sched.h 92 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427242824292430243124322433243424352436243724382439244024412442244324442445244624472448244924502451245224532454245524562457245824592460246124622463246424652466246724682469247024712472247324742475247624772478247924802481248224832484248524862487248824892490249124922493249424952496249724982499250025012502250325042505250625072508250925102511251225132514251525162517251825192520252125222523252425252526252725282529253025312532253325342535253625372538253925402541254225432544254525462547254825492550255125522553255425552556255725582559256025612562256325642565256625672568256925702571257225732574257525762577257825792580258125822583258425852586258725882589259025912592259325942595259625972598259926002601260226032604260526062607260826092610261126122613261426152616261726182619262026212622262326242625262626272628262926302631263226332634263526362637263826392640264126422643264426452646264726482649265026512652265326542655265626572658265926602661266226632664266526662667266826692670267126722673267426752676267726782679268026812682268326842685268626872688268926902691269226932694269526962697269826992700270127022703270427052706270727082709271027112712271327142715271627172718271927202721272227232724272527262727272827292730273127322733273427352736273727382739274027412742274327442745274627472748274927502751275227532754275527562757275827592760276127622763276427652766276727682769277027712772277327742775277627772778277927802781278227832784278527862787278827892790279127922793279427952796279727982799280028012802280328042805280628072808280928102811281228132814281528162817281828192820282128222823282428252826282728282829283028312832283328342835283628372838283928402841284228432844284528462847284828492850285128522853285428552856285728582859286028612862286328642865286628672868286928702871287228732874287528762877287828792880288128822883288428852886288728882889289028912892289328942895289628972898289929002901290229032904290529062907290829092910291129122913291429152916291729182919292029212922292329242925292629272928292929302931293229332934293529362937293829392940294129422943294429452946294729482949295029512952295329542955295629572958295929602961296229632964296529662967296829692970297129722973297429752976297729782979298029812982298329842985298629872988298929902991299229932994299529962997299829993000300130023003300430053006300730083009301030113012301330143015301630173018301930203021302230233024302530263027302830293030303130323033303430353036303730383039304030413042304330443045304630473048304930503051305230533054305530563057305830593060306130623063306430653066306730683069307030713072307330743075307630773078307930803081308230833084308530863087308830893090309130923093309430953096309730983099310031013102310331043105310631073108310931103111311231133114311531163117311831193120312131223123312431253126312731283129313031313132313331343135313631373138313931403141314231433144314531463147314831493150315131523153315431553156315731583159316031613162316331643165316631673168316931703171317231733174317531763177317831793180318131823183318431853186318731883189319031913192319331943195319631973198319932003201320232033204320532063207320832093210321132123213321432153216321732183219322032213222322332243225322632273228322932303231323232333234323532363237323832393240324132423243324432453246
  1. #ifndef _LINUX_SCHED_H
  2. #define _LINUX_SCHED_H
  3. #include <uapi/linux/sched.h>
  4. #include <linux/sched/prio.h>
  5. struct sched_param {
  6. int sched_priority;
  7. };
  8. #include <asm/param.h> /* for HZ */
  9. #include <linux/capability.h>
  10. #include <linux/threads.h>
  11. #include <linux/kernel.h>
  12. #include <linux/types.h>
  13. #include <linux/timex.h>
  14. #include <linux/jiffies.h>
  15. #include <linux/plist.h>
  16. #include <linux/rbtree.h>
  17. #include <linux/thread_info.h>
  18. #include <linux/cpumask.h>
  19. #include <linux/errno.h>
  20. #include <linux/nodemask.h>
  21. #include <linux/mm_types.h>
  22. #include <linux/preempt.h>
  23. #include <asm/page.h>
  24. #include <asm/ptrace.h>
  25. #include <linux/cputime.h>
  26. #include <linux/smp.h>
  27. #include <linux/sem.h>
  28. #include <linux/shm.h>
  29. #include <linux/signal.h>
  30. #include <linux/compiler.h>
  31. #include <linux/completion.h>
  32. #include <linux/pid.h>
  33. #include <linux/percpu.h>
  34. #include <linux/topology.h>
  35. #include <linux/proportions.h>
  36. #include <linux/seccomp.h>
  37. #include <linux/rcupdate.h>
  38. #include <linux/rculist.h>
  39. #include <linux/rtmutex.h>
  40. #include <linux/time.h>
  41. #include <linux/param.h>
  42. #include <linux/resource.h>
  43. #include <linux/timer.h>
  44. #include <linux/hrtimer.h>
  45. #include <linux/kcov.h>
  46. #include <linux/task_io_accounting.h>
  47. #include <linux/latencytop.h>
  48. #include <linux/cred.h>
  49. #include <linux/llist.h>
  50. #include <linux/uidgid.h>
  51. #include <linux/gfp.h>
  52. #include <linux/magic.h>
  53. #include <linux/cgroup-defs.h>
  54. #include <asm/processor.h>
  55. #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
  56. /*
  57. * Extended scheduling parameters data structure.
  58. *
  59. * This is needed because the original struct sched_param can not be
  60. * altered without introducing ABI issues with legacy applications
  61. * (e.g., in sched_getparam()).
  62. *
  63. * However, the possibility of specifying more than just a priority for
  64. * the tasks may be useful for a wide variety of application fields, e.g.,
  65. * multimedia, streaming, automation and control, and many others.
  66. *
  67. * This variant (sched_attr) is meant at describing a so-called
  68. * sporadic time-constrained task. In such model a task is specified by:
  69. * - the activation period or minimum instance inter-arrival time;
  70. * - the maximum (or average, depending on the actual scheduling
  71. * discipline) computation time of all instances, a.k.a. runtime;
  72. * - the deadline (relative to the actual activation time) of each
  73. * instance.
  74. * Very briefly, a periodic (sporadic) task asks for the execution of
  75. * some specific computation --which is typically called an instance--
  76. * (at most) every period. Moreover, each instance typically lasts no more
  77. * than the runtime and must be completed by time instant t equal to
  78. * the instance activation time + the deadline.
  79. *
  80. * This is reflected by the actual fields of the sched_attr structure:
  81. *
  82. * @size size of the structure, for fwd/bwd compat.
  83. *
  84. * @sched_policy task's scheduling policy
  85. * @sched_flags for customizing the scheduler behaviour
  86. * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
  87. * @sched_priority task's static priority (SCHED_FIFO/RR)
  88. * @sched_deadline representative of the task's deadline
  89. * @sched_runtime representative of the task's runtime
  90. * @sched_period representative of the task's period
  91. *
  92. * Given this task model, there are a multiplicity of scheduling algorithms
  93. * and policies, that can be used to ensure all the tasks will make their
  94. * timing constraints.
  95. *
  96. * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
  97. * only user of this new interface. More information about the algorithm
  98. * available in the scheduling class file or in Documentation/.
  99. */
  100. struct sched_attr {
  101. u32 size;
  102. u32 sched_policy;
  103. u64 sched_flags;
  104. /* SCHED_NORMAL, SCHED_BATCH */
  105. s32 sched_nice;
  106. /* SCHED_FIFO, SCHED_RR */
  107. u32 sched_priority;
  108. /* SCHED_DEADLINE */
  109. u64 sched_runtime;
  110. u64 sched_deadline;
  111. u64 sched_period;
  112. };
  113. struct futex_pi_state;
  114. struct robust_list_head;
  115. struct bio_list;
  116. struct fs_struct;
  117. struct perf_event_context;
  118. struct blk_plug;
  119. struct filename;
  120. struct nameidata;
  121. #define VMACACHE_BITS 2
  122. #define VMACACHE_SIZE (1U << VMACACHE_BITS)
  123. #define VMACACHE_MASK (VMACACHE_SIZE - 1)
  124. /*
  125. * These are the constant used to fake the fixed-point load-average
  126. * counting. Some notes:
  127. * - 11 bit fractions expand to 22 bits by the multiplies: this gives
  128. * a load-average precision of 10 bits integer + 11 bits fractional
  129. * - if you want to count load-averages more often, you need more
  130. * precision, or rounding will get you. With 2-second counting freq,
  131. * the EXP_n values would be 1981, 2034 and 2043 if still using only
  132. * 11 bit fractions.
  133. */
  134. extern unsigned long avenrun[]; /* Load averages */
  135. extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
  136. #define FSHIFT 11 /* nr of bits of precision */
  137. #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
  138. #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
  139. #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
  140. #define EXP_5 2014 /* 1/exp(5sec/5min) */
  141. #define EXP_15 2037 /* 1/exp(5sec/15min) */
  142. #define CALC_LOAD(load,exp,n) \
  143. load *= exp; \
  144. load += n*(FIXED_1-exp); \
  145. load >>= FSHIFT;
  146. extern unsigned long total_forks;
  147. extern int nr_threads;
  148. DECLARE_PER_CPU(unsigned long, process_counts);
  149. extern int nr_processes(void);
  150. extern unsigned long nr_running(void);
  151. extern bool single_task_running(void);
  152. extern unsigned long nr_iowait(void);
  153. extern unsigned long nr_iowait_cpu(int cpu);
  154. extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
  155. extern void calc_global_load(unsigned long ticks);
  156. #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
  157. extern void update_cpu_load_nohz(int active);
  158. #else
  159. static inline void update_cpu_load_nohz(int active) { }
  160. #endif
  161. extern void dump_cpu_task(int cpu);
  162. struct seq_file;
  163. struct cfs_rq;
  164. struct task_group;
  165. #ifdef CONFIG_SCHED_DEBUG
  166. extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
  167. extern void proc_sched_set_task(struct task_struct *p);
  168. #endif
  169. /*
  170. * Task state bitmask. NOTE! These bits are also
  171. * encoded in fs/proc/array.c: get_task_state().
  172. *
  173. * We have two separate sets of flags: task->state
  174. * is about runnability, while task->exit_state are
  175. * about the task exiting. Confusing, but this way
  176. * modifying one set can't modify the other one by
  177. * mistake.
  178. */
  179. #define TASK_RUNNING 0
  180. #define TASK_INTERRUPTIBLE 1
  181. #define TASK_UNINTERRUPTIBLE 2
  182. #define __TASK_STOPPED 4
  183. #define __TASK_TRACED 8
  184. /* in tsk->exit_state */
  185. #define EXIT_DEAD 16
  186. #define EXIT_ZOMBIE 32
  187. #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
  188. /* in tsk->state again */
  189. #define TASK_DEAD 64
  190. #define TASK_WAKEKILL 128
  191. #define TASK_WAKING 256
  192. #define TASK_PARKED 512
  193. #define TASK_NOLOAD 1024
  194. #define TASK_STATE_MAX 2048
  195. #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
  196. extern char ___assert_task_state[1 - 2*!!(
  197. sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
  198. /* Convenience macros for the sake of set_task_state */
  199. #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
  200. #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
  201. #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
  202. #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
  203. /* Convenience macros for the sake of wake_up */
  204. #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
  205. #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
  206. /* get_task_state() */
  207. #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
  208. TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
  209. __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
  210. #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
  211. #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
  212. #define task_is_stopped_or_traced(task) \
  213. ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
  214. #define task_contributes_to_load(task) \
  215. ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
  216. (task->flags & PF_FROZEN) == 0 && \
  217. (task->state & TASK_NOLOAD) == 0)
  218. #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
  219. #define __set_task_state(tsk, state_value) \
  220. do { \
  221. (tsk)->task_state_change = _THIS_IP_; \
  222. (tsk)->state = (state_value); \
  223. } while (0)
  224. #define set_task_state(tsk, state_value) \
  225. do { \
  226. (tsk)->task_state_change = _THIS_IP_; \
  227. smp_store_mb((tsk)->state, (state_value)); \
  228. } while (0)
  229. /*
  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. */
  240. #define __set_current_state(state_value) \
  241. do { \
  242. current->task_state_change = _THIS_IP_; \
  243. current->state = (state_value); \
  244. } while (0)
  245. #define set_current_state(state_value) \
  246. do { \
  247. current->task_state_change = _THIS_IP_; \
  248. smp_store_mb(current->state, (state_value)); \
  249. } while (0)
  250. #else
  251. #define __set_task_state(tsk, state_value) \
  252. do { (tsk)->state = (state_value); } while (0)
  253. #define set_task_state(tsk, state_value) \
  254. smp_store_mb((tsk)->state, (state_value))
  255. /*
  256. * set_current_state() includes a barrier so that the write of current->state
  257. * is correctly serialised wrt the caller's subsequent test of whether to
  258. * actually sleep:
  259. *
  260. * set_current_state(TASK_UNINTERRUPTIBLE);
  261. * if (do_i_need_to_sleep())
  262. * schedule();
  263. *
  264. * If the caller does not need such serialisation then use __set_current_state()
  265. */
  266. #define __set_current_state(state_value) \
  267. do { current->state = (state_value); } while (0)
  268. #define set_current_state(state_value) \
  269. smp_store_mb(current->state, (state_value))
  270. #endif
  271. /* Task command name length */
  272. #define TASK_COMM_LEN 16
  273. #include <linux/spinlock.h>
  274. /*
  275. * This serializes "schedule()" and also protects
  276. * the run-queue from deletions/modifications (but
  277. * _adding_ to the beginning of the run-queue has
  278. * a separate lock).
