blob: 6f7bba93929bf5c55434b6fb196af320087e9c11 [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 */
24#define CLONE_STOPPED 0x02000000 /* Start in stopped state */
Serge E. Hallyn071df102006-10-02 02:18:17 -070025#define CLONE_NEWUTS 0x04000000 /* New utsname group? */
Kirill Korotaev25b21cb2006-10-02 02:18:19 -070026#define CLONE_NEWIPC 0x08000000 /* New ipcs */
Serge E. Hallyn77ec7392007-07-15 23:41:01 -070027#define CLONE_NEWUSER 0x10000000 /* New user namespace */
Pavel Emelyanov30e49c22007-10-18 23:40:10 -070028#define CLONE_NEWPID 0x20000000 /* New pid namespace */
Eric W. Biederman169e3672007-09-27 17:10:06 -070029#define CLONE_NEWNET 0x40000000 /* New network namespace */
Jens Axboefadad8782008-01-24 08:54:47 +010030#define CLONE_IO 0x80000000 /* Clone io context */
David Woodhouseb7b3c762006-04-27 00:12:56 +010031
32/*
33 * Scheduling policies
34 */
35#define SCHED_NORMAL 0
36#define SCHED_FIFO 1
37#define SCHED_RR 2
38#define SCHED_BATCH 3
Ingo Molnar0e6aca42007-07-09 18:51:57 +020039/* SCHED_ISO: reserved but not implemented yet */
40#define SCHED_IDLE 5
Lennart Poetteringca94c442009-06-15 17:17:47 +020041/* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
42#define SCHED_RESET_ON_FORK 0x40000000
David Woodhouseb7b3c762006-04-27 00:12:56 +010043
David Woodhousea3b67142006-04-25 14:54:40 +010044#ifdef __KERNEL__
David Woodhouseb7b3c762006-04-27 00:12:56 +010045
46struct sched_param {
47 int sched_priority;
48};
49
Linus Torvalds1da177e2005-04-16 15:20:36 -070050#include <asm/param.h> /* for HZ */
51
Linus Torvalds1da177e2005-04-16 15:20:36 -070052#include <linux/capability.h>
53#include <linux/threads.h>
54#include <linux/kernel.h>
55#include <linux/types.h>
56#include <linux/timex.h>
57#include <linux/jiffies.h>
58#include <linux/rbtree.h>
59#include <linux/thread_info.h>
60#include <linux/cpumask.h>
61#include <linux/errno.h>
62#include <linux/nodemask.h>
Martin Schwidefskyc92ff1b2007-10-16 01:24:43 -070063#include <linux/mm_types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070064
65#include <asm/system.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070066#include <asm/page.h>
67#include <asm/ptrace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070068#include <asm/cputime.h>
69
70#include <linux/smp.h>
71#include <linux/sem.h>
72#include <linux/signal.h>
Al Viro5ad4e532009-03-29 19:50:06 -040073#include <linux/path.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>
Dhaval Giani5cb350b2007-10-15 17:00:14 +020091#include <linux/kobject.h>
Arjan van de Ven97455122008-01-25 21:08:34 +010092#include <linux/latencytop.h>
David Howells9e2b2dc2008-08-13 16:20:04 +010093#include <linux/cred.h>
David Woodhousea3b67142006-04-25 14:54:40 +010094
95#include <asm/processor.h>
H. J. Lu36d57ac2005-09-06 15:16:49 -070096
Linus Torvalds1da177e2005-04-16 15:20:36 -070097struct exec_domain;
Ingo Molnarc87e2832006-06-27 02:54:58 -070098struct futex_pi_state;
Alexey Dobriyan286100a2008-01-25 21:08:34 +010099struct robust_list_head;
Neil Brownd89d8792007-05-01 09:53:42 +0200100struct bio;
Al Viro5ad4e532009-03-29 19:50:06 -0400101struct fs_struct;
Markus Metzgere2b371f2009-04-03 16:43:35 +0200102struct bts_context;
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200103struct perf_event_context;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700104
105/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700106 * List of flags we want to share for kernel threads,
107 * if only because they are not used by them anyway.
108 */
109#define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110
111/*
112 * These are the constant used to fake the fixed-point load-average
113 * counting. Some notes:
114 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
115 * a load-average precision of 10 bits integer + 11 bits fractional
116 * - if you want to count load-averages more often, you need more
117 * precision, or rounding will get you. With 2-second counting freq,
118 * the EXP_n values would be 1981, 2034 and 2043 if still using only
119 * 11 bit fractions.
120 */
121extern unsigned long avenrun[]; /* Load averages */
Thomas Gleixner2d024942009-05-02 20:08:52 +0200122extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700123
124#define FSHIFT 11 /* nr of bits of precision */
125#define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
Linus Torvalds0c2043a2007-10-07 16:17:38 -0700126#define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700127#define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
128#define EXP_5 2014 /* 1/exp(5sec/5min) */
129#define EXP_15 2037 /* 1/exp(5sec/15min) */
130
131#define CALC_LOAD(load,exp,n) \
132 load *= exp; \
133 load += n*(FIXED_1-exp); \
134 load >>= FSHIFT;
135
136extern unsigned long total_forks;
137extern int nr_threads;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700138DECLARE_PER_CPU(unsigned long, process_counts);
139extern int nr_processes(void);
140extern unsigned long nr_running(void);
141extern unsigned long nr_uninterruptible(void);
142extern unsigned long nr_iowait(void);
Arjan van de Ven69d25872009-09-21 17:04:08 -0700143extern unsigned long nr_iowait_cpu(void);
144extern unsigned long this_cpu_load(void);
145
146
Thomas Gleixnerdce48a82009-04-11 10:43:41 +0200147extern void calc_global_load(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148
Steven Rostedt7e49fcc2009-01-22 19:01:40 -0500149extern unsigned long get_parent_ip(unsigned long addr);
150
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200151struct seq_file;
152struct cfs_rq;
Ingo Molnar4cf86d72007-10-15 17:00:14 +0200153struct task_group;
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200154#ifdef CONFIG_SCHED_DEBUG
155extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
156extern void proc_sched_set_task(struct task_struct *p);
157extern void
Ingo Molnar5cef9ec2007-08-09 11:16:47 +0200158print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200159#else
160static inline void
161proc_sched_show_task(struct task_struct *p, struct seq_file *m)
162{
163}
164static inline void proc_sched_set_task(struct task_struct *p)
165{
166}
167static inline void
Ingo Molnar5cef9ec2007-08-09 11:16:47 +0200168print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
Ingo Molnar43ae34c2007-07-09 18:52:00 +0200169{
170}
171#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700172
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700173/*
174 * Task state bitmask. NOTE! These bits are also
175 * encoded in fs/proc/array.c: get_task_state().
176 *
177 * We have two separate sets of flags: task->state
178 * is about runnability, while task->exit_state are
179 * about the task exiting. Confusing, but this way
180 * modifying one set can't modify the other one by
181 * mistake.
182 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700183#define TASK_RUNNING 0
184#define TASK_INTERRUPTIBLE 1
185#define TASK_UNINTERRUPTIBLE 2
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500186#define __TASK_STOPPED 4
187#define __TASK_TRACED 8
Linus Torvalds4a8342d2005-09-29 15:18:21 -0700188/* in tsk->exit_state */
189#define EXIT_ZOMBIE 16
190#define EXIT_DEAD 32
191/* in tsk->state again */
Mike Galbraithaf927232007-10-15 17:00:13 +0200192#define TASK_DEAD 64
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500193#define TASK_WAKEKILL 128
Peter Zijlstrae9c84312009-09-15 14:43:03 +0200194#define TASK_WAKING 256
Peter Zijlstrae1781532009-12-17 13:16:30 +0100195#define TASK_STATE_MAX 512
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500196
Peter Zijlstra44d90df2009-12-17 13:16:28 +0100197#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
Peter Zijlstra73342152009-12-17 13:16:27 +0100198
Peter Zijlstrae1781532009-12-17 13:16:30 +0100199extern char ___assert_task_state[1 - 2*!!(
200 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500201
202/* Convenience macros for the sake of set_task_state */
203#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
204#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
205#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700206
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500207/* Convenience macros for the sake of wake_up */
208#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500209#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500210
211/* get_task_state() */
212#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500213 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
214 __TASK_TRACED)
Matthew Wilcox92a1f4b2007-12-06 10:55:25 -0500215
Matthew Wilcoxf021a3c2007-12-06 11:13:16 -0500216#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
217#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 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
261extern void sched_init(void);
262extern void sched_init_smp(void);
Harvey Harrison2d07b252008-02-15 09:56:34 -0800263extern asmlinkage void schedule_tail(struct task_struct *prev);
Ingo Molnar36c8b582006-07-03 00:25:41 -0700264extern void init_idle(struct task_struct *idle, int cpu);
Ingo Molnar1df21052007-07-09 18:51:58 +0200265extern void init_idle_bootup_task(struct task_struct *idle);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700266
Andrew Morton89f19f02009-09-19 11:55:44 -0700267extern int runqueue_is_locked(int cpu);
Oleg Nesterovad474ca2008-11-10 15:39:30 +0100268extern void task_rq_unlock_wait(struct task_struct *p);
Ingo Molnar017730c2008-05-12 21:20:52 +0200269
Rusty Russell6a7b3dc2008-11-25 02:35:04 +1030270extern cpumask_var_t nohz_cpu_mask;
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700271#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
272extern int select_nohz_load_balancer(int cpu);
Arun R Bharadwajeea08f32009-04-16 12:16:41 +0530273extern int get_nohz_load_balancer(void);
Siddha, Suresh B46cb4b72007-05-08 00:32:51 -0700274#else
275static inline int select_nohz_load_balancer(int cpu)
276{
277 return 0;
278}
279#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
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700310#ifdef CONFIG_DETECT_SOFTLOCKUP
Ingo Molnar6687a972006-03-24 03:18:41 -0800311extern void softlockup_tick(void);
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700312extern void touch_softlockup_watchdog(void);
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700313extern void touch_all_softlockup_watchdogs(void);
Mandeep Singh Bainesbaf48f62009-01-12 21:15:17 -0800314extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
Alexey Dobriyan8d65af72009-09-23 15:57:19 -0700315 void __user *buffer,
Mandeep Singh Bainesbaf48f62009-01-12 21:15:17 -0800316 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;
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700319#else
Ingo Molnar6687a972006-03-24 03:18:41 -0800320static inline void softlockup_tick(void)
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700321{
322}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700323static inline void touch_softlockup_watchdog(void)
324{
325}
Jeremy Fitzhardinge04c91672007-05-08 00:28:05 -0700326static inline void touch_all_softlockup_watchdogs(void)
327{
328}
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700329#endif
330
Mandeep Singh Bainese162b392009-01-15 11:08:40 -0800331#ifdef CONFIG_DETECT_HUNG_TASK
332extern unsigned int sysctl_hung_task_panic;
333extern unsigned long sysctl_hung_task_check_count;
334extern unsigned long sysctl_hung_task_timeout_secs;
335extern unsigned long sysctl_hung_task_warnings;
336extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
Alexey Dobriyan8d65af72009-09-23 15:57:19 -0700337 void __user *buffer,
Mandeep Singh Bainese162b392009-01-15 11:08:40 -0800338 size_t *lenp, loff_t *ppos);
339#endif
Ingo Molnar8446f1d2005-09-06 15:16:27 -0700340
Linus Torvalds1da177e2005-04-16 15:20:36 -0700341/* Attach to any functions which should be ignored in wchan output. */
342#define __sched __attribute__((__section__(".sched.text")))
Ingo Molnardeaf2222007-11-28 15:52:56 +0100343
344/* Linker adds these: start and end of __sched functions */
345extern char __sched_text_start[], __sched_text_end[];
346
Linus Torvalds1da177e2005-04-16 15:20:36 -0700347/* Is this address in the __sched functions? */
348extern int in_sched_functions(unsigned long addr);
349
350#define MAX_SCHEDULE_TIMEOUT LONG_MAX
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800351extern signed long schedule_timeout(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700352extern signed long schedule_timeout_interruptible(signed long timeout);
Matthew Wilcox294d5cc2007-12-06 11:59:46 -0500353extern signed long schedule_timeout_killable(signed long timeout);
Nishanth Aravamudan64ed93a2005-09-10 00:27:21 -0700354extern signed long schedule_timeout_uninterruptible(signed long timeout);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700355asmlinkage void schedule(void);
Peter Zijlstra0d66bf62009-01-12 14:01:47 +0100356extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700357
Serge E. Hallynab516012006-10-02 02:18:06 -0700358struct nsproxy;
Cedric Le Goateracce2922007-07-15 23:40:59 -0700359struct user_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700360
KAMEZAWA Hiroyuki341c87b2009-06-30 11:41:23 -0700361/*
362 * Default maximum number of active map areas, this limits the number of vmas
363 * per mm struct. Users can overwrite this number by sysctl but there is a
364 * problem.
365 *
366 * When a program's coredump is generated as ELF format, a section is created
367 * per a vma. In ELF, the number of sections is represented in unsigned short.
368 * This means the number of sections should be smaller than 65535 at coredump.
369 * Because the kernel adds some informative sections to a image of program at
370 * generating coredump, we need some margin. The number of extra sections is
371 * 1-3 now and depends on arch. We use "5" as safe margin, here.
372 */
373#define MAPCOUNT_ELF_CORE_MARGIN (5)
374#define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700375
376extern int sysctl_max_map_count;
377
378#include <linux/aio.h>
379
David Howellsefc1a3b2010-01-15 17:01:35 -0800380#ifdef CONFIG_MMU
381extern void arch_pick_mmap_layout(struct mm_struct *mm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382extern unsigned long
383arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
384 unsigned long, unsigned long);
385extern unsigned long
386arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
387 unsigned long len, unsigned long pgoff,
388 unsigned long flags);
Wolfgang Wander1363c3c2005-06-21 17:14:49 -0700389extern void arch_unmap_area(struct mm_struct *, unsigned long);
390extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
David Howellsefc1a3b2010-01-15 17:01:35 -0800391#else
392static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
393#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700394
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700395#if USE_SPLIT_PTLOCKS
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700396/*
397 * The mm counters are not protected by its page_table_lock,
398 * so must be incremented atomically.
399 */
Christoph Lameterd3cb4872006-01-06 00:11:20 -0800400#define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
401#define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
402#define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
403#define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
404#define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700405
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700406#else /* !USE_SPLIT_PTLOCKS */
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700407/*
408 * The mm counters are protected by its page_table_lock,
409 * so can be incremented directly.
