blob: 5a9168b01db807e9b8a056dda626a5cf319ad862 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
David Woodhouseb7b3c762006-04-27 00:12:56 +01004/*
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 */
Dave Jones43bb40c2011-01-13 15:45:40 -080024/* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
25 and is now available for re-use. */
Serge E. Hallyn071df102006-10-02 02:18:17 -070026#define CLONE_NEWUTS 0x04000000 /* New utsname group? */
Kirill Korotaev25b21cb2006-10-02 02:18:19 -070027#define CLONE_NEWIPC 0x08000000 /* New ipcs */
Serge E. Hallyn77ec7392007-07-15 23:41:01 -070028#define CLONE_NEWUSER 0x10000000 /* New user namespace */
Pavel Emelyanov30e49c22007-10-18 23:40:10 -070029#define CLONE_NEWPID 0x20000000 /* New pid namespace */
Eric W. Biederman169e3672007-09-27 17:10:06 -070030#define CLONE_NEWNET 0x40000000 /* New network namespace */
Jens Axboefadad8782008-01-24 08:54:47 +010031#define CLONE_IO 0x80000000 /* Clone io context */
David Woodhouseb7b3c762006-04-27 00:12:56 +010032
33/*
34 * Scheduling policies
35 */
36#define SCHED_NORMAL 0
37#define SCHED_FIFO 1
38#define SCHED_RR 2
39#define SCHED_BATCH 3
Ingo Molnar0e6aca42007-07-09 18:51:57 +020040/* SCHED_ISO: reserved but not implemented yet */
41#define SCHED_IDLE 5
Lennart Poetteringca94c442009-06-15 17:17:47 +020042/* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
43#define SCHED_RESET_ON_FORK 0x40000000
David Woodhouseb7b3c762006-04-27 00:12:56 +010044
David Woodhousea3b67142006-04-25 14:54:40 +010045#ifdef __KERNEL__
David Woodhouseb7b3c762006-04-27 00:12:56 +010046
47struct sched_param {
48 int sched_priority;
49};
50
Linus Torvalds1da177e2005-04-16 15:20:36 -070051#include <asm/param.h> /* for HZ */
52
Linus Torvalds1da177e2005-04-16 15:20:36 -070053#include <linux/capability.h>
54#include <linux/threads.h>
55#include <linux/kernel.h>
56#include <linux/types.h>
57#include <linux/timex.h>
58#include <linux/jiffies.h>
59#include <linux/rbtree.h>
60#include <linux/thread_info.h>
61#include <linux/cpumask.h>
62#include <linux/errno.h>
63#include <linux/nodemask.h>
Martin Schwidefskyc92ff1b2007-10-16 01:24:43 -070064#include <linux/mm_types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070065
66#include <asm/system.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070067#include <asm/page.h>
68#include <asm/ptrace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070069#include <asm/cputime.h>
70
71#include <linux/smp.h>
72#include <linux/sem.h>
73#include <linux/signal.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070074#include <linux/compiler.h>
75#include <linux/completion.h>
76#include <linux/pid.h>
77#include <linux/percpu.h>
78#include <linux/topology.h>
Peter Zijlstra3e26c142007-10-16 23:25:50 -070079#include <linux/proportions.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070080#include <linux/seccomp.h>
Ingo Molnare56d0902006-01-08 01:01:37 -080081#include <linux/rcupdate.h>
Jiri Pirko05725f72009-04-14 20:17:16 +020082#include <linux/rculist.h>
Ingo Molnar23f78d4a2006-06-27 02:54:53 -070083#include <linux/rtmutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070084
David Woodhousea3b67142006-04-25 14:54:40 +010085#include <linux/time.h>
86#include <linux/param.h>
87#include <linux/resource.h>
88#include <linux/timer.h>
89#include <linux/hrtimer.h>
Andrew Morton7c3ab732006-12-10 02:19:19 -080090#include <linux/task_io_accounting.h>
Arjan van de Ven97455122008-01-25 21:08:34 +010091#include <linux/latencytop.h>
David Howells9e2b2dc2008-08-13 16:20:04 +010092#include <linux/cred.h>
David Woodhousea3b67142006-04-25 14:54:40 +010093
94#include <asm/processor.h>
H. J. Lu36d57ac2005-09-06 15:16:49 -070095
Linus Torvalds1da177e2005-04-16 15:20:36 -070096struct exec_domain;
Ingo Molnarc87e2832006-06-27 02:54:58 -070097struct futex_pi_state;
Alexey Dobriyan286100a2008-01-25 21:08:34 +010098struct robust_list_head;
Akinobu Mitabddd87c2010-02-23 08:55:42 +010099struct bio_list;
Al Viro5ad4e532009-03-29 19:50:06 -0400100struct fs_struct;
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200101struct perf_event_context;
Jens Axboe73c10102011-03-08 13:19:51 +0100102struct blk_plug;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103
104/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105 * List of flags we want to share for kernel threads,
106 * if only because they are not used by them anyway.
107 */
108#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
109
110/*
111 * These are the constant used to fake the fixed-point load-average
112 * counting. Some notes:
113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
114 * a load-average precision of 10 bits integer + 11 bits fractional
115 * - if you want to count load-averages more often, you need more
116 * precision, or rounding will get you. With 2-second counting freq,
117 * the EXP_n values would be 1981, 2034 and 2043 if still using only
118 * 11 bit fractions.
119 */
120extern unsigned long avenrun[]; /* Load averages */
Thomas Gleixner2d024942009-05-02 20:08:52 +0200121extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700122
123#define FSHIFT 11 /* nr of bits of precision */
124#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
Linus Torvalds0c2043a2007-10-07 16:17:38 -0700125#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
127#define EXP_5 2014 /* 1/exp(5sec/5min) */
128#define EXP_15 2037 /* 1/exp(5sec/15min) */
129
130#define CALC_LOAD(load,exp,n) \
131 load *= exp; \
132 load += n*(FIXED_1-exp); \
133 load >>= FSHIFT;
134
135extern unsigned long total_forks;
136extern int nr_threads;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700137DECLARE_PER_CPU(unsigned long, process_counts);
138extern int nr_processes(void);
139extern unsigned long nr_running(void);
140extern unsigned long nr_uninterruptible(void);
141extern unsigned long nr_iowait(void);
Peter Zijlstra8c215bd2010-07-01 09:07:17 +0200142extern unsigned long nr_iowait_cpu(int cpu);
Arjan van de Ven69d25872009-09-21 17:04:08 -0700143extern unsigned long this_cpu_load(void);
144
145
Peter Zijlstra0f004f52010-11-30 19:48:45 +0100146extern void calc_global_load(unsigned long ticks);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147
Steven Rostedt7e49fcc2009-01-22 19:01:40 -0500148extern unsigned long get_parent_ip(unsigned long addr);
149
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200150struct seq_file;
151struct cfs_rq;
Ingo Molnar4cf86d72007-10-15 17:00:14 +0200152struct task_group;
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200153#ifdef CONFIG_SCHED_DEBUG
154extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
155extern void proc_sched_set_task(struct task_struct *p);
156extern void
Ingo Molnar5cef9ec2007-08-09 11:16:47 +0200157print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200158#else
159static inline void
160proc_sched_show_task(struct task_struct *p, struct seq_file *m)
161{
162}
163static inline void proc_sched_set_task(struct task_struct *p)
164{
165}
166static inline void
Ingo Molnar5cef9ec2007-08-09 11:16:47 +0200167print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200168{
169}
170#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700172/*
173 * Task state bitmask. NOTE! These bits are also
174 * encoded in fs/proc/array.c: get_task_state().
175 *
176 * We have two separate sets of flags: task->state
177 * is about runnability, while task->exit_state are
178 * about the task exiting. Confusing, but this way
179 * modifying one set can't modify the other one by
180 * mistake.
181 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700182#define TASK_RUNNING 0
183#define TASK_INTERRUPTIBLE 1
184#define TASK_UNINTERRUPTIBLE 2
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500185#define __TASK_STOPPED 4
186#define __TASK_TRACED 8
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700187/* in tsk->exit_state */
188#define EXIT_ZOMBIE 16
189#define EXIT_DEAD 32
190/* in tsk->state again */
Mike Galbraithaf927232007-10-15 17:00:13 +0200191#define TASK_DEAD 64
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500192#define TASK_WAKEKILL 128
Peter Zijlstrae9c84312009-09-15 14:43:03 +0200193#define TASK_WAKING 256
Peter Zijlstrae1781532009-12-17 13:16:30 +0100194#define TASK_STATE_MAX 512
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500195
Peter Zijlstra44d90df2009-12-17 13:16:28 +0100196#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
Peter Zijlstra73342152009-12-17 13:16:27 +0100197
Peter Zijlstrae1781532009-12-17 13:16:30 +0100198extern char ___assert_task_state[1 - 2*!!(
199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500200
201/* Convenience macros for the sake of set_task_state */
202#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
204#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700205
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500206/* Convenience macros for the sake of wake_up */
207#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500208#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500209
210/* get_task_state() */
211#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
213 __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500214
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500215#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
216#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
David Howells8f920542010-07-29 12:45:55 +0100217#define task_is_dead(task) ((task)->exit_state != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500218#define task_is_stopped_or_traced(task) \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500220#define task_contributes_to_load(task) \
Nathan Lynche3c8ca82009-04-08 19:45:12 -0500221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
Thomas Gleixner6301cb92009-07-17 14:15:47 +0200222 (task->flags & PF_FREEZING) == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700223
224#define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226#define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
228
Andrew Morton498d0c52005-09-13 01:25:14 -0700229/*
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
232 * actually sleep:
233 *
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
236 * schedule();
237 *
238 * If the caller does not need such serialisation then use __set_current_state()
239 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700240#define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242#define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
244
245/* Task command name length */
246#define TASK_COMM_LEN 16
247
Linus Torvalds1da177e2005-04-16 15:20:36 -0700248#include <linux/spinlock.h>
249
250/*
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
254 * a separate lock).
255 */
256extern rwlock_t tasklist_lock;
257extern spinlock_t mmlist_lock;
258
Ingo Molnar36c8b582006-07-03 00:25:41 -0700259struct task_struct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700260
Paul E. McKenneydb1466b2010-03-03 07:46:56 -0800261#ifdef CONFIG_PROVE_RCU
262extern int lockdep_tasklist_lock_is_held(void);
263#endif /* #ifdef CONFIG_PROVE_RCU */
264
Linus Torvalds1da177e2005-04-16 15:20:36 -0700265extern void sched_init(void);
266extern void sched_init_smp(void);
Harvey Harrison2d07b252008-02-15 09:56:34 -0800267extern asmlinkage void schedule_tail(struct task_struct *prev);
Ingo Molnar36c8b582006-07-03 00:25:41 -0700268extern void init_idle(struct task_struct *idle, int cpu);
Ingo Molnar1df21052007-07-09 18:51:58 +0200269extern void init_idle_bootup_task(struct task_struct *idle);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700270
Andrew Morton89f19f02009-09-19 11:55:44 -0700271extern int runqueue_is_locked(int cpu);
Ingo Molnar017730c2008-05-12 21:20:52 +0200272
Rusty Russell6a7b3dc2008-11-25 02:35:04 +1030273extern cpumask_var_t nohz_cpu_mask;
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700274#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
Venkatesh Pallipadi83cd4fe2010-05-21 17:09:41 -0700275extern void select_nohz_load_balancer(int stop_tick);
276extern int get_nohz_timer_target(void);
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700277#else
Venkatesh Pallipadi83cd4fe2010-05-21 17:09:41 -0700278static inline void select_nohz_load_balancer(int stop_tick) { }
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700279#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800281/*
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700282 * Only dump TASK_* tasks. (0 for all tasks)
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800283 */
284extern void show_state_filter(unsigned long state_filter);
285
286static inline void show_state(void)
287{
Ingo Molnar39bc89f2007-04-25 20:50:03 -0700288 show_state_filter(0);
Ingo Molnare59e2ae2006-12-06 20:35:59 -0800289}
290
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291extern void show_regs(struct pt_regs *);
292
293/*
294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
295 * task), SP is the stack pointer of the first frame that should be shown in the back
296 * trace (or NULL if the entire call-chain of the task should be shown).
297 */
298extern void show_stack(struct task_struct *task, unsigned long *sp);
299
300void io_schedule(void);
301long io_schedule_timeout(long timeout);
302
303extern void cpu_init (void);
304extern void trap_init(void);
305extern void update_process_times(int user);
306extern void scheduler_tick(void);
307
Ingo Molnar82a1fcb2008-01-25 21:08:02 +0100308extern void sched_show_task(struct task_struct *p);
309
Frederic Weisbecker19cc36c2010-05-13 02:30:49 +0200310#ifdef CONFIG_LOCKUP_DETECTOR
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700311extern void touch_softlockup_watchdog(void);
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600312extern void touch_softlockup_watchdog_sync(void);
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700313extern void touch_all_softlockup_watchdogs(void);
Don Zickus332fbdb2010-05-07 17:11:45 -0400314extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
315 void __user *buffer,
316 size_t *lenp, loff_t *ppos);
Ingo Molnar9c44bc02008-05-12 21:21:04 +0200317extern unsigned int softlockup_panic;
Dimitri Sivanich9383d962008-05-12 21:21:14 +0200318extern int softlockup_thresh;
Peter Zijlstra004417a2010-11-25 18:38:29 +0100319void lockup_detector_init(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700320#else
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700321static inline void touch_softlockup_watchdog(void)
322{
323}
Jason Wesseld6ad3e22010-01-27 16:25:22 -0600324static inline void touch_softlockup_watchdog_sync(void)
325{
326}
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700327static inline void touch_all_softlockup_watchdogs(void)
328{
329}
Peter Zijlstra004417a2010-11-25 18:38:29 +0100330static inline void lockup_detector_init(void)
331{
332}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700333#endif
334
Mandeep Singh Bainese162b392009-01-15 11:08:40 -0800335#ifdef CONFIG_DETECT_HUNG_TASK
336extern unsigned int sysctl_hung_task_panic;
337extern unsigned long sysctl_hung_task_check_count;
338extern unsigned long sysctl_hung_task_timeout_secs;
339extern unsigned long sysctl_hung_task_warnings;
340extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
Alexey Dobriyan8d65af72009-09-23 15:57:19 -0700341 void __user *buffer,
Mandeep Singh Bainese162b392009-01-15 11:08:40 -0800342 size_t *lenp, loff_t *ppos);
Mark Lorde4ecda12010-09-25 11:17:22 +0200343#else
344/* Avoid need for ifdefs elsewhere in the code */
345enum { sysctl_hung_task_timeout_secs = 0 };
Mandeep Singh Bainese162b392009-01-15 11:08:40 -0800346#endif
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700347
Linus Torvalds1da177e2005-04-16 15:20:36 -0700348/* Attach to any functions which should be ignored in wchan output. */
349#define __sched __attribute__((__section__(".sched.text")))
Ingo Molnardeaf2222007-11-28 15:52:56 +0100350
351/* Linker adds these: start and end of __sched functions */
352extern char __sched_text_start[], __sched_text_end[];
353
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354/* Is this address in the __sched functions? */
355extern int in_sched_functions(unsigned long addr);
356
357#define MAX_SCHEDULE_TIMEOUT LONG_MAX
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800358extern signed long schedule_timeout(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700359extern signed long schedule_timeout_interruptible(signed long timeout);
Matthew Wilcox294d5cc2007-12-06 11:59:46 -0500360extern signed long schedule_timeout_killable(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700361extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700362asmlinkage void schedule(void);
Peter Zijlstra0d66bf62009-01-12 14:01:47 +0100363extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700364
Serge E. Hallynab516012006-10-02 02:18:06 -0700365struct nsproxy;
Cedric Le Goateracce2922007-07-15 23:40:59 -0700366struct user_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700367
KAMEZAWA Hiroyuki341c87b2009-06-30 11:41:23 -0700368/*
369 * Default maximum number of active map areas, this limits the number of vmas
370 * per mm struct. Users can overwrite this number by sysctl but there is a
371 * problem.