  279. */
  280. extern rwlock_t tasklist_lock;
  281. extern spinlock_t mmlist_lock;
  282. struct task_struct;
  283. #ifdef CONFIG_PROVE_RCU
  284. extern int lockdep_tasklist_lock_is_held(void);
  285. #endif /* #ifdef CONFIG_PROVE_RCU */
  286. extern void sched_init(void);
  287. extern void sched_init_smp(void);
  288. extern asmlinkage void schedule_tail(struct task_struct *prev);
  289. extern void init_idle(struct task_struct *idle, int cpu);
  290. extern void init_idle_bootup_task(struct task_struct *idle);
  291. extern cpumask_var_t cpu_isolated_map;
  292. extern int runqueue_is_locked(int cpu);
  293. #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
  294. extern void nohz_balance_enter_idle(int cpu);
  295. extern void set_cpu_sd_state_idle(void);
  296. extern int get_nohz_timer_target(void);
  297. #else
  298. static inline void nohz_balance_enter_idle(int cpu) { }
  299. static inline void set_cpu_sd_state_idle(void) { }
  300. #endif
  301. /*
  302. * Only dump TASK_* tasks. (0 for all tasks)
  303. */
  304. extern void show_state_filter(unsigned long state_filter);
  305. static inline void show_state(void)
  306. {
  307. show_state_filter(0);
  308. }
  309. extern void show_regs(struct pt_regs *);
  310. /*
  311. * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
  312. * task), SP is the stack pointer of the first frame that should be shown in the back
  313. * trace (or NULL if the entire call-chain of the task should be shown).
  314. */
  315. extern void show_stack(struct task_struct *task, unsigned long *sp);
  316. extern void cpu_init (void);
  317. extern void trap_init(void);
  318. extern void update_process_times(int user);
  319. extern void scheduler_tick(void);
  320. extern void sched_show_task(struct task_struct *p);
  321. #ifdef CONFIG_LOCKUP_DETECTOR
  322. extern void touch_softlockup_watchdog_sched(void);
  323. extern void touch_softlockup_watchdog(void);
  324. extern void touch_softlockup_watchdog_sync(void);
  325. extern void touch_all_softlockup_watchdogs(void);
  326. extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
  327. void __user *buffer,
  328. size_t *lenp, loff_t *ppos);
  329. extern unsigned int softlockup_panic;
  330. extern unsigned int hardlockup_panic;
  331. void lockup_detector_init(void);
  332. #else
  333. static inline void touch_softlockup_watchdog_sched(void)
  334. {
  335. }
  336. static inline void touch_softlockup_watchdog(void)
  337. {
  338. }
  339. static inline void touch_softlockup_watchdog_sync(void)
  340. {
  341. }
  342. static inline void touch_all_softlockup_watchdogs(void)
  343. {
  344. }
  345. static inline void lockup_detector_init(void)
  346. {
  347. }
  348. #endif
  349. #ifdef CONFIG_DETECT_HUNG_TASK
  350. void reset_hung_task_detector(void);
  351. #else
  352. static inline void reset_hung_task_detector(void)
  353. {
  354. }
  355. #endif
  356. /* Attach to any functions which should be ignored in wchan output. */
  357. #define __sched __attribute__((__section__(".sched.text")))
  358. /* Linker adds these: start and end of __sched functions */
  359. extern char __sched_text_start[], __sched_text_end[];
  360. /* Is this address in the __sched functions? */
  361. extern int in_sched_functions(unsigned long addr);
  362. #define MAX_SCHEDULE_TIMEOUT LONG_MAX
  363. extern signed long schedule_timeout(signed long timeout);
  364. extern signed long schedule_timeout_interruptible(signed long timeout);
  365. extern signed long schedule_timeout_killable(signed long timeout);
  366. extern signed long schedule_timeout_uninterruptible(signed long timeout);
  367. extern signed long schedule_timeout_idle(signed long timeout);
  368. asmlinkage void schedule(void);
  369. extern void schedule_preempt_disabled(void);
  370. extern long io_schedule_timeout(long timeout);
  371. static inline void io_schedule(void)
  372. {
  373. io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
  374. }
  375. struct nsproxy;
  376. struct user_namespace;
  377. #ifdef CONFIG_MMU
  378. extern void arch_pick_mmap_layout(struct mm_struct *mm);
  379. extern unsigned long
  380. arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
  381. unsigned long, unsigned long);
  382. extern unsigned long
  383. arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
  384. unsigned long len, unsigned long pgoff,
  385. unsigned long flags);
  386. #else
  387. static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
  388. #endif
  389. #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
  390. #define SUID_DUMP_USER 1 /* Dump as user of process */
  391. #define SUID_DUMP_ROOT 2 /* Dump as root */
  392. /* mm flags */
  393. /* for SUID_DUMP_* above */
  394. #define MMF_DUMPABLE_BITS 2
  395. #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
  396. extern void set_dumpable(struct mm_struct *mm, int value);
  397. /*
  398. * This returns the actual value of the suid_dumpable flag. For things
  399. * that are using this for checking for privilege transitions, it must
  400. * test against SUID_DUMP_USER rather than treating it as a boolean
  401. * value.
  402. */
  403. static inline int __get_dumpable(unsigned long mm_flags)
  404. {
  405. return mm_flags & MMF_DUMPABLE_MASK;
  406. }
  407. static inline int get_dumpable(struct mm_struct *mm)
  408. {
  409. return __get_dumpable(mm->flags);
  410. }
  411. /* coredump filter bits */
  412. #define MMF_DUMP_ANON_PRIVATE 2
  413. #define MMF_DUMP_ANON_SHARED 3
  414. #define MMF_DUMP_MAPPED_PRIVATE 4
  415. #define MMF_DUMP_MAPPED_SHARED 5
  416. #define MMF_DUMP_ELF_HEADERS 6
  417. #define MMF_DUMP_HUGETLB_PRIVATE 7
  418. #define MMF_DUMP_HUGETLB_SHARED 8
  419. #define MMF_DUMP_DAX_PRIVATE 9
  420. #define MMF_DUMP_DAX_SHARED 10
  421. #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
  422. #define MMF_DUMP_FILTER_BITS 9
  423. #define MMF_DUMP_FILTER_MASK \
  424. (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
  425. #define MMF_DUMP_FILTER_DEFAULT \
  426. ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
  427. (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
  428. #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
  429. # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
  430. #else
  431. # define MMF_DUMP_MASK_DEFAULT_ELF 0
  432. #endif
  433. /* leave room for more dump flags */
  434. #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
  435. #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
  436. #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
  437. #define MMF_HAS_UPROBES 19 /* has uprobes */
  438. #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
  439. #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
  440. struct sighand_struct {
  441. atomic_t count;
  442. struct k_sigaction action[_NSIG];
  443. spinlock_t siglock;
  444. wait_queue_head_t signalfd_wqh;
  445. };
  446. struct pacct_struct {
  447. int ac_flag;
  448. long ac_exitcode;
  449. unsigned long ac_mem;
  450. cputime_t ac_utime, ac_stime;
  451. unsigned long ac_minflt, ac_majflt;
  452. };
  453. struct cpu_itimer {
  454. cputime_t expires;
  455. cputime_t incr;
  456. u32 error;
  457. u32 incr_error;
  458. };
  459. /**
  460. * struct prev_cputime - snaphsot of system and user cputime
  461. * @utime: time spent in user mode
  462. * @stime: time spent in system mode
  463. * @lock: protects the above two fields
  464. *
  465. * Stores previous user/system time values such that we can guarantee
  466. * monotonicity.
  467. */
  468. struct prev_cputime {
  469. #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
  470. cputime_t utime;
  471. cputime_t stime;
  472. raw_spinlock_t lock;
  473. #endif
  474. };
  475. static inline void prev_cputime_init(struct prev_cputime *prev)
  476. {
  477. #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
  478. prev->utime = prev->stime = 0;
  479. raw_spin_lock_init(&prev->lock);
  480. #endif
  481. }
  482. /**
  483. * struct task_cputime - collected CPU time counts
  484. * @utime: time spent in user mode, in &cputime_t units
  485. * @stime: time spent in kernel mode, in &cputime_t units
  486. * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
  487. *
  488. * This structure groups together three kinds of CPU time that are tracked for
  489. * threads and thread groups. Most things considering CPU time want to group
  490. * these counts together and treat all three of them in parallel.
  491. */
  492. struct task_cputime {
  493. cputime_t utime;
  494. cputime_t stime;
  495. unsigned long long sum_exec_runtime;
  496. };
  497. /* Alternate field names when used to cache expirations. */
  498. #define virt_exp utime
  499. #define prof_exp stime
  500. #define sched_exp sum_exec_runtime
  501. #define INIT_CPUTIME \
  502. (struct task_cputime) { \
  503. .utime = 0, \
  504. .stime = 0, \
  505. .sum_exec_runtime = 0, \
  506. }
  507. /*
  508. * This is the atomic variant of task_cputime, which can be used for
  509. * storing and updating task_cputime statistics without locking.
  510. */
  511. struct task_cputime_atomic {
  512. atomic64_t utime;
  513. atomic64_t stime;
  514. atomic64_t sum_exec_runtime;
  515. };
  516. #define INIT_CPUTIME_ATOMIC \
  517. (struct task_cputime_atomic) { \
  518. .utime = ATOMIC64_INIT(0), \
  519. .stime = ATOMIC64_INIT(0), \
  520. .sum_exec_runtime = ATOMIC64_INIT(0), \
  521. }
  522. #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
  523. /*
  524. * Disable preemption until the scheduler is running -- use an unconditional
  525. * value so that it also works on !PREEMPT_COUNT kernels.
  526. *
  527. * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
  528. */
  529. #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
  530. /*
  531. * Initial preempt_count value; reflects the preempt_count schedule invariant
  532. * which states that during context switches:
  533. *
  534. * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
  535. *
  536. * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
  537. * Note: See finish_task_switch().
  538. */
  539. #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
  540. /**
  541. * struct thread_group_cputimer - thread group interval timer counts
  542. * @cputime_atomic: atomic thread group interval timers.
  543. * @running: true when there are timers running and
  544. * @cputime_atomic receives updates.
  545. * @checking_timer: true when a thread in the group is in the
  546. * process of checking for thread group timers.
  547. *
  548. * This structure contains the version of task_cputime, above, that is
  549. * used for thread group CPU timer calculations.
  550. */
  551. struct thread_group_cputimer {
  552. struct task_cputime_atomic cputime_atomic;
  553. bool running;
  554. bool checking_timer;
  555. };
  556. #include <linux/rwsem.h>
  557. struct autogroup;
  558. /*
  559. * NOTE! "signal_struct" does not have its own
  560. * locking, because a shared signal_struct always
  561. * implies a shared sighand_struct, so locking
  562. * sighand_struct is always a proper superset of
  563. * the locking of signal_struct.
  564. */
  565. struct signal_struct {
  566. atomic_t sigcnt;
  567. atomic_t live;
  568. int nr_threads;
  569. struct list_head thread_head;
  570. wait_queue_head_t wait_chldexit; /* for wait4() */
  571. /* current thread group signal load-balancing target: */
  572. struct task_struct *curr_target;
  573. /* shared signal handling: */
  574. struct sigpending shared_pending;
  575. /* thread group exit support */
  576. int group_exit_code;
  577. /* overloaded:
  578. * - notify group_exit_task when ->count is equal to notify_count
  579. * - everyone except group_exit_task is stopped during signal delivery
  580. * of fatal signals, group_exit_task processes the signal.
  581. */
  582. int notify_count;
  583. struct task_struct *group_exit_task;
  584. /* thread group stop support, overloads group_exit_code too */
  585. int group_stop_count;
  586. unsigned int flags; /* see SIGNAL_* flags below */
  587. /*
  588. * PR_SET_CHILD_SUBREAPER marks a process, like a service
  589. * manager, to re-parent orphan (double-forking) child processes
  590. * to this process instead of 'init'. The service manager is
  591. * able to receive SIGCHLD signals and is able to investigate
  592. * the process until it calls wait(). All children of this
  593. * process will inherit a flag if they should look for a
  594. * child_subreaper process at exit.
  595. */
  596. unsigned int is_child_subreaper:1;
  597. unsigned int has_child_subreaper:1;
  598. /* POSIX.1b Interval Timers */
  599. int posix_timer_id;
  600. struct list_head posix_timers;
  601. /* ITIMER_REAL timer for the process */
  602. struct hrtimer real_timer;
  603. struct pid *leader_pid;
  604. ktime_t it_real_incr;
  605. /*
  606. * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
  607. * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
  608. * values are defined to 0 and 1 respectively
  609. */
  610. struct cpu_itimer it[2];
  611. /*
  612. * Thread group totals for process CPU timers.
  613. * See thread_group_cputimer(), et al, for details.
  614. */
  615. struct thread_group_cputimer cputimer;
  616. /* Earliest-expiration cache. */
  617. struct task_cputime cputime_expires;
  618. #ifdef CONFIG_NO_HZ_FULL
  619. unsigned long tick_dep_mask;
  620. #endif
  621. struct list_head cpu_timers[3];
  622. struct pid *tty_old_pgrp;
  623. /* boolean value for session group leader */
  624. int leader;
  625. struct tty_struct *tty; /* NULL if no tty */
  626. #ifdef CONFIG_SCHED_AUTOGROUP
  627. struct autogroup *autogroup;
  628. #endif
  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. */
  635. seqlock_t stats_lock;
  636. cputime_t utime, stime, cutime, cstime;
  637. cputime_t gtime;
  638. cputime_t cgtime;
  639. struct prev_cputime prev_cputime;
  640. unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
  641. unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
  642. unsigned long inblock, oublock, cinblock, coublock;
  643. unsigned long maxrss, cmaxrss;
  644. struct task_io_accounting ioac;
  645. /*
  646. * Cumulative ns of schedule CPU time fo dead threads in the
  647. * group, not including a zombie group leader, (This only differs
  648. * from jiffies_to_ns(utime + stime) if sched_clock uses something
  649. * other than jiffies.)
  650. */
  651. unsigned long long sum_sched_runtime;
  652. /*
  653. * We don't bother to synchronize most readers of this at all,
  654. * because there is no reader checking a limit that actually needs
  655. * to get both rlim_cur and rlim_max atomically, and either one
  656. * alone is a single word that can safely be read normally.
  657. * getrlimit/setrlimit use task_lock(current->group_leader) to
  658. * protect this instead of the siglock, because they really
  659. * have no need to disable irqs.