410 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700411#define set_mm_counter(mm, member, value) (mm)->_##member = (value)
412#define get_mm_counter(mm, member) ((mm)->_##member)
413#define add_mm_counter(mm, member, value) (mm)->_##member += (value)
414#define inc_mm_counter(mm, member) (mm)->_##member++
415#define dec_mm_counter(mm, member) (mm)->_##member--
Hugh Dickins42946212005-10-29 18:16:05 -0700416
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700417#endif /* !USE_SPLIT_PTLOCKS */
Hugh Dickinsf412ac02005-10-29 18:16:41 -0700418
419#define get_mm_rss(mm) \
420 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
Hugh Dickins365e9c872005-10-29 18:16:18 -0700421#define update_hiwater_rss(mm) do { \
422 unsigned long _rss = get_mm_rss(mm); \
423 if ((mm)->hiwater_rss < _rss) \
424 (mm)->hiwater_rss = _rss; \
425} while (0)
426#define update_hiwater_vm(mm) do { \
427 if ((mm)->hiwater_vm < (mm)->total_vm) \
428 (mm)->hiwater_vm = (mm)->total_vm; \
429} while (0)
430
Oleg Nesterov9de15812009-03-31 15:19:29 -0700431static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
432{
433 return max(mm->hiwater_rss, get_mm_rss(mm));
434}
435
Jiri Pirko1f102062009-09-22 16:44:10 -0700436static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
437 struct mm_struct *mm)
438{
439 unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
440
441 if (*maxrss < hiwater_rss)
442 *maxrss = hiwater_rss;
443}
444
Oleg Nesterov9de15812009-03-31 15:19:29 -0700445static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
446{
447 return max(mm->hiwater_vm, mm->total_vm);
448}
Oleg Nesterov901608d2009-01-06 14:40:29 -0800449
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700450extern void set_dumpable(struct mm_struct *mm, int value);
451extern int get_dumpable(struct mm_struct *mm);
452
453/* mm flags */
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700454/* dumpable bits */
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700455#define MMF_DUMPABLE 0 /* core dump is permitted */
456#define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700457
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700458#define MMF_DUMPABLE_BITS 2
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700459#define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700460
461/* coredump filter bits */
462#define MMF_DUMP_ANON_PRIVATE 2
463#define MMF_DUMP_ANON_SHARED 3
464#define MMF_DUMP_MAPPED_PRIVATE 4
465#define MMF_DUMP_MAPPED_SHARED 5
Roland McGrath82df3972007-10-16 23:27:02 -0700466#define MMF_DUMP_ELF_HEADERS 6
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700467#define MMF_DUMP_HUGETLB_PRIVATE 7
468#define MMF_DUMP_HUGETLB_SHARED 8
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700469
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700470#define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700471#define MMF_DUMP_FILTER_BITS 7
Kawai, Hidehiro3cb4a0b2007-07-19 01:48:28 -0700472#define MMF_DUMP_FILTER_MASK \
473 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
474#define MMF_DUMP_FILTER_DEFAULT \
KOSAKI Motohiroe575f112008-10-18 20:27:08 -0700475 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
Roland McGrath656eb2c2008-10-18 20:28:23 -0700476 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
477
478#ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
479# define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
480#else
481# define MMF_DUMP_MASK_DEFAULT_ELF 0
482#endif
Hugh Dickinsf8af4da2009-09-21 17:01:57 -0700483 /* leave room for more dump flags */
484#define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
485
486#define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
Kawai, Hidehiro6c5d5232007-07-19 01:48:27 -0700487
Linus Torvalds1da177e2005-04-16 15:20:36 -0700488struct sighand_struct {
489 atomic_t count;
490 struct k_sigaction action[_NSIG];
491 spinlock_t siglock;
Davide Libenzib8fceee2007-09-20 12:40:16 -0700492 wait_queue_head_t signalfd_wqh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700493};
494
KaiGai Kohei0e464812006-06-25 05:49:24 -0700495struct pacct_struct {
KaiGai Koheif6ec29a2006-06-25 05:49:25 -0700496 int ac_flag;
497 long ac_exitcode;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700498 unsigned long ac_mem;
KaiGai Kohei77787bf2006-06-25 05:49:26 -0700499 cputime_t ac_utime, ac_stime;
500 unsigned long ac_minflt, ac_majflt;
KaiGai Kohei0e464812006-06-25 05:49:24 -0700501};
502
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200503struct cpu_itimer {
504 cputime_t expires;
505 cputime_t incr;
Stanislaw Gruszka8356b5f2009-07-29 12:15:27 +0200506 u32 error;
507 u32 incr_error;
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200508};
509
Frank Mayharf06febc2008-09-12 09:54:39 -0700510/**
511 * struct task_cputime - collected CPU time counts
512 * @utime: time spent in user mode, in &cputime_t units
513 * @stime: time spent in kernel mode, in &cputime_t units
514 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
Ingo Molnar5ce73a42008-09-14 17:11:46 +0200515 *
Frank Mayharf06febc2008-09-12 09:54:39 -0700516 * This structure groups together three kinds of CPU time that are
517 * tracked for threads and thread groups. Most things considering
518 * CPU time want to group these counts together and treat all three
519 * of them in parallel.
520 */
521struct task_cputime {
522 cputime_t utime;
523 cputime_t stime;
524 unsigned long long sum_exec_runtime;
525};
526/* Alternate field names when used to cache expirations. */
527#define prof_exp stime
528#define virt_exp utime
529#define sched_exp sum_exec_runtime
530
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100531#define INIT_CPUTIME \
532 (struct task_cputime) { \
533 .utime = cputime_zero, \
534 .stime = cputime_zero, \
535 .sum_exec_runtime = 0, \
536 }
537
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200538/*
539 * Disable preemption until the scheduler is running.
540 * Reset by start_kernel()->sched_init()->init_idle().
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200541 *
542 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
543 * before the scheduler is active -- see should_resched().
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200544 */
Peter Zijlstrad86ee482009-07-10 14:57:57 +0200545#define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
Peter Zijlstrac99e6ef2009-07-10 14:57:56 +0200546
Frank Mayharf06febc2008-09-12 09:54:39 -0700547/**
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100548 * struct thread_group_cputimer - thread group interval timer counts
549 * @cputime: thread group interval timers.
550 * @running: non-zero when there are timers running and
551 * @cputime receives updates.
552 * @lock: lock for fields in this struct.
Frank Mayharf06febc2008-09-12 09:54:39 -0700553 *
554 * This structure contains the version of task_cputime, above, that is
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100555 * used for thread group CPU timer calculations.
Frank Mayharf06febc2008-09-12 09:54:39 -0700556 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100557struct thread_group_cputimer {
558 struct task_cputime cputime;
559 int running;
560 spinlock_t lock;
Frank Mayharf06febc2008-09-12 09:54:39 -0700561};
Frank Mayharf06febc2008-09-12 09:54:39 -0700562
Linus Torvalds1da177e2005-04-16 15:20:36 -0700563/*
564 * NOTE! "signal_struct" does not have it's own
565 * locking, because a shared signal_struct always
566 * implies a shared sighand_struct, so locking
567 * sighand_struct is always a proper superset of
568 * the locking of signal_struct.
569 */
570struct signal_struct {
571 atomic_t count;
572 atomic_t live;
573
574 wait_queue_head_t wait_chldexit; /* for wait4() */
575
576 /* current thread group signal load-balancing target: */
Ingo Molnar36c8b582006-07-03 00:25:41 -0700577 struct task_struct *curr_target;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700578
579 /* shared signal handling: */
580 struct sigpending shared_pending;
581
582 /* thread group exit support */
583 int group_exit_code;
584 /* overloaded:
585 * - notify group_exit_task when ->count is equal to notify_count
586 * - everyone except group_exit_task is stopped during signal delivery
587 * of fatal signals, group_exit_task processes the signal.
588 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700589 int notify_count;
Richard Kennedy07dd20e2008-08-01 13:18:04 +0100590 struct task_struct *group_exit_task;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700591
592 /* thread group stop support, overloads group_exit_code too */
593 int group_stop_count;
594 unsigned int flags; /* see SIGNAL_* flags below */
595
596 /* POSIX.1b Interval Timers */
597 struct list_head posix_timers;
598
599 /* ITIMER_REAL timer for the process */
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800600 struct hrtimer real_timer;
Oleg Nesterovfea9d172008-02-08 04:19:19 -0800601 struct pid *leader_pid;
Thomas Gleixner2ff678b2006-01-09 20:52:34 -0800602 ktime_t it_real_incr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700603
Stanislaw Gruszka42c4ab42009-07-29 12:15:26 +0200604 /*
605 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
606 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
607 * values are defined to 0 and 1 respectively
608 */
609 struct cpu_itimer it[2];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610
Frank Mayharf06febc2008-09-12 09:54:39 -0700611 /*
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100612 * Thread group totals for process CPU timers.
613 * See thread_group_cputimer(), et al, for details.
Frank Mayharf06febc2008-09-12 09:54:39 -0700614 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +0100615 struct thread_group_cputimer cputimer;
Frank Mayharf06febc2008-09-12 09:54:39 -0700616
617 /* Earliest-expiration cache. */
618 struct task_cputime cputime_expires;
619
620 struct list_head cpu_timers[3];
621
Eric W. Biedermanab521dc2007-02-12 00:53:00 -0800622 struct pid *tty_old_pgrp;
Cedric Le Goater1ec320a2006-12-08 02:37:55 -0800623
Linus Torvalds1da177e2005-04-16 15:20:36 -0700624 /* boolean value for session group leader */
625 int leader;
626
627 struct tty_struct *tty; /* NULL if no tty */
628
629 /*
630 * Cumulative resource counters for dead threads in the group,
631 * and for reaped dead child processes forked by this group.
632 * Live threads maintain their own counters and add to these
633 * in __exit_signal, except for the group leader.
634 */
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100635 cputime_t utime, stime, cutime, cstime;
Laurent Vivier9ac52312007-10-15 17:00:19 +0200636 cputime_t gtime;
637 cputime_t cgtime;
Hidetoshi Seto0cf55e12009-12-02 17:28:07 +0900638#ifndef CONFIG_VIRT_CPU_ACCOUNTING
639 cputime_t prev_utime, prev_stime;
640#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700641 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
642 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
Eric Dumazet6eaeeab2007-05-10 22:22:37 -0700643 unsigned long inblock, oublock, cinblock, coublock;
Jiri Pirko1f102062009-09-22 16:44:10 -0700644 unsigned long maxrss, cmaxrss;
Andrea Righi940389b2008-07-28 00:48:12 +0200645 struct task_io_accounting ioac;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700646
647 /*
Peter Zijlstra32bd6712009-02-05 12:24:15 +0100648 * Cumulative ns of schedule CPU time fo dead threads in the
649 * group, not including a zombie group leader, (This only differs
650 * from jiffies_to_ns(utime + stime) if sched_clock uses something
651 * other than jiffies.)
652 */
653 unsigned long long sum_sched_runtime;
654
655 /*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700656 * We don't bother to synchronize most readers of this at all,
657 * because there is no reader checking a limit that actually needs
658 * to get both rlim_cur and rlim_max atomically, and either one
659 * alone is a single word that can safely be read normally.
660 * getrlimit/setrlimit use task_lock(current->group_leader) to
661 * protect this instead of the siglock, because they really
662 * have no need to disable irqs.
663 */
664 struct rlimit rlim[RLIM_NLIMITS];
665
KaiGai Kohei0e464812006-06-25 05:49:24 -0700666#ifdef CONFIG_BSD_PROCESS_ACCT
667 struct pacct_struct pacct; /* per-process accounting information */
668#endif
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700669#ifdef CONFIG_TASKSTATS
Shailabh Nagarad4ecbc2006-07-14 00:24:44 -0700670 struct taskstats *stats;
671#endif
Miloslav Trmac522ed772007-07-15 23:40:56 -0700672#ifdef CONFIG_AUDIT
673 unsigned audit_tty;
674 struct tty_audit_buf *tty_audit_buf;
675#endif
KOSAKI Motohiro28b83c52009-09-21 17:03:13 -0700676
677 int oom_adj; /* OOM kill score adjustment (bit shift) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700678};
679
Nick Piggin4866cde2005-06-25 14:57:23 -0700680/* Context switch must be unlocked if interrupts are to be enabled */
681#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
682# define __ARCH_WANT_UNLOCKED_CTXSW
683#endif
684
Linus Torvalds1da177e2005-04-16 15:20:36 -0700685/*
686 * Bits in flags field of signal_struct.
687 */
688#define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
689#define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
690#define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
691#define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
Oleg Nesterove4420552008-04-30 00:52:44 -0700692/*
693 * Pending notifications to parent.
694 */
695#define SIGNAL_CLD_STOPPED 0x00000010
696#define SIGNAL_CLD_CONTINUED 0x00000020
697#define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698
Oleg Nesterovfae5fa42008-04-30 00:53:03 -0700699#define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
700
Oleg Nesteroved5d2ca2008-02-04 22:27:24 -0800701/* If true, all threads except ->group_exit_task have pending SIGKILL */
702static inline int signal_group_exit(const struct signal_struct *sig)
703{
704 return (sig->flags & SIGNAL_GROUP_EXIT) ||
705 (sig->group_exit_task != NULL);
706}
707
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708/*
709 * Some day this will be a full-fledged user tracking system..
