sched.h 89 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522252325242525252625272528252925302531253225332534253525362537253825392540254125422543254425452546254725482549255025512552255325542555255625572558255925602561256225632564256525662567256825692570257125722573257425752576257725782579258025812582258325842585258625872588258925902591259225932594259525962597259825992600260126022603260426052606260726082609261026112612261326142615261626172618261926202621262226232624262526262627262826292630263126322633263426352636263726382639264026412642264326442645264626472648264926502651265226532654265526562657265826592660266126622663266426652666266726682669267026712672267326742675267626772678267926802681268226832684268526862687268826892690269126922693269426952696269726982699270027012702270327042705270627072708270927102711271227132714271527162717271827192720272127222723272427252726272727282729273027312732273327342735273627372738273927402741274227432744274527462747274827492750275127522753275427552756275727582759276027612762276327642765276627672768276927702771277227732774277527762777277827792780278127822783278427852786278727882789279027912792279327942795279627972798279928002801280228032804280528062807280828092810281128122813281428152816281728182819282028212822282328242825282628272828282928302831283228332834283528362837283828392840284128422843284428452846284728482849285028512852285328542855285628572858285928602861286228632864286528662867286828692870287128722873287428752876287728782879288028812882288328842885288628872888288928902891289228932894289528962897289828992900290129022903290429052906290729082909291029112912291329142915291629172918291929202921292229232924292529262927292829292930293129322933293429352936293729382939294029412942294329442945294629472948294929502951295229532954295529562957295829592960296129622963296429652966296729682969297029712972297329742975297629772978297929802981298229832984298529862987298829892990299129922993299429952996299729982999300030013002300330043005300630073008300930103011301230133014301530163017301830193020302130223023302430253026302730283029303030313032303330343035303630373038303930403041304230433044304530463047304830493050305130523053305430553056305730583059306030613062306330643065306630673068306930703071307230733074307530763077307830793080308130823083308430853086308730883089309030913092309330943095309630973098309931003101310231033104310531063107310831093110311131123113311431153116311731183119312031213122312331243125312631273128312931303131313231333134313531363137313831393140314131423143
  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/task_io_accounting.h>
  46. #include <linux/latencytop.h>
  47. #include <linux/cred.h>
  48. #include <linux/llist.h>
  49. #include <linux/uidgid.h>
  50. #include <linux/gfp.h>
  51. #include <linux/magic.h>
  52. #include <linux/cgroup-defs.h>
  53. #include <asm/processor.h>
  54. #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
  55. /*
  56. * Extended scheduling parameters data structure.
  57. *
  58. * This is needed because the original struct sched_param can not be
  59. * altered without introducing ABI issues with legacy applications
  60. * (e.g., in sched_getparam()).
  61. *
  62. * However, the possibility of specifying more than just a priority for
  63. * the tasks may be useful for a wide variety of application fields, e.g.,
  64. * multimedia, streaming, automation and control, and many others.
  65. *
  66. * This variant (sched_attr) is meant at describing a so-called
  67. * sporadic time-constrained task. In such model a task is specified by:
  68. * - the activation period or minimum instance inter-arrival time;
  69. * - the maximum (or average, depending on the actual scheduling
  70. * discipline) computation time of all instances, a.k.a. runtime;
  71. * - the deadline (relative to the actual activation time) of each
  72. * instance.
  73. * Very briefly, a periodic (sporadic) task asks for the execution of
  74. * some specific computation --which is typically called an instance--
  75. * (at most) every period. Moreover, each instance typically lasts no more
  76. * than the runtime and must be completed by time instant t equal to
  77. * the instance activation time + the deadline.
  78. *
  79. * This is reflected by the actual fields of the sched_attr structure:
  80. *
  81. * @size size of the structure, for fwd/bwd compat.
  82. *
  83. * @sched_policy task's scheduling policy
  84. * @sched_flags for customizing the scheduler behaviour
  85. * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
  86. * @sched_priority task's static priority (SCHED_FIFO/RR)
  87. * @sched_deadline representative of the task's deadline
  88. * @sched_runtime representative of the task's runtime
  89. * @sched_period representative of the task's period
  90. *
  91. * Given this task model, there are a multiplicity of scheduling algorithms
  92. * and policies, that can be used to ensure all the tasks will make their
  93. * timing constraints.
  94. *
  95. * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
  96. * only user of this new interface. More information about the algorithm
  97. * available in the scheduling class file or in Documentation/.