372 *
373 * When a program's coredump is generated as ELF format, a section is created
374 * per a vma. In ELF, the number of sections is represented in unsigned short.
375 * This means the number of sections should be smaller than 65535 at coredump.
376 * Because the kernel adds some informative sections to a image of program at
377 * generating coredump, we need some margin. The number of extra sections is
378 * 1-3 now and depends on arch. We use "5" as safe margin, here.
379 */
380#define MAPCOUNT_ELF_CORE_MARGIN (5)
Alexey Dobriyan4be929b2010-05-24 14:33:03 -0700381#define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382
383extern int sysctl_max_map_count;
384
385#include <linux/aio.h>
386
David Howellsefc1a3b2010-01-15 17:01:35 -0800387#ifdef CONFIG_MMU
388extern void arch_pick_mmap_layout(struct mm_struct *mm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389extern unsigned long
390arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
392extern unsigned long
393arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700396extern void arch_unmap_area(struct mm_struct *, unsigned long);
397extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
David Howellsefc1a3b2010-01-15 17:01:35 -0800398#else
399static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
400#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700401
Oleg Nesterov901608d2009-01-06 14:40:29 -0800402
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700403extern void set_dumpable(struct mm_struct *mm, int value);
404extern int get_dumpable(struct mm_struct *mm);
405
406/* mm flags */
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700407/* dumpable bits */
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700408#define MMF_DUMPABLE 0 /* core dump is permitted */
409#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700410
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700411#define MMF_DUMPABLE_BITS 2
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700412#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700413
414/* coredump filter bits */
415#define MMF_DUMP_ANON_PRIVATE 2
416#define MMF_DUMP_ANON_SHARED 3
417#define MMF_DUMP_MAPPED_PRIVATE 4
418#define MMF_DUMP_MAPPED_SHARED 5
Roland McGrath82df3972007-10-16 23:27:02 -0700419#define MMF_DUMP_ELF_HEADERS 6
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700420#define MMF_DUMP_HUGETLB_PRIVATE 7
421#define MMF_DUMP_HUGETLB_SHARED 8
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700422
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700423#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700424#define MMF_DUMP_FILTER_BITS 7
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700425#define MMF_DUMP_FILTER_MASK \
426 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
427#define MMF_DUMP_FILTER_DEFAULT \
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700428 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
Roland McGrath656eb2c2008-10-18 20:28:23 -0700429 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
430
431#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
432# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
433#else
434# define MMF_DUMP_MASK_DEFAULT_ELF 0
435#endif
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700436 /* leave room for more dump flags */
437#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
Andrea Arcangeliba761492011-01-13 15:46:58 -0800438#define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700439
440#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700441
Linus Torvalds1da177e2005-04-16 15:20:36 -0700442struct sighand_struct {
443 atomic_t count;
444 struct k_sigaction action[_NSIG];
445 spinlock_t siglock;
Davide Libenzib8fceee2007-09-20 12:40:16 -0700446 wait_queue_head_t signalfd_wqh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700447};
448
KaiGai Kohei0e464812006-06-25 05:49:24 -0700449struct pacct_struct {
KaiGai Koheif6ec29a2006-06-25 05:49:25 -0700450 int ac_flag;
451 long ac_exitcode;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700452 unsigned long ac_mem;
KaiGai Kohei77787bf2006-06-25 05:49:26 -0700453 cputime_t ac_utime, ac_stime;
454 unsigned long ac_minflt, ac_majflt;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700455};
456
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200457struct cpu_itimer {
458 cputime_t expires;
459 cputime_t incr;
Stanislaw Gruszka8356b5f2009-07-29 12:15:27 +0200460 u32 error;
461 u32 incr_error;
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200462};
463
Frank Mayharf06febc2008-09-12 09:54:39 -0700464/**
465 * struct task_cputime - collected CPU time counts
466 * @utime: time spent in user mode, in &cputime_t units
467 * @stime: time spent in kernel mode, in &cputime_t units
468 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
Ingo Molnar5ce73a42008-09-14 17:11:46 +0200469 *
Frank Mayharf06febc2008-09-12 09:54:39 -0700470 * This structure groups together three kinds of CPU time that are
471 * tracked for threads and thread groups. Most things considering
472 * CPU time want to group these counts together and treat all three
473 * of them in parallel.
474 */
475struct task_cputime {
476 cputime_t utime;
477 cputime_t stime;
478 unsigned long long sum_exec_runtime;
479};
480/* Alternate field names when used to cache expirations. */
481#define prof_exp stime
482#define virt_exp utime
483#define sched_exp sum_exec_runtime
484
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100485#define INIT_CPUTIME \
486 (struct task_cputime) { \
487 .utime = cputime_zero, \
488 .stime = cputime_zero, \
489 .sum_exec_runtime = 0, \
490 }
491
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200492/*
493 * Disable preemption until the scheduler is running.
494 * Reset by start_kernel()->sched_init()->init_idle().
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200495 *
496 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
497 * before the scheduler is active -- see should_resched().
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200498 */
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200499#define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200500
Frank Mayharf06febc2008-09-12 09:54:39 -0700501/**
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100502 * struct thread_group_cputimer - thread group interval timer counts
503 * @cputime: thread group interval timers.
504 * @running: non-zero when there are timers running and
505 * @cputime receives updates.
506 * @lock: lock for fields in this struct.
Frank Mayharf06febc2008-09-12 09:54:39 -0700507 *
508 * This structure contains the version of task_cputime, above, that is
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100509 * used for thread group CPU timer calculations.
Frank Mayharf06febc2008-09-12 09:54:39 -0700510 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100511struct thread_group_cputimer {
512 struct task_cputime cputime;
513 int running;
514 spinlock_t lock;
Frank Mayharf06febc2008-09-12 09:54:39 -0700515};
Frank Mayharf06febc2008-09-12 09:54:39 -0700516
Mike Galbraith5091faa2010-11-30 14:18:03 +0100517struct autogroup;
518
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519/*
Jonathan Neuschäfere815f0a2011-03-21 20:24:47 +0100520 * NOTE! "signal_struct" does not have its own
Linus Torvalds1da177e2005-04-16 15:20:36 -0700521 * locking, because a shared signal_struct always
522 * implies a shared sighand_struct, so locking
523 * sighand_struct is always a proper superset of
524 * the locking of signal_struct.
525 */
526struct signal_struct {
Oleg Nesterovea6d2902010-05-26 14:43:16 -0700527 atomic_t sigcnt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700528 atomic_t live;
Oleg Nesterovb3ac0222010-05-26 14:43:24 -0700529 int nr_threads;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700530
531 wait_queue_head_t wait_chldexit; /* for wait4() */
532
533 /* current thread group signal load-balancing target: */
Ingo Molnar36c8b582006-07-03 00:25:41 -0700534 struct task_struct *curr_target;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700535
536 /* shared signal handling: */
537 struct sigpending shared_pending;
538
539 /* thread group exit support */
540 int group_exit_code;
541 /* overloaded:
542 * - notify group_exit_task when ->count is equal to notify_count
543 * - everyone except group_exit_task is stopped during signal delivery
544 * of fatal signals, group_exit_task processes the signal.
545 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700546 int notify_count;
Richard Kennedy07dd20e2008-08-01 13:18:04 +0100547 struct task_struct *group_exit_task;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700548
549 /* thread group stop support, overloads group_exit_code too */
550 int group_stop_count;
551 unsigned int flags; /* see SIGNAL_* flags below */
552
553 /* POSIX.1b Interval Timers */
554 struct list_head posix_timers;
555
556 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800557 struct hrtimer real_timer;
Oleg Nesterovfea9d172008-02-08 04:19:19 -0800558 struct pid *leader_pid;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800559 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700560
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200561 /*
562 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
563 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
564 * values are defined to 0 and 1 respectively
565 */
566 struct cpu_itimer it[2];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700567
Frank Mayharf06febc2008-09-12 09:54:39 -0700568 /*
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100569 * Thread group totals for process CPU timers.
570 * See thread_group_cputimer(), et al, for details.
Frank Mayharf06febc2008-09-12 09:54:39 -0700571 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100572 struct thread_group_cputimer cputimer;
Frank Mayharf06febc2008-09-12 09:54:39 -0700573
574 /* Earliest-expiration cache. */
575 struct task_cputime cputime_expires;
576
577 struct list_head cpu_timers[3];
578
Eric W. Biedermanab521dc2007-02-12 00:53:00 -0800579 struct pid *tty_old_pgrp;
Cedric Le Goater1ec320a2006-12-08 02:37:55 -0800580
Linus Torvalds1da177e2005-04-16 15:20:36 -0700581 /* boolean value for session group leader */
582 int leader;
583
584 struct tty_struct *tty; /* NULL if no tty */
585
Mike Galbraith5091faa2010-11-30 14:18:03 +0100586#ifdef CONFIG_SCHED_AUTOGROUP
587 struct autogroup *autogroup;
588#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700589 /*
590 * Cumulative resource counters for dead threads in the group,
591 * and for reaped dead child processes forked by this group.
592 * Live threads maintain their own counters and add to these
593 * in __exit_signal, except for the group leader.
594 */
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100595 cputime_t utime, stime, cutime, cstime;
Laurent Vivier9ac52312007-10-15 17:00:19 +0200596 cputime_t gtime;
597 cputime_t cgtime;
Hidetoshi Seto0cf55e12009-12-02 17:28:07 +0900598#ifndef CONFIG_VIRT_CPU_ACCOUNTING
599 cputime_t prev_utime, prev_stime;
600#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700601 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
602 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
Eric Dumazet6eaeeab2007-05-10 22:22:37 -0700603 unsigned long inblock, oublock, cinblock, coublock;
Jiri Pirko1f102062009-09-22 16:44:10 -0700604 unsigned long maxrss, cmaxrss;
Andrea Righi940389b2008-07-28 00:48:12 +0200605 struct task_io_accounting ioac;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700606
607 /*
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100608 * Cumulative ns of schedule CPU time fo dead threads in the
609 * group, not including a zombie group leader, (This only differs
610 * from jiffies_to_ns(utime + stime) if sched_clock uses something
611 * other than jiffies.)
612 */
613 unsigned long long sum_sched_runtime;
614
615 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700616 * We don't bother to synchronize most readers of this at all,
617 * because there is no reader checking a limit that actually needs
618 * to get both rlim_cur and rlim_max atomically, and either one
619 * alone is a single word that can safely be read normally.
620 * getrlimit/setrlimit use task_lock(current->group_leader) to
621 * protect this instead of the siglock, because they really
622 * have no need to disable irqs.
623 */
624 struct rlimit rlim[RLIM_NLIMITS];
625
KaiGai Kohei0e464812006-06-25 05:49:24 -0700626#ifdef CONFIG_BSD_PROCESS_ACCT
627 struct pacct_struct pacct; /* per-process accounting information */
628#endif
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700629#ifdef CONFIG_TASKSTATS
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700630 struct taskstats *stats;
631#endif
Miloslav Trmac522ed772007-07-15 23:40:56 -0700632#ifdef CONFIG_AUDIT
633 unsigned audit_tty;
634 struct tty_audit_buf *tty_audit_buf;
635#endif
KOSAKI Motohiro28b83c52009-09-21 17:03:13 -0700636
David Rientjesa63d83f2010-08-09 17:19:46 -0700637 int oom_adj; /* OOM kill score adjustment (bit shift) */
638 int oom_score_adj; /* OOM kill score adjustment */
Mandeep Singh Bainesdabb16f2011-01-13 15:46:05 -0800639 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
640 * Only settable by CAP_SYS_RESOURCE. */
KOSAKI Motohiro9b1bf122010-10-27 15:34:08 -0700641
642 struct mutex cred_guard_mutex; /* guard against foreign influences on
643 * credential calculations
644 * (notably. ptrace) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700645};
646
Nick Piggin4866cde2005-06-25 14:57:23 -0700647/* Context switch must be unlocked if interrupts are to be enabled */
648#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
649# define __ARCH_WANT_UNLOCKED_CTXSW
650#endif
651
Linus Torvalds1da177e2005-04-16 15:20:36 -0700652/*
653 * Bits in flags field of signal_struct.
654 */
655#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
656#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
657#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
658#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
Oleg Nesterove4420552008-04-30 00:52:44 -0700659/*
660 * Pending notifications to parent.
661 */
662#define SIGNAL_CLD_STOPPED 0x00000010
663#define SIGNAL_CLD_CONTINUED 0x00000020
664#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700665
Oleg Nesterovfae5fa42008-04-30 00:53:03 -0700666#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
667
Oleg Nesteroved5d2ca2008-02-04 22:27:24 -0800668/* If true, all threads except ->group_exit_task have pending SIGKILL */
669static inline int signal_group_exit(const struct signal_struct *sig)
670{
671 return (sig->flags & SIGNAL_GROUP_EXIT) ||
672 (sig->group_exit_task != NULL);
673}
674
Linus Torvalds1da177e2005-04-16 15:20:36 -0700675/*
676 * Some day this will be a full-fledged user tracking system..