710 */
711struct user_struct {
712 atomic_t __count; /* reference count */
713 atomic_t processes; /* How many processes does this user have? */
714 atomic_t files; /* How many open files does this user have? */
715 atomic_t sigpending; /* How many pending signals does this user have? */
Amy Griffis2d9048e2006-06-01 13:10:59 -0700716#ifdef CONFIG_INOTIFY_USER
Robert Love0eeca282005-07-12 17:06:03 -0400717 atomic_t inotify_watches; /* How many inotify watches does this user have? */
718 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
719#endif
Davide Libenzi7ef99642008-12-01 13:13:55 -0800720#ifdef CONFIG_EPOLL
Davide Libenzi7ef99642008-12-01 13:13:55 -0800721 atomic_t epoll_watches; /* The number of file descriptors currently watched */
722#endif
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700723#ifdef CONFIG_POSIX_MQUEUE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700724 /* protected by mq_lock */
725 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
Alexey Dobriyan970a8642007-10-16 23:30:09 -0700726#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700727 unsigned long locked_shm; /* How many pages of mlocked shm ? */
728
729#ifdef CONFIG_KEYS
730 struct key *uid_keyring; /* UID specific keyring */
731 struct key *session_keyring; /* UID's default session keyring */
732#endif
733
734 /* Hash table maintenance information */
Pavel Emelyanov735de222007-09-18 22:46:44 -0700735 struct hlist_node uidhash_node;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700736 uid_t uid;
Serge Hallyn18b6e042008-10-15 16:38:45 -0500737 struct user_namespace *user_ns;
Srivatsa Vaddagiri24e377a2007-10-15 17:00:09 +0200738
Peter Zijlstra052f1dc2008-02-13 15:45:40 +0100739#ifdef CONFIG_USER_SCHED
Ingo Molnar4cf86d72007-10-15 17:00:14 +0200740 struct task_group *tg;
Dhaval Gianib1a8c172007-10-17 16:55:11 +0200741#ifdef CONFIG_SYSFS
Kay Sieverseb41d942007-11-02 13:47:53 +0100742 struct kobject kobj;
Kay Sievers39592142009-03-24 15:43:30 +0100743 struct delayed_work work;
Srivatsa Vaddagiri24e377a2007-10-15 17:00:09 +0200744#endif
Dhaval Gianib1a8c172007-10-17 16:55:11 +0200745#endif
Peter Zijlstra789f90f2009-05-15 15:19:27 +0200746
Ingo Molnarcdd6c482009-09-21 12:02:48 +0200747#ifdef CONFIG_PERF_EVENTS
Peter Zijlstra789f90f2009-05-15 15:19:27 +0200748 atomic_long_t locked_vm;
749#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700750};
751
Kay Sieverseb41d942007-11-02 13:47:53 +0100752extern int uids_sysfs_init(void);
Dhaval Giani5cb350b2007-10-15 17:00:14 +0200753
Linus Torvalds1da177e2005-04-16 15:20:36 -0700754extern struct user_struct *find_user(uid_t);
755
756extern struct user_struct root_user;
757#define INIT_USER (&root_user)
758
David Howellsb6dff3e2008-11-14 10:39:16 +1100759
Linus Torvalds1da177e2005-04-16 15:20:36 -0700760struct backing_dev_info;
761struct reclaim_state;
762
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700763#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700764struct sched_info {
765 /* cumulative counters */
Ingo Molnar2d723762007-10-15 17:00:12 +0200766 unsigned long pcount; /* # of times run on this cpu */
Ken Chen9c2c4802008-12-16 23:41:22 -0800767 unsigned long long run_delay; /* time spent waiting on a runqueue */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700768
769 /* timestamps */
Balbir Singh172ba842007-07-09 18:52:00 +0200770 unsigned long long last_arrival,/* when we last ran on a cpu */
771 last_queued; /* when we were last queued to run */
Ingo Molnarb8efb562007-10-15 17:00:10 +0200772#ifdef CONFIG_SCHEDSTATS
773 /* BKL stats */
Ken Chen480b9432007-10-18 21:32:56 +0200774 unsigned int bkl_count;
Ingo Molnarb8efb562007-10-15 17:00:10 +0200775#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700777#endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700778
Shailabh Nagarca74e922006-07-14 00:24:36 -0700779#ifdef CONFIG_TASK_DELAY_ACCT
780struct task_delay_info {
781 spinlock_t lock;
782 unsigned int flags; /* Private per-task flags */
783
784 /* For each stat XXX, add following, aligned appropriately
785 *
786 * struct timespec XXX_start, XXX_end;
787 * u64 XXX_delay;
788 * u32 XXX_count;
789 *
790 * Atomicity of updates to XXX_delay, XXX_count protected by
791 * single lock above (split into XXX_lock if contention is an issue).
792 */
Shailabh Nagar0ff92242006-07-14 00:24:37 -0700793
794 /*
795 * XXX_count is incremented on every XXX operation, the delay
796 * associated with the operation is added to XXX_delay.
797 * XXX_delay contains the accumulated delay time in nanoseconds.
798 */
799 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
800 u64 blkio_delay; /* wait for sync block io completion */
801 u64 swapin_delay; /* wait for swapin block io completion */
802 u32 blkio_count; /* total count of the number of sync block */
803 /* io operations performed */
804 u32 swapin_count; /* total count of the number of swapin block */
805 /* io operations performed */
Keika Kobayashi873b4772008-07-25 01:48:52 -0700806
807 struct timespec freepages_start, freepages_end;
808 u64 freepages_delay; /* wait for memory reclaim */
809 u32 freepages_count; /* total count of memory reclaim */
Shailabh Nagarca74e922006-07-14 00:24:36 -0700810};
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700811#endif /* CONFIG_TASK_DELAY_ACCT */
812
813static inline int sched_info_on(void)
814{
815#ifdef CONFIG_SCHEDSTATS
816 return 1;
817#elif defined(CONFIG_TASK_DELAY_ACCT)
818 extern int delayacct_on;
819 return delayacct_on;
820#else
821 return 0;
Shailabh Nagarca74e922006-07-14 00:24:36 -0700822#endif
Chandra Seetharaman52f17b62006-07-14 00:24:38 -0700823}
Shailabh Nagarca74e922006-07-14 00:24:36 -0700824
Ingo Molnard15bcfd2007-07-09 18:51:57 +0200825enum cpu_idle_type {
826 CPU_IDLE,
827 CPU_NOT_IDLE,
828 CPU_NEWLY_IDLE,
829 CPU_MAX_IDLE_TYPES
Linus Torvalds1da177e2005-04-16 15:20:36 -0700830};
831
832/*
833 * sched-domains (multiprocessor balancing) declarations:
834 */
Ingo Molnar9aa7b362007-07-09 18:51:58 +0200835
836/*
837 * Increase resolution of nice-level calculations:
838 */
839#define SCHED_LOAD_SHIFT 10
840#define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
841
Suresh Siddhaf8700df2007-08-23 15:18:02 +0200842#define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
Linus Torvalds1da177e2005-04-16 15:20:36 -0700843
Peter Williams2dd73a42006-06-27 02:54:34 -0700844#ifdef CONFIG_SMP
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200845#define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
846#define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
847#define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
848#define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
Peter Zijlstrac88d5912009-09-10 13:50:02 +0200849#define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200850#define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
Peter Zijlstra59abf022009-09-16 08:28:30 +0200851#define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200852#define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
853#define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
854#define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
855#define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
Peter Zijlstrac88d5912009-09-10 13:50:02 +0200856
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200857#define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700858
Gautham R Shenoyafb8a9b2008-12-18 23:26:09 +0530859enum powersavings_balance_level {
860 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
861 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
862 * first for long running threads
863 */
864 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
865 * cpu package for power savings
866 */
867 MAX_POWERSAVINGS_BALANCE_LEVELS
868};
Siddha, Suresh B89c47102006-10-03 01:14:09 -0700869
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530870extern int sched_mc_power_savings, sched_smt_power_savings;
Siddha, Suresh B89c47102006-10-03 01:14:09 -0700871
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530872static inline int sd_balance_for_mc_power(void)
873{
874 if (sched_smt_power_savings)
875 return SD_POWERSAVINGS_BALANCE;
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -0700876
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200877 return SD_PREFER_SIBLING;
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530878}
879
880static inline int sd_balance_for_package_power(void)
881{
882 if (sched_mc_power_savings | sched_smt_power_savings)
883 return SD_POWERSAVINGS_BALANCE;
884
Peter Zijlstrab5d978e2009-09-01 10:34:33 +0200885 return SD_PREFER_SIBLING;
Vaidyanathan Srinivasan716707b2008-12-18 23:26:02 +0530886}
Nick Piggin147cbb42005-06-25 14:57:19 -0700887
Vaidyanathan Srinivasan100fdae2008-12-18 23:26:47 +0530888/*
889 * Optimise SD flags for power savings:
890 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
891 * Keep default SD flags if sched_{smt,mc}_power_saving=0
892 */
893
894static inline int sd_power_saving_flags(void)
895{
896 if (sched_mc_power_savings | sched_smt_power_savings)
897 return SD_BALANCE_NEWIDLE;
898
899 return 0;
900}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700901
902struct sched_group {
903 struct sched_group *next; /* Must be a circular list */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700904
905 /*
906 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
Peter Zijlstra18a38852009-09-01 10:34:39 +0200907 * single CPU.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700908 */
Peter Zijlstra18a38852009-09-01 10:34:39 +0200909 unsigned int cpu_power;
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030910
Ingo Molnar4200efd2009-05-19 09:22:19 +0200911 /*
912 * The CPUs this group covers.
913 *
914 * NOTE: this field is variable length. (Allocated dynamically
915 * by attaching extra space to the end of the structure,
916 * depending on how many CPUs the kernel has booted up with)
917 *
918 * It is also be embedded into static data structures at build
919 * time. (See 'struct static_sched_group' in kernel/sched.c)
920 */
921 unsigned long cpumask[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700922};
923
Rusty Russell758b2cd2008-11-25 02:35:04 +1030924static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
925{
Rusty Russell6c99e9a2008-11-25 02:35:04 +1030926 return to_cpumask(sg->cpumask);
Rusty Russell758b2cd2008-11-25 02:35:04 +1030927}
928
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900929enum sched_domain_level {
930 SD_LV_NONE = 0,
931 SD_LV_SIBLING,
932 SD_LV_MC,
933 SD_LV_CPU,
934 SD_LV_NODE,
935 SD_LV_ALLNODES,
936 SD_LV_MAX
937};
938
939struct sched_domain_attr {
940 int relax_domain_level;
941};
942
943#define SD_ATTR_INIT (struct sched_domain_attr) { \
944 .relax_domain_level = -1, \
945}
946
Linus Torvalds1da177e2005-04-16 15:20:36 -0700947struct sched_domain {
948 /* These fields must be setup */
949 struct sched_domain *parent; /* top domain must be null terminated */
Siddha, Suresh B1a848872006-10-03 01:14:08 -0700950 struct sched_domain *child; /* bottom domain must be null terminated */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951 struct sched_group *groups; /* the balancing groups of the domain */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700952 unsigned long min_interval; /* Minimum balance interval ms */
953 unsigned long max_interval; /* Maximum balance interval ms */
954 unsigned int busy_factor; /* less balancing by factor if busy */
955 unsigned int imbalance_pct; /* No balance until over watermark */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700956 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
Nick Piggin78979862005-06-25 14:57:13 -0700957 unsigned int busy_idx;
958 unsigned int idle_idx;
959 unsigned int newidle_idx;
960 unsigned int wake_idx;
Nick Piggin147cbb42005-06-25 14:57:19 -0700961 unsigned int forkexec_idx;
Peter Zijlstraa52bfd732009-09-01 10:34:35 +0200962 unsigned int smt_gain;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700963 int flags; /* See SD_* */
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +0900964 enum sched_domain_level level;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700965
966 /* Runtime fields. */
967 unsigned long last_balance; /* init to jiffies. units in jiffies */
968 unsigned int balance_interval; /* initialise to 1. units in ms. */
969 unsigned int nr_balance_failed; /* initialise to 0 */
970
Peter Zijlstra2398f2c2008-06-27 13:41:35 +0200971 u64 last_update;
972
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973#ifdef CONFIG_SCHEDSTATS
974 /* load_balance() stats */
Ken Chen480b9432007-10-18 21:32:56 +0200975 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
976 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
977 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
978 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
979 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
980 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
981 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
982 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700983
984 /* Active load balancing */
Ken Chen480b9432007-10-18 21:32:56 +0200985 unsigned int alb_count;
986 unsigned int alb_failed;
987 unsigned int alb_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700988
Nick Piggin68767a02005-06-25 14:57:20 -0700989 /* SD_BALANCE_EXEC stats */
Ken Chen480b9432007-10-18 21:32:56 +0200990 unsigned int sbe_count;
991 unsigned int sbe_balanced;
992 unsigned int sbe_pushed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700993
Nick Piggin68767a02005-06-25 14:57:20 -0700994 /* SD_BALANCE_FORK stats */
Ken Chen480b9432007-10-18 21:32:56 +0200995 unsigned int sbf_count;
996 unsigned int sbf_balanced;
997 unsigned int sbf_pushed;
Nick Piggin68767a02005-06-25 14:57:20 -0700998
Linus Torvalds1da177e2005-04-16 15:20:36 -0700999 /* try_to_wake_up() stats */
Ken Chen480b9432007-10-18 21:32:56 +02001000 unsigned int ttwu_wake_remote;
1001 unsigned int ttwu_move_affine;
1002 unsigned int ttwu_move_balance;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001003#endif
Ingo Molnara5d8c342008-10-09 11:35:51 +02001004#ifdef CONFIG_SCHED_DEBUG
1005 char *name;
1006#endif
Rusty Russell6c99e9a2008-11-25 02:35:04 +10301007
Ingo Molnar4200efd2009-05-19 09:22:19 +02001008 /*
1009 * Span of all CPUs in this domain.