  660. */
  661. struct rlimit rlim[RLIM_NLIMITS];
  662. #ifdef CONFIG_BSD_PROCESS_ACCT
  663. struct pacct_struct pacct; /* per-process accounting information */
  664. #endif
  665. #ifdef CONFIG_TASKSTATS
  666. struct taskstats *stats;
  667. #endif
  668. #ifdef CONFIG_AUDIT
  669. unsigned audit_tty;
  670. struct tty_audit_buf *tty_audit_buf;
  671. #endif
  672. oom_flags_t oom_flags;
  673. short oom_score_adj; /* OOM kill score adjustment */
  674. short oom_score_adj_min; /* OOM kill score adjustment min value.
  675. * Only settable by CAP_SYS_RESOURCE. */
  676. struct mutex cred_guard_mutex; /* guard against foreign influences on
  677. * credential calculations
  678. * (notably. ptrace) */
  679. };
  680. /*
  681. * Bits in flags field of signal_struct.
  682. */
  683. #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
  684. #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
  685. #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
  686. #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
  687. /*
  688. * Pending notifications to parent.
  689. */
  690. #define SIGNAL_CLD_STOPPED 0x00000010
  691. #define SIGNAL_CLD_CONTINUED 0x00000020
  692. #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
  693. #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
  694. /* If true, all threads except ->group_exit_task have pending SIGKILL */
  695. static inline int signal_group_exit(const struct signal_struct *sig)
  696. {
  697. return (sig->flags & SIGNAL_GROUP_EXIT) ||
  698. (sig->group_exit_task != NULL);
  699. }
  700. /*
  701. * Some day this will be a full-fledged user tracking system..
  702. */
  703. struct user_struct {
  704. atomic_t __count; /* reference count */
  705. atomic_t processes; /* How many processes does this user have? */
  706. atomic_t sigpending; /* How many pending signals does this user have? */
  707. #ifdef CONFIG_INOTIFY_USER
  708. atomic_t inotify_watches; /* How many inotify watches does this user have? */
  709. atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
  710. #endif
  711. #ifdef CONFIG_FANOTIFY
  712. atomic_t fanotify_listeners;
  713. #endif
  714. #ifdef CONFIG_EPOLL
  715. atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
  716. #endif
  717. #ifdef CONFIG_POSIX_MQUEUE
  718. /* protected by mq_lock */
  719. unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
  720. #endif
  721. unsigned long locked_shm; /* How many pages of mlocked shm ? */
  722. unsigned long unix_inflight; /* How many files in flight in unix sockets */
  723. atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
  724. #ifdef CONFIG_KEYS
  725. struct key *uid_keyring; /* UID specific keyring */
  726. struct key *session_keyring; /* UID's default session keyring */
  727. #endif
  728. /* Hash table maintenance information */
  729. struct hlist_node uidhash_node;
  730. kuid_t uid;
  731. #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
  732. atomic_long_t locked_vm;
  733. #endif
  734. };
  735. extern int uids_sysfs_init(void);
  736. extern struct user_struct *find_user(kuid_t);
  737. extern struct user_struct root_user;
  738. #define INIT_USER (&root_user)
  739. struct backing_dev_info;
  740. struct reclaim_state;
  741. #ifdef CONFIG_SCHED_INFO
  742. struct sched_info {
  743. /* cumulative counters */
  744. unsigned long pcount; /* # of times run on this cpu */
  745. unsigned long long run_delay; /* time spent waiting on a runqueue */
  746. /* timestamps */
  747. unsigned long long last_arrival,/* when we last ran on a cpu */
  748. last_queued; /* when we were last queued to run */
  749. };
  750. #endif /* CONFIG_SCHED_INFO */
  751. #ifdef CONFIG_TASK_DELAY_ACCT
  752. struct task_delay_info {
  753. spinlock_t lock;
  754. unsigned int flags; /* Private per-task flags */
  755. /* For each stat XXX, add following, aligned appropriately
  756. *
  757. * struct timespec XXX_start, XXX_end;
  758. * u64 XXX_delay;
  759. * u32 XXX_count;
  760. *
  761. * Atomicity of updates to XXX_delay, XXX_count protected by
  762. * single lock above (split into XXX_lock if contention is an issue).
  763. */
  764. /*
  765. * XXX_count is incremented on every XXX operation, the delay
  766. * associated with the operation is added to XXX_delay.
  767. * XXX_delay contains the accumulated delay time in nanoseconds.
  768. */
  769. u64 blkio_start; /* Shared by blkio, swapin */
  770. u64 blkio_delay; /* wait for sync block io completion */
  771. u64 swapin_delay; /* wait for swapin block io completion */
  772. u32 blkio_count; /* total count of the number of sync block */
  773. /* io operations performed */
  774. u32 swapin_count; /* total count of the number of swapin block */
  775. /* io operations performed */
  776. u64 freepages_start;
  777. u64 freepages_delay; /* wait for memory reclaim */
  778. u32 freepages_count; /* total count of memory reclaim */
  779. };
  780. #endif /* CONFIG_TASK_DELAY_ACCT */
  781. static inline int sched_info_on(void)
  782. {
  783. #ifdef CONFIG_SCHEDSTATS
  784. return 1;
  785. #elif defined(CONFIG_TASK_DELAY_ACCT)
  786. extern int delayacct_on;
  787. return delayacct_on;
  788. #else
  789. return 0;
  790. #endif
  791. }
  792. #ifdef CONFIG_SCHEDSTATS
  793. void force_schedstat_enabled(void);
  794. #endif
  795. enum cpu_idle_type {
  796. CPU_IDLE,
  797. CPU_NOT_IDLE,
  798. CPU_NEWLY_IDLE,
  799. CPU_MAX_IDLE_TYPES
  800. };
  801. /*
  802. * Increase resolution of cpu_capacity calculations
  803. */
  804. #define SCHED_CAPACITY_SHIFT 10
  805. #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
  806. /*
  807. * Wake-queues are lists of tasks with a pending wakeup, whose
  808. * callers have already marked the task as woken internally,
  809. * and can thus carry on. A common use case is being able to
  810. * do the wakeups once the corresponding user lock as been
  811. * released.
  812. *
  813. * We hold reference to each task in the list across the wakeup,
  814. * thus guaranteeing that the memory is still valid by the time
  815. * the actual wakeups are performed in wake_up_q().
  816. *
  817. * One per task suffices, because there's never a need for a task to be
  818. * in two wake queues simultaneously; it is forbidden to abandon a task
  819. * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
  820. * already in a wake queue, the wakeup will happen soon and the second
  821. * waker can just skip it.
  822. *
  823. * The WAKE_Q macro declares and initializes the list head.
  824. * wake_up_q() does NOT reinitialize the list; it's expected to be
  825. * called near the end of a function, where the fact that the queue is
  826. * not used again will be easy to see by inspection.
  827. *
  828. * Note that this can cause spurious wakeups. schedule() callers
  829. * must ensure the call is done inside a loop, confirming that the
  830. * wakeup condition has in fact occurred.
  831. */
  832. struct wake_q_node {
  833. struct wake_q_node *next;
  834. };
  835. struct wake_q_head {
  836. struct wake_q_node *first;
  837. struct wake_q_node **lastp;
  838. };
  839. #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
  840. #define WAKE_Q(name) \
  841. struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
  842. extern void wake_q_add(struct wake_q_head *head,
  843. struct task_struct *task);
  844. extern void wake_up_q(struct wake_q_head *head);
  845. /*
  846. * sched-domains (multiprocessor balancing) declarations:
  847. */
  848. #ifdef CONFIG_SMP
  849. #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
  850. #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
  851. #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
  852. #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
  853. #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
  854. #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
  855. #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
  856. #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
  857. #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
  858. #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
  859. #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
  860. #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
  861. #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
  862. #define SD_NUMA 0x4000 /* cross-node balancing */
  863. #ifdef CONFIG_SCHED_SMT
  864. static inline int cpu_smt_flags(void)
  865. {
  866. return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
  867. }
  868. #endif
  869. #ifdef CONFIG_SCHED_MC
  870. static inline int cpu_core_flags(void)
  871. {
  872. return SD_SHARE_PKG_RESOURCES;
  873. }
  874. #endif
  875. #ifdef CONFIG_NUMA
  876. static inline int cpu_numa_flags(void)
  877. {
  878. return SD_NUMA;
  879. }
  880. #endif
  881. struct sched_domain_attr {
  882. int relax_domain_level;
  883. };
  884. #define SD_ATTR_INIT (struct sched_domain_attr) { \
  885. .relax_domain_level = -1, \
  886. }
  887. extern int sched_domain_level_max;
  888. struct sched_group;
  889. struct sched_domain {
  890. /* These fields must be setup */
  891. struct sched_domain *parent; /* top domain must be null terminated */
  892. struct sched_domain *child; /* bottom domain must be null terminated */
  893. struct sched_group *groups; /* the balancing groups of the domain */
  894. unsigned long min_interval; /* Minimum balance interval ms */
  895. unsigned long max_interval; /* Maximum balance interval ms */
  896. unsigned int busy_factor; /* less balancing by factor if busy */
  897. unsigned int imbalance_pct; /* No balance until over watermark */
  898. unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
  899. unsigned int busy_idx;
  900. unsigned int idle_idx;
  901. unsigned int newidle_idx;
  902. unsigned int wake_idx;
  903. unsigned int forkexec_idx;
  904. unsigned int smt_gain;
  905. int nohz_idle; /* NOHZ IDLE status */
  906. int flags; /* See SD_* */
  907. int level;
  908. /* Runtime fields. */
  909. unsigned long last_balance; /* init to jiffies. units in jiffies */
  910. unsigned int balance_interval; /* initialise to 1. units in ms. */
  911. unsigned int nr_balance_failed; /* initialise to 0 */
  912. /* idle_balance() stats */
  913. u64 max_newidle_lb_cost;
  914. unsigned long next_decay_max_lb_cost;
  915. #ifdef CONFIG_SCHEDSTATS
  916. /* load_balance() stats */
  917. unsigned int lb_count[CPU_MAX_IDLE_TYPES];
  918. unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
  919. unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
  920. unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
  921. unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
  922. unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
  923. unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
  924. unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
  925. /* Active load balancing */
  926. unsigned int alb_count;
  927. unsigned int alb_failed;
  928. unsigned int alb_pushed;
  929. /* SD_BALANCE_EXEC stats */
  930. unsigned int sbe_count;
  931. unsigned int sbe_balanced;
  932. unsigned int sbe_pushed;
  933. /* SD_BALANCE_FORK stats */
  934. unsigned int sbf_count;
  935. unsigned int sbf_balanced;
  936. unsigned int sbf_pushed;
  937. /* try_to_wake_up() stats */
  938. unsigned int ttwu_wake_remote;
  939. unsigned int ttwu_move_affine;
  940. unsigned int ttwu_move_balance;
  941. #endif
  942. #ifdef CONFIG_SCHED_DEBUG
  943. char *name;
  944. #endif
  945. union {
  946. void *private; /* used during construction */
  947. struct rcu_head rcu; /* used during destruction */
  948. };
  949. unsigned int span_weight;
  950. /*
  951. * Span of all CPUs in this domain.
  952. *
  953. * NOTE: this field is variable length. (Allocated dynamically
  954. * by attaching extra space to the end of the structure,
  955. * depending on how many CPUs the kernel has booted up with)
  956. */
  957. unsigned long span[0];
  958. };
  959. static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
  960. {
  961. return to_cpumask(sd->span);
  962. }
  963. extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  964. struct sched_domain_attr *dattr_new);
  965. /* Allocate an array of sched domains, for partition_sched_domains(). */
  966. cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
  967. void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
  968. bool cpus_share_cache(int this_cpu, int that_cpu);
  969. typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
  970. typedef int (*sched_domain_flags_f)(void);
  971. #define SDTL_OVERLAP 0x01
  972. struct sd_data {
  973. struct sched_domain **__percpu sd;
  974. struct sched_group **__percpu sg;
  975. struct sched_group_capacity **__percpu sgc;
  976. };
  977. struct sched_domain_topology_level {
  978. sched_domain_mask_f mask;
  979. sched_domain_flags_f sd_flags;
  980. int flags;
  981. int numa_level;
  982. struct sd_data data;
  983. #ifdef CONFIG_SCHED_DEBUG
  984. char *name;
  985. #endif
  986. };
  987. extern void set_sched_topology(struct sched_domain_topology_level *tl);
  988. extern void wake_up_if_idle(int cpu);
  989. #ifdef CONFIG_SCHED_DEBUG
  990. # define SD_INIT_NAME(type) .name = #type
  991. #else
  992. # define SD_INIT_NAME(type)
  993. #endif
  994. #else /* CONFIG_SMP */
  995. struct sched_domain_attr;
  996. static inline void
  997. partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  998. struct sched_domain_attr *dattr_new)
  999. {
  1000. }
  1001. static inline bool cpus_share_cache(int this_cpu, int that_cpu)
  1002. {
  1003. return true;
  1004. }
  1005. #endif /* !CONFIG_SMP */
  1006. struct io_context; /* See blkdev.h */
  1007. #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
  1008. extern void prefetch_stack(struct task_struct *t);
  1009. #else
  1010. static inline void prefetch_stack(struct task_struct *t) { }
  1011. #endif
  1012. struct audit_context; /* See audit.c */
  1013. struct mempolicy;
  1014. struct pipe_inode_info;
  1015. struct uts_namespace;
  1016. struct load_weight {
  1017. unsigned long weight;
  1018. u32 inv_weight;
  1019. };
  1020. /*
  1021. * The load_avg/util_avg accumulates an infinite geometric series.
  1022. * 1) load_avg factors frequency scaling into the amount of time that a
  1023. * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
  1024. * aggregated such weights of all runnable and blocked sched_entities.
  1025. * 2) util_avg factors frequency and cpu scaling into the amount of time
  1026. * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
  1027. * For cfs_rq, it is the aggregated such times of all runnable and
  1028. * blocked sched_entities.