677 */
678struct user_struct {
679 atomic_t __count; /* reference count */
680 atomic_t processes; /* How many processes does this user have? */
681 atomic_t files; /* How many open files does this user have? */
682 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700683#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400684 atomic_t inotify_watches; /* How many inotify watches does this user have? */
685 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
686#endif
Eric Paris4afeff82010-10-28 17:21:58 -0400687#ifdef CONFIG_FANOTIFY
688 atomic_t fanotify_listeners;
689#endif
Davide Libenzi7ef99642008-12-01 13:13:55 -0800690#ifdef CONFIG_EPOLL
Robin Holt52bd19f72011-01-12 17:00:01 -0800691 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
Davide Libenzi7ef99642008-12-01 13:13:55 -0800692#endif
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700693#ifdef CONFIG_POSIX_MQUEUE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700694 /* protected by mq_lock */
695 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700696#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700697 unsigned long locked_shm; /* How many pages of mlocked shm ? */
698
699#ifdef CONFIG_KEYS
700 struct key *uid_keyring; /* UID specific keyring */
701 struct key *session_keyring; /* UID's default session keyring */
702#endif
703
704 /* Hash table maintenance information */
Pavel Emelyanov735de222007-09-18 22:46:44 -0700705 struct hlist_node uidhash_node;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700706 uid_t uid;
Serge Hallyn18b6e042008-10-15 16:38:45 -0500707 struct user_namespace *user_ns;
Srivatsa Vaddagiri24e377a2007-10-15 17:00:09 +0200708
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200709#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra789f90f2009-05-15 15:19:27 +0200710 atomic_long_t locked_vm;
711#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700712};
713
Kay Sieverseb41d942007-11-02 13:47:53 +0100714extern int uids_sysfs_init(void);
Dhaval Giani5cb350b2007-10-15 17:00:14 +0200715
Linus Torvalds1da177e2005-04-16 15:20:36 -0700716extern struct user_struct *find_user(uid_t);
717
718extern struct user_struct root_user;
719#define INIT_USER (&root_user)
720
David Howellsb6dff3e2008-11-14 10:39:16 +1100721
Linus Torvalds1da177e2005-04-16 15:20:36 -0700722struct backing_dev_info;
723struct reclaim_state;
724
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700725#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700726struct sched_info {
727 /* cumulative counters */
Ingo Molnar2d723762007-10-15 17:00:12 +0200728 unsigned long pcount; /* # of times run on this cpu */
Ken Chen9c2c4802008-12-16 23:41:22 -0800729 unsigned long long run_delay; /* time spent waiting on a runqueue */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700730
731 /* timestamps */
Balbir Singh172ba842007-07-09 18:52:00 +0200732 unsigned long long last_arrival,/* when we last ran on a cpu */
733 last_queued; /* when we were last queued to run */
Ingo Molnarb8efb562007-10-15 17:00:10 +0200734#ifdef CONFIG_SCHEDSTATS
735 /* BKL stats */
Ken Chen480b9432007-10-18 21:32:56 +0200736 unsigned int bkl_count;
Ingo Molnarb8efb562007-10-15 17:00:10 +0200737#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700738};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700739#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700740
Shailabh Nagarca74e922006-07-14 00:24:36 -0700741#ifdef CONFIG_TASK_DELAY_ACCT
742struct task_delay_info {
743 spinlock_t lock;
744 unsigned int flags; /* Private per-task flags */
745
746 /* For each stat XXX, add following, aligned appropriately
747 *
748 * struct timespec XXX_start, XXX_end;
749 * u64 XXX_delay;
750 * u32 XXX_count;
751 *
752 * Atomicity of updates to XXX_delay, XXX_count protected by
753 * single lock above (split into XXX_lock if contention is an issue).
754 */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700755
756 /*
757 * XXX_count is incremented on every XXX operation, the delay
758 * associated with the operation is added to XXX_delay.
759 * XXX_delay contains the accumulated delay time in nanoseconds.
760 */
761 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
762 u64 blkio_delay; /* wait for sync block io completion */
763 u64 swapin_delay; /* wait for swapin block io completion */
764 u32 blkio_count; /* total count of the number of sync block */
765 /* io operations performed */
766 u32 swapin_count; /* total count of the number of swapin block */
767 /* io operations performed */
Keika Kobayashi873b4772008-07-25 01:48:52 -0700768
769 struct timespec freepages_start, freepages_end;
770 u64 freepages_delay; /* wait for memory reclaim */
771 u32 freepages_count; /* total count of memory reclaim */
Shailabh Nagarca74e922006-07-14 00:24:36 -0700772};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700773#endif /* CONFIG_TASK_DELAY_ACCT */
774
775static inline int sched_info_on(void)
776{
777#ifdef CONFIG_SCHEDSTATS
778 return 1;
779#elif defined(CONFIG_TASK_DELAY_ACCT)
780 extern int delayacct_on;
781 return delayacct_on;
782#else
783 return 0;
Shailabh Nagarca74e922006-07-14 00:24:36 -0700784#endif
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700785}
Shailabh Nagarca74e922006-07-14 00:24:36 -0700786
Ingo Molnard15bcfd2007-07-09 18:51:57 +0200787enum cpu_idle_type {
788 CPU_IDLE,
789 CPU_NOT_IDLE,
790 CPU_NEWLY_IDLE,
791 CPU_MAX_IDLE_TYPES
Linus Torvalds1da177e2005-04-16 15:20:36 -0700792};
793
794/*
795 * sched-domains (multiprocessor balancing) declarations:
796 */
Ingo Molnar9aa7b362007-07-09 18:51:58 +0200797
798/*
799 * Increase resolution of nice-level calculations:
800 */
801#define SCHED_LOAD_SHIFT 10
802#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
803
Suresh Siddhaf8700df2007-08-23 15:18:02 +0200804#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700805
Peter Williams2dd73a42006-06-27 02:54:34 -0700806#ifdef CONFIG_SMP
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200807#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
808#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
809#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
810#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
Peter Zijlstrac88d5912009-09-10 13:50:02 +0200811#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200812#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
Peter Zijlstra59abf022009-09-16 08:28:30 +0200813#define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200814#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
815#define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
816#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
817#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
Michael Neuling532cb4c2010-06-08 14:57:02 +1000818#define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200819#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700820
Gautham R Shenoyafb8a9b2008-12-18 23:26:09 +0530821enum powersavings_balance_level {
822 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
823 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
824 * first for long running threads
825 */
826 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
827 * cpu package for power savings
828 */
829 MAX_POWERSAVINGS_BALANCE_LEVELS
830};
Siddha, Suresh B89c47102006-10-03 01:14:09 -0700831
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530832extern int sched_mc_power_savings, sched_smt_power_savings;
Siddha, Suresh B89c47102006-10-03 01:14:09 -0700833
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530834static inline int sd_balance_for_mc_power(void)
835{
836 if (sched_smt_power_savings)
837 return SD_POWERSAVINGS_BALANCE;
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700838
Vaidyanathan Srinivasan28f53182010-02-08 15:35:55 +0530839 if (!sched_mc_power_savings)
840 return SD_PREFER_SIBLING;
841
842 return 0;
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530843}
844
845static inline int sd_balance_for_package_power(void)
846{
847 if (sched_mc_power_savings | sched_smt_power_savings)
848 return SD_POWERSAVINGS_BALANCE;
849
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200850 return SD_PREFER_SIBLING;
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530851}
Nick Piggin147cbb42005-06-25 14:57:19 -0700852
Michael Neuling532cb4c2010-06-08 14:57:02 +1000853extern int __weak arch_sd_sibiling_asym_packing(void);
854
Vaidyanathan Srinivasan100fdae2008-12-18 23:26:47 +0530855/*
856 * Optimise SD flags for power savings:
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300857 * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
Vaidyanathan Srinivasan100fdae2008-12-18 23:26:47 +0530858 * Keep default SD flags if sched_{smt,mc}_power_saving=0
859 */
860
861static inline int sd_power_saving_flags(void)
862{
863 if (sched_mc_power_savings | sched_smt_power_savings)
864 return SD_BALANCE_NEWIDLE;
865
866 return 0;
867}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700868
869struct sched_group {
870 struct sched_group *next; /* Must be a circular list */
Peter Zijlstradce840a2011-04-07 14:09:50 +0200871 atomic_t ref;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700872
873 /*
874 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
Peter Zijlstra18a38852009-09-01 10:34:39 +0200875 * single CPU.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700876 */
Srivatsa Vaddagiri9d5efe02010-06-08 14:57:02 +1000877 unsigned int cpu_power, cpu_power_orig;
Suresh Siddhaaae6d3d2010-09-17 15:02:32 -0700878 unsigned int group_weight;
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030879
Ingo Molnar4200efd2009-05-19 09:22:19 +0200880 /*
881 * The CPUs this group covers.
882 *
883 * NOTE: this field is variable length. (Allocated dynamically
884 * by attaching extra space to the end of the structure,
885 * depending on how many CPUs the kernel has booted up with)
886 *
887 * It is also be embedded into static data structures at build
888 * time. (See 'struct static_sched_group' in kernel/sched.c)
889 */
890 unsigned long cpumask[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700891};
892
Rusty Russell758b2cd2008-11-25 02:35:04 +1030893static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
894{
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030895 return to_cpumask(sg->cpumask);
Rusty Russell758b2cd2008-11-25 02:35:04 +1030896}
897
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900898enum sched_domain_level {
899 SD_LV_NONE = 0,
Peter Zijlstra38591732011-04-07 14:09:53 +0200900#ifdef CONFIG_SCHED_SMT
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900901 SD_LV_SIBLING,
Peter Zijlstra38591732011-04-07 14:09:53 +0200902#endif
903#ifdef CONFIG_SCHED_MC
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900904 SD_LV_MC,
Peter Zijlstra38591732011-04-07 14:09:53 +0200905#endif
906#ifdef CONFIG_SCHED_BOOK
Heiko Carstens01a08542010-08-31 10:28:16 +0200907 SD_LV_BOOK,
Peter Zijlstra38591732011-04-07 14:09:53 +0200908#endif
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900909 SD_LV_CPU,
Peter Zijlstra38591732011-04-07 14:09:53 +0200910#ifdef CONFIG_NUMA
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900911 SD_LV_NODE,
912 SD_LV_ALLNODES,
Peter Zijlstra38591732011-04-07 14:09:53 +0200913#endif
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900914 SD_LV_MAX
915};
916
917struct sched_domain_attr {
918 int relax_domain_level;
919};
920
921#define SD_ATTR_INIT (struct sched_domain_attr) { \
922 .relax_domain_level = -1, \
923}
924
Linus Torvalds1da177e2005-04-16 15:20:36 -0700925struct sched_domain {
926 /* These fields must be setup */
927 struct sched_domain *parent; /* top domain must be null terminated */
Siddha, Suresh B1a848872006-10-03 01:14:08 -0700928 struct sched_domain *child; /* bottom domain must be null terminated */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700929 struct sched_group *groups; /* the balancing groups of the domain */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700930 unsigned long min_interval; /* Minimum balance interval ms */
931 unsigned long max_interval; /* Maximum balance interval ms */
932 unsigned int busy_factor; /* less balancing by factor if busy */
933 unsigned int imbalance_pct; /* No balance until over watermark */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700934 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
Nick Piggin78979862005-06-25 14:57:13 -0700935 unsigned int busy_idx;
936 unsigned int idle_idx;
937 unsigned int newidle_idx;
938 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700939 unsigned int forkexec_idx;
Peter Zijlstraa52bfd732009-09-01 10:34:35 +0200940 unsigned int smt_gain;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700941 int flags; /* See SD_* */
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900942 enum sched_domain_level level;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700943
944 /* Runtime fields. */
945 unsigned long last_balance; /* init to jiffies. units in jiffies */
946 unsigned int balance_interval; /* initialise to 1. units in ms. */
947 unsigned int nr_balance_failed; /* initialise to 0 */
948
Peter Zijlstra2398f2c2008-06-27 13:41:35 +0200949 u64 last_update;
950
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951#ifdef CONFIG_SCHEDSTATS
952 /* load_balance() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200953 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
954 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
955 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
956 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
957 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
958 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
959 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
960 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700961
962 /* Active load balancing */
Ken Chen480b9432007-10-18 21:32:56 +0200963 unsigned int alb_count;
964 unsigned int alb_failed;
965 unsigned int alb_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966
Nick Piggin68767a02005-06-25 14:57:20 -0700967 /* SD_BALANCE_EXEC stats */
Ken Chen480b9432007-10-18 21:32:56 +0200968 unsigned int sbe_count;
969 unsigned int sbe_balanced;
970 unsigned int sbe_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700971
Nick Piggin68767a02005-06-25 14:57:20 -0700972 /* SD_BALANCE_FORK stats */
Ken Chen480b9432007-10-18 21:32:56 +0200973 unsigned int sbf_count;
974 unsigned int sbf_balanced;
975 unsigned int sbf_pushed;
Nick Piggin68767a02005-06-25 14:57:20 -0700976
Linus Torvalds1da177e2005-04-16 15:20:36 -0700977 /* try_to_wake_up() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200978 unsigned int ttwu_wake_remote;
979 unsigned int ttwu_move_affine;
980 unsigned int ttwu_move_balance;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700981#endif
Ingo Molnara5d8c342008-10-09 11:35:51 +0200982#ifdef CONFIG_SCHED_DEBUG
983 char *name;
984#endif
Peter Zijlstradce840a2011-04-07 14:09:50 +0200985 union {
986 void *private; /* used during construction */
987 struct rcu_head rcu; /* used during destruction */
988 };
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030989
Peter Zijlstra669c55e2010-04-16 14:59:29 +0200990 unsigned int span_weight;
Ingo Molnar4200efd2009-05-19 09:22:19 +0200991 /*
992 * Span of all CPUs in this domain.