  98. */
  99. struct sched_attr {
  100. u32 size;
  101. u32 sched_policy;
  102. u64 sched_flags;
  103. /* SCHED_NORMAL, SCHED_BATCH */
  104. s32 sched_nice;
  105. /* SCHED_FIFO, SCHED_RR */
  106. u32 sched_priority;
  107. /* SCHED_DEADLINE */
  108. u64 sched_runtime;
  109. u64 sched_deadline;
  110. u64 sched_period;
  111. };
  112. struct futex_pi_state;
  113. struct robust_list_head;
  114. struct bio_list;
  115. struct fs_struct;
  116. struct perf_event_context;
  117. struct blk_plug;
  118. struct filename;
  119. struct nameidata;
  120. #define VMACACHE_BITS 2
  121. #define VMACACHE_SIZE (1U << VMACACHE_BITS)
  122. #define VMACACHE_MASK (VMACACHE_SIZE - 1)
  123. /*
  124. * These are the constant used to fake the fixed-point load-average
  125. * counting. Some notes:
  126. * - 11 bit fractions expand to 22 bits by the multiplies: this gives
  127. * a load-average precision of 10 bits integer + 11 bits fractional
  128. * - if you want to count load-averages more often, you need more
  129. * precision, or rounding will get you. With 2-second counting freq,
  130. * the EXP_n values would be 1981, 2034 and 2043 if still using only
  131. * 11 bit fractions.
  132. */
  133. extern unsigned long avenrun[]; /* Load averages */
  134. extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
  135. #define FSHIFT 11 /* nr of bits of precision */
  136. #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
  137. #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
  138. #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
  139. #define EXP_5 2014 /* 1/exp(5sec/5min) */
  140. #define EXP_15 2037 /* 1/exp(5sec/15min) */
  141. #define CALC_LOAD(load,exp,n) \
  142. load *= exp; \
  143. load += n*(FIXED_1-exp); \
  144. load >>= FSHIFT;
  145. extern unsigned long total_forks;
  146. extern int nr_threads;
  147. DECLARE_PER_CPU(unsigned long, process_counts);
  148. extern int nr_processes(void);
  149. extern unsigned long nr_running(void);
  150. extern bool single_task_running(void);
  151. extern unsigned long nr_iowait(void);
  152. extern unsigned long nr_iowait_cpu(int cpu);
  153. extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
  154. extern void calc_global_load(unsigned long ticks);
  155. #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
  156. extern void update_cpu_load_nohz(void);
  157. #else
  158. static inline void update_cpu_load_nohz(void) { }
  159. #endif
  160. extern unsigned long get_parent_ip(unsigned long addr);
  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(void);
  323. extern void touch_softlockup_watchdog_sync(void);
  324. extern void touch_all_softlockup_watchdogs(void);
  325. extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
  326. void __user *buffer,
  327. size_t *lenp, loff_t *ppos);
  328. extern unsigned int softlockup_panic;
  329. void lockup_detector_init(void);
  330. #else
  331. static inline void touch_softlockup_watchdog(void)
  332. {
  333. }
  334. static inline void touch_softlockup_watchdog_sync(void)
  335. {
  336. }
  337. static inline void touch_all_softlockup_watchdogs(void)
  338. {
  339. }
  340. static inline void lockup_detector_init(void)
  341. {
  342. }
  343. #endif
  344. #ifdef CONFIG_DETECT_HUNG_TASK
  345. void reset_hung_task_detector(void);
  346. #else
  347. static inline void reset_hung_task_detector(void)
  348. {
  349. }
  350. #endif
  351. /* Attach to any functions which should be ignored in wchan output. */
  352. #define __sched __attribute__((__section__(".sched.text")))
  353. /* Linker adds these: start and end of __sched functions */
  354. extern char __sched_text_start[], __sched_text_end[];
  355. /* Is this address in the __sched functions? */
  356. extern int in_sched_functions(unsigned long addr);
  357. #define MAX_SCHEDULE_TIMEOUT LONG_MAX
  358. extern signed long schedule_timeout(signed long timeout);
  359. extern signed long schedule_timeout_interruptible(signed long timeout);
  360. extern signed long schedule_timeout_killable(signed long timeout);
  361. extern signed long schedule_timeout_uninterruptible(signed long timeout);
  362. asmlinkage void schedule(void);
  363. extern void schedule_preempt_disabled(void);
  364. extern long io_schedule_timeout(long timeout);
  365. static inline void io_schedule(void)
  366. {
  367. io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
  368. }
  369. struct nsproxy;
  370. struct user_namespace;
  371. #ifdef CONFIG_MMU
  372. extern void arch_pick_mmap_layout(struct mm_struct *mm);
  373. extern unsigned long
  374. arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
  375. unsigned long, unsigned long);
  376. extern unsigned long
  377. arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
  378. unsigned long len, unsigned long pgoff,
  379. unsigned long flags);
  380. #else
  381. static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
  382. #endif
  383. #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
  384. #define SUID_DUMP_USER 1 /* Dump as user of process */
  385. #define SUID_DUMP_ROOT 2 /* Dump as root */
  386. /* mm flags */
  387. /* for SUID_DUMP_* above */
  388. #define MMF_DUMPABLE_BITS 2
  389. #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
  390. extern void set_dumpable(struct mm_struct *mm, int value);
  391. /*
  392. * This returns the actual value of the suid_dumpable flag. For things
  393. * that are using this for checking for privilege transitions, it must
  394. * test against SUID_DUMP_USER rather than treating it as a boolean
  395. * value.