1010 *
1011 * NOTE: this field is variable length. (Allocated dynamically
1012 * by attaching extra space to the end of the structure,
1013 * depending on how many CPUs the kernel has booted up with)
1014 *
1015 * It is also be embedded into static data structures at build
1016 * time. (See 'struct static_sched_domain' in kernel/sched.c)
1017 */
1018 unsigned long span[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -07001019};
1020
Rusty Russell758b2cd2008-11-25 02:35:04 +10301021static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1022{
Rusty Russell6c99e9a2008-11-25 02:35:04 +10301023 return to_cpumask(sd->span);
Rusty Russell758b2cd2008-11-25 02:35:04 +10301024}
1025
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301026extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Hidetoshi Seto1d3504f2008-04-15 14:04:23 +09001027 struct sched_domain_attr *dattr_new);
Paul Jackson029190c2007-10-18 23:40:20 -07001028
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301029/* Allocate an array of sched domains, for partition_sched_domains(). */
1030cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1031void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1032
Ingo Molnar06aaf762008-12-18 21:30:23 +01001033/* Test a flag in parent sched domain */
1034static inline int test_sd_parent(struct sched_domain *sd, int flag)
1035{
1036 if (sd->parent && (sd->parent->flags & flag))
1037 return 1;
1038
1039 return 0;
1040}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001041
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001042unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1043unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1044
Ingo Molnar1b427c12008-07-18 14:01:39 +02001045#else /* CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001046
Ingo Molnar1b427c12008-07-18 14:01:39 +02001047struct sched_domain_attr;
1048
1049static inline void
Rusty Russellacc3f5d2009-11-03 14:53:40 +10301050partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
Ingo Molnar1b427c12008-07-18 14:01:39 +02001051 struct sched_domain_attr *dattr_new)
Con Kolivasd02c7a82007-07-26 13:40:43 +02001052{
Con Kolivasd02c7a82007-07-26 13:40:43 +02001053}
Ingo Molnar1b427c12008-07-18 14:01:39 +02001054#endif /* !CONFIG_SMP */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001055
Peter Zijlstra47fe38f2009-09-02 13:49:18 +02001056
Linus Torvalds1da177e2005-04-16 15:20:36 -07001057struct io_context; /* See blkdev.h */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001058
Linus Torvalds1da177e2005-04-16 15:20:36 -07001059
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001060#ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
Ingo Molnar36c8b582006-07-03 00:25:41 -07001061extern void prefetch_stack(struct task_struct *t);
Chen, Kenneth W383f2832005-09-09 13:02:02 -07001062#else
1063static inline void prefetch_stack(struct task_struct *t) { }
1064#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001065
1066struct audit_context; /* See audit.c */
1067struct mempolicy;
Jens Axboeb92ce552006-04-11 13:52:07 +02001068struct pipe_inode_info;
Serge E. Hallyn4865ecf2006-10-02 02:18:14 -07001069struct uts_namespace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001070
Ingo Molnar20b8a592007-07-09 18:51:58 +02001071struct rq;
1072struct sched_domain;
1073
Peter Zijlstra7d478722009-09-14 19:55:44 +02001074/*
1075 * wake flags
1076 */
1077#define WF_SYNC 0x01 /* waker goes to sleep after wakup */
Peter Zijlstraa7558e02009-09-14 20:02:34 +02001078#define WF_FORK 0x02 /* child wakeup after fork */
Peter Zijlstra7d478722009-09-14 19:55:44 +02001079
Ingo Molnar20b8a592007-07-09 18:51:58 +02001080struct sched_class {
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001081 const struct sched_class *next;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001082
Ingo Molnarfd390f62007-08-09 11:16:48 +02001083 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
Ingo Molnarf02231e2007-08-09 11:16:48 +02001084 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
Dmitry Adamushko4530d7a2007-10-15 17:00:08 +02001085 void (*yield_task) (struct rq *rq);
Ingo Molnar20b8a592007-07-09 18:51:58 +02001086
Peter Zijlstra7d478722009-09-14 19:55:44 +02001087 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
Ingo Molnar20b8a592007-07-09 18:51:58 +02001088
Ingo Molnarfb8d4722007-08-09 11:16:48 +02001089 struct task_struct * (*pick_next_task) (struct rq *rq);
Ingo Molnar31ee5292007-08-09 11:16:49 +02001090 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
Ingo Molnar20b8a592007-07-09 18:51:58 +02001091
Peter Williams681f3e62007-10-24 18:23:51 +02001092#ifdef CONFIG_SMP
Peter Zijlstra7d478722009-09-14 19:55:44 +02001093 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
Li Zefan4ce72a22008-10-22 15:25:26 +08001094
Peter Williams43010652007-08-09 11:16:46 +02001095 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
Peter Williamse1d14842007-10-24 18:23:51 +02001096 struct rq *busiest, unsigned long max_load_move,
Ingo Molnar20b8a592007-07-09 18:51:58 +02001097 struct sched_domain *sd, enum cpu_idle_type idle,
Peter Williamsa4ac01c2007-08-09 11:16:46 +02001098 int *all_pinned, int *this_best_prio);
Ingo Molnar20b8a592007-07-09 18:51:58 +02001099
Peter Williamse1d14842007-10-24 18:23:51 +02001100 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1101 struct rq *busiest, struct sched_domain *sd,
1102 enum cpu_idle_type idle);
Steven Rostedt9a897c52008-01-25 21:08:22 +01001103 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1104 void (*post_schedule) (struct rq *this_rq);
Peter Zijlstraefbbd052009-12-16 18:04:40 +01001105 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1106 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
Peter Williamse1d14842007-10-24 18:23:51 +02001107
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001108 void (*set_cpus_allowed)(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10301109 const struct cpumask *newmask);
Gregory Haskins57d885f2008-01-25 21:08:18 +01001110
Gregory Haskins1f11eb62008-06-04 15:04:05 -04001111 void (*rq_online)(struct rq *rq);
1112 void (*rq_offline)(struct rq *rq);
Li Zefan4ce72a22008-10-22 15:25:26 +08001113#endif
1114
1115 void (*set_curr_task) (struct rq *rq);
1116 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
Peter Zijlstracd29fe62009-11-27 17:32:46 +01001117 void (*task_fork) (struct task_struct *p);
Steven Rostedtcb469842008-01-25 21:08:22 +01001118
1119 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1120 int running);
1121 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1122 int running);
1123 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1124 int oldprio, int running);
Peter Zijlstra810b3812008-02-29 15:21:01 -05001125
Thomas Gleixnerdba091b2009-12-09 09:32:03 +01001126 unsigned int (*get_rr_interval) (struct rq *rq,
1127 struct task_struct *task);
Peter Williams0d721ce2009-09-21 01:31:53 +00001128
Peter Zijlstra810b3812008-02-29 15:21:01 -05001129#ifdef CONFIG_FAIR_GROUP_SCHED
Peter Zijlstra88ec22d2009-12-16 18:04:41 +01001130 void (*moved_group) (struct task_struct *p, int on_rq);
Peter Zijlstra810b3812008-02-29 15:21:01 -05001131#endif
Ingo Molnar20b8a592007-07-09 18:51:58 +02001132};
1133
1134struct load_weight {
1135 unsigned long weight, inv_weight;
1136};
1137
1138/*
1139 * CFS stats for a schedulable entity (task, task-group etc)
1140 *
1141 * Current field usage histogram:
1142 *
1143 * 4 se->block_start
1144 * 4 se->run_node
1145 * 4 se->sleep_start
Ingo Molnar20b8a592007-07-09 18:51:58 +02001146 * 6 se->load.weight
Ingo Molnar20b8a592007-07-09 18:51:58 +02001147 */
1148struct sched_entity {
Ingo Molnar20b8a592007-07-09 18:51:58 +02001149 struct load_weight load; /* for load-balancing */
1150 struct rb_node run_node;
Peter Zijlstra4a55bd52008-04-19 19:45:00 +02001151 struct list_head group_node;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001152 unsigned int on_rq;
1153
Ingo Molnar20b8a592007-07-09 18:51:58 +02001154 u64 exec_start;
Ingo Molnar94c18222007-08-02 17:41:40 +02001155 u64 sum_exec_runtime;
Ingo Molnare9acbff2007-10-15 17:00:04 +02001156 u64 vruntime;
Ingo Molnarf6cf8912007-08-28 12:53:24 +02001157 u64 prev_sum_exec_runtime;
Ingo Molnar94c18222007-08-02 17:41:40 +02001158
Ingo Molnar4ae7d5c2008-03-19 01:42:00 +01001159 u64 last_wakeup;
1160 u64 avg_overlap;
1161
Ingo Molnar6c594c22008-12-14 12:34:15 +01001162 u64 nr_migrations;
1163
Ingo Molnar34cb6132009-01-16 13:36:06 +01001164 u64 start_runtime;
1165 u64 avg_wakeup;
Ingo Molnar34cb6132009-01-16 13:36:06 +01001166
Ingo Molnar94c18222007-08-02 17:41:40 +02001167#ifdef CONFIG_SCHEDSTATS
1168 u64 wait_start;
1169 u64 wait_max;
Arjan van de Ven6d082592008-01-25 21:08:35 +01001170 u64 wait_count;
1171 u64 wait_sum;
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001172 u64 iowait_count;
1173 u64 iowait_sum;
Ingo Molnar94c18222007-08-02 17:41:40 +02001174
1175 u64 sleep_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001176 u64 sleep_max;
Ingo Molnar94c18222007-08-02 17:41:40 +02001177 s64 sum_sleep_runtime;
1178
1179 u64 block_start;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001180 u64 block_max;
1181 u64 exec_max;
Ingo Molnareba1ed42007-10-15 17:00:02 +02001182 u64 slice_max;
Ingo Molnarcc367732007-10-15 17:00:18 +02001183
Ingo Molnarcc367732007-10-15 17:00:18 +02001184 u64 nr_migrations_cold;
1185 u64 nr_failed_migrations_affine;
1186 u64 nr_failed_migrations_running;
1187 u64 nr_failed_migrations_hot;
1188 u64 nr_forced_migrations;
Ingo Molnarcc367732007-10-15 17:00:18 +02001189
1190 u64 nr_wakeups;
1191 u64 nr_wakeups_sync;
1192 u64 nr_wakeups_migrate;
1193 u64 nr_wakeups_local;
1194 u64 nr_wakeups_remote;
1195 u64 nr_wakeups_affine;
1196 u64 nr_wakeups_affine_attempts;
1197 u64 nr_wakeups_passive;
1198 u64 nr_wakeups_idle;
Ingo Molnar94c18222007-08-02 17:41:40 +02001199#endif
1200
Ingo Molnar20b8a592007-07-09 18:51:58 +02001201#ifdef CONFIG_FAIR_GROUP_SCHED
1202 struct sched_entity *parent;
1203 /* rq on which this entity is (to be) queued: */
1204 struct cfs_rq *cfs_rq;
1205 /* rq "owned" by this entity/group: */
1206 struct cfs_rq *my_q;
1207#endif
1208};
Ingo Molnar70b97a72006-07-03 00:25:42 -07001209
Peter Zijlstrafa717062008-01-25 21:08:27 +01001210struct sched_rt_entity {
1211 struct list_head run_list;
Peter Zijlstra78f2c7d2008-01-25 21:08:27 +01001212 unsigned long timeout;
Richard Kennedybee367e2008-08-01 13:24:08 +01001213 unsigned int time_slice;
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001214 int nr_cpus_allowed;
1215
Peter Zijlstra58d6c2d2008-04-19 19:45:00 +02001216 struct sched_rt_entity *back;
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01001217#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra6f505b12008-01-25 21:08:30 +01001218 struct sched_rt_entity *parent;
1219 /* rq on which this entity is (to be) queued: */
1220 struct rt_rq *rt_rq;
1221 /* rq "owned" by this entity/group: */
1222 struct rt_rq *my_q;
1223#endif
Peter Zijlstrafa717062008-01-25 21:08:27 +01001224};
1225
Paul E. McKenney86848962009-08-27 15:00:12 -07001226struct rcu_node;
1227
Linus Torvalds1da177e2005-04-16 15:20:36 -07001228struct task_struct {
1229 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
Roman Zippelf7e42172007-05-09 02:35:17 -07001230 void *stack;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001231 atomic_t usage;
William Cohen97dc32c2007-05-08 00:23:41 -07001232 unsigned int flags; /* per process flags, defined below */
1233 unsigned int ptrace;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001234
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001235 int lock_depth; /* BKL lock depth */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001236
Peter Williams2dd73a42006-06-27 02:54:34 -07001237#ifdef CONFIG_SMP
1238#ifdef __ARCH_WANT_UNLOCKED_CTXSW
Nick Piggin4866cde2005-06-25 14:57:23 -07001239 int oncpu;
1240#endif
Peter Williams2dd73a42006-06-27 02:54:34 -07001241#endif
Ingo Molnar50e645a2007-07-09 18:52:00 +02001242
Ingo Molnarb29739f2006-06-27 02:54:51 -07001243 int prio, static_prio, normal_prio;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001244 unsigned int rt_priority;
Ingo Molnar5522d5d2007-10-15 17:00:12 +02001245 const struct sched_class *sched_class;
Ingo Molnar20b8a592007-07-09 18:51:58 +02001246 struct sched_entity se;
Peter Zijlstrafa717062008-01-25 21:08:27 +01001247 struct sched_rt_entity rt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001248
Avi Kivitye107be32007-07-26 13:40:43 +02001249#ifdef CONFIG_PREEMPT_NOTIFIERS
1250 /* list of struct preempt_notifier: */
1251 struct hlist_head preempt_notifiers;
1252#endif
1253
Alexey Dobriyan18796aa2007-10-16 23:30:26 -07001254 /*
1255 * fpu_counter contains the number of consecutive context switches
1256 * that the FPU is used. If this is over a threshold, the lazy fpu
1257 * saving becomes unlazy to save the trap. This is an unsigned char
1258 * so that after 256 times the counter wraps and the behavior turns
1259 * lazy again; this to deal with bursty apps that only use FPU for
1260 * a short time
1261 */
1262 unsigned char fpu_counter;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001263#ifdef CONFIG_BLK_DEV_IO_TRACE
Jens Axboe2056a782006-03-23 20:00:26 +01001264 unsigned int btrace_seq;
Alexey Dobriyan6c5c9342006-09-29 01:59:40 -07001265#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001266
William Cohen97dc32c2007-05-08 00:23:41 -07001267 unsigned int policy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001268 cpumask_t cpus_allowed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001269
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001270#ifdef CONFIG_TREE_PREEMPT_RCU
Paul E. McKenneye260be62008-01-25 21:08:24 +01001271 int rcu_read_lock_nesting;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001272 char rcu_read_unlock_special;
Paul E. McKenney86848962009-08-27 15:00:12 -07001273 struct rcu_node *rcu_blocked_node;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001274 struct list_head rcu_node_entry;
1275#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
Paul E. McKenneye260be62008-01-25 21:08:24 +01001276
Chandra Seetharaman52f17b62006-07-14 00:24:38 -07001277#if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001278 struct sched_info sched_info;
1279#endif
1280
1281 struct list_head tasks;
Gregory Haskins917b6272008-12-29 09:39:53 -05001282 struct plist_node pushable_tasks;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001283
1284 struct mm_struct *mm, *active_mm;
1285
1286/* task state */
William Cohen97dc32c2007-05-08 00:23:41 -07001287 int exit_state;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001288 int exit_code, exit_signal;
1289 int pdeath_signal; /* The signal sent when the parent dies */
1290 /* ??? */
William Cohen97dc32c2007-05-08 00:23:41 -07001291 unsigned int personality;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001292 unsigned did_exec:1;
Kentaro Takedaf9ce1f12009-02-05 17:18:11 +09001293 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1294 * execve */
Arjan van de Ven8f0dfc32009-07-20 11:26:58 -07001295 unsigned in_iowait:1;
1296
Lennart Poetteringca94c442009-06-15 17:17:47 +02001297
1298 /* Revert to default priority/policy when forking */
1299 unsigned sched_reset_on_fork:1;
1300
Linus Torvalds1da177e2005-04-16 15:20:36 -07001301 pid_t pid;
1302 pid_t tgid;
Arjan van de Ven0a425402006-09-26 10:52:38 +02001303
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001304#ifdef CONFIG_CC_STACKPROTECTOR
Arjan van de Ven0a425402006-09-26 10:52:38 +02001305 /* Canary value for the -fstack-protector gcc feature */
1306 unsigned long stack_canary;
Hiroshi Shimamoto13145622009-08-18 15:06:02 +09001307#endif
Ingo Molnare0032082008-02-14 08:48:23 +01001308
Linus Torvalds1da177e2005-04-16 15:20:36 -07001309 /*
1310 * pointers to (original) parent process, youngest child, younger sibling,
1311 * older sibling, respectively. (p->father can be replaced with
Roland McGrathf4700212008-03-24 18:36:23 -07001312 * p->real_parent->pid)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001313 */
Roland McGrathf4700212008-03-24 18:36:23 -07001314 struct task_struct *real_parent; /* real parent process */
1315 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001316 /*
Roland McGrathf4700212008-03-24 18:36:23 -07001317 * children/sibling forms the list of my natural children
Linus Torvalds1da177e2005-04-16 15:20:36 -07001318 */
1319 struct list_head children; /* list of my children */
1320 struct list_head sibling; /* linkage in my parent's children list */
1321 struct task_struct *group_leader; /* threadgroup leader */
1322
Roland McGrathf4700212008-03-24 18:36:23 -07001323 /*
1324 * ptraced is the list of tasks this task is using ptrace on.
1325 * This includes both natural children and PTRACE_ATTACH targets.
1326 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1327 */
1328 struct list_head ptraced;
1329 struct list_head ptrace_entry;
1330
Markus Metzgerca0002a2008-11-25 09:01:25 +01001331 /*
1332 * This is the tracer handle for the ptrace BTS extension.