993 *
994 * NOTE: this field is variable length. (Allocated dynamically
995 * by attaching extra space to the end of the structure,
996 * depending on how many CPUs the kernel has booted up with)
997 *
998 * It is also be embedded into static data structures at build
999 * time. (See 'struct static_sched_domain' in kernel/sched.c)
1000 */
1001 unsigned long span[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -07001002};
1003
Rusty Russell758b2cd2008-11-25 02:35:04 +10301004static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1005{
Rusty Russell6c99e9a2008-11-25 02:35:04 +10301006 return to_cpumask(sd->span);
Rusty Russell758b2cd2008-11-25 02:35:04 +10301007}
1008
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301009extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +09001010 struct sched_domain_attr *dattr_new);
Paul Jackson029190c2007-10-18 23:40:20 -07001011
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301012/* Allocate an array of sched domains, for partition_sched_domains(). */
1013cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1014void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1015
Ingo Molnar06aaf762008-12-18 21:30:23 +01001016/* Test a flag in parent sched domain */
1017static inline int test_sd_parent(struct sched_domain *sd, int flag)
1018{
1019 if (sd->parent && (sd->parent->flags & flag))
1020 return 1;
1021
1022 return 0;
1023}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001024
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001025unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1026unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1027
Ingo Molnar1b427c12008-07-18 14:01:39 +02001028#else /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001029
Ingo Molnar1b427c12008-07-18 14:01:39 +02001030struct sched_domain_attr;
1031
1032static inline void
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301033partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Ingo Molnar1b427c12008-07-18 14:01:39 +02001034 struct sched_domain_attr *dattr_new)
Con Kolivasd02c7a82007-07-26 13:40:43 +02001035{
Con Kolivasd02c7a82007-07-26 13:40:43 +02001036}
Ingo Molnar1b427c12008-07-18 14:01:39 +02001037#endif /* !CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001038
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001039
Linus Torvalds1da177e2005-04-16 15:20:36 -07001040struct io_context; /* See blkdev.h */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001041
Linus Torvalds1da177e2005-04-16 15:20:36 -07001042
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001043#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
Ingo Molnar36c8b582006-07-03 00:25:41 -07001044extern void prefetch_stack(struct task_struct *t);
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001045#else
1046static inline void prefetch_stack(struct task_struct *t) { }
1047#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001048
1049struct audit_context; /* See audit.c */
1050struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +02001051struct pipe_inode_info;
Serge E. Hallyn4865ecf2006-10-02 02:18:14 -07001052struct uts_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001053
Ingo Molnar20b8a592007-07-09 18:51:58 +02001054struct rq;
1055struct sched_domain;
1056
Peter Zijlstra7d478722009-09-14 19:55:44 +02001057/*
1058 * wake flags
1059 */
1060#define WF_SYNC 0x01 /* waker goes to sleep after wakup */
Peter Zijlstraa7558e02009-09-14 20:02:34 +02001061#define WF_FORK 0x02 /* child wakeup after fork */
Peter Zijlstra7d478722009-09-14 19:55:44 +02001062
Peter Zijlstra371fd7e2010-03-24 16:38:48 +01001063#define ENQUEUE_WAKEUP 1
1064#define ENQUEUE_WAKING 2
1065#define ENQUEUE_HEAD 4
1066
1067#define DEQUEUE_SLEEP 1
1068
Ingo Molnar20b8a592007-07-09 18:51:58 +02001069struct sched_class {
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001070 const struct sched_class *next;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001071
Peter Zijlstra371fd7e2010-03-24 16:38:48 +01001072 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1073 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
Dmitry Adamushko4530d7a2007-10-15 17:00:08 +02001074 void (*yield_task) (struct rq *rq);
Mike Galbraithd95f4122011-02-01 09:50:51 -05001075 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
Ingo Molnar20b8a592007-07-09 18:51:58 +02001076
Peter Zijlstra7d478722009-09-14 19:55:44 +02001077 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
Ingo Molnar20b8a592007-07-09 18:51:58 +02001078
Ingo Molnarfb8d4722007-08-09 11:16:48 +02001079 struct task_struct * (*pick_next_task) (struct rq *rq);
Ingo Molnar31ee5292007-08-09 11:16:49 +02001080 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
Ingo Molnar20b8a592007-07-09 18:51:58 +02001081
Peter Williams681f3e62007-10-24 18:23:51 +02001082#ifdef CONFIG_SMP
Peter Zijlstra0017d732010-03-24 18:34:10 +01001083 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1084 int sd_flag, int flags);
Li Zefan4ce72a22008-10-22 15:25:26 +08001085
Steven Rostedt9a897c52008-01-25 21:08:22 +01001086 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1087 void (*post_schedule) (struct rq *this_rq);
Peter Zijlstraefbbd052009-12-16 18:04:40 +01001088 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1089 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
Peter Williamse1d14842007-10-24 18:23:51 +02001090
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001091 void (*set_cpus_allowed)(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10301092 const struct cpumask *newmask);
Gregory Haskins57d885f2008-01-25 21:08:18 +01001093
Gregory Haskins1f11eb62008-06-04 15:04:05 -04001094 void (*rq_online)(struct rq *rq);
1095 void (*rq_offline)(struct rq *rq);
Li Zefan4ce72a22008-10-22 15:25:26 +08001096#endif
1097
1098 void (*set_curr_task) (struct rq *rq);
1099 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
Peter Zijlstracd29fe62009-11-27 17:32:46 +01001100 void (*task_fork) (struct task_struct *p);
Steven Rostedtcb469842008-01-25 21:08:22 +01001101
Peter Zijlstrada7a7352011-01-17 17:03:27 +01001102 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1103 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
Steven Rostedtcb469842008-01-25 21:08:22 +01001104 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
Peter Zijlstrada7a7352011-01-17 17:03:27 +01001105 int oldprio);
Peter Zijlstra810b3812008-02-29 15:21:01 -05001106
Thomas Gleixnerdba091b2009-12-09 09:32:03 +01001107 unsigned int (*get_rr_interval) (struct rq *rq,
1108 struct task_struct *task);
Peter Williams0d721ce2009-09-21 01:31:53 +00001109
Peter Zijlstra810b3812008-02-29 15:21:01 -05001110#ifdef CONFIG_FAIR_GROUP_SCHED
Peter Zijlstrab2b5ce02010-10-15 15:24:15 +02001111 void (*task_move_group) (struct task_struct *p, int on_rq);
Peter Zijlstra810b3812008-02-29 15:21:01 -05001112#endif
Ingo Molnar20b8a592007-07-09 18:51:58 +02001113};
1114
1115struct load_weight {
1116 unsigned long weight, inv_weight;
1117};
1118
Ingo Molnar94c18222007-08-02 17:41:40 +02001119#ifdef CONFIG_SCHEDSTATS
Lucas De Marchi41acab82010-03-10 23:37:45 -03001120struct sched_statistics {
Ingo Molnar94c18222007-08-02 17:41:40 +02001121 u64 wait_start;
1122 u64 wait_max;
Arjan van de Ven6d082592008-01-25 21:08:35 +01001123 u64 wait_count;
1124 u64 wait_sum;
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001125 u64 iowait_count;
1126 u64 iowait_sum;
Ingo Molnar94c18222007-08-02 17:41:40 +02001127
1128 u64 sleep_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001129 u64 sleep_max;
Ingo Molnar94c18222007-08-02 17:41:40 +02001130 s64 sum_sleep_runtime;
1131
1132 u64 block_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001133 u64 block_max;
1134 u64 exec_max;
Ingo Molnareba1ed42007-10-15 17:00:02 +02001135 u64 slice_max;
Ingo Molnarcc367732007-10-15 17:00:18 +02001136
Ingo Molnarcc367732007-10-15 17:00:18 +02001137 u64 nr_migrations_cold;
1138 u64 nr_failed_migrations_affine;
1139 u64 nr_failed_migrations_running;
1140 u64 nr_failed_migrations_hot;
1141 u64 nr_forced_migrations;
Ingo Molnarcc367732007-10-15 17:00:18 +02001142
1143 u64 nr_wakeups;
1144 u64 nr_wakeups_sync;
1145 u64 nr_wakeups_migrate;
1146 u64 nr_wakeups_local;
1147 u64 nr_wakeups_remote;
1148 u64 nr_wakeups_affine;
1149 u64 nr_wakeups_affine_attempts;
1150 u64 nr_wakeups_passive;
1151 u64 nr_wakeups_idle;
Lucas De Marchi41acab82010-03-10 23:37:45 -03001152};
1153#endif
1154
1155struct sched_entity {
1156 struct load_weight load; /* for load-balancing */
1157 struct rb_node run_node;
1158 struct list_head group_node;
1159 unsigned int on_rq;
1160
1161 u64 exec_start;
1162 u64 sum_exec_runtime;
1163 u64 vruntime;
1164 u64 prev_sum_exec_runtime;
1165
Lucas De Marchi41acab82010-03-10 23:37:45 -03001166 u64 nr_migrations;
1167
Lucas De Marchi41acab82010-03-10 23:37:45 -03001168#ifdef CONFIG_SCHEDSTATS
1169 struct sched_statistics statistics;
Ingo Molnar94c18222007-08-02 17:41:40 +02001170#endif
1171
Ingo Molnar20b8a592007-07-09 18:51:58 +02001172#ifdef CONFIG_FAIR_GROUP_SCHED
1173 struct sched_entity *parent;
1174 /* rq on which this entity is (to be) queued: */
1175 struct cfs_rq *cfs_rq;
1176 /* rq "owned" by this entity/group: */
1177 struct cfs_rq *my_q;
1178#endif
1179};
Ingo Molnar70b97a72006-07-03 00:25:42 -07001180
Peter Zijlstrafa717062008-01-25 21:08:27 +01001181struct sched_rt_entity {
1182 struct list_head run_list;
Peter Zijlstra78f2c7d2008-01-25 21:08:27 +01001183 unsigned long timeout;
Richard Kennedybee367e2008-08-01 13:24:08 +01001184 unsigned int time_slice;
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001185 int nr_cpus_allowed;
1186
Peter Zijlstra58d6c2d2008-04-19 19:45:00 +02001187 struct sched_rt_entity *back;
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01001188#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001189 struct sched_rt_entity *parent;
1190 /* rq on which this entity is (to be) queued: */
1191 struct rt_rq *rt_rq;
1192 /* rq "owned" by this entity/group: */
1193 struct rt_rq *my_q;
1194#endif
Peter Zijlstrafa717062008-01-25 21:08:27 +01001195};
1196
Paul E. McKenney86848962009-08-27 15:00:12 -07001197struct rcu_node;
1198
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001199enum perf_event_task_context {
1200 perf_invalid_context = -1,
1201 perf_hw_context = 0,
Peter Zijlstra89a1e182010-09-07 17:34:50 +02001202 perf_sw_context,
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001203 perf_nr_task_contexts,
1204};
1205
Linus Torvalds1da177e2005-04-16 15:20:36 -07001206struct task_struct {
1207 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
Roman Zippelf7e42172007-05-09 02:35:17 -07001208 void *stack;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001209 atomic_t usage;
William Cohen97dc32c2007-05-08 00:23:41 -07001210 unsigned int flags; /* per process flags, defined below */
1211 unsigned int ptrace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001212
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001213 int lock_depth; /* BKL lock depth */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001214
Peter Williams2dd73a42006-06-27 02:54:34 -07001215#ifdef CONFIG_SMP
1216#ifdef __ARCH_WANT_UNLOCKED_CTXSW
Nick Piggin4866cde2005-06-25 14:57:23 -07001217 int oncpu;
1218#endif
Peter Williams2dd73a42006-06-27 02:54:34 -07001219#endif
Ingo Molnar50e645a2007-07-09 18:52:00 +02001220
Ingo Molnarb29739f2006-06-27 02:54:51 -07001221 int prio, static_prio, normal_prio;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001222 unsigned int rt_priority;
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001223 const struct sched_class *sched_class;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001224 struct sched_entity se;
Peter Zijlstrafa717062008-01-25 21:08:27 +01001225 struct sched_rt_entity rt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001226
Avi Kivitye107be32007-07-26 13:40:43 +02001227#ifdef CONFIG_PREEMPT_NOTIFIERS
1228 /* list of struct preempt_notifier: */
1229 struct hlist_head preempt_notifiers;
1230#endif
1231
Alexey Dobriyan18796aa2007-10-16 23:30:26 -07001232 /*
1233 * fpu_counter contains the number of consecutive context switches
1234 * that the FPU is used. If this is over a threshold, the lazy fpu
1235 * saving becomes unlazy to save the trap. This is an unsigned char
1236 * so that after 256 times the counter wraps and the behavior turns
1237 * lazy again; this to deal with bursty apps that only use FPU for
1238 * a short time
1239 */
1240 unsigned char fpu_counter;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001241#ifdef CONFIG_BLK_DEV_IO_TRACE
Jens Axboe2056a782006-03-23 20:00:26 +01001242 unsigned int btrace_seq;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001243#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001244
William Cohen97dc32c2007-05-08 00:23:41 -07001245 unsigned int policy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001246 cpumask_t cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001247
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001248#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneye260be62008-01-25 21:08:24 +01001249 int rcu_read_lock_nesting;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001250 char rcu_read_unlock_special;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001251 struct list_head rcu_node_entry;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001252#endif /* #ifdef CONFIG_PREEMPT_RCU */
1253#ifdef CONFIG_TREE_PREEMPT_RCU
1254 struct rcu_node *rcu_blocked_node;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001255#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
Paul E. McKenney24278d12010-09-27 17:25:23 -07001256#ifdef CONFIG_RCU_BOOST
1257 struct rt_mutex *rcu_boost_mutex;
1258#endif /* #ifdef CONFIG_RCU_BOOST */
Paul E. McKenneye260be62008-01-25 21:08:24 +01001259
Chandra Seetharaman52f17b62006-07-14 00:24:38 -07001260#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001261 struct sched_info sched_info;
1262#endif
1263
1264 struct list_head tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001265#ifdef CONFIG_SMP
Gregory Haskins917b6272008-12-29 09:39:53 -05001266 struct plist_node pushable_tasks;
Dario Faggioli806c09a2010-11-30 19:51:33 +01001267#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001268
1269 struct mm_struct *mm, *active_mm;
KAMEZAWA Hiroyuki34e55232010-03-05 13:41:40 -08001270#if defined(SPLIT_RSS_COUNTING)
1271 struct task_rss_stat rss_stat;
1272#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001273/* task state */
William Cohen97dc32c2007-05-08 00:23:41 -07001274 int exit_state;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001275 int exit_code, exit_signal;
1276 int pdeath_signal; /* The signal sent when the parent dies */
1277 /* ??? */
William Cohen97dc32c2007-05-08 00:23:41 -07001278 unsigned int personality;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001279 unsigned did_exec:1;
Kentaro Takedaf9ce1f12009-02-05 17:18:11 +09001280 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1281 * execve */
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001282 unsigned in_iowait:1;
1283
Lennart Poetteringca94c442009-06-15 17:17:47 +02001284
1285 /* Revert to default priority/policy when forking */
1286 unsigned sched_reset_on_fork:1;
1287
Linus Torvalds1da177e2005-04-16 15:20:36 -07001288 pid_t pid;
1289 pid_t tgid;
Arjan van de Ven0a425402006-09-26 10:52:38 +02001290
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001291#ifdef CONFIG_CC_STACKPROTECTOR
Arjan van de Ven0a425402006-09-26 10:52:38 +02001292 /* Canary value for the -fstack-protector gcc feature */
1293 unsigned long stack_canary;
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001294#endif
Ingo Molnare0032082008-02-14 08:48:23 +01001295
Linus Torvalds1da177e2005-04-16 15:20:36 -07001296 /*
1297 * pointers to (original) parent process, youngest child, younger sibling,
1298 * older sibling, respectively. (p->father can be replaced with
Roland McGrathf4700212008-03-24 18:36:23 -07001299 * p->real_parent->pid)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001300 */
Roland McGrathf4700212008-03-24 18:36:23 -07001301 struct task_struct *real_parent; /* real parent process */
1302 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001303 /*
Roland McGrathf4700212008-03-24 18:36:23 -07001304 * children/sibling forms the list of my natural children
Linus Torvalds1da177e2005-04-16 15:20:36 -07001305 */
1306 struct list_head children; /* list of my children */
1307 struct list_head sibling; /* linkage in my parent's children list */
1308 struct task_struct *group_leader; /* threadgroup leader */
1309
Roland McGrathf4700212008-03-24 18:36:23 -07001310 /*
1311 * ptraced is the list of tasks this task is using ptrace on.