  396. */
  397. static inline int __get_dumpable(unsigned long mm_flags)
  398. {
  399. return mm_flags & MMF_DUMPABLE_MASK;
  400. }
  401. static inline int get_dumpable(struct mm_struct *mm)
  402. {
  403. return __get_dumpable(mm->flags);
  404. }
  405. /* coredump filter bits */
  406. #define MMF_DUMP_ANON_PRIVATE 2
  407. #define MMF_DUMP_ANON_SHARED 3
  408. #define MMF_DUMP_MAPPED_PRIVATE 4
  409. #define MMF_DUMP_MAPPED_SHARED 5
  410. #define MMF_DUMP_ELF_HEADERS 6
  411. #define MMF_DUMP_HUGETLB_PRIVATE 7
  412. #define MMF_DUMP_HUGETLB_SHARED 8
  413. #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
  414. #define MMF_DUMP_FILTER_BITS 7
  415. #define MMF_DUMP_FILTER_MASK \
  416. (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
  417. #define MMF_DUMP_FILTER_DEFAULT \
  418. ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
  419. (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
  420. #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
  421. # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
  422. #else
  423. # define MMF_DUMP_MASK_DEFAULT_ELF 0
  424. #endif
  425. /* leave room for more dump flags */
  426. #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
  427. #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
  428. #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
  429. #define MMF_HAS_UPROBES 19 /* has uprobes */
  430. #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
  431. #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
  432. struct sighand_struct {
  433. atomic_t count;
  434. struct k_sigaction action[_NSIG];
  435. spinlock_t siglock;
  436. wait_queue_head_t signalfd_wqh;
  437. };
  438. struct pacct_struct {
  439. int ac_flag;
  440. long ac_exitcode;
  441. unsigned long ac_mem;
  442. cputime_t ac_utime, ac_stime;
  443. unsigned long ac_minflt, ac_majflt;
  444. };
  445. struct cpu_itimer {
  446. cputime_t expires;
  447. cputime_t incr;
  448. u32 error;
  449. u32 incr_error;
  450. };
  451. /**
  452. * struct prev_cputime - snaphsot of system and user cputime
  453. * @utime: time spent in user mode
  454. * @stime: time spent in system mode
  455. * @lock: protects the above two fields
  456. *
  457. * Stores previous user/system time values such that we can guarantee
  458. * monotonicity.
  459. */
  460. struct prev_cputime {
  461. #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
  462. cputime_t utime;
  463. cputime_t stime;
  464. raw_spinlock_t lock;
  465. #endif
  466. };
  467. static inline void prev_cputime_init(struct prev_cputime *prev)
  468. {
  469. #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
  470. prev->utime = prev->stime = 0;
  471. raw_spin_lock_init(&prev->lock);
  472. #endif
  473. }
  474. /**
  475. * struct task_cputime - collected CPU time counts
  476. * @utime: time spent in user mode, in &cputime_t units
  477. * @stime: time spent in kernel mode, in &cputime_t units
  478. * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
  479. *
  480. * This structure groups together three kinds of CPU time that are tracked for
  481. * threads and thread groups. Most things considering CPU time want to group
  482. * these counts together and treat all three of them in parallel.