  1029. * The 64 bit load_sum can:
  1030. * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
  1031. * the highest weight (=88761) always runnable, we should not overflow
  1032. * 2) for entity, support any load.weight always runnable
  1033. */
  1034. struct sched_avg {
  1035. u64 last_update_time, load_sum;
  1036. u32 util_sum, period_contrib;
  1037. unsigned long load_avg, util_avg;
  1038. };
  1039. #ifdef CONFIG_SCHEDSTATS
  1040. struct sched_statistics {
  1041. u64 wait_start;
  1042. u64 wait_max;
  1043. u64 wait_count;
  1044. u64 wait_sum;
  1045. u64 iowait_count;
  1046. u64 iowait_sum;
  1047. u64 sleep_start;
  1048. u64 sleep_max;
  1049. s64 sum_sleep_runtime;
  1050. u64 block_start;
  1051. u64 block_max;
  1052. u64 exec_max;
  1053. u64 slice_max;
  1054. u64 nr_migrations_cold;
  1055. u64 nr_failed_migrations_affine;
  1056. u64 nr_failed_migrations_running;
  1057. u64 nr_failed_migrations_hot;
  1058. u64 nr_forced_migrations;
  1059. u64 nr_wakeups;
  1060. u64 nr_wakeups_sync;
  1061. u64 nr_wakeups_migrate;
  1062. u64 nr_wakeups_local;
  1063. u64 nr_wakeups_remote;
  1064. u64 nr_wakeups_affine;
  1065. u64 nr_wakeups_affine_attempts;
  1066. u64 nr_wakeups_passive;
  1067. u64 nr_wakeups_idle;
  1068. };
  1069. #endif
  1070. struct sched_entity {
  1071. struct load_weight load; /* for load-balancing */
  1072. struct rb_node run_node;
  1073. struct list_head group_node;
  1074. unsigned int on_rq;
  1075. u64 exec_start;
  1076. u64 sum_exec_runtime;
  1077. u64 vruntime;
  1078. u64 prev_sum_exec_runtime;
  1079. u64 nr_migrations;
  1080. #ifdef CONFIG_SCHEDSTATS
  1081. struct sched_statistics statistics;
  1082. #endif
  1083. #ifdef CONFIG_FAIR_GROUP_SCHED
  1084. int depth;
  1085. struct sched_entity *parent;
  1086. /* rq on which this entity is (to be) queued: */
  1087. struct cfs_rq *cfs_rq;
  1088. /* rq "owned" by this entity/group: */
  1089. struct cfs_rq *my_q;
  1090. #endif
  1091. #ifdef CONFIG_SMP
  1092. /*
  1093. * Per entity load average tracking.
  1094. *
  1095. * Put into separate cache line so it does not
  1096. * collide with read-mostly values above.
  1097. */
  1098. struct sched_avg avg ____cacheline_aligned_in_smp;
  1099. #endif
  1100. };
  1101. struct sched_rt_entity {
  1102. struct list_head run_list;
  1103. unsigned long timeout;
  1104. unsigned long watchdog_stamp;
  1105. unsigned int time_slice;
  1106. unsigned short on_rq;
  1107. unsigned short on_list;
  1108. struct sched_rt_entity *back;
  1109. #ifdef CONFIG_RT_GROUP_SCHED
  1110. struct sched_rt_entity *parent;
  1111. /* rq on which this entity is (to be) queued: */
  1112. struct rt_rq *rt_rq;
  1113. /* rq "owned" by this entity/group: */
  1114. struct rt_rq *my_q;
  1115. #endif
  1116. };
  1117. struct sched_dl_entity {
  1118. struct rb_node rb_node;
  1119. /*
  1120. * Original scheduling parameters. Copied here from sched_attr
  1121. * during sched_setattr(), they will remain the same until
  1122. * the next sched_setattr().
  1123. */
  1124. u64 dl_runtime; /* maximum runtime for each instance */
  1125. u64 dl_deadline; /* relative deadline of each instance */
  1126. u64 dl_period; /* separation of two instances (period) */
  1127. u64 dl_bw; /* dl_runtime / dl_deadline */
  1128. /*
  1129. * Actual scheduling parameters. Initialized with the values above,
  1130. * they are continously updated during task execution. Note that
  1131. * the remaining runtime could be < 0 in case we are in overrun.
  1132. */
  1133. s64 runtime; /* remaining runtime for this instance */
  1134. u64 deadline; /* absolute deadline for this instance */
  1135. unsigned int flags; /* specifying the scheduler behaviour */
  1136. /*
  1137. * Some bool flags:
  1138. *
  1139. * @dl_throttled tells if we exhausted the runtime. If so, the
  1140. * task has to wait for a replenishment to be performed at the
  1141. * next firing of dl_timer.
  1142. *
  1143. * @dl_boosted tells if we are boosted due to DI. If so we are
  1144. * outside bandwidth enforcement mechanism (but only until we
  1145. * exit the critical section);
  1146. *
  1147. * @dl_yielded tells if task gave up the cpu before consuming
  1148. * all its available runtime during the last job.
  1149. */
  1150. int dl_throttled, dl_boosted, dl_yielded;
  1151. /*
  1152. * Bandwidth enforcement timer. Each -deadline task has its
  1153. * own bandwidth to be enforced, thus we need one timer per task.
  1154. */
  1155. struct hrtimer dl_timer;
  1156. };
  1157. union rcu_special {
  1158. struct {
  1159. u8 blocked;
  1160. u8 need_qs;
  1161. u8 exp_need_qs;
  1162. u8 pad; /* Otherwise the compiler can store garbage here. */
  1163. } b; /* Bits. */
  1164. u32 s; /* Set of bits. */
  1165. };
  1166. struct rcu_node;
  1167. enum perf_event_task_context {
  1168. perf_invalid_context = -1,
  1169. perf_hw_context = 0,
  1170. perf_sw_context,
  1171. perf_nr_task_contexts,
  1172. };
  1173. /* Track pages that require TLB flushes */
  1174. struct tlbflush_unmap_batch {
  1175. /*
  1176. * Each bit set is a CPU that potentially has a TLB entry for one of
  1177. * the PFNs being flushed. See set_tlb_ubc_flush_pending().
  1178. */
  1179. struct cpumask cpumask;
  1180. /* True if any bit in cpumask is set */
  1181. bool flush_required;
  1182. /*
  1183. * If true then the PTE was dirty when unmapped. The entry must be
  1184. * flushed before IO is initiated or a stale TLB entry potentially
  1185. * allows an update without redirtying the page.
  1186. */
  1187. bool writable;
  1188. };
  1189. struct task_struct {
  1190. volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
  1191. void *stack;
  1192. atomic_t usage;
  1193. unsigned int flags; /* per process flags, defined below */
  1194. unsigned int ptrace;
  1195. #ifdef CONFIG_SMP
  1196. struct llist_node wake_entry;
  1197. int on_cpu;
  1198. unsigned int wakee_flips;
  1199. unsigned long wakee_flip_decay_ts;
  1200. struct task_struct *last_wakee;
  1201. int wake_cpu;
  1202. #endif
  1203. int on_rq;
  1204. int prio, static_prio, normal_prio;
  1205. unsigned int rt_priority;
  1206. const struct sched_class *sched_class;
  1207. struct sched_entity se;
  1208. struct sched_rt_entity rt;
  1209. #ifdef CONFIG_CGROUP_SCHED
  1210. struct task_group *sched_task_group;
  1211. #endif
  1212. struct sched_dl_entity dl;
  1213. #ifdef CONFIG_PREEMPT_NOTIFIERS
  1214. /* list of struct preempt_notifier: */
  1215. struct hlist_head preempt_notifiers;
  1216. #endif
  1217. #ifdef CONFIG_BLK_DEV_IO_TRACE
  1218. unsigned int btrace_seq;
  1219. #endif
  1220. unsigned int policy;
  1221. int nr_cpus_allowed;
  1222. cpumask_t cpus_allowed;
  1223. #ifdef CONFIG_PREEMPT_RCU
  1224. int rcu_read_lock_nesting;
  1225. union rcu_special rcu_read_unlock_special;
  1226. struct list_head rcu_node_entry;
  1227. struct rcu_node *rcu_blocked_node;
  1228. #endif /* #ifdef CONFIG_PREEMPT_RCU */
  1229. #ifdef CONFIG_TASKS_RCU
  1230. unsigned long rcu_tasks_nvcsw;
  1231. bool rcu_tasks_holdout;
  1232. struct list_head rcu_tasks_holdout_list;
  1233. int rcu_tasks_idle_cpu;
  1234. #endif /* #ifdef CONFIG_TASKS_RCU */
  1235. #ifdef CONFIG_SCHED_INFO
  1236. struct sched_info sched_info;
  1237. #endif
  1238. struct list_head tasks;
  1239. #ifdef CONFIG_SMP
  1240. struct plist_node pushable_tasks;
  1241. struct rb_node pushable_dl_tasks;
  1242. #endif
  1243. struct mm_struct *mm, *active_mm;
  1244. /* per-thread vma caching */
  1245. u32 vmacache_seqnum;
  1246. struct vm_area_struct *vmacache[VMACACHE_SIZE];
  1247. #if defined(SPLIT_RSS_COUNTING)
  1248. struct task_rss_stat rss_stat;
  1249. #endif
  1250. /* task state */
  1251. int exit_state;
  1252. int exit_code, exit_signal;
  1253. int pdeath_signal; /* The signal sent when the parent dies */
  1254. unsigned long jobctl; /* JOBCTL_*, siglock protected */
  1255. /* Used for emulating ABI behavior of previous Linux versions */
  1256. unsigned int personality;
  1257. /* scheduler bits, serialized by scheduler locks */
  1258. unsigned sched_reset_on_fork:1;
  1259. unsigned sched_contributes_to_load:1;
  1260. unsigned sched_migrated:1;
  1261. unsigned :0; /* force alignment to the next boundary */
  1262. /* unserialized, strictly 'current' */
  1263. unsigned in_execve:1; /* bit to tell LSMs we're in execve */
  1264. unsigned in_iowait:1;
  1265. #ifdef CONFIG_MEMCG
  1266. unsigned memcg_may_oom:1;
  1267. #ifndef CONFIG_SLOB
  1268. unsigned memcg_kmem_skip_account:1;
  1269. #endif
  1270. #endif
  1271. #ifdef CONFIG_COMPAT_BRK
  1272. unsigned brk_randomized:1;
  1273. #endif
  1274. unsigned long atomic_flags; /* Flags needing atomic access. */
  1275. struct restart_block restart_block;
  1276. pid_t pid;
  1277. pid_t tgid;
  1278. #ifdef CONFIG_CC_STACKPROTECTOR
  1279. /* Canary value for the -fstack-protector gcc feature */
  1280. unsigned long stack_canary;
  1281. #endif
  1282. /*
  1283. * pointers to (original) parent process, youngest child, younger sibling,
  1284. * older sibling, respectively. (p->father can be replaced with
  1285. * p->real_parent->pid)
  1286. */
  1287. struct task_struct __rcu *real_parent; /* real parent process */
  1288. struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
  1289. /*
  1290. * children/sibling forms the list of my natural children
  1291. */
  1292. struct list_head children; /* list of my children */
  1293. struct list_head sibling; /* linkage in my parent's children list */
  1294. struct task_struct *group_leader; /* threadgroup leader */
  1295. /*
  1296. * ptraced is the list of tasks this task is using ptrace on.
  1297. * This includes both natural children and PTRACE_ATTACH targets.
  1298. * p->ptrace_entry is p's link on the p->parent->ptraced list.