1312 * This includes both natural children and PTRACE_ATTACH targets.
1313 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1314 */
1315 struct list_head ptraced;
1316 struct list_head ptrace_entry;
1317
Linus Torvalds1da177e2005-04-16 15:20:36 -07001318 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -08001319 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -08001320 struct list_head thread_group;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001321
1322 struct completion *vfork_done; /* for vfork() */
1323 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1324 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1325
Michael Neulingc66f08b2007-10-18 03:06:34 -07001326 cputime_t utime, stime, utimescaled, stimescaled;
Laurent Vivier9ac52312007-10-15 17:00:19 +02001327 cputime_t gtime;
Hidetoshi Setod99ca3b2009-12-02 17:26:47 +09001328#ifndef CONFIG_VIRT_CPU_ACCOUNTING
Balbir Singh93018992007-10-30 00:26:32 +01001329 cputime_t prev_utime, prev_stime;
Hidetoshi Setod99ca3b2009-12-02 17:26:47 +09001330#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001331 unsigned long nvcsw, nivcsw; /* context switch counts */
Tomas Janousek924b42d2007-07-15 23:39:42 -07001332 struct timespec start_time; /* monotonic time */
1333 struct timespec real_start_time; /* boot based time */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001334/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1335 unsigned long min_flt, maj_flt;
1336
Frank Mayharf06febc2008-09-12 09:54:39 -07001337 struct task_cputime cputime_expires;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001338 struct list_head cpu_timers[3];
1339
1340/* process credentials */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001341 const struct cred __rcu *real_cred; /* objective and real subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001342 * credentials (COW) */
Arnd Bergmann1b0ba1c2010-02-24 19:45:09 +01001343 const struct cred __rcu *cred; /* effective (overridable) subjective task
David Howells3b11a1d2008-11-14 10:39:26 +11001344 * credentials (COW) */
David Howellsee18d642009-09-02 09:14:21 +01001345 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
David Howellsb6dff3e2008-11-14 10:39:16 +11001346
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001347 char comm[TASK_COMM_LEN]; /* executable name excluding path
1348 - access with [gs]et_task_comm (which lock
1349 it with task_lock())
Linus Torvalds221af7f2010-01-28 22:14:42 -08001350 - initialized normally by setup_new_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001351/* file system info */
1352 int link_count, total_link_count;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001353#ifdef CONFIG_SYSVIPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001354/* ipc stuff */
1355 struct sysv_sem sysvsem;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001356#endif
Mandeep Singh Bainese162b392009-01-15 11:08:40 -08001357#ifdef CONFIG_DETECT_HUNG_TASK
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001358/* hung task detection */
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001359 unsigned long last_switch_count;
1360#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001361/* CPU-specific state of this task */
1362 struct thread_struct thread;
1363/* filesystem information */
1364 struct fs_struct *fs;
1365/* open file information */
1366 struct files_struct *files;
Serge E. Hallyn1651e142006-10-02 02:18:08 -07001367/* namespaces */
Serge E. Hallynab516012006-10-02 02:18:06 -07001368 struct nsproxy *nsproxy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001369/* signal handlers */
1370 struct signal_struct *signal;
1371 struct sighand_struct *sighand;
1372
1373 sigset_t blocked, real_blocked;
Roland McGrathf3de2722008-04-30 00:53:09 -07001374 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001375 struct sigpending pending;
1376
1377 unsigned long sas_ss_sp;
1378 size_t sas_ss_size;
1379 int (*notifier)(void *priv);
1380 void *notifier_data;
1381 sigset_t *notifier_mask;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001382 struct audit_context *audit_context;
Al Virobfef93a2008-01-10 04:53:18 -05001383#ifdef CONFIG_AUDITSYSCALL
1384 uid_t loginuid;
Eric Paris4746ec52008-01-08 10:06:53 -05001385 unsigned int sessionid;
Al Virobfef93a2008-01-10 04:53:18 -05001386#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001387 seccomp_t seccomp;
1388
1389/* Thread group tracking */
1390 u32 parent_exec_id;
1391 u32 self_exec_id;
Miao Xie58568d22009-06-16 15:31:49 -07001392/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1393 * mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001394 spinlock_t alloc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001395
Thomas Gleixner3aa551c2009-03-23 18:28:15 +01001396#ifdef CONFIG_GENERIC_HARDIRQS
1397 /* IRQ handler threads */
1398 struct irqaction *irqaction;
1399#endif
1400
Ingo Molnarb29739f2006-06-27 02:54:51 -07001401 /* Protection of the PI data structures: */
Thomas Gleixner1d615482009-11-17 14:54:03 +01001402 raw_spinlock_t pi_lock;
Ingo Molnarb29739f2006-06-27 02:54:51 -07001403
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001404#ifdef CONFIG_RT_MUTEXES
1405 /* PI waiters blocked on a rt_mutex held by this task */
1406 struct plist_head pi_waiters;
1407 /* Deadlock detection and priority inheritance handling */
1408 struct rt_mutex_waiter *pi_blocked_on;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001409#endif
1410
Ingo Molnar408894e2006-01-09 15:59:20 -08001411#ifdef CONFIG_DEBUG_MUTEXES
1412 /* mutex deadlock detection */
1413 struct mutex_waiter *blocked_on;
1414#endif
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001415#ifdef CONFIG_TRACE_IRQFLAGS
1416 unsigned int irq_events;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001417 unsigned long hardirq_enable_ip;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001418 unsigned long hardirq_disable_ip;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001419 unsigned int hardirq_enable_event;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001420 unsigned int hardirq_disable_event;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001421 int hardirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001422 int hardirq_context;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001423 unsigned long softirq_disable_ip;
1424 unsigned long softirq_enable_ip;
1425 unsigned int softirq_disable_event;
1426 unsigned int softirq_enable_event;
1427 int softirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001428 int softirq_context;
1429#endif
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001430#ifdef CONFIG_LOCKDEP
Peter Zijlstrabdb94412008-02-25 23:02:48 +01001431# define MAX_LOCK_DEPTH 48UL
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001432 u64 curr_chain_key;
1433 int lockdep_depth;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001434 unsigned int lockdep_recursion;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001435 struct held_lock held_locks[MAX_LOCK_DEPTH];
Nick Piggincf40bd12009-01-21 08:12:39 +01001436 gfp_t lockdep_reclaim_gfp;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001437#endif
Ingo Molnar408894e2006-01-09 15:59:20 -08001438
Linus Torvalds1da177e2005-04-16 15:20:36 -07001439/* journalling filesystem info */
1440 void *journal_info;
1441
Neil Brownd89d8792007-05-01 09:53:42 +02001442/* stacked block device info */
Akinobu Mitabddd87c2010-02-23 08:55:42 +01001443 struct bio_list *bio_list;
Neil Brownd89d8792007-05-01 09:53:42 +02001444
Jens Axboe73c10102011-03-08 13:19:51 +01001445#ifdef CONFIG_BLOCK
1446/* stack plugging */
1447 struct blk_plug *plug;
1448#endif
1449
Linus Torvalds1da177e2005-04-16 15:20:36 -07001450/* VM state */
1451 struct reclaim_state *reclaim_state;
1452
Linus Torvalds1da177e2005-04-16 15:20:36 -07001453 struct backing_dev_info *backing_dev_info;
1454
1455 struct io_context *io_context;
1456
1457 unsigned long ptrace_message;
1458 siginfo_t *last_siginfo; /* For ptrace use. */
Andrew Morton7c3ab732006-12-10 02:19:19 -08001459 struct task_io_accounting ioac;
Jay Lan8f0ab512006-09-30 23:28:59 -07001460#if defined(CONFIG_TASK_XACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001461 u64 acct_rss_mem1; /* accumulated rss usage */
1462 u64 acct_vm_mem1; /* accumulated virtual memory usage */
Jonathan Lim49b5cf32008-07-25 01:48:40 -07001463 cputime_t acct_timexpd; /* stime + utime since last update */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001464#endif
1465#ifdef CONFIG_CPUSETS
Miao Xie58568d22009-06-16 15:31:49 -07001466 nodemask_t mems_allowed; /* Protected by alloc_lock */
Miao Xiec0ff7452010-05-24 14:32:08 -07001467 int mems_allowed_change_disable;
Paul Jackson825a46a2006-03-24 03:16:03 -08001468 int cpuset_mem_spread_rotor;
Jack Steiner6adef3e2010-05-26 14:42:49 -07001469 int cpuset_slab_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001470#endif
Paul Menageddbcc7e2007-10-18 23:39:30 -07001471#ifdef CONFIG_CGROUPS
Paul Menage817929e2007-10-18 23:39:36 -07001472 /* Control Group info protected by css_set_lock */
Arnd Bergmann2c392b82010-02-24 19:41:39 +01001473 struct css_set __rcu *cgroups;
Paul Menage817929e2007-10-18 23:39:36 -07001474 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1475 struct list_head cg_list;
Paul Menageddbcc7e2007-10-18 23:39:30 -07001476#endif
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001477#ifdef CONFIG_FUTEX
Ingo Molnar0771dfe2006-03-27 01:16:22 -08001478 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -08001479#ifdef CONFIG_COMPAT
1480 struct compat_robust_list_head __user *compat_robust_list;
1481#endif
Ingo Molnarc87e2832006-06-27 02:54:58 -07001482 struct list_head pi_state_list;
1483 struct futex_pi_state *pi_state_cache;
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001484#endif
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001485#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra8dc85d52010-09-02 16:50:03 +02001486 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001487 struct mutex perf_event_mutex;
1488 struct list_head perf_event_list;
Paul Mackerrasa63eaf32009-05-22 14:17:31 +10001489#endif
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001490#ifdef CONFIG_NUMA
Miao Xie58568d22009-06-16 15:31:49 -07001491 struct mempolicy *mempolicy; /* Protected by alloc_lock */
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001492 short il_next;
Eric Dumazet207205a2011-03-22 16:30:44 -07001493 short pref_node_fork;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001494#endif
Jens Axboe22e2c502005-06-27 10:55:12 +02001495 atomic_t fs_excl; /* holding fs exclusive resources */
Ingo Molnare56d0902006-01-08 01:01:37 -08001496 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +02001497
1498 /*
1499 * cache last used pipe for splice
1500 */
1501 struct pipe_inode_info *splice_pipe;
Shailabh Nagarca74e922006-07-14 00:24:36 -07001502#ifdef CONFIG_TASK_DELAY_ACCT
1503 struct task_delay_info *delays;
1504#endif
Akinobu Mitaf4f154f2006-12-08 02:39:47 -08001505#ifdef CONFIG_FAULT_INJECTION
1506 int make_it_fail;
1507#endif
Peter Zijlstra3e26c142007-10-16 23:25:50 -07001508 struct prop_local_single dirties;
Arjan van de Ven97455122008-01-25 21:08:34 +01001509#ifdef CONFIG_LATENCYTOP
1510 int latency_record_count;
1511 struct latency_record latency_record[LT_SAVECOUNT];
1512#endif
Arjan van de Ven69766752008-09-01 15:52:40 -07001513 /*
1514 * time slack values; these are used to round up poll() and
1515 * select() etc timeout values. These are in nanoseconds.
1516 */
1517 unsigned long timer_slack_ns;
1518 unsigned long default_timer_slack_ns;
David Millerf8d570a2008-11-06 00:37:40 -08001519
1520 struct list_head *scm_work_list;
Frederic Weisbeckerfb526072008-11-25 21:07:04 +01001521#ifdef CONFIG_FUNCTION_GRAPH_TRACER
Daniel Mack3ad2f3f2010-02-03 08:01:28 +08001522 /* Index of current stored address in ret_stack */
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001523 int curr_ret_stack;
1524 /* Stack of return addresses for return function tracing */
1525 struct ftrace_ret_stack *ret_stack;
Steven Rostedt8aef2d22009-03-24 01:10:15 -04001526 /* time stamp for last schedule */
1527 unsigned long long ftrace_timestamp;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001528 /*
1529 * Number of functions that haven't been traced
1530 * because of depth overrun.
1531 */
1532 atomic_t trace_overrun;
Frederic Weisbecker380c4b12008-12-06 03:43:41 +01001533 /* Pause for the tracing */
1534 atomic_t tracing_graph_pause;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001535#endif
Steven Rostedtea4e2bc2008-12-03 15:36:57 -05001536#ifdef CONFIG_TRACING
1537 /* state flags for use by tracers */
1538 unsigned long trace;
Steven Rostedt261842b2009-04-16 21:41:52 -04001539 /* bitmask of trace recursion */
1540 unsigned long trace_recursion;
1541#endif /* CONFIG_TRACING */
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001542#ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1543 struct memcg_batch_info {
1544 int do_batch; /* incremented when batch uncharge started */
1545 struct mem_cgroup *memcg; /* target memcg of uncharge */
Johannes Weiner7ffd4ca2011-03-23 16:42:35 -07001546 unsigned long nr_pages; /* uncharged usage */
1547 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001548 } memcg_batch;
1549#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001550};
1551
Rusty Russell76e6eee2009-03-12 14:35:43 -06001552/* Future-safe accessor for struct task_struct's cpus_allowed. */
Rusty Russella4636812009-12-17 11:43:29 -06001553#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
Rusty Russell76e6eee2009-03-12 14:35:43 -06001554
Ingo Molnare05606d2007-07-09 18:51:59 +02001555/*
1556 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1557 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1558 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1559 * values are inverted: lower p->prio value means higher priority.
