003 2004 #ifndef __HAVE_THREAD_FUNCTIONS 2005 2006 #define task_thread_info(task) ((struct thread_info *)(task)->stack) 2007 #define task_stack_page(task) ((task)->stack) 2008 2009 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 2010 { 2011 *task_thread_info(p) = *task_thread_info(org); 2012 task_thread_info(p)->task = p; 2013 } 2014 2015 static inline unsigned long *end_of_stack(struct task_struct *p) 2016 { 2017 return (unsigned long *)(task_thread_info(p) + 1); 2018 } 2019 2020 #endif 2021 2022 static inline int object_is_on_stack(void *obj) 2023 { 2024 void *stack = task_stack_page(current); 2025 2026 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 2027 } 2028 2029 extern void thread_info_cache_init(void); 2030 2031 /* set thread flags in other task's structures 2032 * - see asm/thread_info.h for TIF_xxxx flags available 2033 */ 2034 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 2035 { 2036 set_ti_thread_flag(task_thread_info(tsk), flag); 2037 } 2038 2039 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2040 { 2041 clear_ti_thread_flag(task_thread_info(tsk), flag); 2042 } 2043 2044 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2045 { 2046 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2047 } 2048 2049 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2050 { 2051 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2052 } 2053 2054 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2055 { 2056 return test_ti_thread_flag(task_thread_info(tsk), flag); 2057 } 2058 2059 static inline void set_tsk_need_resched(struct task_struct *tsk) 2060 { 2061 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2062 } 2063 2064 static inline void clear_tsk_need_resched(struct task_struct *tsk) 2065 { 2066 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2067 } 2068 2069 static inline int test_tsk_need_resched(struct task_struct *tsk) 2070 { 2071 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2072 } 2073 2074 static inline int signal_pending(struct task_struct *p) 2075 { 2076 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2077 } 2078 2079 extern int __fatal_signal_pending(struct task_struct *p); 2080 2081 static inline int fatal_signal_pending(struct task_struct *p) 2082 { 2083 return signal_pending(p) && __fatal_signal_pending(p); 2084 } 2085 2086 static inline int signal_pending_state(long state, struct task_struct *p) 2087 { 2088 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2089 return 0; 2090 if (!signal_pending(p)) 2091 return 0; 2092 2093 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2094 } 2095 2096 static inline int need_resched(void) 2097 { 2098 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2099 } 2100 2101 /* 2102 * cond_resched() and cond_resched_lock(): latency reduction via 2103 * explicit rescheduling in places that are safe. The return 2104 * value indicates whether a reschedule was done in fact. 2105 * cond_resched_lock() will drop the spinlock before scheduling, 2106 * cond_resched_softirq() will enable bhs before scheduling. 2107 */ 2108 extern int _cond_resched(void); 2109 #ifdef CONFIG_PREEMPT_BKL 2110 static inline int cond_resched(void) 2111 { 2112 return 0; 2113 } 2114 #else 2115 static inline int cond_resched(void) 2116 { 2117 return _cond_resched(); 2118 } 2119 #endif 2120 extern int cond_resched_lock(spinlock_t * lock); 2121 extern int cond_resched_softirq(void); 2122 static inline int cond_resched_bkl(void) 2123 { 2124 return _cond_resched(); 2125 } 2126 2127 /* 2128 * Does a critical section need to be broken due to another 2129 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2130 * but a general need for low latency) 2131 */ 2132 static inline int spin_needbreak(spinlock_t *lock) 2133 { 2134 #ifdef CONFIG_PREEMPT 2135 return spin_is_contended(lock); 2136 #else 2137 return 0; 2138 #endif 2139 } 2140 2141 /* 2142 * Thread group CPU time accounting. 