  1299. */
  1300. struct list_head ptraced;
  1301. struct list_head ptrace_entry;
  1302. /* PID/PID hash table linkage. */
  1303. struct pid_link pids[PIDTYPE_MAX];
  1304. struct list_head thread_group;
  1305. struct list_head thread_node;
  1306. struct completion *vfork_done; /* for vfork() */
  1307. int __user *set_child_tid; /* CLONE_CHILD_SETTID */
  1308. int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
  1309. cputime_t utime, stime, utimescaled, stimescaled;
  1310. cputime_t gtime;
  1311. struct prev_cputime prev_cputime;
  1312. #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
  1313. seqcount_t vtime_seqcount;
  1314. unsigned long long vtime_snap;
  1315. enum {
  1316. /* Task is sleeping or running in a CPU with VTIME inactive */
  1317. VTIME_INACTIVE = 0,
  1318. /* Task runs in userspace in a CPU with VTIME active */
  1319. VTIME_USER,
  1320. /* Task runs in kernelspace in a CPU with VTIME active */
  1321. VTIME_SYS,
  1322. } vtime_snap_whence;
  1323. #endif
  1324. #ifdef CONFIG_NO_HZ_FULL
  1325. unsigned long tick_dep_mask;
  1326. #endif
  1327. unsigned long nvcsw, nivcsw; /* context switch counts */
  1328. u64 start_time; /* monotonic time in nsec */
  1329. u64 real_start_time; /* boot based time in nsec */
  1330. /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
  1331. unsigned long min_flt, maj_flt;
  1332. struct task_cputime cputime_expires;
  1333. struct list_head cpu_timers[3];
  1334. /* process credentials */
  1335. const struct cred __rcu *real_cred; /* objective and real subjective task
  1336. * credentials (COW) */
  1337. const struct cred __rcu *cred; /* effective (overridable) subjective task
  1338. * credentials (COW) */
  1339. char comm[TASK_COMM_LEN]; /* executable name excluding path
  1340. - access with [gs]et_task_comm (which lock
  1341. it with task_lock())
  1342. - initialized normally by setup_new_exec */
  1343. /* file system info */
  1344. struct nameidata *nameidata;
  1345. #ifdef CONFIG_SYSVIPC
  1346. /* ipc stuff */
  1347. struct sysv_sem sysvsem;
  1348. struct sysv_shm sysvshm;
  1349. #endif
  1350. #ifdef CONFIG_DETECT_HUNG_TASK
  1351. /* hung task detection */
  1352. unsigned long last_switch_count;
  1353. #endif
  1354. /* filesystem information */
  1355. struct fs_struct *fs;
  1356. /* open file information */
  1357. struct files_struct *files;
  1358. /* namespaces */
  1359. struct nsproxy *nsproxy;
  1360. /* signal handlers */
  1361. struct signal_struct *signal;
  1362. struct sighand_struct *sighand;
  1363. sigset_t blocked, real_blocked;
  1364. sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
  1365. struct sigpending pending;
  1366. unsigned long sas_ss_sp;
  1367. size_t sas_ss_size;
  1368. struct callback_head *task_works;
  1369. struct audit_context *audit_context;
  1370. #ifdef CONFIG_AUDITSYSCALL
  1371. kuid_t loginuid;
  1372. unsigned int sessionid;
  1373. #endif
  1374. struct seccomp seccomp;
  1375. /* Thread group tracking */
  1376. u32 parent_exec_id;
  1377. u32 self_exec_id;
  1378. /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
  1379. * mempolicy */
  1380. spinlock_t alloc_lock;
  1381. /* Protection of the PI data structures: */
  1382. raw_spinlock_t pi_lock;
  1383. struct wake_q_node wake_q;
  1384. #ifdef CONFIG_RT_MUTEXES
  1385. /* PI waiters blocked on a rt_mutex held by this task */
  1386. struct rb_root pi_waiters;
  1387. struct rb_node *pi_waiters_leftmost;
  1388. /* Deadlock detection and priority inheritance handling */
  1389. struct rt_mutex_waiter *pi_blocked_on;
  1390. #endif
  1391. #ifdef CONFIG_DEBUG_MUTEXES
  1392. /* mutex deadlock detection */
  1393. struct mutex_waiter *blocked_on;
  1394. #endif
  1395. #ifdef CONFIG_TRACE_IRQFLAGS
  1396. unsigned int irq_events;
  1397. unsigned long hardirq_enable_ip;
  1398. unsigned long hardirq_disable_ip;
  1399. unsigned int hardirq_enable_event;
  1400. unsigned int hardirq_disable_event;
  1401. int hardirqs_enabled;
  1402. int hardirq_context;
  1403. unsigned long softirq_disable_ip;
  1404. unsigned long softirq_enable_ip;
  1405. unsigned int softirq_disable_event;
  1406. unsigned int softirq_enable_event;
  1407. int softirqs_enabled;
  1408. int softirq_context;
  1409. #endif
  1410. #ifdef CONFIG_LOCKDEP
  1411. # define MAX_LOCK_DEPTH 48UL
  1412. u64 curr_chain_key;
  1413. int lockdep_depth;
  1414. unsigned int lockdep_recursion;
  1415. struct held_lock held_locks[MAX_LOCK_DEPTH];
  1416. gfp_t lockdep_reclaim_gfp;
  1417. #endif
  1418. #ifdef CONFIG_UBSAN
  1419. unsigned int in_ubsan;
  1420. #endif
  1421. /* journalling filesystem info */
  1422. void *journal_info;
  1423. /* stacked block device info */
  1424. struct bio_list *bio_list;
  1425. #ifdef CONFIG_BLOCK
  1426. /* stack plugging */
  1427. struct blk_plug *plug;
  1428. #endif
  1429. /* VM state */
  1430. struct reclaim_state *reclaim_state;
  1431. struct backing_dev_info *backing_dev_info;
  1432. struct io_context *io_context;
  1433. unsigned long ptrace_message;
  1434. siginfo_t *last_siginfo; /* For ptrace use. */
  1435. struct task_io_accounting ioac;
  1436. #if defined(CONFIG_TASK_XACCT)
  1437. u64 acct_rss_mem1; /* accumulated rss usage */
  1438. u64 acct_vm_mem1; /* accumulated virtual memory usage */
  1439. cputime_t acct_timexpd; /* stime + utime since last update */
  1440. #endif
  1441. #ifdef CONFIG_CPUSETS
  1442. nodemask_t mems_allowed; /* Protected by alloc_lock */
  1443. seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
  1444. int cpuset_mem_spread_rotor;
  1445. int cpuset_slab_spread_rotor;
  1446. #endif
  1447. #ifdef CONFIG_CGROUPS
  1448. /* Control Group info protected by css_set_lock */
  1449. struct css_set __rcu *cgroups;
  1450. /* cg_list protected by css_set_lock and tsk->alloc_lock */
  1451. struct list_head cg_list;
  1452. #endif
  1453. #ifdef CONFIG_FUTEX
  1454. struct robust_list_head __user *robust_list;
  1455. #ifdef CONFIG_COMPAT
  1456. struct compat_robust_list_head __user *compat_robust_list;
  1457. #endif
  1458. struct list_head pi_state_list;
  1459. struct futex_pi_state *pi_state_cache;
  1460. #endif
  1461. #ifdef CONFIG_PERF_EVENTS
  1462. struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
  1463. struct mutex perf_event_mutex;
  1464. struct list_head perf_event_list;
  1465. #endif
  1466. #ifdef CONFIG_DEBUG_PREEMPT
  1467. unsigned long preempt_disable_ip;
  1468. #endif
  1469. #ifdef CONFIG_NUMA
  1470. struct mempolicy *mempolicy; /* Protected by alloc_lock */
  1471. short il_next;
  1472. short pref_node_fork;
  1473. #endif
  1474. #ifdef CONFIG_NUMA_BALANCING
  1475. int numa_scan_seq;
  1476. unsigned int numa_scan_period;
  1477. unsigned int numa_scan_period_max;
  1478. int numa_preferred_nid;
  1479. unsigned long numa_migrate_retry;
  1480. u64 node_stamp; /* migration stamp */
  1481. u64 last_task_numa_placement;
  1482. u64 last_sum_exec_runtime;
  1483. struct callback_head numa_work;
  1484. struct list_head numa_entry;
  1485. struct numa_group *numa_group;
  1486. /*
  1487. * numa_faults is an array split into four regions:
  1488. * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
  1489. * in this precise order.
  1490. *
  1491. * faults_memory: Exponential decaying average of faults on a per-node
  1492. * basis. Scheduling placement decisions are made based on these
  1493. * counts. The values remain static for the duration of a PTE scan.
  1494. * faults_cpu: Track the nodes the process was running on when a NUMA
  1495. * hinting fault was incurred.
  1496. * faults_memory_buffer and faults_cpu_buffer: Record faults per node
  1497. * during the current scan window. When the scan completes, the counts
  1498. * in faults_memory and faults_cpu decay and these values are copied.
  1499. */
  1500. unsigned long *numa_faults;
  1501. unsigned long total_numa_faults;
  1502. /*
  1503. * numa_faults_locality tracks if faults recorded during the last
  1504. * scan window were remote/local or failed to migrate. The task scan
  1505. * period is adapted based on the locality of the faults with different
  1506. * weights depending on whether they were shared or private faults
  1507. */
  1508. unsigned long numa_faults_locality[3];
  1509. unsigned long numa_pages_migrated;
  1510. #endif /* CONFIG_NUMA_BALANCING */
  1511. #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
  1512. struct tlbflush_unmap_batch tlb_ubc;
  1513. #endif
  1514. struct rcu_head rcu;
  1515. /*
  1516. * cache last used pipe for splice
  1517. */
  1518. struct pipe_inode_info *splice_pipe;
  1519. struct page_frag task_frag;
  1520. #ifdef CONFIG_TASK_DELAY_ACCT
  1521. struct task_delay_info *delays;
  1522. #endif
  1523. #ifdef CONFIG_FAULT_INJECTION
  1524. int make_it_fail;
  1525. #endif
  1526. /*
  1527. * when (nr_dirtied >= nr_dirtied_pause), it's time to call
  1528. * balance_dirty_pages() for some dirty throttling pause
  1529. */
  1530. int nr_dirtied;
  1531. int nr_dirtied_pause;
  1532. unsigned long dirty_paused_when; /* start of a write-and-pause period */
  1533. #ifdef CONFIG_LATENCYTOP
  1534. int latency_record_count;
  1535. struct latency_record latency_record[LT_SAVECOUNT];
  1536. #endif
  1537. /*
  1538. * time slack values; these are used to round up poll() and
  1539. * select() etc timeout values. These are in nanoseconds.
  1540. */
  1541. u64 timer_slack_ns;
  1542. u64 default_timer_slack_ns;
  1543. #ifdef CONFIG_KASAN
  1544. unsigned int kasan_depth;
  1545. #endif
  1546. #ifdef CONFIG_FUNCTION_GRAPH_TRACER
  1547. /* Index of current stored address in ret_stack */
  1548. int curr_ret_stack;
  1549. /* Stack of return addresses for return function tracing */
  1550. struct ftrace_ret_stack *ret_stack;
  1551. /* time stamp for last schedule */
  1552. unsigned long long ftrace_timestamp;
  1553. /*
  1554. * Number of functions that haven't been traced
  1555. * because of depth overrun.
  1556. */
  1557. atomic_t trace_overrun;
  1558. /* Pause for the tracing */
  1559. atomic_t tracing_graph_pause;
  1560. #endif
  1561. #ifdef CONFIG_TRACING
  1562. /* state flags for use by tracers */
  1563. unsigned long trace;
  1564. /* bitmask and counter of trace recursion */
  1565. unsigned long trace_recursion;
  1566. #endif /* CONFIG_TRACING */
  1567. #ifdef CONFIG_KCOV
  1568. /* Coverage collection mode enabled for this task (0 if disabled). */
  1569. enum kcov_mode kcov_mode;
  1570. /* Size of the kcov_area. */
  1571. unsigned kcov_size;
  1572. /* Buffer for coverage collection. */
  1573. void *kcov_area;
  1574. /* kcov desciptor wired with this task or NULL. */
  1575. struct kcov *kcov;
  1576. #endif
  1577. #ifdef CONFIG_MEMCG
  1578. struct mem_cgroup *memcg_in_oom;
  1579. gfp_t memcg_oom_gfp_mask;
  1580. int memcg_oom_order;
  1581. /* number of pages to reclaim on returning to userland */
  1582. unsigned int memcg_nr_pages_over_high;
  1583. #endif
  1584. #ifdef CONFIG_UPROBES
  1585. struct uprobe_task *utask;
  1586. #endif
  1587. #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
  1588. unsigned int sequential_io;
  1589. unsigned int sequential_io_avg;
  1590. #endif
  1591. #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
  1592. unsigned long task_state_change;
  1593. #endif
  1594. int pagefault_disabled;
  1595. #ifdef CONFIG_MMU
  1596. struct list_head oom_reaper_list;
  1597. #endif
  1598. /* CPU-specific state of this task */
  1599. struct thread_struct thread;
  1600. /*
  1601. * WARNING: on x86, 'thread_struct' contains a variable-sized
  1602. * structure. It *MUST* be at the end of 'task_struct'.
  1603. *
  1604. * Do not put anything below here!
  1605. */
  1606. };
  1607. #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
  1608. extern int arch_task_struct_size __read_mostly;
  1609. #else
  1610. # define arch_task_struct_size (sizeof(struct task_struct))
  1611. #endif
  1612. /* Future-safe accessor for struct task_struct's cpus_allowed. */
  1613. #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
  1614. #define TNF_MIGRATED 0x01
  1615. #define TNF_NO_GROUP 0x02
  1616. #define TNF_SHARED 0x04
  1617. #define TNF_FAULT_LOCAL 0x08
  1618. #define TNF_MIGRATE_FAIL 0x10
  1619. #ifdef CONFIG_NUMA_BALANCING
  1620. extern void task_numa_fault(int last_node, int node, int pages, int flags);
  1621. extern pid_t task_numa_group_id(struct task_struct *p);
  1622. extern void set_numabalancing_state(bool enabled);
  1623. extern void task_numa_free(struct task_struct *p);
  1624. extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
  1625. int src_nid, int dst_cpu);
  1626. #else
  1627. static inline void task_numa_fault(int last_node, int node, int pages,
  1628. int flags)
  1629. {
  1630. }
  1631. static inline pid_t task_numa_group_id(struct task_struct *p)
  1632. {
  1633. return 0;
  1634. }
  1635. static inline void set_numabalancing_state(bool enabled)
  1636. {
  1637. }
  1638. static inline void task_numa_free(struct task_struct *p)
  1639. {
  1640. }
  1641. static inline bool should_numa_migrate_memory(struct task_struct *p,
  1642. struct page *page, int src_nid, int dst_cpu)
  1643. {
  1644. return true;
  1645. }
  1646. #endif
  1647. static inline struct pid *task_pid(struct task_struct *task)
  1648. {
  1649. return task->pids[PIDTYPE_PID].pid;
  1650. }
  1651. static inline struct pid *task_tgid(struct task_struct *task)
  1652. {
  1653. return task->group_leader->pids[PIDTYPE_PID].pid;
  1654. }
  1655. /*
  1656. * Without tasklist or rcu lock it is not safe to dereference
  1657. * the result of task_pgrp/task_session even if task == current,
  1658. * we can race with another thread doing sys_setsid/sys_setpgid.
  1659. */
  1660. static inline struct pid *task_pgrp(struct task_struct *task)
  1661. {
  1662. return task->group_leader->pids[PIDTYPE_PGID].pid;
  1663. }
  1664. static inline struct pid *task_session(struct task_struct *task)
  1665. {
  1666. return task->group_leader->pids[PIDTYPE_SID].pid;
  1667. }
  1668. struct pid_namespace;
  1669. /*
  1670. * the helpers to get the task's different pids as they are seen
  1671. * from various namespaces
  1672. *
  1673. * task_xid_nr() : global id, i.e. the id seen from the init namespace;
  1674. * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
  1675. * current.