1333 * This field actually belongs to the ptracer task.
1334 */
Markus Metzgere2b371f2009-04-03 16:43:35 +02001335 struct bts_context *bts;
Markus Metzgerca0002a2008-11-25 09:01:25 +01001336
Linus Torvalds1da177e2005-04-16 15:20:36 -07001337 /* PID/PID hash table linkage. */
Eric W. Biederman92476d72006-03-31 02:31:42 -08001338 struct pid_link pids[PIDTYPE_MAX];
Oleg Nesterov47e65322006-03-28 16:11:25 -08001339 struct list_head thread_group;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001340
1341 struct completion *vfork_done; /* for vfork() */
1342 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1343 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1344
Michael Neulingc66f08b2007-10-18 03:06:34 -07001345 cputime_t utime, stime, utimescaled, stimescaled;
Laurent Vivier9ac52312007-10-15 17:00:19 +02001346 cputime_t gtime;
Hidetoshi Setod99ca3b2009-12-02 17:26:47 +09001347#ifndef CONFIG_VIRT_CPU_ACCOUNTING
Balbir Singh93018992007-10-30 00:26:32 +01001348 cputime_t prev_utime, prev_stime;
Hidetoshi Setod99ca3b2009-12-02 17:26:47 +09001349#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001350 unsigned long nvcsw, nivcsw; /* context switch counts */
Tomas Janousek924b42d2007-07-15 23:39:42 -07001351 struct timespec start_time; /* monotonic time */
1352 struct timespec real_start_time; /* boot based time */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001353/* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1354 unsigned long min_flt, maj_flt;
1355
Frank Mayharf06febc2008-09-12 09:54:39 -07001356 struct task_cputime cputime_expires;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001357 struct list_head cpu_timers[3];
1358
1359/* process credentials */
David Howells3b11a1d2008-11-14 10:39:26 +11001360 const struct cred *real_cred; /* objective and real subjective task
1361 * credentials (COW) */
1362 const struct cred *cred; /* effective (overridable) subjective task
1363 * credentials (COW) */
David Howells5e751e92009-05-08 13:55:22 +01001364 struct mutex cred_guard_mutex; /* guard against foreign influences on
1365 * credential calculations
1366 * (notably. ptrace) */
David Howellsee18d642009-09-02 09:14:21 +01001367 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
David Howellsb6dff3e2008-11-14 10:39:16 +11001368
Paolo 'Blaisorblade' Giarrusso36772092005-05-05 16:16:12 -07001369 char comm[TASK_COMM_LEN]; /* executable name excluding path
1370 - access with [gs]et_task_comm (which lock
1371 it with task_lock())
1372 - initialized normally by flush_old_exec */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001373/* file system info */
1374 int link_count, total_link_count;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001375#ifdef CONFIG_SYSVIPC
Linus Torvalds1da177e2005-04-16 15:20:36 -07001376/* ipc stuff */
1377 struct sysv_sem sysvsem;
Alexey Dobriyan3d5b6fc2006-09-29 01:59:40 -07001378#endif
Mandeep Singh Bainese162b392009-01-15 11:08:40 -08001379#ifdef CONFIG_DETECT_HUNG_TASK
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001380/* hung task detection */
Ingo Molnar82a1fcb2008-01-25 21:08:02 +01001381 unsigned long last_switch_count;
1382#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001383/* CPU-specific state of this task */
1384 struct thread_struct thread;
1385/* filesystem information */
1386 struct fs_struct *fs;
1387/* open file information */
1388 struct files_struct *files;
Serge E. Hallyn1651e142006-10-02 02:18:08 -07001389/* namespaces */
Serge E. Hallynab516012006-10-02 02:18:06 -07001390 struct nsproxy *nsproxy;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001391/* signal handlers */
1392 struct signal_struct *signal;
1393 struct sighand_struct *sighand;
1394
1395 sigset_t blocked, real_blocked;
Roland McGrathf3de2722008-04-30 00:53:09 -07001396 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001397 struct sigpending pending;
1398
1399 unsigned long sas_ss_sp;
1400 size_t sas_ss_size;
1401 int (*notifier)(void *priv);
1402 void *notifier_data;
1403 sigset_t *notifier_mask;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001404 struct audit_context *audit_context;
Al Virobfef93a2008-01-10 04:53:18 -05001405#ifdef CONFIG_AUDITSYSCALL
1406 uid_t loginuid;
Eric Paris4746ec52008-01-08 10:06:53 -05001407 unsigned int sessionid;
Al Virobfef93a2008-01-10 04:53:18 -05001408#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001409 seccomp_t seccomp;
1410
1411/* Thread group tracking */
1412 u32 parent_exec_id;
1413 u32 self_exec_id;
Miao Xie58568d22009-06-16 15:31:49 -07001414/* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1415 * mempolicy */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001416 spinlock_t alloc_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001417
Thomas Gleixner3aa551c2009-03-23 18:28:15 +01001418#ifdef CONFIG_GENERIC_HARDIRQS
1419 /* IRQ handler threads */
1420 struct irqaction *irqaction;
1421#endif
1422
Ingo Molnarb29739f2006-06-27 02:54:51 -07001423 /* Protection of the PI data structures: */
Thomas Gleixner1d615482009-11-17 14:54:03 +01001424 raw_spinlock_t pi_lock;
Ingo Molnarb29739f2006-06-27 02:54:51 -07001425
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001426#ifdef CONFIG_RT_MUTEXES
1427 /* PI waiters blocked on a rt_mutex held by this task */
1428 struct plist_head pi_waiters;
1429 /* Deadlock detection and priority inheritance handling */
1430 struct rt_mutex_waiter *pi_blocked_on;
Ingo Molnar23f78d4a2006-06-27 02:54:53 -07001431#endif
1432
Ingo Molnar408894e2006-01-09 15:59:20 -08001433#ifdef CONFIG_DEBUG_MUTEXES
1434 /* mutex deadlock detection */
1435 struct mutex_waiter *blocked_on;
1436#endif
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001437#ifdef CONFIG_TRACE_IRQFLAGS
1438 unsigned int irq_events;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001439 unsigned long hardirq_enable_ip;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001440 unsigned long hardirq_disable_ip;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001441 unsigned int hardirq_enable_event;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001442 unsigned int hardirq_disable_event;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001443 int hardirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001444 int hardirq_context;
Hiroshi Shimamotofa1452e2009-11-30 14:59:44 +09001445 unsigned long softirq_disable_ip;
1446 unsigned long softirq_enable_ip;
1447 unsigned int softirq_disable_event;
1448 unsigned int softirq_enable_event;
1449 int softirqs_enabled;
Ingo Molnarde30a2b2006-07-03 00:24:42 -07001450 int softirq_context;
1451#endif
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001452#ifdef CONFIG_LOCKDEP
Peter Zijlstrabdb94412008-02-25 23:02:48 +01001453# define MAX_LOCK_DEPTH 48UL
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001454 u64 curr_chain_key;
1455 int lockdep_depth;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001456 unsigned int lockdep_recursion;
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001457 struct held_lock held_locks[MAX_LOCK_DEPTH];
Nick Piggincf40bd12009-01-21 08:12:39 +01001458 gfp_t lockdep_reclaim_gfp;
Ingo Molnarfbb9ce952006-07-03 00:24:50 -07001459#endif
Ingo Molnar408894e2006-01-09 15:59:20 -08001460
Linus Torvalds1da177e2005-04-16 15:20:36 -07001461/* journalling filesystem info */
1462 void *journal_info;
1463
Neil Brownd89d8792007-05-01 09:53:42 +02001464/* stacked block device info */
1465 struct bio *bio_list, **bio_tail;
1466
Linus Torvalds1da177e2005-04-16 15:20:36 -07001467/* VM state */
1468 struct reclaim_state *reclaim_state;
1469
Linus Torvalds1da177e2005-04-16 15:20:36 -07001470 struct backing_dev_info *backing_dev_info;
1471
1472 struct io_context *io_context;
1473
1474 unsigned long ptrace_message;
1475 siginfo_t *last_siginfo; /* For ptrace use. */
Andrew Morton7c3ab732006-12-10 02:19:19 -08001476 struct task_io_accounting ioac;
Jay Lan8f0ab512006-09-30 23:28:59 -07001477#if defined(CONFIG_TASK_XACCT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001478 u64 acct_rss_mem1; /* accumulated rss usage */
1479 u64 acct_vm_mem1; /* accumulated virtual memory usage */
Jonathan Lim49b5cf32008-07-25 01:48:40 -07001480 cputime_t acct_timexpd; /* stime + utime since last update */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001481#endif
1482#ifdef CONFIG_CPUSETS
Miao Xie58568d22009-06-16 15:31:49 -07001483 nodemask_t mems_allowed; /* Protected by alloc_lock */
Paul Jackson825a46a2006-03-24 03:16:03 -08001484 int cpuset_mem_spread_rotor;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001485#endif
Paul Menageddbcc7e2007-10-18 23:39:30 -07001486#ifdef CONFIG_CGROUPS
Paul Menage817929e2007-10-18 23:39:36 -07001487 /* Control Group info protected by css_set_lock */
1488 struct css_set *cgroups;
1489 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1490 struct list_head cg_list;
Paul Menageddbcc7e2007-10-18 23:39:30 -07001491#endif
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001492#ifdef CONFIG_FUTEX
Ingo Molnar0771dfe2006-03-27 01:16:22 -08001493 struct robust_list_head __user *robust_list;
Ingo Molnar34f192c2006-03-27 01:16:24 -08001494#ifdef CONFIG_COMPAT
1495 struct compat_robust_list_head __user *compat_robust_list;
1496#endif
Ingo Molnarc87e2832006-06-27 02:54:58 -07001497 struct list_head pi_state_list;
1498 struct futex_pi_state *pi_state_cache;
Alexey Dobriyan42b2dd02007-10-16 23:27:30 -07001499#endif
Ingo Molnarcdd6c482009-09-21 12:02:48 +02001500#ifdef CONFIG_PERF_EVENTS
1501 struct perf_event_context *perf_event_ctxp;
1502 struct mutex perf_event_mutex;
1503 struct list_head perf_event_list;
Paul Mackerrasa63eaf32009-05-22 14:17:31 +10001504#endif
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001505#ifdef CONFIG_NUMA
Miao Xie58568d22009-06-16 15:31:49 -07001506 struct mempolicy *mempolicy; /* Protected by alloc_lock */
Richard Kennedyc7aceab2008-05-15 12:09:15 +01001507 short il_next;
1508#endif
Jens Axboe22e2c502005-06-27 10:55:12 +02001509 atomic_t fs_excl; /* holding fs exclusive resources */
Ingo Molnare56d0902006-01-08 01:01:37 -08001510 struct rcu_head rcu;
Jens Axboeb92ce552006-04-11 13:52:07 +02001511
1512 /*
1513 * cache last used pipe for splice
1514 */
1515 struct pipe_inode_info *splice_pipe;
Shailabh Nagarca74e922006-07-14 00:24:36 -07001516#ifdef CONFIG_TASK_DELAY_ACCT
1517 struct task_delay_info *delays;
1518#endif
Akinobu Mitaf4f154f2006-12-08 02:39:47 -08001519#ifdef CONFIG_FAULT_INJECTION
1520 int make_it_fail;
1521#endif
Peter Zijlstra3e26c142007-10-16 23:25:50 -07001522 struct prop_local_single dirties;
Arjan van de Ven97455122008-01-25 21:08:34 +01001523#ifdef CONFIG_LATENCYTOP
1524 int latency_record_count;
1525 struct latency_record latency_record[LT_SAVECOUNT];
1526#endif
Arjan van de Ven69766752008-09-01 15:52:40 -07001527 /*
1528 * time slack values; these are used to round up poll() and
1529 * select() etc timeout values. These are in nanoseconds.
1530 */
1531 unsigned long timer_slack_ns;
1532 unsigned long default_timer_slack_ns;
David Millerf8d570a2008-11-06 00:37:40 -08001533
1534 struct list_head *scm_work_list;
Frederic Weisbeckerfb526072008-11-25 21:07:04 +01001535#ifdef CONFIG_FUNCTION_GRAPH_TRACER
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001536 /* Index of current stored adress in ret_stack */
1537 int curr_ret_stack;
1538 /* Stack of return addresses for return function tracing */
1539 struct ftrace_ret_stack *ret_stack;
Steven Rostedt8aef2d22009-03-24 01:10:15 -04001540 /* time stamp for last schedule */
1541 unsigned long long ftrace_timestamp;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001542 /*
1543 * Number of functions that haven't been traced
1544 * because of depth overrun.
1545 */
1546 atomic_t trace_overrun;
Frederic Weisbecker380c4b12008-12-06 03:43:41 +01001547 /* Pause for the tracing */
1548 atomic_t tracing_graph_pause;
Frederic Weisbeckerf201ae22008-11-23 06:22:56 +01001549#endif
Steven Rostedtea4e2bc2008-12-03 15:36:57 -05001550#ifdef CONFIG_TRACING
1551 /* state flags for use by tracers */
1552 unsigned long trace;
Steven Rostedt261842b2009-04-16 21:41:52 -04001553 /* bitmask of trace recursion */
1554 unsigned long trace_recursion;
1555#endif /* CONFIG_TRACING */
Stefani Seiboldd899bf72009-09-22 16:45:40 -07001556 unsigned long stack_start;
KAMEZAWA Hiroyuki569b8462009-12-15 16:47:03 -08001557#ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1558 struct memcg_batch_info {
1559 int do_batch; /* incremented when batch uncharge started */
1560 struct mem_cgroup *memcg; /* target memcg of uncharge */
1561 unsigned long bytes; /* uncharged usage */
1562 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1563 } memcg_batch;
1564#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001565};
1566
Rusty Russell76e6eee2009-03-12 14:35:43 -06001567/* Future-safe accessor for struct task_struct's cpus_allowed. */
Rusty Russella4636812009-12-17 11:43:29 -06001568#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
Rusty Russell76e6eee2009-03-12 14:35:43 -06001569
Ingo Molnare05606d2007-07-09 18:51:59 +02001570/*
1571 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1572 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1573 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1574 * values are inverted: lower p->prio value means higher priority.