  483. */
  484. struct task_cputime {
  485. cputime_t utime;
  486. cputime_t stime;
  487. unsigned long long sum_exec_runtime;
  488. };
  489. /* Alternate field names when used to cache expirations. */
  490. #define virt_exp utime
  491. #define prof_exp stime
  492. #define sched_exp sum_exec_runtime
  493. #define INIT_CPUTIME \
  494. (struct task_cputime) { \
  495. .utime = 0, \
  496. .stime = 0, \
  497. .sum_exec_runtime = 0, \
  498. }
  499. /*
  500. * This is the atomic variant of task_cputime, which can be used for
  501. * storing and updating task_cputime statistics without locking.
  502. */
  503. struct task_cputime_atomic {
  504. atomic64_t utime;
  505. atomic64_t stime;
  506. atomic64_t sum_exec_runtime;
  507. };
  508. #define INIT_CPUTIME_ATOMIC \
  509. (struct task_cputime_atomic) { \
  510. .utime = ATOMIC64_INIT(0), \
  511. .stime = ATOMIC64_INIT(0), \
  512. .sum_exec_runtime = ATOMIC64_INIT(0), \
  513. }
  514. #ifdef CONFIG_PREEMPT_COUNT
  515. #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
  516. #else
  517. #define PREEMPT_DISABLED PREEMPT_ENABLED
  518. #endif
  519. /*
  520. * Disable preemption until the scheduler is running.
  521. * Reset by start_kernel()->sched_init()->init_idle().
  522. *
  523. * We include PREEMPT_ACTIVE to avoid cond_resched() from working
  524. * before the scheduler is active -- see should_resched().
  525. */
  526. #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
  527. /**
  528. * struct thread_group_cputimer - thread group interval timer counts
  529. * @cputime_atomic: atomic thread group interval timers.
  530. * @running: non-zero when there are timers running and
  531. * @cputime receives updates.
  532. *
  533. * This structure contains the version of task_cputime, above, that is
  534. * used for thread group CPU timer calculations.
  535. */
  536. struct thread_group_cputimer {
  537. struct task_cputime_atomic cputime_atomic;
  538. int running;
  539. };
  540. #include <linux/rwsem.h>
  541. struct autogroup;
  542. /*
  543. * NOTE! "signal_struct" does not have its own
  544. * locking, because a shared signal_struct always
  545. * implies a shared sighand_struct, so locking
  546. * sighand_struct is always a proper superset of
  547. * the locking of signal_struct.
  548. */
  549. struct signal_struct {
  550. atomic_t sigcnt;
  551. atomic_t live;
  552. int nr_threads;
  553. struct list_head thread_head;
  554. wait_queue_head_t wait_chldexit; /* for wait4() */
  555. /* current thread group signal load-balancing target: */
  556. struct task_struct *curr_target;
  557. /* shared signal handling: */
  558. struct sigpending shared_pending;
  559. /* thread group exit support */
  560. int group_exit_code;
  561. /* overloaded:
  562. * - notify group_exit_task when ->count is equal to notify_count
  563. * - everyone except group_exit_task is stopped during signal delivery
  564. * of fatal signals, group_exit_task processes the signal.
  565. */
  566. int notify_count;
  567. struct task_struct *group_exit_task;
  568. /* thread group stop support, overloads group_exit_code too */
  569. int group_stop_count;
  570. unsigned int flags; /* see SIGNAL_* flags below */
  571. /*
  572. * PR_SET_CHILD_SUBREAPER marks a process, like a service
  573. * manager, to re-parent orphan (double-forking) child processes
  574. * to this process instead of 'init'. The service manager is
  575. * able to receive SIGCHLD signals and is able to investigate
  576. * the process until it calls wait(). All children of this
  577. * process will inherit a flag if they should look for a
  578. * child_subreaper process at exit.
  579. */
  580. unsigned int is_child_subreaper:1;
  581. unsigned int has_child_subreaper:1;
  582. /* POSIX.1b Interval Timers */
  583. int posix_timer_id;
  584. struct list_head posix_timers;
  585. /* ITIMER_REAL timer for the process */
  586. struct hrtimer real_timer;
  587. struct pid *leader_pid;
  588. ktime_t it_real_incr;
  589. /*
  590. * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
  591. * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
  592. * values are defined to 0 and 1 respectively
  593. */
  594. struct cpu_itimer it[2];
  595. /*
  596. * Thread group totals for process CPU timers.
  597. * See thread_group_cputimer(), et al, for details.