  1676. * task_xid_nr_ns() : id seen from the ns specified;
  1677. *
  1678. * set_task_vxid() : assigns a virtual id to a task;
  1679. *
  1680. * see also pid_nr() etc in include/linux/pid.h
  1681. */
  1682. pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
  1683. struct pid_namespace *ns);
  1684. static inline pid_t task_pid_nr(struct task_struct *tsk)
  1685. {
  1686. return tsk->pid;
  1687. }
  1688. static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
  1689. struct pid_namespace *ns)
  1690. {
  1691. return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
  1692. }
  1693. static inline pid_t task_pid_vnr(struct task_struct *tsk)
  1694. {
  1695. return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
  1696. }
  1697. static inline pid_t task_tgid_nr(struct task_struct *tsk)
  1698. {
  1699. return tsk->tgid;
  1700. }
  1701. pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
  1702. static inline pid_t task_tgid_vnr(struct task_struct *tsk)
  1703. {
  1704. return pid_vnr(task_tgid(tsk));
  1705. }
  1706. static inline int pid_alive(const struct task_struct *p);
  1707. static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
  1708. {
  1709. pid_t pid = 0;
  1710. rcu_read_lock();
  1711. if (pid_alive(tsk))
  1712. pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
  1713. rcu_read_unlock();
  1714. return pid;
  1715. }
  1716. static inline pid_t task_ppid_nr(const struct task_struct *tsk)
  1717. {
  1718. return task_ppid_nr_ns(tsk, &init_pid_ns);
  1719. }
  1720. static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
  1721. struct pid_namespace *ns)
  1722. {
  1723. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
  1724. }
  1725. static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
  1726. {
  1727. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
  1728. }
  1729. static inline pid_t task_session_nr_ns(struct task_struct *tsk,
  1730. struct pid_namespace *ns)
  1731. {
  1732. return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
  1733. }
  1734. static inline pid_t task_session_vnr(struct task_struct *tsk)
  1735. {
  1736. return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
  1737. }
  1738. /* obsolete, do not use */
  1739. static inline pid_t task_pgrp_nr(struct task_struct *tsk)
  1740. {
  1741. return task_pgrp_nr_ns(tsk, &init_pid_ns);
  1742. }
  1743. /**
  1744. * pid_alive - check that a task structure is not stale
  1745. * @p: Task structure to be checked.
  1746. *
  1747. * Test if a process is not yet dead (at most zombie state)
  1748. * If pid_alive fails, then pointers within the task structure
  1749. * can be stale and must not be dereferenced.
  1750. *
  1751. * Return: 1 if the process is alive. 0 otherwise.
  1752. */
  1753. static inline int pid_alive(const struct task_struct *p)
  1754. {
  1755. return p->pids[PIDTYPE_PID].pid != NULL;
  1756. }
  1757. /**
  1758. * is_global_init - check if a task structure is init. Since init
  1759. * is free to have sub-threads we need to check tgid.
  1760. * @tsk: Task structure to be checked.
  1761. *
  1762. * Check if a task structure is the first user space task the kernel created.
  1763. *
  1764. * Return: 1 if the task structure is init. 0 otherwise.
  1765. */
  1766. static inline int is_global_init(struct task_struct *tsk)
  1767. {
  1768. return task_tgid_nr(tsk) == 1;
  1769. }
  1770. extern struct pid *cad_pid;
  1771. extern void free_task(struct task_struct *tsk);
  1772. #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
  1773. extern void __put_task_struct(struct task_struct *t);
  1774. static inline void put_task_struct(struct task_struct *t)
  1775. {
  1776. if (atomic_dec_and_test(&t->usage))
  1777. __put_task_struct(t);
  1778. }
  1779. #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
  1780. extern void task_cputime(struct task_struct *t,
  1781. cputime_t *utime, cputime_t *stime);
  1782. extern void task_cputime_scaled(struct task_struct *t,
  1783. cputime_t *utimescaled, cputime_t *stimescaled);
  1784. extern cputime_t task_gtime(struct task_struct *t);
  1785. #else
  1786. static inline void task_cputime(struct task_struct *t,
  1787. cputime_t *utime, cputime_t *stime)
  1788. {
  1789. if (utime)
  1790. *utime = t->utime;
  1791. if (stime)
  1792. *stime = t->stime;
  1793. }
  1794. static inline void task_cputime_scaled(struct task_struct *t,
  1795. cputime_t *utimescaled,
  1796. cputime_t *stimescaled)
  1797. {
  1798. if (utimescaled)
  1799. *utimescaled = t->utimescaled;
  1800. if (stimescaled)
  1801. *stimescaled = t->stimescaled;
  1802. }
  1803. static inline cputime_t task_gtime(struct task_struct *t)
  1804. {
  1805. return t->gtime;
  1806. }
  1807. #endif
  1808. extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
  1809. extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
  1810. /*
  1811. * Per process flags
  1812. */
  1813. #define PF_EXITING 0x00000004 /* getting shut down */
  1814. #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
  1815. #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
  1816. #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
  1817. #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
  1818. #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
  1819. #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
  1820. #define PF_DUMPCORE 0x00000200 /* dumped core */
  1821. #define PF_SIGNALED 0x00000400 /* killed by a signal */
  1822. #define PF_MEMALLOC 0x00000800 /* Allocating memory */
  1823. #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
  1824. #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
  1825. #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
  1826. #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
  1827. #define PF_FROZEN 0x00010000 /* frozen for system suspend */
  1828. #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
  1829. #define PF_KSWAPD 0x00040000 /* I am kswapd */
  1830. #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
  1831. #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
  1832. #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
  1833. #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
  1834. #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
  1835. #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
  1836. #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
  1837. #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
  1838. #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
  1839. #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
  1840. /*
  1841. * Only the _current_ task can read/write to tsk->flags, but other
  1842. * tasks can access tsk->flags in readonly mode for example
  1843. * with tsk_used_math (like during threaded core dumping).
  1844. * There is however an exception to this rule during ptrace
  1845. * or during fork: the ptracer task is allowed to write to the
  1846. * child->flags of its traced child (same goes for fork, the parent
  1847. * can write to the child->flags), because we're guaranteed the
  1848. * child is not running and in turn not changing child->flags
  1849. * at the same time the parent does it.
  1850. */
  1851. #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
  1852. #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
  1853. #define clear_used_math() clear_stopped_child_used_math(current)
  1854. #define set_used_math() set_stopped_child_used_math(current)
  1855. #define conditional_stopped_child_used_math(condition, child) \
  1856. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
  1857. #define conditional_used_math(condition) \
  1858. conditional_stopped_child_used_math(condition, current)
  1859. #define copy_to_stopped_child_used_math(child) \
  1860. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
  1861. /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
  1862. #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
  1863. #define used_math() tsk_used_math(current)
  1864. /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
  1865. * __GFP_FS is also cleared as it implies __GFP_IO.
  1866. */
  1867. static inline gfp_t memalloc_noio_flags(gfp_t flags)
  1868. {
  1869. if (unlikely(current->flags & PF_MEMALLOC_NOIO))
  1870. flags &= ~(__GFP_IO | __GFP_FS);
  1871. return flags;
  1872. }
  1873. static inline unsigned int memalloc_noio_save(void)
  1874. {
  1875. unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
  1876. current->flags |= PF_MEMALLOC_NOIO;
  1877. return flags;
  1878. }
  1879. static inline void memalloc_noio_restore(unsigned int flags)
  1880. {
  1881. current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
  1882. }
  1883. /* Per-process atomic flags. */
  1884. #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
  1885. #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
  1886. #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
  1887. #define TASK_PFA_TEST(name, func) \
  1888. static inline bool task_##func(struct task_struct *p) \
  1889. { return test_bit(PFA_##name, &p->atomic_flags); }
  1890. #define TASK_PFA_SET(name, func) \
  1891. static inline void task_set_##func(struct task_struct *p) \
  1892. { set_bit(PFA_##name, &p->atomic_flags); }
  1893. #define TASK_PFA_CLEAR(name, func) \
  1894. static inline void task_clear_##func(struct task_struct *p) \
  1895. { clear_bit(PFA_##name, &p->atomic_flags); }
  1896. TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
  1897. TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
  1898. TASK_PFA_TEST(SPREAD_PAGE, spread_page)
  1899. TASK_PFA_SET(SPREAD_PAGE, spread_page)
  1900. TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
  1901. TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
  1902. TASK_PFA_SET(SPREAD_SLAB, spread_slab)
  1903. TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
  1904. /*
  1905. * task->jobctl flags
  1906. */
  1907. #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
  1908. #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
  1909. #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
  1910. #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
  1911. #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
  1912. #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
  1913. #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
  1914. #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
  1915. #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
  1916. #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
  1917. #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
  1918. #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
  1919. #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
  1920. #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
  1921. #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
  1922. #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
  1923. #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
  1924. extern bool task_set_jobctl_pending(struct task_struct *task,
  1925. unsigned long mask);
  1926. extern void task_clear_jobctl_trapping(struct task_struct *task);
  1927. extern void task_clear_jobctl_pending(struct task_struct *task,
  1928. unsigned long mask);
  1929. static inline void rcu_copy_process(struct task_struct *p)
  1930. {
  1931. #ifdef CONFIG_PREEMPT_RCU
  1932. p->rcu_read_lock_nesting = 0;
  1933. p->rcu_read_unlock_special.s = 0;
  1934. p->rcu_blocked_node = NULL;
  1935. INIT_LIST_HEAD(&p->rcu_node_entry);
  1936. #endif /* #ifdef CONFIG_PREEMPT_RCU */
  1937. #ifdef CONFIG_TASKS_RCU
  1938. p->rcu_tasks_holdout = false;
  1939. INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
  1940. p->rcu_tasks_idle_cpu = -1;
  1941. #endif /* #ifdef CONFIG_TASKS_RCU */
  1942. }
  1943. static inline void tsk_restore_flags(struct task_struct *task,
  1944. unsigned long orig_flags, unsigned long flags)
  1945. {
  1946. task->flags &= ~flags;
  1947. task->flags |= orig_flags & flags;
  1948. }
  1949. extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
  1950. const struct cpumask *trial);
  1951. extern int task_can_attach(struct task_struct *p,
  1952. const struct cpumask *cs_cpus_allowed);
  1953. #ifdef CONFIG_SMP
  1954. extern void do_set_cpus_allowed(struct task_struct *p,
  1955. const struct cpumask *new_mask);
  1956. extern int set_cpus_allowed_ptr(struct task_struct *p,
  1957. const struct cpumask *new_mask);
  1958. #else
  1959. static inline void do_set_cpus_allowed(struct task_struct *p,
  1960. const struct cpumask *new_mask)
  1961. {
  1962. }
  1963. static inline int set_cpus_allowed_ptr(struct task_struct *p,
  1964. const struct cpumask *new_mask)
  1965. {
  1966. if (!cpumask_test_cpu(0, new_mask))
  1967. return -EINVAL;
  1968. return 0;
  1969. }
  1970. #endif
  1971. #ifdef CONFIG_NO_HZ_COMMON
  1972. void calc_load_enter_idle(void);
  1973. void calc_load_exit_idle(void);
  1974. #else
  1975. static inline void calc_load_enter_idle(void) { }
  1976. static inline void calc_load_exit_idle(void) { }
  1977. #endif /* CONFIG_NO_HZ_COMMON */
  1978. /*
  1979. * Do not use outside of architecture code which knows its limitations.
  1980. *
  1981. * sched_clock() has no promise of monotonicity or bounded drift between
  1982. * CPUs, use (which you should not) requires disabling IRQs.
  1983. *
  1984. * Please use one of the three interfaces below.
  1985. */
  1986. extern unsigned long long notrace sched_clock(void);
  1987. /*
  1988. * See the comment in kernel/sched/clock.c
  1989. */
  1990. extern u64 cpu_clock(int cpu);
  1991. extern u64 local_clock(void);
  1992. extern u64 running_clock(void);
  1993. extern u64 sched_clock_cpu(int cpu);
  1994. extern void sched_clock_init(void);
  1995. #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
  1996. static inline void sched_clock_tick(void)
  1997. {
  1998. }
  1999. static inline void sched_clock_idle_sleep_event(void)
  2000. {
  2001. }
  2002. static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
  2003. {
  2004. }
  2005. #else
  2006. /*
  2007. * Architectures can set this to 1 if they have specified
  2008. * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
  2009. * but then during bootup it turns out that sched_clock()
  2010. * is reliable after all:
  2011. */
  2012. extern int sched_clock_stable(void);
  2013. extern void set_sched_clock_stable(void);
  2014. extern void clear_sched_clock_stable(void);
  2015. extern void sched_clock_tick(void);
  2016. extern void sched_clock_idle_sleep_event(void);
  2017. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  2018. #endif
  2019. #ifdef CONFIG_IRQ_TIME_ACCOUNTING
  2020. /*
  2021. * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
  2022. * The reason for this explicit opt-in is not to have perf penalty with
  2023. * slow sched_clocks.
  2024. */
  2025. extern void enable_sched_clock_irqtime(void);
  2026. extern void disable_sched_clock_irqtime(void);
  2027. #else
  2028. static inline void enable_sched_clock_irqtime(void) {}
  2029. static inline void disable_sched_clock_irqtime(void) {}
  2030. #endif
  2031. extern unsigned long long
  2032. task_sched_runtime(struct task_struct *task);
  2033. /* sched_exec is called by processes performing an exec */
  2034. #ifdef CONFIG_SMP
  2035. extern void sched_exec(void);
  2036. #else
  2037. #define sched_exec() {}
  2038. #endif
  2039. extern void sched_clock_idle_sleep_event(void);
  2040. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  2041. #ifdef CONFIG_HOTPLUG_CPU
  2042. extern void idle_task_exit(void);
  2043. #else
  2044. static inline void idle_task_exit(void) {}
  2045. #endif
  2046. #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
  2047. extern void wake_up_nohz_cpu(int cpu);
  2048. #else
  2049. static inline void wake_up_nohz_cpu(int cpu) { }
  2050. #endif
  2051. #ifdef CONFIG_NO_HZ_FULL
  2052. extern u64 scheduler_tick_max_deferment(void);
  2053. #endif
  2054. #ifdef CONFIG_SCHED_AUTOGROUP
  2055. extern void sched_autogroup_create_attach(struct task_struct *p);
  2056. extern void sched_autogroup_detach(struct task_struct *p);
  2057. extern void sched_autogroup_fork(struct signal_struct *sig);
  2058. extern void sched_autogroup_exit(struct signal_struct *sig);
  2059. #ifdef CONFIG_PROC_FS
  2060. extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
  2061. extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
  2062. #endif
  2063. #else
  2064. static inline void sched_autogroup_create_attach(struct task_struct *p) { }
  2065. static inline void sched_autogroup_detach(struct task_struct *p) { }
  2066. static inline void sched_autogroup_fork(struct signal_struct *sig) { }
  2067. static inline void sched_autogroup_exit(struct signal_struct *sig) { }
  2068. #endif
  2069. extern int yield_to(struct task_struct *p, bool preempt);
  2070. extern void set_user_nice(struct task_struct *p, long nice);
  2071. extern int task_prio(const struct task_struct *p);
  2072. /**
  2073. * task_nice - return the nice value of a given task.