1560 *
1561 * The MAX_USER_RT_PRIO value allows the actual maximum
1562 * RT priority to be separate from the value exported to
1563 * user-space. This allows kernel threads to set their
1564 * priority to a value higher than any user task. Note:
1565 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1566 */
1567
1568#define MAX_USER_RT_PRIO 100
1569#define MAX_RT_PRIO MAX_USER_RT_PRIO
1570
1571#define MAX_PRIO (MAX_RT_PRIO + 40)
1572#define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1573
1574static inline int rt_prio(int prio)
1575{
1576 if (unlikely(prio < MAX_RT_PRIO))
1577 return 1;
1578 return 0;
1579}
1580
Alexey Dobriyane8681712007-10-26 12:17:22 +04001581static inline int rt_task(struct task_struct *p)
Ingo Molnare05606d2007-07-09 18:51:59 +02001582{
1583 return rt_prio(p->prio);
1584}
1585
Alexey Dobriyane8681712007-10-26 12:17:22 +04001586static inline struct pid *task_pid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001587{
1588 return task->pids[PIDTYPE_PID].pid;
1589}
1590
Alexey Dobriyane8681712007-10-26 12:17:22 +04001591static inline struct pid *task_tgid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001592{
1593 return task->group_leader->pids[PIDTYPE_PID].pid;
1594}
1595
Oleg Nesterov6dda81f2009-04-02 16:58:35 -07001596/*
1597 * Without tasklist or rcu lock it is not safe to dereference
1598 * the result of task_pgrp/task_session even if task == current,
1599 * we can race with another thread doing sys_setsid/sys_setpgid.
1600 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001601static inline struct pid *task_pgrp(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001602{
1603 return task->group_leader->pids[PIDTYPE_PGID].pid;
1604}
1605
Alexey Dobriyane8681712007-10-26 12:17:22 +04001606static inline struct pid *task_session(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001607{
1608 return task->group_leader->pids[PIDTYPE_SID].pid;
1609}
1610
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001611struct pid_namespace;
1612
1613/*
1614 * the helpers to get the task's different pids as they are seen
1615 * from various namespaces
1616 *
1617 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
Eric W. Biederman44c4e1b2008-02-08 04:19:15 -08001618 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1619 * current.
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001620 * task_xid_nr_ns() : id seen from the ns specified;
1621 *
1622 * set_task_vxid() : assigns a virtual id to a task;
1623 *
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001624 * see also pid_nr() etc in include/linux/pid.h
1625 */
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001626pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1627 struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001628
Alexey Dobriyane8681712007-10-26 12:17:22 +04001629static inline pid_t task_pid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001630{
1631 return tsk->pid;
1632}
1633
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001634static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1635 struct pid_namespace *ns)
1636{
1637 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1638}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001639
1640static inline pid_t task_pid_vnr(struct task_struct *tsk)
1641{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001642 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001643}
1644
1645
Alexey Dobriyane8681712007-10-26 12:17:22 +04001646static inline pid_t task_tgid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001647{
1648 return tsk->tgid;
1649}
1650
Pavel Emelyanov2f2a3a42007-10-18 23:40:19 -07001651pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001652
1653static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1654{
1655 return pid_vnr(task_tgid(tsk));
1656}
1657
1658
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001659static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1660 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001661{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001662 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001663}
1664
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001665static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1666{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001667 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001668}
1669
1670
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001671static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1672 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001673{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001674 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001675}
1676
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001677static inline pid_t task_session_vnr(struct task_struct *tsk)
1678{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001679 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001680}
1681
Oleg Nesterov1b0f7ff2009-04-02 16:58:39 -07001682/* obsolete, do not use */
1683static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1684{
1685 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1686}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001687
Linus Torvalds1da177e2005-04-16 15:20:36 -07001688/**
1689 * pid_alive - check that a task structure is not stale
1690 * @p: Task structure to be checked.
1691 *
1692 * Test if a process is not yet dead (at most zombie state)
1693 * If pid_alive fails, then pointers within the task structure
1694 * can be stale and must not be dereferenced.
1695 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001696static inline int pid_alive(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001697{
Eric W. Biederman92476d72006-03-31 02:31:42 -08001698 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001699}
1700
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001701/**
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001702 * is_global_init - check if a task structure is init
Henne32602592006-10-06 00:44:01 -07001703 * @tsk: Task structure to be checked.
1704 *
1705 * Check if a task structure is the first user space task the kernel created.
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001706 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001707static inline int is_global_init(struct task_struct *tsk)
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07001708{
1709 return tsk->pid == 1;
1710}
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001711
1712/*
1713 * is_container_init:
1714 * check whether in the task is init in its own pid namespace.
1715 */
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07001716extern int is_container_init(struct task_struct *tsk);
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001717
Cedric Le Goater9ec52092006-10-02 02:19:00 -07001718extern struct pid *cad_pid;
1719
Linus Torvalds1da177e2005-04-16 15:20:36 -07001720extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001721#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -08001722
Andrew Morton158d9eb2006-03-31 02:31:34 -08001723extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001724
1725static inline void put_task_struct(struct task_struct *t)
1726{
1727 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -08001728 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001729}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001730
Hidetoshi Setod180c5b2009-11-26 14:48:30 +09001731extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
Hidetoshi Seto0cf55e12009-12-02 17:28:07 +09001732extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
Balbir Singh49048622008-09-05 18:12:23 +02001733
Linus Torvalds1da177e2005-04-16 15:20:36 -07001734/*
1735 * Per process flags
1736 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001737#define PF_STARTING 0x00000002 /* being created */
1738#define PF_EXITING 0x00000004 /* getting shut down */
Alexey Kuznetsov778e9a92007-06-08 13:47:00 -07001739#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
Laurent Vivier94886b82007-10-15 17:00:19 +02001740#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
Tejun Heo21aa9af2010-06-08 21:40:37 +02001741#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001742#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001743#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001744#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1745#define PF_DUMPCORE 0x00000200 /* dumped core */
1746#define PF_SIGNALED 0x00000400 /* killed by a signal */
1747#define PF_MEMALLOC 0x00000800 /* Allocating memory */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001748#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
Thomas Gleixner6301cb92009-07-17 14:15:47 +02001749#define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001750#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1751#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1752#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1753#define PF_KSWAPD 0x00040000 /* I am kswapd */
Hugh Dickins35451be2009-09-21 17:02:27 -07001754#define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001755#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Oleg Nesterov246bb0b2008-07-25 01:47:38 -07001756#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
Jens Axboeb31dc662006-06-13 08:26:10 +02001757#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1758#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1759#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1760#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
David Rientjes9985b0b2008-06-05 12:57:11 -07001761#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001762#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
Paul Jacksonc61afb12006-03-24 03:16:08 -08001763#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
Thomas Gleixner61a87122006-06-27 02:54:56 -07001764#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
Tejun Heo58a69cb2011-02-16 09:25:31 +01001765#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
Rafael J. Wysockiebb12db2008-06-11 22:04:29 +02001766#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001767
1768/*
1769 * Only the _current_ task can read/write to tsk->flags, but other
1770 * tasks can access tsk->flags in readonly mode for example
1771 * with tsk_used_math (like during threaded core dumping).
1772 * There is however an exception to this rule during ptrace
1773 * or during fork: the ptracer task is allowed to write to the
1774 * child->flags of its traced child (same goes for fork, the parent
1775 * can write to the child->flags), because we're guaranteed the
1776 * child is not running and in turn not changing child->flags
1777 * at the same time the parent does it.
1778 */
1779#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1780#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1781#define clear_used_math() clear_stopped_child_used_math(current)
1782#define set_used_math() set_stopped_child_used_math(current)
1783#define conditional_stopped_child_used_math(condition, child) \
1784 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1785#define conditional_used_math(condition) \
1786 conditional_stopped_child_used_math(condition, current)
1787#define copy_to_stopped_child_used_math(child) \
1788 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1789/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1790#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1791#define used_math() tsk_used_math(current)
1792
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001793#ifdef CONFIG_PREEMPT_RCU
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001794
1795#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
Paul E. McKenney24278d12010-09-27 17:25:23 -07001796#define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1797#define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001798
1799static inline void rcu_copy_process(struct task_struct *p)
1800{
1801 p->rcu_read_lock_nesting = 0;
1802 p->rcu_read_unlock_special = 0;
Paul E. McKenneya57eb942010-06-29 16:49:16 -07001803#ifdef CONFIG_TREE_PREEMPT_RCU
Paul E. McKenneydd5d19b2009-08-27 14:58:16 -07001804 p->rcu_blocked_node = NULL;
Paul E. McKenney24278d12010-09-27 17:25:23 -07001805#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1806#ifdef CONFIG_RCU_BOOST
1807 p->rcu_boost_mutex = NULL;
1808#endif /* #ifdef CONFIG_RCU_BOOST */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001809 INIT_LIST_HEAD(&p->rcu_node_entry);
1810}
1811
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001812#else
1813
1814static inline void rcu_copy_process(struct task_struct *p)
1815{
1816}
1817
1818#endif
1819
Linus Torvalds1da177e2005-04-16 15:20:36 -07001820#ifdef CONFIG_SMP
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001821extern int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10301822 const struct cpumask *new_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001823#else
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001824static inline int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10301825 const struct cpumask *new_mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001826{
Rusty Russell96f874e2008-11-25 02:35:14 +10301827 if (!cpumask_test_cpu(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001828 return -EINVAL;
1829 return 0;
1830}
1831#endif
Rusty Russelle0ad9552009-09-24 09:34:38 -06001832
1833#ifndef CONFIG_CPUMASK_OFFSTACK
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001834static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1835{
1836 return set_cpus_allowed_ptr(p, &new_mask);
1837}
Rusty Russelle0ad9552009-09-24 09:34:38 -06001838#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001839
Ingo Molnarb3425012009-02-26 20:20:29 +01001840/*
Peter Zijlstrac6763292010-05-25 10:48:51 +02001841 * Do not use outside of architecture code which knows its limitations.
1842 *
1843 * sched_clock() has no promise of monotonicity or bounded drift between
1844 * CPUs, use (which you should not) requires disabling IRQs.
1845 *
1846 * Please use one of the three interfaces below.
Ingo Molnarb3425012009-02-26 20:20:29 +01001847 */
Mike Frysinger1bbfa6f2009-12-09 20:07:03 -05001848extern unsigned long long notrace sched_clock(void);
Peter Zijlstrac6763292010-05-25 10:48:51 +02001849/*
1850 * See the comment in kernel/sched_clock.c
1851 */
1852extern u64 cpu_clock(int cpu);
1853extern u64 local_clock(void);
1854extern u64 sched_clock_cpu(int cpu);
1855
Ingo Molnare436d802007-07-19 21:28:35 +02001856
Peter Zijlstrac1955a32008-08-11 08:59:03 +02001857extern void sched_clock_init(void);
Peter Zijlstrac1955a32008-08-11 08:59:03 +02001858
Peter Zijlstra3e51f332008-05-03 18:29:28 +02001859#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
Peter Zijlstra3e51f332008-05-03 18:29:28 +02001860static inline void sched_clock_tick(void)
1861{
1862}
1863
1864static inline void sched_clock_idle_sleep_event(void)
1865{
1866}
1867
1868static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1869{
1870}
1871#else
Peter Zijlstrac6763292010-05-25 10:48:51 +02001872/*
1873 * Architectures can set this to 1 if they have specified
1874 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1875 * but then during bootup it turns out that sched_clock()
1876 * is reliable after all:
1877 */
1878extern int sched_clock_stable;
1879
Peter Zijlstra3e51f332008-05-03 18:29:28 +02001880extern void sched_clock_tick(void);
1881extern void sched_clock_idle_sleep_event(void);
1882extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1883#endif
1884
Venkatesh Pallipadib52bfee2010-10-04 17:03:19 -07001885#ifdef CONFIG_IRQ_TIME_ACCOUNTING
1886/*
1887 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1888 * The reason for this explicit opt-in is not to have perf penalty with
1889 * slow sched_clocks.