  598. */
  599. struct thread_group_cputimer cputimer;
  600. /* Earliest-expiration cache. */
  601. struct task_cputime cputime_expires;
  602. struct list_head cpu_timers[3];
  603. struct pid *tty_old_pgrp;
  604. /* boolean value for session group leader */
  605. int leader;
  606. struct tty_struct *tty; /* NULL if no tty */
  607. #ifdef CONFIG_SCHED_AUTOGROUP
  608. struct autogroup *autogroup;
  609. #endif
  610. /*
  611. * Cumulative resource counters for dead threads in the group,
  612. * and for reaped dead child processes forked by this group.
  613. * Live threads maintain their own counters and add to these
  614. * in __exit_signal, except for the group leader.
  615. */
  616. seqlock_t stats_lock;
  617. cputime_t utime, stime, cutime, cstime;
  618. cputime_t gtime;
  619. cputime_t cgtime;
  620. struct prev_cputime prev_cputime;
  621. unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
  622. unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
  623. unsigned long inblock, oublock, cinblock, coublock;
  624. unsigned long maxrss, cmaxrss;
  625. struct task_io_accounting ioac;
  626. /*
  627. * Cumulative ns of schedule CPU time fo dead threads in the
  628. * group, not including a zombie group leader, (This only differs
  629. * from jiffies_to_ns(utime + stime) if sched_clock uses something
  630. * other than jiffies.)
  631. */
  632. unsigned long long sum_sched_runtime;
  633. /*
  634. * We don't bother to synchronize most readers of this at all,
  635. * because there is no reader checking a limit that actually needs
  636. * to get both rlim_cur and rlim_max atomically, and either one
  637. * alone is a single word that can safely be read normally.
  638. * getrlimit/setrlimit use task_lock(current->group_leader) to
  639. * protect this instead of the siglock, because they really
  640. * have no need to disable irqs.
  641. */
  642. struct rlimit rlim[RLIM_NLIMITS];
  643. #ifdef CONFIG_BSD_PROCESS_ACCT
  644. struct pacct_struct pacct; /* per-process accounting information */
  645. #endif
  646. #ifdef CONFIG_TASKSTATS
  647. struct taskstats *stats;
  648. #endif
  649. #ifdef CONFIG_AUDIT
  650. unsigned audit_tty;
  651. unsigned audit_tty_log_passwd;
  652. struct tty_audit_buf *tty_audit_buf;
  653. #endif
  654. oom_flags_t oom_flags;
  655. short oom_score_adj; /* OOM kill score adjustment */
  656. short oom_score_adj_min; /* OOM kill score adjustment min value.
  657. * Only settable by CAP_SYS_RESOURCE. */
  658. struct mutex cred_guard_mutex; /* guard against foreign influences on
  659. * credential calculations
  660. * (notably. ptrace) */
  661. };
  662. /*
  663. * Bits in flags field of signal_struct.
  664. */
  665. #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
  666. #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
  667. #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
  668. #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
  669. /*
  670. * Pending notifications to parent.
  671. */
  672. #define SIGNAL_CLD_STOPPED 0x00000010
  673. #define SIGNAL_CLD_CONTINUED 0x00000020
  674. #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
  675. #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
  676. /* If true, all threads except ->group_exit_task have pending SIGKILL */
  677. static inline int signal_group_exit(const struct signal_struct *sig)
  678. {
  679. return (sig->flags & SIGNAL_GROUP_EXIT) ||
  680. (sig->group_exit_task != NULL);
  681. }
  682. /*
  683. * Some day this will be a full-fledged user tracking system..