1575 *
1576 * The MAX_USER_RT_PRIO value allows the actual maximum
1577 * RT priority to be separate from the value exported to
1578 * user-space. This allows kernel threads to set their
1579 * priority to a value higher than any user task. Note:
1580 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1581 */
1582
1583#define MAX_USER_RT_PRIO 100
1584#define MAX_RT_PRIO MAX_USER_RT_PRIO
1585
1586#define MAX_PRIO (MAX_RT_PRIO + 40)
1587#define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1588
1589static inline int rt_prio(int prio)
1590{
1591 if (unlikely(prio < MAX_RT_PRIO))
1592 return 1;
1593 return 0;
1594}
1595
Alexey Dobriyane8681712007-10-26 12:17:22 +04001596static inline int rt_task(struct task_struct *p)
Ingo Molnare05606d2007-07-09 18:51:59 +02001597{
1598 return rt_prio(p->prio);
1599}
1600
Alexey Dobriyane8681712007-10-26 12:17:22 +04001601static inline struct pid *task_pid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001602{
1603 return task->pids[PIDTYPE_PID].pid;
1604}
1605
Alexey Dobriyane8681712007-10-26 12:17:22 +04001606static inline struct pid *task_tgid(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001607{
1608 return task->group_leader->pids[PIDTYPE_PID].pid;
1609}
1610
Oleg Nesterov6dda81f2009-04-02 16:58:35 -07001611/*
1612 * Without tasklist or rcu lock it is not safe to dereference
1613 * the result of task_pgrp/task_session even if task == current,
1614 * we can race with another thread doing sys_setsid/sys_setpgid.
1615 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001616static inline struct pid *task_pgrp(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001617{
1618 return task->group_leader->pids[PIDTYPE_PGID].pid;
1619}
1620
Alexey Dobriyane8681712007-10-26 12:17:22 +04001621static inline struct pid *task_session(struct task_struct *task)
Eric W. Biederman22c935f2006-10-02 02:17:09 -07001622{
1623 return task->group_leader->pids[PIDTYPE_SID].pid;
1624}
1625
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001626struct pid_namespace;
1627
1628/*
1629 * the helpers to get the task's different pids as they are seen
1630 * from various namespaces
1631 *
1632 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
Eric W. Biederman44c4e1b2008-02-08 04:19:15 -08001633 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1634 * current.
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001635 * task_xid_nr_ns() : id seen from the ns specified;
1636 *
1637 * set_task_vxid() : assigns a virtual id to a task;
1638 *
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001639 * see also pid_nr() etc in include/linux/pid.h
1640 */
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001641pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1642 struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001643
Alexey Dobriyane8681712007-10-26 12:17:22 +04001644static inline pid_t task_pid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001645{
1646 return tsk->pid;
1647}
1648
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001649static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1650 struct pid_namespace *ns)
1651{
1652 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1653}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001654
1655static inline pid_t task_pid_vnr(struct task_struct *tsk)
1656{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001657 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001658}
1659
1660
Alexey Dobriyane8681712007-10-26 12:17:22 +04001661static inline pid_t task_tgid_nr(struct task_struct *tsk)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001662{
1663 return tsk->tgid;
1664}
1665
Pavel Emelyanov2f2a3a42007-10-18 23:40:19 -07001666pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001667
1668static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1669{
1670 return pid_vnr(task_tgid(tsk));
1671}
1672
1673
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001674static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1675 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001676{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001677 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001678}
1679
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001680static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1681{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001682 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001683}
1684
1685
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001686static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1687 struct pid_namespace *ns)
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001688{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001689 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001690}
1691
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001692static inline pid_t task_session_vnr(struct task_struct *tsk)
1693{
Oleg Nesterov52ee2df2009-04-02 16:58:38 -07001694 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001695}
1696
Oleg Nesterov1b0f7ff2009-04-02 16:58:39 -07001697/* obsolete, do not use */
1698static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1699{
1700 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1701}
Pavel Emelyanov7af57292007-10-18 23:40:06 -07001702
Linus Torvalds1da177e2005-04-16 15:20:36 -07001703/**
1704 * pid_alive - check that a task structure is not stale
1705 * @p: Task structure to be checked.
1706 *
1707 * Test if a process is not yet dead (at most zombie state)
1708 * If pid_alive fails, then pointers within the task structure
1709 * can be stale and must not be dereferenced.
1710 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001711static inline int pid_alive(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001712{
Eric W. Biederman92476d72006-03-31 02:31:42 -08001713 return p->pids[PIDTYPE_PID].pid != NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001714}
1715
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001716/**
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001717 * is_global_init - check if a task structure is init
Henne32602592006-10-06 00:44:01 -07001718 * @tsk: Task structure to be checked.
1719 *
1720 * Check if a task structure is the first user space task the kernel created.
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001721 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04001722static inline int is_global_init(struct task_struct *tsk)
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07001723{
1724 return tsk->pid == 1;
1725}
Serge E. Hallynb460cbc2007-10-18 23:39:52 -07001726
1727/*
1728 * is_container_init:
1729 * check whether in the task is init in its own pid namespace.
1730 */
Pavel Emelyanovb461cc02007-10-18 23:40:09 -07001731extern int is_container_init(struct task_struct *tsk);
Sukadev Bhattiproluf400e192006-09-29 02:00:07 -07001732
Cedric Le Goater9ec52092006-10-02 02:19:00 -07001733extern struct pid *cad_pid;
1734
Linus Torvalds1da177e2005-04-16 15:20:36 -07001735extern void free_task(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001736#define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
Ingo Molnare56d0902006-01-08 01:01:37 -08001737
Andrew Morton158d9eb2006-03-31 02:31:34 -08001738extern void __put_task_struct(struct task_struct *t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001739
1740static inline void put_task_struct(struct task_struct *t)
1741{
1742 if (atomic_dec_and_test(&t->usage))
Eric W. Biederman8c7904a2006-03-31 02:31:37 -08001743 __put_task_struct(t);
Ingo Molnare56d0902006-01-08 01:01:37 -08001744}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001745
Hidetoshi Setod180c5b2009-11-26 14:48:30 +09001746extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
Hidetoshi Seto0cf55e12009-12-02 17:28:07 +09001747extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
Balbir Singh49048622008-09-05 18:12:23 +02001748
Linus Torvalds1da177e2005-04-16 15:20:36 -07001749/*
1750 * Per process flags
1751 */
1752#define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1753 /* Not implemented yet, only for 486*/
1754#define PF_STARTING 0x00000002 /* being created */
1755#define PF_EXITING 0x00000004 /* getting shut down */
Alexey Kuznetsov778e9a92007-06-08 13:47:00 -07001756#define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
Laurent Vivier94886b82007-10-15 17:00:19 +02001757#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001758#define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001759#define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001760#define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1761#define PF_DUMPCORE 0x00000200 /* dumped core */
1762#define PF_SIGNALED 0x00000400 /* killed by a signal */
1763#define PF_MEMALLOC 0x00000800 /* Allocating memory */
1764#define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1765#define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
Thomas Gleixner6301cb92009-07-17 14:15:47 +02001766#define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001767#define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1768#define PF_FROZEN 0x00010000 /* frozen for system suspend */
1769#define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1770#define PF_KSWAPD 0x00040000 /* I am kswapd */
Hugh Dickins35451be2009-09-21 17:02:27 -07001771#define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001772#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
Oleg Nesterov246bb0b2008-07-25 01:47:38 -07001773#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
Jens Axboeb31dc662006-06-13 08:26:10 +02001774#define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1775#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1776#define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1777#define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
David Rientjes9985b0b2008-06-05 12:57:11 -07001778#define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
Andi Kleen4db96cf2009-09-16 11:50:14 +02001779#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
Paul Jacksonc61afb12006-03-24 03:16:08 -08001780#define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
Thomas Gleixner61a87122006-06-27 02:54:56 -07001781#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
Rafael J. Wysockiba96a0c2007-05-23 13:57:25 -07001782#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
Rafael J. Wysockiebb12db2008-06-11 22:04:29 +02001783#define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001784
1785/*
1786 * Only the _current_ task can read/write to tsk->flags, but other
1787 * tasks can access tsk->flags in readonly mode for example
1788 * with tsk_used_math (like during threaded core dumping).
1789 * There is however an exception to this rule during ptrace
1790 * or during fork: the ptracer task is allowed to write to the
1791 * child->flags of its traced child (same goes for fork, the parent
1792 * can write to the child->flags), because we're guaranteed the
1793 * child is not running and in turn not changing child->flags
1794 * at the same time the parent does it.
1795 */
1796#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1797#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1798#define clear_used_math() clear_stopped_child_used_math(current)
1799#define set_used_math() set_stopped_child_used_math(current)
1800#define conditional_stopped_child_used_math(condition, child) \
1801 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1802#define conditional_used_math(condition) \
1803 conditional_stopped_child_used_math(condition, current)
1804#define copy_to_stopped_child_used_math(child) \
1805 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1806/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1807#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1808#define used_math() tsk_used_math(current)
1809
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001810#ifdef CONFIG_TREE_PREEMPT_RCU
1811
1812#define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1813#define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001814
1815static inline void rcu_copy_process(struct task_struct *p)
1816{
1817 p->rcu_read_lock_nesting = 0;
1818 p->rcu_read_unlock_special = 0;
Paul E. McKenneydd5d19b2009-08-27 14:58:16 -07001819 p->rcu_blocked_node = NULL;
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001820 INIT_LIST_HEAD(&p->rcu_node_entry);
1821}
1822
Paul E. McKenneyf41d9112009-08-22 13:56:52 -07001823#else
1824
1825static inline void rcu_copy_process(struct task_struct *p)
1826{
1827}
1828
1829#endif
1830
Linus Torvalds1da177e2005-04-16 15:20:36 -07001831#ifdef CONFIG_SMP
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001832extern int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10301833 const struct cpumask *new_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001834#else
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001835static inline int set_cpus_allowed_ptr(struct task_struct *p,
Rusty Russell96f874e2008-11-25 02:35:14 +10301836 const struct cpumask *new_mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001837{
Rusty Russell96f874e2008-11-25 02:35:14 +10301838 if (!cpumask_test_cpu(0, new_mask))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001839 return -EINVAL;
1840 return 0;
1841}
1842#endif
Rusty Russelle0ad9552009-09-24 09:34:38 -06001843
1844#ifndef CONFIG_CPUMASK_OFFSTACK
Mike Traviscd8ba7c2008-03-26 14:23:49 -07001845static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1846{
1847 return set_cpus_allowed_ptr(p, &new_mask);
1848}
Rusty Russelle0ad9552009-09-24 09:34:38 -06001849#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001850
Ingo Molnarb3425012009-02-26 20:20:29 +01001851/*
1852 * Architectures can set this to 1 if they have specified
1853 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1854 * but then during bootup it turns out that sched_clock()
1855 * is reliable after all:
1856 */
1857#ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1858extern int sched_clock_stable;
1859#endif
1860
Mike Frysinger1bbfa6f2009-12-09 20:07:03 -05001861/* ftrace calls sched_clock() directly */
1862extern unsigned long long notrace sched_clock(void);
Ingo Molnare436d802007-07-19 21:28:35 +02001863
Peter Zijlstrac1955a32008-08-11 08:59:03 +02001864extern void sched_clock_init(void);
1865extern u64 sched_clock_cpu(int cpu);
1866
Peter Zijlstra3e51f332008-05-03 18:29:28 +02001867#ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
Peter Zijlstra3e51f332008-05-03 18:29:28 +02001868static inline void sched_clock_tick(void)
1869{
1870}
1871
1872static inline void sched_clock_idle_sleep_event(void)
1873{
1874}
1875
1876static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1877{
1878}
1879#else
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
Ingo Molnare436d802007-07-19 21:28:35 +02001885/*
1886 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1887 * clock constructed from sched_clock():
1888 */
1889extern unsigned long long cpu_clock(int cpu);
1890
Ingo Molnar36c8b582006-07-03 00:25:41 -07001891extern unsigned long long
Ingo Molnar41b86e92007-07-09 18:51:58 +02001892task_sched_runtime(struct task_struct *task);
Frank Mayharf06febc2008-09-12 09:54:39 -07001893extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001894
1895/* sched_exec is called by processes performing an exec */
1896#ifdef CONFIG_SMP
1897extern void sched_exec(void);
1898#else
1899#define sched_exec() {}
1900#endif
1901
Ingo Molnar2aa44d02007-08-23 15:18:02 +02001902extern void sched_clock_idle_sleep_event(void);
1903extern void sched_clock_idle_wakeup_event(u64 delta_ns);
Ingo Molnarbb29ab22007-07-09 18:51:59 +02001904
Linus Torvalds1da177e2005-04-16 15:20:36 -07001905#ifdef CONFIG_HOTPLUG_CPU
1906extern void idle_task_exit(void);
1907#else
1908static inline void idle_task_exit(void) {}
1909#endif
1910
1911extern void sched_idle_next(void);
Ingo Molnarb29739f2006-06-27 02:54:51 -07001912
Thomas Gleixner06d83082008-03-22 09:20:24 +01001913#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1914extern void wake_up_idle_cpu(int cpu);
1915#else
1916static inline void wake_up_idle_cpu(int cpu) { }
1917#endif
1918
Peter Zijlstra21805082007-08-25 18:41:53 +02001919extern unsigned int sysctl_sched_latency;
Peter Zijlstrab2be5e92007-11-09 22:39:37 +01001920extern unsigned int sysctl_sched_min_granularity;
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02001921extern unsigned int sysctl_sched_wakeup_granularity;
Jaswinder Singh Rajput47fea2a2008-12-29 23:39:17 +05301922extern unsigned int sysctl_sched_shares_ratelimit;
1923extern unsigned int sysctl_sched_shares_thresh;
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02001924extern unsigned int sysctl_sched_child_runs_first;
Christian Ehrhardt1983a922009-11-30 12:16:47 +01001925
1926enum sched_tunable_scaling {
1927 SCHED_TUNABLESCALING_NONE,
1928 SCHED_TUNABLESCALING_LOG,
1929 SCHED_TUNABLESCALING_LINEAR,
1930 SCHED_TUNABLESCALING_END,
1931};
1932extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1933
Mike Galbraith2bba22c2009-09-09 15:41:37 +02001934#ifdef CONFIG_SCHED_DEBUG
Ingo Molnarda84d962007-10-15 17:00:18 +02001935extern unsigned int sysctl_sched_migration_cost;
Peter Zijlstrab82d9fd2007-11-09 22:39:39 +01001936extern unsigned int sysctl_sched_nr_migrate;
Peter Zijlstrae9e92502009-09-01 10:34:37 +02001937extern unsigned int sysctl_sched_time_avg;
Arun R Bharadwajcd1bb942009-04-16 12:15:34 +05301938extern unsigned int sysctl_timer_migration;
Peter Zijlstrab2be5e92007-11-09 22:39:37 +01001939
Christian Ehrhardt1983a922009-11-30 12:16:47 +01001940int sched_proc_update_handler(struct ctl_table *table, int write,
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07001941 void __user *buffer, size_t *length,
Peter Zijlstrab2be5e92007-11-09 22:39:37 +01001942 loff_t *ppos);
Ingo Molnar2bd8e6d2007-10-15 17:00:02 +02001943#endif
Arun R Bharadwajeea08f32009-04-16 12:16:41 +05301944#ifdef CONFIG_SCHED_DEBUG
1945static inline unsigned int get_sysctl_timer_migration(void)
1946{
1947 return sysctl_timer_migration;
1948}
1949#else
1950static inline unsigned int get_sysctl_timer_migration(void)
1951{
1952 return 1;
1953}
1954#endif
Peter Zijlstra9f0c1e52008-02-13 15:45:39 +01001955extern unsigned int sysctl_sched_rt_period;
1956extern int sysctl_sched_rt_runtime;
Ingo Molnar2bd8e6d2007-10-15 17:00:02 +02001957
Peter Zijlstrad0b27fa2008-04-19 19:44:57 +02001958int sched_rt_handler(struct ctl_table *table, int write,
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07001959 void __user *buffer, size_t *lenp,
Peter Zijlstrad0b27fa2008-04-19 19:44:57 +02001960 loff_t *ppos);
1961
Ingo Molnar2bd8e6d2007-10-15 17:00:02 +02001962extern unsigned int sysctl_sched_compat_yield;
Ingo Molnarbf0f6f22007-07-09 18:51:58 +02001963
Ingo Molnarb29739f2006-06-27 02:54:51 -07001964#ifdef CONFIG_RT_MUTEXES
Ingo Molnar36c8b582006-07-03 00:25:41 -07001965extern int rt_mutex_getprio(struct task_struct *p);
1966extern void rt_mutex_setprio(struct task_struct *p, int prio);
1967extern void rt_mutex_adjust_pi(struct task_struct *p);
Ingo Molnarb29739f2006-06-27 02:54:51 -07001968#else
Alexey Dobriyane8681712007-10-26 12:17:22 +04001969static inline int rt_mutex_getprio(struct task_struct *p)
Ingo Molnarb29739f2006-06-27 02:54:51 -07001970{
1971 return p->normal_prio;
1972}
Thomas Gleixner95e02ca2006-06-27 02:55:02 -07001973# define rt_mutex_adjust_pi(p) do { } while (0)
Ingo Molnarb29739f2006-06-27 02:54:51 -07001974#endif
1975
Ingo Molnar36c8b582006-07-03 00:25:41 -07001976extern void set_user_nice(struct task_struct *p, long nice);
1977extern int task_prio(const struct task_struct *p);
1978extern int task_nice(const struct task_struct *p);
1979extern int can_nice(const struct task_struct *p, const int nice);
1980extern int task_curr(const struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001981extern int idle_cpu(int cpu);
1982extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
Rusty Russell961ccdd2008-06-23 13:55:38 +10001983extern int sched_setscheduler_nocheck(struct task_struct *, int,
1984 struct sched_param *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07001985extern struct task_struct *idle_task(int cpu);
1986extern struct task_struct *curr_task(int cpu);
1987extern void set_curr_task(int cpu, struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001988
1989void yield(void);
1990
1991/*
1992 * The default (Linux) execution domain.