1890 */
1891extern void enable_sched_clock_irqtime(void);
1892extern void disable_sched_clock_irqtime(void);
1893#else
1894static inline void enable_sched_clock_irqtime(void) {}
1895static inline void disable_sched_clock_irqtime(void) {}
1896#endif
1897
Ingo Molnar36c8b582006-07-03 00:25:41 -07001898extern unsigned long long
Ingo Molnar41b86e92007-07-09 18:51:58 +02001899task_sched_runtime(struct task_struct *task);
Frank Mayharf06febc2008-09-12 09:54:39 -07001900extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001901
1902/* sched_exec is called by processes performing an exec */
1903#ifdef CONFIG_SMP
1904extern void sched_exec(void);
1905#else
1906#define sched_exec() {}
1907#endif
1908
Ingo Molnar2aa44d02007-08-23 15:18:02 +02001909extern void sched_clock_idle_sleep_event(void);
1910extern void sched_clock_idle_wakeup_event(u64 delta_ns);
Ingo Molnarbb29ab22007-07-09 18:51:59 +02001911
Linus Torvalds1da177e2005-04-16 15:20:36 -07001912#ifdef CONFIG_HOTPLUG_CPU
1913extern void idle_task_exit(void);
1914#else
1915static inline void idle_task_exit(void) {}
1916#endif
1917
Thomas Gleixner06d83082008-03-22 09:20:24 +01001918#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1919extern void wake_up_idle_cpu(int cpu);
1920#else
1921static inline void wake_up_idle_cpu(int cpu) { }
1922#endif
1923
Peter Zijlstra21805082007-08-25 18:41:53 +02001924extern unsigned int sysctl_sched_latency;
Peter Zijlstrab2be5e92007-11-09 22:39:37 +01001925extern unsigned int sysctl_sched_min_granularity;
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02001926extern unsigned int sysctl_sched_wakeup_granularity;
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02001927extern unsigned int sysctl_sched_child_runs_first;
Christian Ehrhardt1983a922009-11-30 12:16:47 +01001928
1929enum sched_tunable_scaling {
1930 SCHED_TUNABLESCALING_NONE,
1931 SCHED_TUNABLESCALING_LOG,
1932 SCHED_TUNABLESCALING_LINEAR,
1933 SCHED_TUNABLESCALING_END,
1934};
1935extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1936
Mike Galbraith2bba22c2009-09-09 15:41:37 +02001937#ifdef CONFIG_SCHED_DEBUG
Ingo Molnarda84d962007-10-15 17:00:18 +02001938extern unsigned int sysctl_sched_migration_cost;
Peter Zijlstrab82d9fd2007-11-09 22:39:39 +01001939extern unsigned int sysctl_sched_nr_migrate;
Peter Zijlstrae9e92502009-09-01 10:34:37 +02001940extern unsigned int sysctl_sched_time_avg;
Arun R Bharadwajcd1bb942009-04-16 12:15:34 +05301941extern unsigned int sysctl_timer_migration;
Paul Turnera7a4f8a2010-11-15 15:47:06 -08001942extern unsigned int sysctl_sched_shares_window;
Peter Zijlstrab2be5e92007-11-09 22:39:37 +01001943
Christian Ehrhardt1983a922009-11-30 12:16:47 +01001944int sched_proc_update_handler(struct ctl_table *table, int write,
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07001945 void __user *buffer, size_t *length,
Peter Zijlstrab2be5e92007-11-09 22:39:37 +01001946 loff_t *ppos);
Ingo Molnar2bd8e6d2007-10-15 17:00:02 +02001947#endif
Arun R Bharadwajeea08f32009-04-16 12:16:41 +05301948#ifdef CONFIG_SCHED_DEBUG
1949static inline unsigned int get_sysctl_timer_migration(void)
1950{
1951 return sysctl_timer_migration;
1952}
1953#else
1954static inline unsigned int get_sysctl_timer_migration(void)
1955{
1956 return 1;
1957}
1958#endif
Peter Zijlstra9f0c1e52008-02-13 15:45:39 +01001959extern unsigned int sysctl_sched_rt_period;
1960extern int sysctl_sched_rt_runtime;
Ingo Molnar2bd8e6d2007-10-15 17:00:02 +02001961
Peter Zijlstrad0b27fa2008-04-19 19:44:57 +02001962int sched_rt_handler(struct ctl_table *table, int write,
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07001963 void __user *buffer, size_t *lenp,
Peter Zijlstrad0b27fa2008-04-19 19:44:57 +02001964 loff_t *ppos);
1965
Mike Galbraith5091faa2010-11-30 14:18:03 +01001966#ifdef CONFIG_SCHED_AUTOGROUP
1967extern unsigned int sysctl_sched_autogroup_enabled;
1968
1969extern void sched_autogroup_create_attach(struct task_struct *p);
1970extern void sched_autogroup_detach(struct task_struct *p);
1971extern void sched_autogroup_fork(struct signal_struct *sig);
1972extern void sched_autogroup_exit(struct signal_struct *sig);
1973#ifdef CONFIG_PROC_FS
1974extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1975extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1976#endif
1977#else
1978static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1979static inline void sched_autogroup_detach(struct task_struct *p) { }
1980static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1981static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1982#endif
1983
Ingo Molnarb29739f2006-06-27 02:54:51 -07001984#ifdef CONFIG_RT_MUTEXES
Ingo Molnar36c8b582006-07-03 00:25:41 -07001985extern int rt_mutex_getprio(struct task_struct *p);
1986extern void rt_mutex_setprio(struct task_struct *p, int prio);
1987extern void rt_mutex_adjust_pi(struct task_struct *p);
Ingo Molnarb29739f2006-06-27 02:54:51 -07001988#else
Alexey Dobriyane8681712007-10-26 12:17:22 +04001989static inline int rt_mutex_getprio(struct task_struct *p)
Ingo Molnarb29739f2006-06-27 02:54:51 -07001990{
1991 return p->normal_prio;
1992}
Thomas Gleixner95e02ca2006-06-27 02:55:02 -07001993# define rt_mutex_adjust_pi(p) do { } while (0)
Ingo Molnarb29739f2006-06-27 02:54:51 -07001994#endif
1995
Mike Galbraithd95f4122011-02-01 09:50:51 -05001996extern bool yield_to(struct task_struct *p, bool preempt);
Ingo Molnar36c8b582006-07-03 00:25:41 -07001997extern void set_user_nice(struct task_struct *p, long nice);
1998extern int task_prio(const struct task_struct *p);
1999extern int task_nice(const struct task_struct *p);
2000extern int can_nice(const struct task_struct *p, const int nice);
2001extern int task_curr(const struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002002extern int idle_cpu(int cpu);
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002003extern int sched_setscheduler(struct task_struct *, int,
2004 const struct sched_param *);
Rusty Russell961ccdd2008-06-23 13:55:38 +10002005extern int sched_setscheduler_nocheck(struct task_struct *, int,
KOSAKI Motohirofe7de492010-10-20 16:01:12 -07002006 const struct sched_param *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002007extern struct task_struct *idle_task(int cpu);
2008extern struct task_struct *curr_task(int cpu);
2009extern void set_curr_task(int cpu, struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002010
2011void yield(void);
2012
2013/*
2014 * The default (Linux) execution domain.
2015 */
2016extern struct exec_domain default_exec_domain;
2017
2018union thread_union {
2019 struct thread_info thread_info;
2020 unsigned long stack[THREAD_SIZE/sizeof(long)];
2021};
2022
2023#ifndef __HAVE_ARCH_KSTACK_END
2024static inline int kstack_end(void *addr)
2025{
2026 /* Reliable end of stack detection:
2027 * Some APM bios versions misalign the stack
2028 */
2029 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2030}
2031#endif
2032
2033extern union thread_union init_thread_union;
2034extern struct task_struct init_task;
2035
2036extern struct mm_struct init_mm;
2037
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002038extern struct pid_namespace init_pid_ns;
2039
2040/*
2041 * find a task by one of its numerical ids
2042 *
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002043 * find_task_by_pid_ns():
2044 * finds a task by its pid in the specified namespace
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002045 * find_task_by_vpid():
2046 * finds a task by its virtual pid
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002047 *
Pavel Emelyanove49859e2008-07-25 01:48:36 -07002048 * see also find_vpid() etc in include/linux/pid.h
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002049 */
2050
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002051extern struct task_struct *find_task_by_vpid(pid_t nr);
2052extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2053 struct pid_namespace *ns);
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002054
Oleg Nesterov8520d7c2008-02-08 04:19:09 -08002055extern void __set_special_pids(struct pid *pid);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002056
2057/* per-UID process charging. */
Cedric Le Goateracce2922007-07-15 23:40:59 -07002058extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002059static inline struct user_struct *get_uid(struct user_struct *u)
2060{
2061 atomic_inc(&u->__count);
2062 return u;
2063}
2064extern void free_uid(struct user_struct *);
Pavel Emelyanov28f300d2007-09-18 22:46:45 -07002065extern void release_uids(struct user_namespace *ns);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002066
2067#include <asm/current.h>
2068
Torben Hohnf0af911a92011-01-27 15:59:10 +01002069extern void xtime_update(unsigned long ticks);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002070
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002071extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2072extern int wake_up_process(struct task_struct *tsk);
2073extern void wake_up_new_task(struct task_struct *tsk,
2074 unsigned long clone_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002075#ifdef CONFIG_SMP
2076 extern void kick_process(struct task_struct *tsk);
2077#else
2078 static inline void kick_process(struct task_struct *tsk) { }
2079#endif
Ingo Molnarad46c2c2007-07-09 18:52:00 +02002080extern void sched_fork(struct task_struct *p, int clone_flags);
2081extern void sched_dead(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002082
Linus Torvalds1da177e2005-04-16 15:20:36 -07002083extern void proc_caches_init(void);
2084extern void flush_signals(struct task_struct *);
David Howells3bcac022009-04-29 13:45:05 +01002085extern void __flush_signals(struct task_struct *);
Oleg Nesterov10ab8252007-05-09 02:34:37 -07002086extern void ignore_signals(struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002087extern void flush_signal_handlers(struct task_struct *, int force_default);
2088extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2089
2090static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2091{
2092 unsigned long flags;
2093 int ret;
2094
2095 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2096 ret = dequeue_signal(tsk, mask, info);
2097 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2098
2099 return ret;
2100}
2101
2102extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2103 sigset_t *mask);
2104extern void unblock_all_signals(void);
2105extern void release_task(struct task_struct * p);
2106extern int send_sig_info(int, struct siginfo *, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002107extern int force_sigsegv(int, struct task_struct *);
2108extern int force_sig_info(int, struct siginfo *, struct task_struct *);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002109extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002110extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
Eric W. Biederman2425c082006-10-02 02:17:28 -07002111extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002112extern int kill_pgrp(struct pid *pid, int sig, int priv);
2113extern int kill_pid(struct pid *pid, int sig, int priv);
Matthew Wilcoxc3de4b32007-02-09 08:11:47 -07002114extern int kill_proc_info(int, struct siginfo *, pid_t);
Roland McGrath2b2a1ff2008-07-25 19:45:54 -07002115extern int do_notify_parent(struct task_struct *, int);
Oleg Nesterova7f07652009-09-23 15:56:44 -07002116extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002117extern void force_sig(int, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002118extern int send_sig(int, struct task_struct *, int);
Oleg Nesterov09faef12010-05-26 14:43:11 -07002119extern int zap_other_threads(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002120extern struct sigqueue *sigqueue_alloc(void);
2121extern void sigqueue_free(struct sigqueue *);
Oleg Nesterovac5c2152008-04-30 00:52:57 -07002122extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03002123extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002124extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2125
Cedric Le Goater9ec52092006-10-02 02:19:00 -07002126static inline int kill_cad_pid(int sig, int priv)
2127{
2128 return kill_pid(cad_pid, sig, priv);
2129}
2130
Linus Torvalds1da177e2005-04-16 15:20:36 -07002131/* These can be the second arg to send_sig_info/send_group_sig_info. */
2132#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2133#define SEND_SIG_PRIV ((struct siginfo *) 1)
2134#define SEND_SIG_FORCED ((struct siginfo *) 2)
2135
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002136/*
2137 * True if we are on the alternate signal stack.
2138 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002139static inline int on_sig_stack(unsigned long sp)
2140{
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002141#ifdef CONFIG_STACK_GROWSUP
2142 return sp >= current->sas_ss_sp &&
2143 sp - current->sas_ss_sp < current->sas_ss_size;
2144#else
2145 return sp > current->sas_ss_sp &&
2146 sp - current->sas_ss_sp <= current->sas_ss_size;
2147#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002148}
2149
2150static inline int sas_ss_flags(unsigned long sp)
2151{
2152 return (current->sas_ss_size == 0 ? SS_DISABLE
2153 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2154}
2155
Linus Torvalds1da177e2005-04-16 15:20:36 -07002156/*
2157 * Routines for handling mm_structs
2158 */
2159extern struct mm_struct * mm_alloc(void);
2160
2161/* mmdrop drops the mm and the page tables */
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002162extern void __mmdrop(struct mm_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002163static inline void mmdrop(struct mm_struct * mm)
2164{
Ingo Molnar6fb43d72007-07-09 18:52:01 +02002165 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002166 __mmdrop(mm);
2167}
2168
2169/* mmput gets rid of the mappings and all user-space */
2170extern void mmput(struct mm_struct *);
2171/* Grab a reference to a task's mm, if it is not already going away */
2172extern struct mm_struct *get_task_mm(struct task_struct *task);
2173/* Remove the current tasks stale references to the old mm_struct */
2174extern void mm_release(struct task_struct *, struct mm_struct *);
Carsten Otte402b0862008-03-25 18:47:10 +01002175/* Allocate a new mm structure and copy contents from tsk->mm */
2176extern struct mm_struct *dup_mm(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002177
Alexey Dobriyan6f2c55b2009-04-02 16:56:59 -07002178extern int copy_thread(unsigned long, unsigned long, unsigned long,
2179 struct task_struct *, struct pt_regs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002180extern void flush_thread(void);
2181extern void exit_thread(void);
2182
Linus Torvalds1da177e2005-04-16 15:20:36 -07002183extern void exit_files(struct task_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08002184extern void __cleanup_sighand(struct sighand_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002185
Linus Torvalds1da177e2005-04-16 15:20:36 -07002186extern void exit_itimers(struct signal_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002187extern void flush_itimer_signals(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002188
2189extern NORET_TYPE void do_group_exit(int);
2190
Linus Torvalds1da177e2005-04-16 15:20:36 -07002191extern void daemonize(const char *, ...);
2192extern int allow_signal(int);
2193extern int disallow_signal(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002194
David Howellsd7627462010-08-17 23:52:56 +01002195extern int do_execve(const char *,
2196 const char __user * const __user *,
2197 const char __user * const __user *, struct pt_regs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002198extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002199struct task_struct *fork_idle(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002200
2201extern void set_task_comm(struct task_struct *tsk, char *from);
Andrew Morton59714d62008-02-04 22:27:21 -08002202extern char *get_task_comm(char *to, struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002203
2204#ifdef CONFIG_SMP
Roland McGrath85ba2d82008-07-25 19:45:58 -07002205extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002206#else
Roland McGrath85ba2d82008-07-25 19:45:58 -07002207static inline unsigned long wait_task_inactive(struct task_struct *p,
2208 long match_state)
2209{
2210 return 1;
2211}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002212#endif
2213
Jiri Pirko05725f72009-04-14 20:17:16 +02002214#define next_task(p) \
2215 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002216
2217#define for_each_process(p) \
2218 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2219
Oleg Nesterov5bb459b2009-07-10 03:48:23 +02002220extern bool current_is_single_threaded(void);
David Howellsd84f4f92008-11-14 10:39:23 +11002221
Linus Torvalds1da177e2005-04-16 15:20:36 -07002222/*
2223 * Careful: do_each_thread/while_each_thread is a double loop so
2224 * 'break' will not work as expected - use goto instead.
2225 */
2226#define do_each_thread(g, t) \
2227 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2228
2229#define while_each_thread(g, t) \
2230 while ((t = next_thread(t)) != g)
2231
Oleg Nesterov7e498272010-05-26 14:43:22 -07002232static inline int get_nr_threads(struct task_struct *tsk)
2233{
Oleg Nesterovb3ac0222010-05-26 14:43:24 -07002234 return tsk->signal->nr_threads;
Oleg Nesterov7e498272010-05-26 14:43:22 -07002235}
2236
Eric W. Biedermande12a782006-04-10 17:16:49 -06002237/* de_thread depends on thread_group_leader not being a pid based check */
2238#define thread_group_leader(p) (p == p->group_leader)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002239
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002240/* Do to the insanities of de_thread it is possible for a process
2241 * to have the pid of the thread group leader without actually being
2242 * the thread group leader. For iteration through the pids in proc
2243 * all we care about is that we have a task with the appropriate
2244 * pid, we don't actually care if we have the right task.