  2074. * @p: the task in question.
  2075. *
  2076. * Return: The nice value [ -20 ... 0 ... 19 ].
  2077. */
  2078. static inline int task_nice(const struct task_struct *p)
  2079. {
  2080. return PRIO_TO_NICE((p)->static_prio);
  2081. }
  2082. extern int can_nice(const struct task_struct *p, const int nice);
  2083. extern int task_curr(const struct task_struct *p);
  2084. extern int idle_cpu(int cpu);
  2085. extern int sched_setscheduler(struct task_struct *, int,
  2086. const struct sched_param *);
  2087. extern int sched_setscheduler_nocheck(struct task_struct *, int,
  2088. const struct sched_param *);
  2089. extern int sched_setattr(struct task_struct *,
  2090. const struct sched_attr *);
  2091. extern struct task_struct *idle_task(int cpu);
  2092. /**
  2093. * is_idle_task - is the specified task an idle task?
  2094. * @p: the task in question.
  2095. *
  2096. * Return: 1 if @p is an idle task. 0 otherwise.
  2097. */
  2098. static inline bool is_idle_task(const struct task_struct *p)
  2099. {
  2100. return p->pid == 0;
  2101. }
  2102. extern struct task_struct *curr_task(int cpu);
  2103. extern void set_curr_task(int cpu, struct task_struct *p);
  2104. void yield(void);
  2105. union thread_union {
  2106. struct thread_info thread_info;
  2107. unsigned long stack[THREAD_SIZE/sizeof(long)];
  2108. };
  2109. #ifndef __HAVE_ARCH_KSTACK_END
  2110. static inline int kstack_end(void *addr)
  2111. {
  2112. /* Reliable end of stack detection:
  2113. * Some APM bios versions misalign the stack
  2114. */
  2115. return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
  2116. }
  2117. #endif
  2118. extern union thread_union init_thread_union;
  2119. extern struct task_struct init_task;
  2120. extern struct mm_struct init_mm;
  2121. extern struct pid_namespace init_pid_ns;
  2122. /*
  2123. * find a task by one of its numerical ids
  2124. *
  2125. * find_task_by_pid_ns():
  2126. * finds a task by its pid in the specified namespace
  2127. * find_task_by_vpid():
  2128. * finds a task by its virtual pid
  2129. *
  2130. * see also find_vpid() etc in include/linux/pid.h
  2131. */
  2132. extern struct task_struct *find_task_by_vpid(pid_t nr);
  2133. extern struct task_struct *find_task_by_pid_ns(pid_t nr,
  2134. struct pid_namespace *ns);
  2135. /* per-UID process charging. */
  2136. extern struct user_struct * alloc_uid(kuid_t);
  2137. static inline struct user_struct *get_uid(struct user_struct *u)
  2138. {
  2139. atomic_inc(&u->__count);
  2140. return u;
  2141. }
  2142. extern void free_uid(struct user_struct *);
  2143. #include <asm/current.h>
  2144. extern void xtime_update(unsigned long ticks);
  2145. extern int wake_up_state(struct task_struct *tsk, unsigned int state);
  2146. extern int wake_up_process(struct task_struct *tsk);
  2147. extern void wake_up_new_task(struct task_struct *tsk);
  2148. #ifdef CONFIG_SMP
  2149. extern void kick_process(struct task_struct *tsk);
  2150. #else
  2151. static inline void kick_process(struct task_struct *tsk) { }
  2152. #endif
  2153. extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
  2154. extern void sched_dead(struct task_struct *p);
  2155. extern void proc_caches_init(void);
  2156. extern void flush_signals(struct task_struct *);
  2157. extern void ignore_signals(struct task_struct *);
  2158. extern void flush_signal_handlers(struct task_struct *, int force_default);
  2159. extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
  2160. static inline int kernel_dequeue_signal(siginfo_t *info)
  2161. {
  2162. struct task_struct *tsk = current;
  2163. siginfo_t __info;
  2164. int ret;
  2165. spin_lock_irq(&tsk->sighand->siglock);
  2166. ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
  2167. spin_unlock_irq(&tsk->sighand->siglock);
  2168. return ret;
  2169. }
  2170. static inline void kernel_signal_stop(void)
  2171. {
  2172. spin_lock_irq(&current->sighand->siglock);
  2173. if (current->jobctl & JOBCTL_STOP_DEQUEUED)
  2174. __set_current_state(TASK_STOPPED);
  2175. spin_unlock_irq(&current->sighand->siglock);
  2176. schedule();
  2177. }
  2178. extern void release_task(struct task_struct * p);
  2179. extern int send_sig_info(int, struct siginfo *, struct task_struct *);
  2180. extern int force_sigsegv(int, struct task_struct *);
  2181. extern int force_sig_info(int, struct siginfo *, struct task_struct *);
  2182. extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
  2183. extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
  2184. extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
  2185. const struct cred *, u32);
  2186. extern int kill_pgrp(struct pid *pid, int sig, int priv);
  2187. extern int kill_pid(struct pid *pid, int sig, int priv);
  2188. extern int kill_proc_info(int, struct siginfo *, pid_t);
  2189. extern __must_check bool do_notify_parent(struct task_struct *, int);
  2190. extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
  2191. extern void force_sig(int, struct task_struct *);
  2192. extern int send_sig(int, struct task_struct *, int);
  2193. extern int zap_other_threads(struct task_struct *p);
  2194. extern struct sigqueue *sigqueue_alloc(void);
  2195. extern void sigqueue_free(struct sigqueue *);
  2196. extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
  2197. extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
  2198. static inline void restore_saved_sigmask(void)
  2199. {
  2200. if (test_and_clear_restore_sigmask())
  2201. __set_current_blocked(&current->saved_sigmask);
  2202. }
  2203. static inline sigset_t *sigmask_to_save(void)
  2204. {
  2205. sigset_t *res = &current->blocked;
  2206. if (unlikely(test_restore_sigmask()))
  2207. res = &current->saved_sigmask;
  2208. return res;
  2209. }
  2210. static inline int kill_cad_pid(int sig, int priv)
  2211. {
  2212. return kill_pid(cad_pid, sig, priv);
  2213. }
  2214. /* These can be the second arg to send_sig_info/send_group_sig_info. */
  2215. #define SEND_SIG_NOINFO ((struct siginfo *) 0)
  2216. #define SEND_SIG_PRIV ((struct siginfo *) 1)
  2217. #define SEND_SIG_FORCED ((struct siginfo *) 2)
  2218. /*
  2219. * True if we are on the alternate signal stack.
  2220. */
  2221. static inline int on_sig_stack(unsigned long sp)
  2222. {
  2223. #ifdef CONFIG_STACK_GROWSUP
  2224. return sp >= current->sas_ss_sp &&
  2225. sp - current->sas_ss_sp < current->sas_ss_size;
  2226. #else
  2227. return sp > current->sas_ss_sp &&
  2228. sp - current->sas_ss_sp <= current->sas_ss_size;
  2229. #endif
  2230. }
  2231. static inline int sas_ss_flags(unsigned long sp)
  2232. {
  2233. if (!current->sas_ss_size)
  2234. return SS_DISABLE;
  2235. return on_sig_stack(sp) ? SS_ONSTACK : 0;
  2236. }
  2237. static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
  2238. {
  2239. if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
  2240. #ifdef CONFIG_STACK_GROWSUP
  2241. return current->sas_ss_sp;
  2242. #else
  2243. return current->sas_ss_sp + current->sas_ss_size;
  2244. #endif
  2245. return sp;
  2246. }
  2247. /*
  2248. * Routines for handling mm_structs
  2249. */
  2250. extern struct mm_struct * mm_alloc(void);
  2251. /* mmdrop drops the mm and the page tables */
  2252. extern void __mmdrop(struct mm_struct *);
  2253. static inline void mmdrop(struct mm_struct * mm)
  2254. {
  2255. if (unlikely(atomic_dec_and_test(&mm->mm_count)))
  2256. __mmdrop(mm);
  2257. }
  2258. /* mmput gets rid of the mappings and all user-space */
  2259. extern void mmput(struct mm_struct *);
  2260. /* Grab a reference to a task's mm, if it is not already going away */
  2261. extern struct mm_struct *get_task_mm(struct task_struct *task);
  2262. /*
  2263. * Grab a reference to a task's mm, if it is not already going away
  2264. * and ptrace_may_access with the mode parameter passed to it
  2265. * succeeds.
  2266. */
  2267. extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
  2268. /* Remove the current tasks stale references to the old mm_struct */
  2269. extern void mm_release(struct task_struct *, struct mm_struct *);
  2270. #ifdef CONFIG_HAVE_COPY_THREAD_TLS
  2271. extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
  2272. struct task_struct *, unsigned long);
  2273. #else
  2274. extern int copy_thread(unsigned long, unsigned long, unsigned long,
  2275. struct task_struct *);
  2276. /* Architectures that haven't opted into copy_thread_tls get the tls argument
  2277. * via pt_regs, so ignore the tls argument passed via C. */
  2278. static inline int copy_thread_tls(
  2279. unsigned long clone_flags, unsigned long sp, unsigned long arg,
  2280. struct task_struct *p, unsigned long tls)
  2281. {
  2282. return copy_thread(clone_flags, sp, arg, p);
  2283. }
  2284. #endif
  2285. extern void flush_thread(void);
  2286. extern void exit_thread(void);
  2287. extern void exit_files(struct task_struct *);
  2288. extern void __cleanup_sighand(struct sighand_struct *);
  2289. extern void exit_itimers(struct signal_struct *);
  2290. extern void flush_itimer_signals(void);
  2291. extern void do_group_exit(int);
  2292. extern int do_execve(struct filename *,
  2293. const char __user * const __user *,
  2294. const char __user * const __user *);
  2295. extern int do_execveat(int, struct filename *,
  2296. const char __user * const __user *,
  2297. const char __user * const __user *,
  2298. int);
  2299. extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
  2300. extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
  2301. struct task_struct *fork_idle(int);
  2302. extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
  2303. extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
  2304. static inline void set_task_comm(struct task_struct *tsk, const char *from)
  2305. {
  2306. __set_task_comm(tsk, from, false);
  2307. }
  2308. extern char *get_task_comm(char *to, struct task_struct *tsk);
  2309. #ifdef CONFIG_SMP
  2310. void scheduler_ipi(void);
  2311. extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
  2312. #else
  2313. static inline void scheduler_ipi(void) { }
  2314. static inline unsigned long wait_task_inactive(struct task_struct *p,
  2315. long match_state)
  2316. {
  2317. return 1;
  2318. }
  2319. #endif
  2320. #define tasklist_empty() \
  2321. list_empty(&init_task.tasks)
  2322. #define next_task(p) \
  2323. list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
  2324. #define for_each_process(p) \
  2325. for (p = &init_task ; (p = next_task(p)) != &init_task ; )
  2326. extern bool current_is_single_threaded(void);
  2327. /*
  2328. * Careful: do_each_thread/while_each_thread is a double loop so
  2329. * 'break' will not work as expected - use goto instead.
  2330. */
  2331. #define do_each_thread(g, t) \
  2332. for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
  2333. #define while_each_thread(g, t) \
  2334. while ((t = next_thread(t)) != g)
  2335. #define __for_each_thread(signal, t) \
  2336. list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
  2337. #define for_each_thread(p, t) \
  2338. __for_each_thread((p)->signal, t)
  2339. /* Careful: this is a double loop, 'break' won't work as expected. */
  2340. #define for_each_process_thread(p, t) \
  2341. for_each_process(p) for_each_thread(p, t)
  2342. static inline int get_nr_threads(struct task_struct *tsk)
  2343. {
  2344. return tsk->signal->nr_threads;
  2345. }
  2346. static inline bool thread_group_leader(struct task_struct *p)
  2347. {
  2348. return p->exit_signal >= 0;
  2349. }
  2350. /* Do to the insanities of de_thread it is possible for a process
  2351. * to have the pid of the thread group leader without actually being
  2352. * the thread group leader. For iteration through the pids in proc
  2353. * all we care about is that we have a task with the appropriate
  2354. * pid, we don't actually care if we have the right task.
  2355. */
  2356. static inline bool has_group_leader_pid(struct task_struct *p)
  2357. {
  2358. return task_pid(p) == p->signal->leader_pid;
  2359. }
  2360. static inline
  2361. bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
  2362. {
  2363. return p1->signal == p2->signal;
  2364. }
  2365. static inline struct task_struct *next_thread(const struct task_struct *p)
  2366. {
  2367. return list_entry_rcu(p->thread_group.next,
  2368. struct task_struct, thread_group);
  2369. }
  2370. static inline int thread_group_empty(struct task_struct *p)
  2371. {
  2372. return list_empty(&p->thread_group);
  2373. }
  2374. #define delay_group_leader(p) \
  2375. (thread_group_leader(p) && !thread_group_empty(p))
  2376. /*
  2377. * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
  2378. * subscriptions and synchronises with wait4(). Also used in procfs. Also
  2379. * pins the final release of task.io_context. Also protects ->cpuset and
  2380. * ->cgroup.subsys[]. And ->vfork_done.