1993 */
1994extern struct exec_domain default_exec_domain;
1995
1996union thread_union {
1997 struct thread_info thread_info;
1998 unsigned long stack[THREAD_SIZE/sizeof(long)];
1999};
2000
2001#ifndef __HAVE_ARCH_KSTACK_END
2002static inline int kstack_end(void *addr)
2003{
2004 /* Reliable end of stack detection:
2005 * Some APM bios versions misalign the stack
2006 */
2007 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2008}
2009#endif
2010
2011extern union thread_union init_thread_union;
2012extern struct task_struct init_task;
2013
2014extern struct mm_struct init_mm;
2015
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002016extern struct pid_namespace init_pid_ns;
2017
2018/*
2019 * find a task by one of its numerical ids
2020 *
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002021 * find_task_by_pid_ns():
2022 * finds a task by its pid in the specified namespace
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002023 * find_task_by_vpid():
2024 * finds a task by its virtual pid
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002025 *
Pavel Emelyanove49859e2008-07-25 01:48:36 -07002026 * see also find_vpid() etc in include/linux/pid.h
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002027 */
2028
Pavel Emelyanov228ebcb2007-10-18 23:40:16 -07002029extern struct task_struct *find_task_by_vpid(pid_t nr);
2030extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2031 struct pid_namespace *ns);
Pavel Emelyanov198fe212007-10-18 23:40:06 -07002032
Oleg Nesterov8520d7c2008-02-08 04:19:09 -08002033extern void __set_special_pids(struct pid *pid);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002034
2035/* per-UID process charging. */
Cedric Le Goateracce2922007-07-15 23:40:59 -07002036extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002037static inline struct user_struct *get_uid(struct user_struct *u)
2038{
2039 atomic_inc(&u->__count);
2040 return u;
2041}
2042extern void free_uid(struct user_struct *);
Pavel Emelyanov28f300d2007-09-18 22:46:45 -07002043extern void release_uids(struct user_namespace *ns);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002044
2045#include <asm/current.h>
2046
Atsushi Nemoto3171a032006-09-29 02:00:32 -07002047extern void do_timer(unsigned long ticks);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002048
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002049extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2050extern int wake_up_process(struct task_struct *tsk);
2051extern void wake_up_new_task(struct task_struct *tsk,
2052 unsigned long clone_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002053#ifdef CONFIG_SMP
2054 extern void kick_process(struct task_struct *tsk);
2055#else
2056 static inline void kick_process(struct task_struct *tsk) { }
2057#endif
Ingo Molnarad46c2c2007-07-09 18:52:00 +02002058extern void sched_fork(struct task_struct *p, int clone_flags);
2059extern void sched_dead(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002060
Linus Torvalds1da177e2005-04-16 15:20:36 -07002061extern void proc_caches_init(void);
2062extern void flush_signals(struct task_struct *);
David Howells3bcac022009-04-29 13:45:05 +01002063extern void __flush_signals(struct task_struct *);
Oleg Nesterov10ab8252007-05-09 02:34:37 -07002064extern void ignore_signals(struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002065extern void flush_signal_handlers(struct task_struct *, int force_default);
2066extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2067
2068static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2069{
2070 unsigned long flags;
2071 int ret;
2072
2073 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2074 ret = dequeue_signal(tsk, mask, info);
2075 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2076
2077 return ret;
2078}
2079
2080extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2081 sigset_t *mask);
2082extern void unblock_all_signals(void);
2083extern void release_task(struct task_struct * p);
2084extern int send_sig_info(int, struct siginfo *, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002085extern int force_sigsegv(int, struct task_struct *);
2086extern int force_sig_info(int, struct siginfo *, struct task_struct *);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002087extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002088extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
Eric W. Biederman2425c082006-10-02 02:17:28 -07002089extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
Eric W. Biedermanc4b92fc2006-10-02 02:17:10 -07002090extern int kill_pgrp(struct pid *pid, int sig, int priv);
2091extern int kill_pid(struct pid *pid, int sig, int priv);
Matthew Wilcoxc3de4b32007-02-09 08:11:47 -07002092extern int kill_proc_info(int, struct siginfo *, pid_t);
Roland McGrath2b2a1ff2008-07-25 19:45:54 -07002093extern int do_notify_parent(struct task_struct *, int);
Oleg Nesterova7f07652009-09-23 15:56:44 -07002094extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002095extern void force_sig(int, struct task_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002096extern int send_sig(int, struct task_struct *, int);
2097extern void zap_other_threads(struct task_struct *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002098extern struct sigqueue *sigqueue_alloc(void);
2099extern void sigqueue_free(struct sigqueue *);
Oleg Nesterovac5c2152008-04-30 00:52:57 -07002100extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
Oleg Nesterov9ac95f22006-02-09 22:41:50 +03002101extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002102extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2103
Cedric Le Goater9ec52092006-10-02 02:19:00 -07002104static inline int kill_cad_pid(int sig, int priv)
2105{
2106 return kill_pid(cad_pid, sig, priv);
2107}
2108
Linus Torvalds1da177e2005-04-16 15:20:36 -07002109/* These can be the second arg to send_sig_info/send_group_sig_info. */
2110#define SEND_SIG_NOINFO ((struct siginfo *) 0)
2111#define SEND_SIG_PRIV ((struct siginfo *) 1)
2112#define SEND_SIG_FORCED ((struct siginfo *) 2)
2113
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002114/*
2115 * True if we are on the alternate signal stack.
2116 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002117static inline int on_sig_stack(unsigned long sp)
2118{
Sebastian Andrzej Siewior2a855dd2009-10-25 15:37:58 +01002119#ifdef CONFIG_STACK_GROWSUP
2120 return sp >= current->sas_ss_sp &&
2121 sp - current->sas_ss_sp < current->sas_ss_size;
2122#else
2123 return sp > current->sas_ss_sp &&
2124 sp - current->sas_ss_sp <= current->sas_ss_size;
2125#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002126}
2127
2128static inline int sas_ss_flags(unsigned long sp)
2129{
2130 return (current->sas_ss_size == 0 ? SS_DISABLE
2131 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2132}
2133
Linus Torvalds1da177e2005-04-16 15:20:36 -07002134/*
2135 * Routines for handling mm_structs
2136 */
2137extern struct mm_struct * mm_alloc(void);
2138
2139/* mmdrop drops the mm and the page tables */
Harvey Harrisonb3c97522008-02-13 15:03:15 -08002140extern void __mmdrop(struct mm_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002141static inline void mmdrop(struct mm_struct * mm)
2142{
Ingo Molnar6fb43d72007-07-09 18:52:01 +02002143 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002144 __mmdrop(mm);
2145}
2146
2147/* mmput gets rid of the mappings and all user-space */
2148extern void mmput(struct mm_struct *);
2149/* Grab a reference to a task's mm, if it is not already going away */
2150extern struct mm_struct *get_task_mm(struct task_struct *task);
2151/* Remove the current tasks stale references to the old mm_struct */
2152extern void mm_release(struct task_struct *, struct mm_struct *);
Carsten Otte402b0862008-03-25 18:47:10 +01002153/* Allocate a new mm structure and copy contents from tsk->mm */
2154extern struct mm_struct *dup_mm(struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002155
Alexey Dobriyan6f2c55b2009-04-02 16:56:59 -07002156extern int copy_thread(unsigned long, unsigned long, unsigned long,
2157 struct task_struct *, struct pt_regs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002158extern void flush_thread(void);
2159extern void exit_thread(void);
2160
Linus Torvalds1da177e2005-04-16 15:20:36 -07002161extern void exit_files(struct task_struct *);
Oleg Nesterov6b3934e2006-03-28 16:11:16 -08002162extern void __cleanup_signal(struct signal_struct *);
Oleg Nesterova7e53282006-03-28 16:11:27 -08002163extern void __cleanup_sighand(struct sighand_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002164
Linus Torvalds1da177e2005-04-16 15:20:36 -07002165extern void exit_itimers(struct signal_struct *);
Oleg Nesterovcbaffba2008-05-26 20:55:42 +04002166extern void flush_itimer_signals(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002167
2168extern NORET_TYPE void do_group_exit(int);
2169
Linus Torvalds1da177e2005-04-16 15:20:36 -07002170extern void daemonize(const char *, ...);
2171extern int allow_signal(int);
2172extern int disallow_signal(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002173
2174extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2175extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
Ingo Molnar36c8b582006-07-03 00:25:41 -07002176struct task_struct *fork_idle(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002177
2178extern void set_task_comm(struct task_struct *tsk, char *from);
Andrew Morton59714d62008-02-04 22:27:21 -08002179extern char *get_task_comm(char *to, struct task_struct *tsk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002180
2181#ifdef CONFIG_SMP
Markus Metzgera26b89f2009-04-03 16:43:34 +02002182extern void wait_task_context_switch(struct task_struct *p);
Roland McGrath85ba2d82008-07-25 19:45:58 -07002183extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002184#else
Markus Metzgera26b89f2009-04-03 16:43:34 +02002185static inline void wait_task_context_switch(struct task_struct *p) {}
Roland McGrath85ba2d82008-07-25 19:45:58 -07002186static inline unsigned long wait_task_inactive(struct task_struct *p,
2187 long match_state)
2188{
2189 return 1;
2190}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002191#endif
2192
Jiri Pirko05725f72009-04-14 20:17:16 +02002193#define next_task(p) \
2194 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002195
2196#define for_each_process(p) \
2197 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2198
Oleg Nesterov5bb459b2009-07-10 03:48:23 +02002199extern bool current_is_single_threaded(void);
David Howellsd84f4f92008-11-14 10:39:23 +11002200
Linus Torvalds1da177e2005-04-16 15:20:36 -07002201/*
2202 * Careful: do_each_thread/while_each_thread is a double loop so
2203 * 'break' will not work as expected - use goto instead.
2204 */
2205#define do_each_thread(g, t) \
2206 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2207
2208#define while_each_thread(g, t) \
2209 while ((t = next_thread(t)) != g)
2210
Eric W. Biedermande12a782006-04-10 17:16:49 -06002211/* de_thread depends on thread_group_leader not being a pid based check */
2212#define thread_group_leader(p) (p == p->group_leader)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002213
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002214/* Do to the insanities of de_thread it is possible for a process
2215 * to have the pid of the thread group leader without actually being
2216 * the thread group leader. For iteration through the pids in proc
2217 * all we care about is that we have a task with the appropriate
2218 * pid, we don't actually care if we have the right task.
2219 */
Alexey Dobriyane8681712007-10-26 12:17:22 +04002220static inline int has_group_leader_pid(struct task_struct *p)
Eric W. Biederman0804ef42006-10-02 02:17:04 -07002221{
2222 return p->pid == p->tgid;
2223}
2224
Pavel Emelyanovbac0abd2007-10-18 23:40:18 -07002225static inline
2226int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2227{
2228 return p1->tgid == p2->tgid;
2229}
2230
Ingo Molnar36c8b582006-07-03 00:25:41 -07002231static inline struct task_struct *next_thread(const struct task_struct *p)
Oleg Nesterov47e65322006-03-28 16:11:25 -08002232{
Jiri Pirko05725f72009-04-14 20:17:16 +02002233 return list_entry_rcu(p->thread_group.next,
2234 struct task_struct, thread_group);
Oleg Nesterov47e65322006-03-28 16:11:25 -08002235}
2236
Alexey Dobriyane8681712007-10-26 12:17:22 +04002237static inline int thread_group_empty(struct task_struct *p)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002238{
Oleg Nesterov47e65322006-03-28 16:11:25 -08002239 return list_empty(&p->thread_group);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002240}
2241
2242#define delay_group_leader(p) \
2243 (thread_group_leader(p) && !thread_group_empty(p))
2244
Oleg Nesterov39c626a2009-04-02 16:58:18 -07002245static inline int task_detached(struct task_struct *p)
2246{
2247 return p->exit_signal == -1;
2248}
2249
Linus Torvalds1da177e2005-04-16 15:20:36 -07002250/*
Eric W. Biederman260ea102006-06-23 02:05:18 -07002251 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
Jens Axboe22e2c502005-06-27 10:55:12 +02002252 * subscriptions and synchronises with wait4(). Also used in procfs. Also
Paul Menageddbcc7e2007-10-18 23:39:30 -07002253 * pins the final release of task.io_context. Also protects ->cpuset and
2254 * ->cgroup.subsys[].