  684. */
  685. struct user_struct {
  686. atomic_t __count; /* reference count */
  687. atomic_t processes; /* How many processes does this user have? */
  688. atomic_t sigpending; /* How many pending signals does this user have? */
  689. #ifdef CONFIG_INOTIFY_USER
  690. atomic_t inotify_watches; /* How many inotify watches does this user have? */
  691. atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
  692. #endif
  693. #ifdef CONFIG_FANOTIFY
  694. atomic_t fanotify_listeners;
  695. #endif
  696. #ifdef CONFIG_EPOLL
  697. atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
  698. #endif
  699. #ifdef CONFIG_POSIX_MQUEUE
  700. /* protected by mq_lock */
  701. unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
  702. #endif
  703. unsigned long locked_shm; /* How many pages of mlocked shm ? */
  704. #ifdef CONFIG_KEYS
  705. struct key *uid_keyring; /* UID specific keyring */
  706. struct key *session_keyring; /* UID's default session keyring */
  707. #endif
  708. /* Hash table maintenance information */
  709. struct hlist_node uidhash_node;
  710. kuid_t uid;
  711. #ifdef CONFIG_PERF_EVENTS
  712. atomic_long_t locked_vm;
  713. #endif
  714. };
  715. extern int uids_sysfs_init(void);
  716. extern struct user_struct *find_user(kuid_t);
  717. extern struct user_struct root_user;
  718. #define INIT_USER (&root_user)
  719. struct backing_dev_info;
  720. struct reclaim_state;
  721. #ifdef CONFIG_SCHED_INFO
  722. struct sched_info {
  723. /* cumulative counters */
  724. unsigned long pcount; /* # of times run on this cpu */
  725. unsigned long long run_delay; /* time spent waiting on a runqueue */
  726. /* timestamps */
  727. unsigned long long last_arrival,/* when we last ran on a cpu */
  728. last_queued; /* when we were last queued to run */
  729. };
  730. #endif /* CONFIG_SCHED_INFO */
  731. #ifdef CONFIG_TASK_DELAY_ACCT
  732. struct task_delay_info {
  733. spinlock_t lock;
  734. unsigned int flags; /* Private per-task flags */
  735. /* For each stat XXX, add following, aligned appropriately
  736. *
  737. * struct timespec XXX_start, XXX_end;
  738. * u64 XXX_delay;
  739. * u32 XXX_count;
  740. *
  741. * Atomicity of updates to XXX_delay, XXX_count protected by
  742. * single lock above (split into XXX_lock if contention is an issue).
  743. */
  744. /*
  745. * XXX_count is incremented on every XXX operation, the delay
  746. * associated with the operation is added to XXX_delay.
  747. * XXX_delay contains the accumulated delay time in nanoseconds.
  748. */
  749. u64 blkio_start; /* Shared by blkio, swapin */
  750. u64 blkio_delay; /* wait for sync block io completion */
  751. u64 swapin_delay; /* wait for swapin block io completion */
  752. u32 blkio_count; /* total count of the number of sync block */
  753. /* io operations performed */
  754. u32 swapin_count; /* total count of the number of swapin block */
  755. /* io operations performed */
  756. u64 freepages_start;
  757. u64 freepages_delay; /* wait for memory reclaim */
  758. u32 freepages_count; /* total count of memory reclaim */
  759. };
  760. #endif /* CONFIG_TASK_DELAY_ACCT */
  761. static inline int sched_info_on(void)
  762. {
  763. #ifdef CONFIG_SCHEDSTATS
  764. return 1;
  765. #elif defined(CONFIG_TASK_DELAY_ACCT)
  766. extern int delayacct_on;
  767. return delayacct_on;
  768. #else
  769. return 0;
  770. #endif
  771. }
  772. enum cpu_idle_type {
  773. CPU_IDLE,
  774. CPU_NOT_IDLE,
  775. CPU_NEWLY_IDLE,
  776. CPU_MAX_IDLE_TYPES
  777. };
  778. /*
  779. * Increase resolution of cpu_capacity calculations
  780. */
  781. #define SCHED_CAPACITY_SHIFT 10
  782. #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
  783. /*
  784. * Wake-queues are lists of tasks with a pending wakeup, whose
  785. * callers have already marked the task as woken internally,
  786. * and can thus carry on. A common use case is being able to
  787. * do the wakeups once the corresponding user lock as been
  788. * released.
  789. *
  790. * We hold reference to each task in the list across the wakeup,
  791. * thus guaranteeing that the memory is still valid by the time
  792. * the actual wakeups are performed in wake_up_q().
  793. *
  794. * One per task suffices, because there's never a need for a task to be
  795. * in two wake queues simultaneously; it is forbidden to abandon a task
  796. * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
  797. * already in a wake queue, the wakeup will happen soon and the second
  798. * waker can just skip it.
  799. *
  800. * The WAKE_Q macro declares and initializes the list head.
  801. * wake_up_q() does NOT reinitialize the list; it's expected to be
  802. * called near the end of a function, where the fact that the queue is
  803. * not used again will be easy to see by inspection.
  804. *
  805. * Note that this can cause spurious wakeups. schedule() callers
  806. * must ensure the call is done inside a loop, confirming that the
  807. * wakeup condition has in fact occurred.