2143 */ 2144 2145 extern int thread_group_cputime_alloc(struct task_struct *); 2146 extern void thread_group_cputime(struct task_struct *, struct task_cputime *); 2147 2148 static inline void thread_group_cputime_init(struct signal_struct *sig) 2149 { 2150 sig->cputime.totals = NULL; 2151 } 2152 2153 static inline int thread_group_cputime_clone_thread(struct task_struct *curr) 2154 { 2155 if (curr->signal->cputime.totals) 2156 return 0; 2157 return thread_group_cputime_alloc(curr); 2158 } 2159 2160 static inline void thread_group_cputime_free(struct signal_struct *sig) 2161 { 2162 free_percpu(sig->cputime.totals); 2163 } 2164 2165 /* 2166 * Reevaluate whether the task has signals pending delivery. 2167 * Wake the task if so. 2168 * This is required every time the blocked sigset_t changes. 2169 * callers must hold sighand->siglock. 2170 */ 2171 extern void recalc_sigpending_and_wake(struct task_struct *t); 2172 extern void recalc_sigpending(void); 2173 2174 extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2175 2176 /* 2177 * Wrappers for p->thread_info->cpu access. No-op on UP. 2178 */ 2179 #ifdef CONFIG_SMP 2180 2181 static inline unsigned int task_cpu(const struct task_struct *p) 2182 { 2183 return task_thread_info(p)->cpu; 2184 } 2185 2186 extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2187 2188 #else 2189 2190 static inline unsigned int task_cpu(const struct task_struct *p) 2191 { 2192 return 0; 2193 } 2194 2195 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2196 { 2197 } 2198 2199 #endif /* CONFIG_SMP */ 2200 2201 extern void arch_pick_mmap_layout(struct mm_struct *mm); 2202 2203 #ifdef CONFIG_TRACING 2204 extern void 2205 __trace_special(void *__tr, void *__data, 2206 unsigned long arg1, unsigned long arg2, unsigned long arg3); 2207 #else 2208 static inline void 2209 __trace_special(void *__tr, void *__data, 2210 unsigned long arg1, unsigned long arg2, unsigned long arg3) 2211 { 2212 } 2213 #endif 2214 2215 extern long sched_setaffinity(pid_t pid, const cpumask_t *new_mask); 2216 extern long sched_getaffinity(pid_t pid, cpumask_t *mask); 2217 2218 extern int sched_mc_power_savings, sched_smt_power_savings; 2219 2220 extern void normalize_rt_tasks(void); 2221 2222 #ifdef CONFIG_GROUP_SCHED 2223 2224 extern struct task_group init_task_group; 2225 #ifdef CONFIG_USER_SCHED 2226 extern struct task_group root_task_group; 2227 #endif 2228 2229 extern struct task_group *sched_create_group(struct task_group *parent); 2230 extern void sched_destroy_group(struct task_group *tg); 2231 extern void sched_move_task(struct task_struct *tsk); 2232 #ifdef CONFIG_FAIR_GROUP_SCHED 2233 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2234 extern unsigned long sched_group_shares(struct task_group *tg); 2235 #endif 2236 #ifdef CONFIG_RT_GROUP_SCHED 2237 extern int sched_group_set_rt_runtime(struct task_group *tg, 2238 long rt_runtime_us); 2239 extern long sched_group_rt_runtime(struct task_group *tg); 2240 extern int sched_group_set_rt_period(struct task_group *tg, 2241 long rt_period_us); 2242 extern long sched_group_rt_period(struct task_group *tg); 2243 #endif 2244 #endif 2245 2246 #ifdef CONFIG_TASK_XACCT 2247 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2248 { 2249 tsk->ioac.rchar += amt; 2250 } 2251 2252 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2253 { 2254 tsk->ioac.wchar += amt; 2255 } 2256 2257 static inline void inc_syscr(struct task_struct *tsk) 2258 { 2259 tsk->ioac.syscr++; 2260 } 2261 2262 static inline void inc_syscw(struct task_struct *tsk) 2263 { 2264 tsk->ioac.syscw++; 2265 } 2266 #else 2267 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2268 { 2269 } 2270 2271 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2272 { 2273 } 2274 2275 static inline void inc_syscr(struct task_struct *tsk) 2276 { 2277 } 2278 2279 static inline void inc_syscw(struct task_struct *tsk) 2280 { 2281 } 2282 #endif 2283 2284 #ifndef TASK_SIZE_OF 2285 #define TASK_SIZE_OF(tsk) TASK_SIZE 2286 #endif 2287 2288 #ifdef CONFIG_MM_OWNER 2289 extern void mm_update_next_owner(struct mm_struct *mm); 2290 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2291 #else 2292 static inline void mm_update_next_owner(struct mm_struct *mm) 2293 { 2294 } 2295 2296 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2297 { 2298 } 2299 #endif /* CONFIG_MM_OWNER */ 2300 2301 #define TASK_STATE_TO_CHAR_STR "RSDTtZX" 2302 2303 #endif /* __KERNEL__ */ 2304 2305 #endif 2306