Linus Torvalds1da177e2005-04-16 15:20:36 -07002255 *
2256 * Nests both inside and outside of read_lock(&tasklist_lock).
2257 * It must not be nested with write_lock_irq(&tasklist_lock),
2258 * neither inside nor outside.
2259 */
2260static inline void task_lock(struct task_struct *p)
2261{
2262 spin_lock(&p->alloc_lock);
2263}
2264
2265static inline void task_unlock(struct task_struct *p)
2266{
2267 spin_unlock(&p->alloc_lock);
2268}
2269
Oleg Nesterovf63ee722006-03-28 16:11:13 -08002270extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2271 unsigned long *flags);
2272
2273static inline void unlock_task_sighand(struct task_struct *tsk,
2274 unsigned long *flags)
2275{
2276 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2277}
2278
Al Virof0373602005-11-13 16:06:57 -08002279#ifndef __HAVE_THREAD_FUNCTIONS
2280
Roman Zippelf7e42172007-05-09 02:35:17 -07002281#define task_thread_info(task) ((struct thread_info *)(task)->stack)
2282#define task_stack_page(task) ((task)->stack)
Al Viroa1261f52005-11-13 16:06:55 -08002283
Al Viro10ebffd2005-11-13 16:06:56 -08002284static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2285{
2286 *task_thread_info(p) = *task_thread_info(org);
2287 task_thread_info(p)->task = p;
2288}
2289
2290static inline unsigned long *end_of_stack(struct task_struct *p)
2291{
Roman Zippelf7e42172007-05-09 02:35:17 -07002292 return (unsigned long *)(task_thread_info(p) + 1);
Al Viro10ebffd2005-11-13 16:06:56 -08002293}
2294
Al Virof0373602005-11-13 16:06:57 -08002295#endif
2296
FUJITA Tomonori8b05c7e2008-07-23 21:26:53 -07002297static inline int object_is_on_stack(void *obj)
2298{
2299 void *stack = task_stack_page(current);
2300
2301 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2302}
2303
Benjamin Herrenschmidt8c9843e2008-04-18 16:56:15 +10002304extern void thread_info_cache_init(void);
2305
Eric Sandeen7c9f8862008-04-22 16:38:23 -05002306#ifdef CONFIG_DEBUG_STACK_USAGE
2307static inline unsigned long stack_not_used(struct task_struct *p)
2308{
2309 unsigned long *n = end_of_stack(p);
2310
2311 do { /* Skip over canary */
2312 n++;
2313 } while (!*n);
2314
2315 return (unsigned long)n - (unsigned long)end_of_stack(p);
2316}
2317#endif
2318
Linus Torvalds1da177e2005-04-16 15:20:36 -07002319/* set thread flags in other task's structures
2320 * - see asm/thread_info.h for TIF_xxxx flags available
2321 */
2322static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2323{
Al Viroa1261f52005-11-13 16:06:55 -08002324 set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002325}
2326
2327static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2328{
Al Viroa1261f52005-11-13 16:06:55 -08002329 clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002330}
2331
2332static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2333{
Al Viroa1261f52005-11-13 16:06:55 -08002334 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002335}
2336
2337static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2338{
Al Viroa1261f52005-11-13 16:06:55 -08002339 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002340}
2341
2342static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2343{
Al Viroa1261f52005-11-13 16:06:55 -08002344 return test_ti_thread_flag(task_thread_info(tsk), flag);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002345}
2346
2347static inline void set_tsk_need_resched(struct task_struct *tsk)
2348{
2349 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2350}
2351
2352static inline void clear_tsk_need_resched(struct task_struct *tsk)
2353{
2354 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2355}
2356
Gregory Haskins8ae121a2008-04-23 07:13:29 -04002357static inline int test_tsk_need_resched(struct task_struct *tsk)
2358{
2359 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2360}
2361
Eric W. Biederman690cc3f2009-05-13 16:55:10 +00002362static inline int restart_syscall(void)
2363{
2364 set_tsk_thread_flag(current, TIF_SIGPENDING);
2365 return -ERESTARTNOINTR;
2366}
2367
Linus Torvalds1da177e2005-04-16 15:20:36 -07002368static inline int signal_pending(struct task_struct *p)
2369{
2370 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2371}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002372
Roland McGrathd9588722009-09-23 15:57:04 -07002373static inline int __fatal_signal_pending(struct task_struct *p)
2374{
2375 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2376}
Matthew Wilcoxf776d122007-12-06 11:15:50 -05002377
2378static inline int fatal_signal_pending(struct task_struct *p)
2379{
2380 return signal_pending(p) && __fatal_signal_pending(p);
2381}
2382
Oleg Nesterov16882c12008-06-08 21:20:41 +04002383static inline int signal_pending_state(long state, struct task_struct *p)
2384{
2385 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2386 return 0;
2387 if (!signal_pending(p))
2388 return 0;
2389
Oleg Nesterov16882c12008-06-08 21:20:41 +04002390 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2391}
2392
Linus Torvalds1da177e2005-04-16 15:20:36 -07002393static inline int need_resched(void)
2394{
Linus Torvalds9404ef02008-05-12 10:14:22 -07002395 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
Linus Torvalds1da177e2005-04-16 15:20:36 -07002396}
2397
2398/*
2399 * cond_resched() and cond_resched_lock(): latency reduction via
2400 * explicit rescheduling in places that are safe. The return
2401 * value indicates whether a reschedule was done in fact.
2402 * cond_resched_lock() will drop the spinlock before scheduling,
2403 * cond_resched_softirq() will enable bhs before scheduling.
2404 */
Linus Torvaldsc3921ab2008-05-11 16:04:48 -07002405extern int _cond_resched(void);
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002406
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002407#define cond_resched() ({ \
2408 __might_sleep(__FILE__, __LINE__, 0); \
2409 _cond_resched(); \
2410})
Frederic Weisbecker6f80bd92009-07-16 15:44:29 +02002411
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002412extern int __cond_resched_lock(spinlock_t *lock);
2413
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002414#ifdef CONFIG_PREEMPT
2415#define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
Herbert Xu02b67cc32008-01-25 21:08:28 +01002416#else
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002417#define PREEMPT_LOCK_OFFSET 0
Herbert Xu02b67cc32008-01-25 21:08:28 +01002418#endif
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002419
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002420#define cond_resched_lock(lock) ({ \
Frederic Weisbecker716a4232009-07-24 20:05:23 +02002421 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
Frederic Weisbecker613afbf2009-07-16 15:44:29 +02002422 __cond_resched_lock(lock); \
2423})
2424
2425extern int __cond_resched_softirq(void);
2426
2427#define cond_resched_softirq() ({ \
2428 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2429 __cond_resched_softirq(); \
2430})
Linus Torvalds1da177e2005-04-16 15:20:36 -07002431
2432/*
2433 * Does a critical section need to be broken due to another
Nick Piggin95c354f2008-01-30 13:31:20 +01002434 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2435 * but a general need for low latency)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002436 */
Nick Piggin95c354f2008-01-30 13:31:20 +01002437static inline int spin_needbreak(spinlock_t *lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002438{
Nick Piggin95c354f2008-01-30 13:31:20 +01002439#ifdef CONFIG_PREEMPT
2440 return spin_is_contended(lock);
2441#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07002442 return 0;
Nick Piggin95c354f2008-01-30 13:31:20 +01002443#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002444}
2445
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002446/*
Frank Mayharf06febc2008-09-12 09:54:39 -07002447 * Thread group CPU time accounting.
2448 */
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +01002449void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
Peter Zijlstra4da94d492009-02-11 11:30:27 +01002450void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
Frank Mayharf06febc2008-09-12 09:54:39 -07002451
2452static inline void thread_group_cputime_init(struct signal_struct *sig)
2453{
Peter Zijlstra4cd4c1b2009-02-05 12:24:16 +01002454 sig->cputimer.cputime = INIT_CPUTIME;
2455 spin_lock_init(&sig->cputimer.lock);
2456 sig->cputimer.running = 0;
Frank Mayharf06febc2008-09-12 09:54:39 -07002457}
2458
2459static inline void thread_group_cputime_free(struct signal_struct *sig)
2460{
Frank Mayharf06febc2008-09-12 09:54:39 -07002461}
2462
Frank Mayharf06febc2008-09-12 09:54:39 -07002463/*
Roland McGrath7bb44ad2007-05-23 13:57:44 -07002464 * Reevaluate whether the task has signals pending delivery.
2465 * Wake the task if so.
2466 * This is required every time the blocked sigset_t changes.
2467 * callers must hold sighand->siglock.
2468 */
2469extern void recalc_sigpending_and_wake(struct task_struct *t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002470extern void recalc_sigpending(void);
2471
2472extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2473
2474/*
2475 * Wrappers for p->thread_info->cpu access. No-op on UP.
2476 */
2477#ifdef CONFIG_SMP
2478
2479static inline unsigned int task_cpu(const struct task_struct *p)
2480{
Al Viroa1261f52005-11-13 16:06:55 -08002481 return task_thread_info(p)->cpu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002482}
2483
Ingo Molnarc65cc872007-07-09 18:51:58 +02002484extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002485
2486#else
2487
2488static inline unsigned int task_cpu(const struct task_struct *p)
2489{
2490 return 0;
2491}
2492
2493static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2494{
2495}
2496
2497#endif /* CONFIG_SMP */
2498
Ingo Molnar1a3c3032008-05-12 21:20:52 +02002499#ifdef CONFIG_TRACING
2500extern void
2501__trace_special(void *__tr, void *__data,
2502 unsigned long arg1, unsigned long arg2, unsigned long arg3);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002503#else
Ingo Molnar1a3c3032008-05-12 21:20:52 +02002504static inline void
2505__trace_special(void *__tr, void *__data,
2506 unsigned long arg1, unsigned long arg2, unsigned long arg3)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002507{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002508}
2509#endif
2510
Rusty Russell96f874e2008-11-25 02:35:14 +10302511extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2512extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
Siddha, Suresh B5c45bf22006-06-27 02:54:42 -07002513
Linus Torvalds1da177e2005-04-16 15:20:36 -07002514extern void normalize_rt_tasks(void);
2515
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01002516#ifdef CONFIG_GROUP_SCHED
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002517
Ingo Molnar4cf86d72007-10-15 17:00:14 +02002518extern struct task_group init_task_group;
Peter Zijlstraeff766a2008-04-19 19:45:00 +02002519#ifdef CONFIG_USER_SCHED
2520extern struct task_group root_task_group;
Arun R Bharadwaj6c415b92008-12-01 20:49:05 +05302521extern void set_tg_uid(struct user_struct *user);
Peter Zijlstraeff766a2008-04-19 19:45:00 +02002522#endif
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002523
Dhaval Gianiec7dc8a2008-04-19 19:44:59 +02002524extern struct task_group *sched_create_group(struct task_group *parent);
Ingo Molnar4cf86d72007-10-15 17:00:14 +02002525extern void sched_destroy_group(struct task_group *tg);
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002526extern void sched_move_task(struct task_struct *tsk);
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01002527#ifdef CONFIG_FAIR_GROUP_SCHED
Ingo Molnar4cf86d72007-10-15 17:00:14 +02002528extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
Dhaval Giani5cb350b2007-10-15 17:00:14 +02002529extern unsigned long sched_group_shares(struct task_group *tg);
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01002530#endif
2531#ifdef CONFIG_RT_GROUP_SCHED
Peter Zijlstra9f0c1e52008-02-13 15:45:39 +01002532extern int sched_group_set_rt_runtime(struct task_group *tg,
2533 long rt_runtime_us);
2534extern long sched_group_rt_runtime(struct task_group *tg);
Peter Zijlstrad0b27fa2008-04-19 19:44:57 +02002535extern int sched_group_set_rt_period(struct task_group *tg,
2536 long rt_period_us);
2537extern long sched_group_rt_period(struct task_group *tg);
Dhaval Giani54e99122009-02-27 15:13:54 +05302538extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
Peter Zijlstra052f1dc2008-02-13 15:45:40 +01002539#endif
Srivatsa Vaddagiri9b5b7752007-10-15 17:00:09 +02002540#endif
2541
Dhaval Giani54e99122009-02-27 15:13:54 +05302542extern int task_can_switch_user(struct user_struct *up,
2543 struct task_struct *tsk);
2544
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002545#ifdef CONFIG_TASK_XACCT
2546static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2547{
Andrea Righi940389b2008-07-28 00:48:12 +02002548 tsk->ioac.rchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002549}
2550
2551static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2552{
Andrea Righi940389b2008-07-28 00:48:12 +02002553 tsk->ioac.wchar += amt;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002554}
2555
2556static inline void inc_syscr(struct task_struct *tsk)
2557{
Andrea Righi940389b2008-07-28 00:48:12 +02002558 tsk->ioac.syscr++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002559}
2560
2561static inline void inc_syscw(struct task_struct *tsk)
2562{
Andrea Righi940389b2008-07-28 00:48:12 +02002563 tsk->ioac.syscw++;
Alexey Dobriyan4b98d112007-02-10 01:46:45 -08002564}
2565#else
2566static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2567{
2568}
2569
2570static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2571{
2572}
2573
2574static inline void inc_syscr(struct task_struct *tsk)
2575{
2576}
2577
2578static inline void inc_syscw(struct task_struct *tsk)
2579{
2580}
2581#endif
2582
Dave Hansen82455252008-02-04 22:28:59 -08002583#ifndef TASK_SIZE_OF
2584#define TASK_SIZE_OF(tsk) TASK_SIZE
2585#endif
2586
Thomas Gleixner0793a612008-12-04 20:12:29 +01002587/*
2588 * Call the function if the target task is executing on a CPU right now:
2589 */
2590extern void task_oncpu_function_call(struct task_struct *p,
2591 void (*func) (void *info), void *info);
2592
2593
Balbir Singhcf475ad2008-04-29 01:00:16 -07002594#ifdef CONFIG_MM_OWNER
2595extern void mm_update_next_owner(struct mm_struct *mm);
2596extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2597#else
2598static inline void mm_update_next_owner(struct mm_struct *mm)
2599{
2600}
2601
2602static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2603{
2604}
2605#endif /* CONFIG_MM_OWNER */
2606
Jiri Slaby3e10e712009-11-19 17:16:37 +01002607static inline unsigned long task_rlimit(const struct task_struct *tsk,
2608 unsigned int limit)
2609{
2610 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2611}
2612
2613static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2614 unsigned int limit)
2615{
2616 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2617}
2618
2619static inline unsigned long rlimit(unsigned int limit)
2620{
2621 return task_rlimit(current, limit);
2622}
2623
2624static inline unsigned long rlimit_max(unsigned int limit)
2625{
2626 return task_rlimit_max(current, limit);
2627}
2628
Linus Torvalds1da177e2005-04-16 15:20:36 -07002629#endif /* __KERNEL__ */
2630
2631#endif