  808. */
  809. struct wake_q_node {
  810. struct wake_q_node *next;
  811. };
  812. struct wake_q_head {
  813. struct wake_q_node *first;
  814. struct wake_q_node **lastp;
  815. };
  816. #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
  817. #define WAKE_Q(name) \
  818. struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
  819. extern void wake_q_add(struct wake_q_head *head,
  820. struct task_struct *task);
  821. extern void wake_up_q(struct wake_q_head *head);
  822. /*
  823. * sched-domains (multiprocessor balancing) declarations:
  824. */
  825. #ifdef CONFIG_SMP
  826. #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
  827. #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
  828. #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
  829. #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
  830. #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
  831. #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
  832. #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
  833. #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
  834. #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
  835. #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
  836. #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
  837. #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
  838. #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
  839. #define SD_NUMA 0x4000 /* cross-node balancing */
  840. #ifdef CONFIG_SCHED_SMT
  841. static inline int cpu_smt_flags(void)
  842. {
  843. return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
  844. }
  845. #endif
  846. #ifdef CONFIG_SCHED_MC
  847. static inline int cpu_core_flags(void)
  848. {
  849. return SD_SHARE_PKG_RESOURCES;
  850. }
  851. #endif
  852. #ifdef CONFIG_NUMA
  853. static inline int cpu_numa_flags(void)
  854. {
  855. return SD_NUMA;
  856. }
  857. #endif
  858. struct sched_domain_attr {
  859. int relax_domain_level;
  860. };
  861. #define SD_ATTR_INIT (struct sched_domain_attr) { \
  862. .relax_domain_level = -1, \
  863. }
  864. extern int sched_domain_level_max;
  865. struct sched_group;
  866. struct sched_domain {
  867. /* These fields must be setup */
  868. struct sched_domain *parent; /* top domain must be null terminated */
  869. struct sched_domain *child; /* bottom domain must be null terminated */
  870. struct sched_group *groups; /* the balancing groups of the domain */
  871. unsigned long min_interval; /* Minimum balance interval ms */
  872. unsigned long max_interval; /* Maximum balance interval ms */
  873. unsigned int busy_factor; /* less balancing by factor if busy */
  874. unsigned int imbalance_pct; /* No balance until over watermark */
  875. unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
  876. unsigned int busy_idx;
  877. unsigned int idle_idx;
  878. unsigned int newidle_idx;
  879. unsigned int wake_idx;
  880. unsigned int forkexec_idx;
  881. unsigned int smt_gain;
  882. int nohz_idle; /* NOHZ IDLE status */
  883. int flags; /* See SD_* */
  884. int level;
  885. /* Runtime fields. */
  886. unsigned long last_balance; /* init to jiffies. units in jiffies */
  887. unsigned int balance_interval; /* initialise to 1. units in ms. */
  888. unsigned int nr_balance_failed; /* initialise to 0 */
  889. /* idle_balance() stats */
  890. u64 max_newidle_lb_cost;
  891. unsigned long next_decay_max_lb_cost;
  892. #ifdef CONFIG_SCHEDSTATS
  893. /* load_balance() stats */
  894. unsigned int lb_count[CPU_MAX_IDLE_TYPES];
  895. unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
  896. unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
  897. unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
  898. unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
  899. unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
  900. unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
  901. unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
  902. /* Active load balancing */
  903. unsigned int alb_count;
  904. unsigned int alb_failed;
  905. unsigned int alb_pushed;
  906. /* SD_BALANCE_EXEC stats */
  907. unsigned int sbe_count;
  908. unsigned int sbe_balanced;
  909. unsigned int sbe_pushed;
  910. /* SD_BALANCE_FORK stats */
  911. unsigned int sbf_count;
  912. unsigned int sbf_balanced;
  913. unsigned int sbf_pushed;
  914. /* try_to_wake_up() stats */
  915. unsigned int ttwu_wake_remote;
  916. unsigned int ttwu_move_affine;
  917. unsigned int ttwu_move_balance;
  918. #endif
  919. #ifdef CONFIG_SCHED_DEBUG
  920. char *name;
  921. #endif
  922. union {
  923. void *private; /* used during construction */
  924. struct rcu_head rcu; /* used during destruction */
  925. };
  926. unsigned int span_weight;
  927. /*
  928. * Span of all CPUs in this domain.