2245 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04002246static inline int has_group_leader_pid(struct task_struct *p)
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002247{
2248 return p->pid == p->tgid;
2249}
2250
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002251static inline
2252int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2253{
2254 return p1->tgid == p2->tgid;
2255}
2256
Ingo Molnar36c8b582006-07-03 00:25:41 -07002257static inline struct task_struct *next_thread(const struct task_struct *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08002258{
Jiri Pirko05725f72009-04-14 20:17:16 +02002259 return list_entry_rcu(p->thread_group.next,
2260 struct task_struct, thread_group);
Oleg Nesterov47e65322006-03-28 16:11:25 -08002261}
2262
Alexey Dobriyane8681712007-10-26 12:17:22 +04002263static inline int thread_group_empty(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002264{
Oleg Nesterov47e65322006-03-28 16:11:25 -08002265 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002266}
2267
2268#define delay_group_leader(p) \
2269 (thread_group_leader(p) && !thread_group_empty(p))
2270
Oleg Nesterov39c626a2009-04-02 16:58:18 -07002271static inline int task_detached(struct task_struct *p)
2272{
2273 return p->exit_signal == -1;
2274}
2275
Linus Torvalds1da177e2005-04-16 15:20:36 -07002276/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07002277 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02002278 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Menageddbcc7e2007-10-18 23:39:30 -07002279 * pins the final release of task.io_context. Also protects ->cpuset and
2280 * ->cgroup.subsys[].
Linus Torvalds1da177e2005-04-16 15:20:36 -07002281 *
2282 * Nests both inside and outside of read_lock(&tasklist_lock).
2283 * It must not be nested with write_lock_irq(&tasklist_lock),
2284 * neither inside nor outside.
2285 */
2286static inline void task_lock(struct task_struct *p)
2287{
2288 spin_lock(&p->alloc_lock);
2289}
2290
2291static inline void task_unlock(struct task_struct *p)
2292{
2293 spin_unlock(&p->alloc_lock);
2294}
2295
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002296extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002297 unsigned long *flags);
2298
Namhyung Kimb8ed3742010-10-27 15:34:06 -07002299#define lock_task_sighand(tsk, flags) \
2300({ struct sighand_struct *__ss; \
2301 __cond_lock(&(tsk)->sighand->siglock, \
2302 (__ss = __lock_task_sighand(tsk, flags))); \
2303 __ss; \
2304}) \
2305
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002306static inline void unlock_task_sighand(struct task_struct *tsk,
2307 unsigned long *flags)
2308{
2309 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2310}
2311
Al Virof0373602005-11-13 16:06:57 -08002312#ifndef __HAVE_THREAD_FUNCTIONS
2313
Roman Zippelf7e42172007-05-09 02:35:17 -07002314#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2315#define task_stack_page(task) ((task)->stack)
Al Viroa1261f52005-11-13 16:06:55 -08002316
Al Viro10ebffd2005-11-13 16:06:56 -08002317static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2318{
2319 *task_thread_info(p) = *task_thread_info(org);
2320 task_thread_info(p)->task = p;
2321}
2322
2323static inline unsigned long *end_of_stack(struct task_struct *p)
2324{
Roman Zippelf7e42172007-05-09 02:35:17 -07002325 return (unsigned long *)(task_thread_info(p) + 1);
Al Viro10ebffd2005-11-13 16:06:56 -08002326}
2327
Al Virof0373602005-11-13 16:06:57 -08002328#endif
2329
FUJITA Tomonori8b05c7e2008-07-23 21:26:53 -07002330static inline int object_is_on_stack(void *obj)
2331{
2332 void *stack = task_stack_page(current);
2333
2334 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2335}
2336
Benjamin Herrenschmidt8c9843e2008-04-18 16:56:15 +10002337extern void thread_info_cache_init(void);
2338
Eric Sandeen7c9f8862008-04-22 16:38:23 -05002339#ifdef CONFIG_DEBUG_STACK_USAGE
2340static inline unsigned long stack_not_used(struct task_struct *p)
2341{
2342 unsigned long *n = end_of_stack(p);
2343
2344 do { /* Skip over canary */
2345 n++;
2346 } while (!*n);
2347
2348 return (unsigned long)n - (unsigned long)end_of_stack(p);
2349}
2350#endif
2351
Linus Torvalds1da177e2005-04-16 15:20:36 -07002352/* set thread flags in other task's structures
2353 * - see asm/thread_info.h for TIF_xxxx flags available
2354 */
2355static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2356{
Al Viroa1261f52005-11-13 16:06:55 -08002357 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002358}
2359
2360static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2361{
Al Viroa1261f52005-11-13 16:06:55 -08002362 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002363}
2364
2365static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2366{
Al Viroa1261f52005-11-13 16:06:55 -08002367 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002368}
2369
2370static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2371{
Al Viroa1261f52005-11-13 16:06:55 -08002372 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002373}
2374
2375static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2376{
Al Viroa1261f52005-11-13 16:06:55 -08002377 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002378}
2379
2380static inline void set_tsk_need_resched(struct task_struct *tsk)
2381{
2382 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2383}
2384
2385static inline void clear_tsk_need_resched(struct task_struct *tsk)
2386{
2387 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2388}
2389
Gregory Haskins8ae121a2008-04-23 07:13:29 -04002390static inline int test_tsk_need_resched(struct task_struct *tsk)
2391{
2392 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2393}
2394
Eric W. Biederman690cc3f2009-05-13 16:55:10 +00002395static inline int restart_syscall(void)
2396{
2397 set_tsk_thread_flag(current, TIF_SIGPENDING);
2398 return -ERESTARTNOINTR;
2399}
2400
Linus Torvalds1da177e2005-04-16 15:20:36 -07002401static inline int signal_pending(struct task_struct *p)
2402{
2403 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2404}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002405
Roland McGrathd9588722009-09-23 15:57:04 -07002406static inline int __fatal_signal_pending(struct task_struct *p)
2407{
2408 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2409}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002410
2411static inline int fatal_signal_pending(struct task_struct *p)
2412{
2413 return signal_pending(p) && __fatal_signal_pending(p);
2414}
2415
Oleg Nesterov16882c12008-06-08 21:20:41 +04002416static inline int signal_pending_state(long state, struct task_struct *p)
2417{
2418 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2419 return 0;
2420 if (!signal_pending(p))
2421 return 0;
2422
Oleg Nesterov16882c12008-06-08 21:20:41 +04002423 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2424}
2425
Linus Torvalds1da177e2005-04-16 15:20:36 -07002426static inline int need_resched(void)
2427{
Linus Torvalds9404ef02008-05-12 10:14:22 -07002428 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
Linus Torvalds1da177e2005-04-16 15:20:36 -07002429}
2430
2431/*
2432 * cond_resched() and cond_resched_lock(): latency reduction via
2433 * explicit rescheduling in places that are safe. The return
2434 * value indicates whether a reschedule was done in fact.
2435 * cond_resched_lock() will drop the spinlock before scheduling,
2436 * cond_resched_softirq() will enable bhs before scheduling.
2437 */
Linus Torvaldsc3921ab2008-05-11 16:04:48 -07002438extern int _cond_resched(void);
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002439
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002440#define cond_resched() ({ \
2441 __might_sleep(__FILE__, __LINE__, 0); \
2442 _cond_resched(); \
2443})
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002444
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002445extern int __cond_resched_lock(spinlock_t *lock);
2446
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002447#ifdef CONFIG_PREEMPT
2448#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
Herbert Xu02b67cc32008-01-25 21:08:28 +01002449#else
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002450#define PREEMPT_LOCK_OFFSET 0
Herbert Xu02b67cc32008-01-25 21:08:28 +01002451#endif
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002452
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002453#define cond_resched_lock(lock) ({ \
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002454 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002455 __cond_resched_lock(lock); \
2456})
2457
2458extern int __cond_resched_softirq(void);
2459
Venkatesh Pallipadi75e10562010-10-04 17:03:16 -07002460#define cond_resched_softirq() ({ \
2461 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2462 __cond_resched_softirq(); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002463})
Linus Torvalds1da177e2005-04-16 15:20:36 -07002464
2465/*
2466 * Does a critical section need to be broken due to another
Nick Piggin95c354f2008-01-30 13:31:20 +01002467 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2468 * but a general need for low latency)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002469 */
Nick Piggin95c354f2008-01-30 13:31:20 +01002470static inline int spin_needbreak(spinlock_t *lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002471{
Nick Piggin95c354f2008-01-30 13:31:20 +01002472#ifdef CONFIG_PREEMPT
2473 return spin_is_contended(lock);
2474#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07002475 return 0;
Nick Piggin95c354f2008-01-30 13:31:20 +01002476#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002477}
2478
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002479/*
Frank Mayharf06febc2008-09-12 09:54:39 -07002480 * Thread group CPU time accounting.
2481 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +01002482void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
Peter Zijlstra4da94d492009-02-11 11:30:27 +01002483void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
Frank Mayharf06febc2008-09-12 09:54:39 -07002484
2485static inline void thread_group_cputime_init(struct signal_struct *sig)
2486{
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +01002487 spin_lock_init(&sig->cputimer.lock);
Frank Mayharf06febc2008-09-12 09:54:39 -07002488}
2489
Frank Mayharf06febc2008-09-12 09:54:39 -07002490/*
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002491 * Reevaluate whether the task has signals pending delivery.
2492 * Wake the task if so.
2493 * This is required every time the blocked sigset_t changes.
2494 * callers must hold sighand->siglock.
2495 */
2496extern void recalc_sigpending_and_wake(struct task_struct *t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002497extern void recalc_sigpending(void);
2498
2499extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2500
2501/*
2502 * Wrappers for p->thread_info->cpu access. No-op on UP.
2503 */
2504#ifdef CONFIG_SMP
2505
2506static inline unsigned int task_cpu(const struct task_struct *p)
2507{
Al Viroa1261f52005-11-13 16:06:55 -08002508 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002509}
2510
Ingo Molnarc65cc872007-07-09 18:51:58 +02002511extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002512
2513#else
2514
2515static inline unsigned int task_cpu(const struct task_struct *p)
2516{
2517 return 0;
2518}
2519
2520static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2521{
2522}
2523
2524#endif /* CONFIG_SMP */
2525
Rusty Russell96f874e2008-11-25 02:35:14 +10302526extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2527extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -07002528
Linus Torvalds1da177e2005-04-16 15:20:36 -07002529extern void normalize_rt_tasks(void);
2530
Dhaval Giani7c941432010-01-20 13:26:18 +01002531#ifdef CONFIG_CGROUP_SCHED
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002532
Yong Zhang07e06b02011-01-07 15:17:36 +08002533extern struct task_group root_task_group;
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002534
Dhaval Gianiec7dc8a2008-04-19 19:44:59 +02002535extern struct task_group *sched_create_group(struct task_group *parent);
Ingo Molnar4cf86d72007-10-15 17:00:14 +02002536extern void sched_destroy_group(struct task_group *tg);
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002537extern void sched_move_task(struct task_struct *tsk);
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01002538#ifdef CONFIG_FAIR_GROUP_SCHED
Ingo Molnar4cf86d72007-10-15 17:00:14 +02002539extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
Dhaval Giani5cb350b2007-10-15 17:00:14 +02002540extern unsigned long sched_group_shares(struct task_group *tg);
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01002541#endif
2542#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra9f0c1e52008-02-13 15:45:39 +01002543extern int sched_group_set_rt_runtime(struct task_group *tg,
2544 long rt_runtime_us);
2545extern long sched_group_rt_runtime(struct task_group *tg);
Peter Zijlstrad0b27fa2008-04-19 19:44:57 +02002546extern int sched_group_set_rt_period(struct task_group *tg,
2547 long rt_period_us);
2548extern long sched_group_rt_period(struct task_group *tg);
Dhaval Giani54e99122009-02-27 15:13:54 +05302549extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01002550#endif
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002551#endif
2552
Dhaval Giani54e99122009-02-27 15:13:54 +05302553extern int task_can_switch_user(struct user_struct *up,
2554 struct task_struct *tsk);
2555
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002556#ifdef CONFIG_TASK_XACCT
2557static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2558{
Andrea Righi940389b2008-07-28 00:48:12 +02002559 tsk->ioac.rchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002560}
2561
2562static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2563{
Andrea Righi940389b2008-07-28 00:48:12 +02002564 tsk->ioac.wchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002565}
2566
2567static inline void inc_syscr(struct task_struct *tsk)
2568{
Andrea Righi940389b2008-07-28 00:48:12 +02002569 tsk->ioac.syscr++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002570}
2571
2572static inline void inc_syscw(struct task_struct *tsk)
2573{
Andrea Righi940389b2008-07-28 00:48:12 +02002574 tsk->ioac.syscw++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002575}
2576#else
2577static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2578{
2579}
2580
2581static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2582{
2583}
2584
2585static inline void inc_syscr(struct task_struct *tsk)
2586{
2587}
2588
2589static inline void inc_syscw(struct task_struct *tsk)
2590{
2591}
2592#endif
2593
Dave Hansen82455252008-02-04 22:28:59 -08002594#ifndef TASK_SIZE_OF
2595#define TASK_SIZE_OF(tsk) TASK_SIZE
2596#endif
2597
Balbir Singhcf475ad2008-04-29 01:00:16 -07002598#ifdef CONFIG_MM_OWNER
2599extern void mm_update_next_owner(struct mm_struct *mm);
2600extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2601#else
2602static inline void mm_update_next_owner(struct mm_struct *mm)
2603{
2604}
2605
2606static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2607{
2608}
2609#endif /* CONFIG_MM_OWNER */
2610
Jiri Slaby3e10e712009-11-19 17:16:37 +01002611static inline unsigned long task_rlimit(const struct task_struct *tsk,
2612 unsigned int limit)
2613{
2614 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2615}
2616
2617static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2618 unsigned int limit)
2619{
2620 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2621}
2622
2623static inline unsigned long rlimit(unsigned int limit)
2624{
2625 return task_rlimit(current, limit);
2626}
2627
2628static inline unsigned long rlimit_max(unsigned int limit)
2629{
2630 return task_rlimit_max(current, limit);
2631}
2632
Linus Torvalds1da177e2005-04-16 15:20:36 -07002633#endif /* __KERNEL__ */
2634
2635#endif