  2381. *
  2382. * Nests both inside and outside of read_lock(&tasklist_lock).
  2383. * It must not be nested with write_lock_irq(&tasklist_lock),
  2384. * neither inside nor outside.
  2385. */
  2386. static inline void task_lock(struct task_struct *p)
  2387. {
  2388. spin_lock(&p->alloc_lock);
  2389. }
  2390. static inline void task_unlock(struct task_struct *p)
  2391. {
  2392. spin_unlock(&p->alloc_lock);
  2393. }
  2394. extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
  2395. unsigned long *flags);
  2396. static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
  2397. unsigned long *flags)
  2398. {
  2399. struct sighand_struct *ret;
  2400. ret = __lock_task_sighand(tsk, flags);
  2401. (void)__cond_lock(&tsk->sighand->siglock, ret);
  2402. return ret;
  2403. }
  2404. static inline void unlock_task_sighand(struct task_struct *tsk,
  2405. unsigned long *flags)
  2406. {
  2407. spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
  2408. }
  2409. /**
  2410. * threadgroup_change_begin - mark the beginning of changes to a threadgroup
  2411. * @tsk: task causing the changes
  2412. *
  2413. * All operations which modify a threadgroup - a new thread joining the
  2414. * group, death of a member thread (the assertion of PF_EXITING) and
  2415. * exec(2) dethreading the process and replacing the leader - are wrapped
  2416. * by threadgroup_change_{begin|end}(). This is to provide a place which
  2417. * subsystems needing threadgroup stability can hook into for
  2418. * synchronization.
  2419. */
  2420. static inline void threadgroup_change_begin(struct task_struct *tsk)
  2421. {
  2422. might_sleep();
  2423. cgroup_threadgroup_change_begin(tsk);
  2424. }
  2425. /**
  2426. * threadgroup_change_end - mark the end of changes to a threadgroup
  2427. * @tsk: task causing the changes
  2428. *
  2429. * See threadgroup_change_begin().
  2430. */
  2431. static inline void threadgroup_change_end(struct task_struct *tsk)
  2432. {
  2433. cgroup_threadgroup_change_end(tsk);
  2434. }
  2435. #ifndef __HAVE_THREAD_FUNCTIONS
  2436. #define task_thread_info(task) ((struct thread_info *)(task)->stack)
  2437. #define task_stack_page(task) ((task)->stack)
  2438. static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
  2439. {
  2440. *task_thread_info(p) = *task_thread_info(org);
  2441. task_thread_info(p)->task = p;
  2442. }
  2443. /*
  2444. * Return the address of the last usable long on the stack.
  2445. *
  2446. * When the stack grows down, this is just above the thread
  2447. * info struct. Going any lower will corrupt the threadinfo.
  2448. *
  2449. * When the stack grows up, this is the highest address.
  2450. * Beyond that position, we corrupt data on the next page.
  2451. */
  2452. static inline unsigned long *end_of_stack(struct task_struct *p)
  2453. {
  2454. #ifdef CONFIG_STACK_GROWSUP
  2455. return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
  2456. #else
  2457. return (unsigned long *)(task_thread_info(p) + 1);
  2458. #endif
  2459. }
  2460. #endif
  2461. #define task_stack_end_corrupted(task) \
  2462. (*(end_of_stack(task)) != STACK_END_MAGIC)
  2463. static inline int object_is_on_stack(void *obj)
  2464. {
  2465. void *stack = task_stack_page(current);
  2466. return (obj >= stack) && (obj < (stack + THREAD_SIZE));
  2467. }
  2468. extern void thread_info_cache_init(void);
  2469. #ifdef CONFIG_DEBUG_STACK_USAGE
  2470. static inline unsigned long stack_not_used(struct task_struct *p)
  2471. {
  2472. unsigned long *n = end_of_stack(p);
  2473. do { /* Skip over canary */
  2474. # ifdef CONFIG_STACK_GROWSUP
  2475. n--;
  2476. # else
  2477. n++;
  2478. # endif
  2479. } while (!*n);
  2480. # ifdef CONFIG_STACK_GROWSUP
  2481. return (unsigned long)end_of_stack(p) - (unsigned long)n;
  2482. # else
  2483. return (unsigned long)n - (unsigned long)end_of_stack(p);
  2484. # endif
  2485. }
  2486. #endif
  2487. extern void set_task_stack_end_magic(struct task_struct *tsk);
  2488. /* set thread flags in other task's structures
  2489. * - see asm/thread_info.h for TIF_xxxx flags available
  2490. */
  2491. static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2492. {
  2493. set_ti_thread_flag(task_thread_info(tsk), flag);
  2494. }
  2495. static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2496. {
  2497. clear_ti_thread_flag(task_thread_info(tsk), flag);
  2498. }
  2499. static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2500. {
  2501. return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
  2502. }
  2503. static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2504. {
  2505. return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
  2506. }
  2507. static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
  2508. {
  2509. return test_ti_thread_flag(task_thread_info(tsk), flag);
  2510. }
  2511. static inline void set_tsk_need_resched(struct task_struct *tsk)
  2512. {
  2513. set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2514. }
  2515. static inline void clear_tsk_need_resched(struct task_struct *tsk)
  2516. {
  2517. clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2518. }
  2519. static inline int test_tsk_need_resched(struct task_struct *tsk)
  2520. {
  2521. return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
  2522. }
  2523. static inline int restart_syscall(void)
  2524. {
  2525. set_tsk_thread_flag(current, TIF_SIGPENDING);
  2526. return -ERESTARTNOINTR;
  2527. }
  2528. static inline int signal_pending(struct task_struct *p)
  2529. {
  2530. return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
  2531. }
  2532. static inline int __fatal_signal_pending(struct task_struct *p)
  2533. {
  2534. return unlikely(sigismember(&p->pending.signal, SIGKILL));
  2535. }
  2536. static inline int fatal_signal_pending(struct task_struct *p)
  2537. {
  2538. return signal_pending(p) && __fatal_signal_pending(p);
  2539. }
  2540. static inline int signal_pending_state(long state, struct task_struct *p)
  2541. {
  2542. if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
  2543. return 0;
  2544. if (!signal_pending(p))
  2545. return 0;
  2546. return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
  2547. }
  2548. /*
  2549. * cond_resched() and cond_resched_lock(): latency reduction via
  2550. * explicit rescheduling in places that are safe. The return
  2551. * value indicates whether a reschedule was done in fact.
  2552. * cond_resched_lock() will drop the spinlock before scheduling,
  2553. * cond_resched_softirq() will enable bhs before scheduling.
  2554. */
  2555. extern int _cond_resched(void);
  2556. #define cond_resched() ({ \
  2557. ___might_sleep(__FILE__, __LINE__, 0); \
  2558. _cond_resched(); \
  2559. })
  2560. extern int __cond_resched_lock(spinlock_t *lock);
  2561. #define cond_resched_lock(lock) ({ \
  2562. ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
  2563. __cond_resched_lock(lock); \
  2564. })
  2565. extern int __cond_resched_softirq(void);
  2566. #define cond_resched_softirq() ({ \
  2567. ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
  2568. __cond_resched_softirq(); \
  2569. })
  2570. static inline void cond_resched_rcu(void)
  2571. {
  2572. #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
  2573. rcu_read_unlock();
  2574. cond_resched();
  2575. rcu_read_lock();
  2576. #endif
  2577. }
  2578. /*
  2579. * Does a critical section need to be broken due to another
  2580. * task waiting?: (technically does not depend on CONFIG_PREEMPT,
  2581. * but a general need for low latency)
  2582. */
  2583. static inline int spin_needbreak(spinlock_t *lock)
  2584. {
  2585. #ifdef CONFIG_PREEMPT
  2586. return spin_is_contended(lock);
  2587. #else
  2588. return 0;
  2589. #endif
  2590. }
  2591. /*
  2592. * Idle thread specific functions to determine the need_resched
  2593. * polling state.
  2594. */
  2595. #ifdef TIF_POLLING_NRFLAG
  2596. static inline int tsk_is_polling(struct task_struct *p)
  2597. {
  2598. return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
  2599. }
  2600. static inline void __current_set_polling(void)
  2601. {
  2602. set_thread_flag(TIF_POLLING_NRFLAG);
  2603. }
  2604. static inline bool __must_check current_set_polling_and_test(void)
  2605. {
  2606. __current_set_polling();
  2607. /*
  2608. * Polling state must be visible before we test NEED_RESCHED,
  2609. * paired by resched_curr()
  2610. */
  2611. smp_mb__after_atomic();
  2612. return unlikely(tif_need_resched());
  2613. }
  2614. static inline void __current_clr_polling(void)
  2615. {
  2616. clear_thread_flag(TIF_POLLING_NRFLAG);
  2617. }
  2618. static inline bool __must_check current_clr_polling_and_test(void)
  2619. {
  2620. __current_clr_polling();
  2621. /*
  2622. * Polling state must be visible before we test NEED_RESCHED,
  2623. * paired by resched_curr()
  2624. */
  2625. smp_mb__after_atomic();
  2626. return unlikely(tif_need_resched());
  2627. }
  2628. #else
  2629. static inline int tsk_is_polling(struct task_struct *p) { return 0; }
  2630. static inline void __current_set_polling(void) { }
  2631. static inline void __current_clr_polling(void) { }
  2632. static inline bool __must_check current_set_polling_and_test(void)
  2633. {
  2634. return unlikely(tif_need_resched());
  2635. }
  2636. static inline bool __must_check current_clr_polling_and_test(void)
  2637. {
  2638. return unlikely(tif_need_resched());
  2639. }
  2640. #endif
  2641. static inline void current_clr_polling(void)
  2642. {
  2643. __current_clr_polling();
  2644. /*
  2645. * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
  2646. * Once the bit is cleared, we'll get IPIs with every new
  2647. * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
  2648. * fold.
  2649. */
  2650. smp_mb(); /* paired with resched_curr() */
  2651. preempt_fold_need_resched();
  2652. }
  2653. static __always_inline bool need_resched(void)
  2654. {
  2655. return unlikely(tif_need_resched());
  2656. }
  2657. /*
  2658. * Thread group CPU time accounting.
  2659. */
  2660. void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
  2661. void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
  2662. /*
  2663. * Reevaluate whether the task has signals pending delivery.
  2664. * Wake the task if so.
  2665. * This is required every time the blocked sigset_t changes.
  2666. * callers must hold sighand->siglock.
  2667. */
  2668. extern void recalc_sigpending_and_wake(struct task_struct *t);
  2669. extern void recalc_sigpending(void);
  2670. extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
  2671. static inline void signal_wake_up(struct task_struct *t, bool resume)
  2672. {
  2673. signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
  2674. }
  2675. static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
  2676. {
  2677. signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
  2678. }
  2679. /*
  2680. * Wrappers for p->thread_info->cpu access. No-op on UP.
  2681. */
  2682. #ifdef CONFIG_SMP
  2683. static inline unsigned int task_cpu(const struct task_struct *p)
  2684. {
  2685. return task_thread_info(p)->cpu;
  2686. }
  2687. static inline int task_node(const struct task_struct *p)
  2688. {
  2689. return cpu_to_node(task_cpu(p));
  2690. }
  2691. extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
  2692. #else
  2693. static inline unsigned int task_cpu(const struct task_struct *p)
  2694. {
  2695. return 0;
  2696. }
  2697. static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
  2698. {
  2699. }
  2700. #endif /* CONFIG_SMP */
  2701. extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
  2702. extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
  2703. #ifdef CONFIG_CGROUP_SCHED
  2704. extern struct task_group root_task_group;
  2705. #endif /* CONFIG_CGROUP_SCHED */
  2706. extern int task_can_switch_user(struct user_struct *up,
  2707. struct task_struct *tsk);
  2708. #ifdef CONFIG_TASK_XACCT
  2709. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  2710. {
  2711. tsk->ioac.rchar += amt;
  2712. }
  2713. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  2714. {
  2715. tsk->ioac.wchar += amt;
  2716. }
  2717. static inline void inc_syscr(struct task_struct *tsk)
  2718. {
  2719. tsk->ioac.syscr++;
  2720. }
  2721. static inline void inc_syscw(struct task_struct *tsk)
  2722. {
  2723. tsk->ioac.syscw++;
  2724. }
  2725. #else
  2726. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  2727. {
  2728. }
  2729. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  2730. {
  2731. }
  2732. static inline void inc_syscr(struct task_struct *tsk)
  2733. {
  2734. }
  2735. static inline void inc_syscw(struct task_struct *tsk)
  2736. {
  2737. }
  2738. #endif
  2739. #ifndef TASK_SIZE_OF
  2740. #define TASK_SIZE_OF(tsk) TASK_SIZE
  2741. #endif
  2742. #ifdef CONFIG_MEMCG
  2743. extern void mm_update_next_owner(struct mm_struct *mm);
  2744. #else
  2745. static inline void mm_update_next_owner(struct mm_struct *mm)
  2746. {
  2747. }
  2748. #endif /* CONFIG_MEMCG */
  2749. static inline unsigned long task_rlimit(const struct task_struct *tsk,
  2750. unsigned int limit)
  2751. {
  2752. return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
  2753. }
  2754. static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
  2755. unsigned int limit)
  2756. {
  2757. return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
  2758. }
  2759. static inline unsigned long rlimit(unsigned int limit)
  2760. {
  2761. return task_rlimit(current, limit);
  2762. }
  2763. static inline unsigned long rlimit_max(unsigned int limit)
  2764. {
  2765. return task_rlimit_max(current, limit);
  2766. }
  2767. #ifdef CONFIG_CPU_FREQ
  2768. struct update_util_data {
  2769. void (*func)(struct update_util_data *data,
  2770. u64 time, unsigned long util, unsigned long max);
  2771. };
  2772. void cpufreq_set_update_util_data(int cpu, struct update_util_data *data);
  2773. #endif /* CONFIG_CPU_FREQ */
  2774. #endif