  929. *
  930. * NOTE: this field is variable length. (Allocated dynamically
  931. * by attaching extra space to the end of the structure,
  932. * depending on how many CPUs the kernel has booted up with)
  933. */
  934. unsigned long span[0];
  935. };
  936. static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
  937. {
  938. return to_cpumask(sd->span);
  939. }
  940. extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  941. struct sched_domain_attr *dattr_new);
  942. /* Allocate an array of sched domains, for partition_sched_domains(). */
  943. cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
  944. void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
  945. bool cpus_share_cache(int this_cpu, int that_cpu);
  946. typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
  947. typedef int (*sched_domain_flags_f)(void);
  948. #define SDTL_OVERLAP 0x01
  949. struct sd_data {
  950. struct sched_domain **__percpu sd;
  951. struct sched_group **__percpu sg;
  952. struct sched_group_capacity **__percpu sgc;
  953. };
  954. struct sched_domain_topology_level {
  955. sched_domain_mask_f mask;
  956. sched_domain_flags_f sd_flags;
  957. int flags;
  958. int numa_level;
  959. struct sd_data data;
  960. #ifdef CONFIG_SCHED_DEBUG
  961. char *name;
  962. #endif
  963. };
  964. extern struct sched_domain_topology_level *sched_domain_topology;
  965. extern void set_sched_topology(struct sched_domain_topology_level *tl);
  966. extern void wake_up_if_idle(int cpu);
  967. #ifdef CONFIG_SCHED_DEBUG
  968. # define SD_INIT_NAME(type) .name = #type
  969. #else
  970. # define SD_INIT_NAME(type)
  971. #endif
  972. #else /* CONFIG_SMP */
  973. struct sched_domain_attr;
  974. static inline void
  975. partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  976. struct sched_domain_attr *dattr_new)
  977. {
  978. }
  979. static inline bool cpus_share_cache(int this_cpu, int that_cpu)
  980. {
  981. return true;
  982. }
  983. #endif /* !CONFIG_SMP */
  984. struct io_context; /* See blkdev.h */
  985. #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
  986. extern void prefetch_stack(struct task_struct *t);
  987. #else
  988. static inline void prefetch_stack(struct task_struct *t) { }
  989. #endif
  990. struct audit_context; /* See audit.c */
  991. struct mempolicy;
  992. struct pipe_inode_info;
  993. struct uts_namespace;
  994. struct load_weight {
  995. unsigned long weight;
  996. u32 inv_weight;
  997. };
  998. /*
  999. * The load_avg/util_avg accumulates an infinite geometric series.
  1000. * 1) load_avg factors the amount of time that a sched_entity is
  1001. * runnable on a rq into its weight. For cfs_rq, it is the aggregated
  1002. * such weights of all runnable and blocked sched_entities.
  1003. * 2) util_avg factors frequency scaling into the amount of time
  1004. * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
  1005. * For cfs_rq, it is the aggregated such times of all runnable and
  1006. * blocked sched_entities.
  1007. * The 64 bit load_sum can:
  1008. * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
  1009. * the highest weight (=88761) always runnable, we should not overflow
  1010. * 2) for entity, support any load.weight always runnable
  1011. */
  1012. struct sched_avg {
  1013. u64 last_update_time, load_sum;
  1014. u32 util_sum, period_contrib;
  1015. unsigned long load_avg, util_avg;
  1016. };
  1017. #ifdef CONFIG_SCHEDSTATS
  1018. struct sched_statistics {
  1019. u64 wait_start;
  1020. u64 wait_max;
  1021. u64 wait_count;
  1022. u64 wait_sum;
  1023. u64 iowait_count;
  1024. u64 iowait_sum;
  1025. u64 sleep_start;
  1026. u64 sleep_max;
  1027. s64 sum_sleep_runtime;
  1028. u64 block_start;
  1029. u64 block_max;
  1030. u64 exec_max;
  1031. u64 slice_max;
  1032. u64 nr_migrations_cold;
  1033. u64 nr_failed_migrations_affine;
  1034. u64 nr_failed_migrations_running;
  1035. u64 nr_failed_migrations_hot;
  1036. u64 nr_forced_migrations;
  1037. u64 nr_wakeups;
  1038. u64 nr_wakeups_sync;
  1039. u64 nr_wakeups_migrate;
  1040. u64 nr_wakeups_local;
  1041. u64 nr_wakeups_remote;
  1042. u64 nr_wakeups_affine;
  1043. u64 nr_wakeups_affine_attempts;
  1044. u64 nr_wakeups_passive;
  1045. u64 nr_wakeups_idle;
  1046. };
  1047. #endif
  1048. struct sched_entity {
  1049. struct load_weight load; /* for load-balancing */
  1050. struct rb_node run_node;
  1051. struct list_head group_node;
  1052. unsigned int on_rq;
  1053. u64 exec_start;
  1054. u64 sum_exec_runtime;
  1055. u64 vruntime;
  1056. u64 prev_sum_exec_runtime;
  1057. u64 nr_migrations;
  1058. #ifdef CONFIG_SCHEDSTATS
  1059. struct sched_statistics statistics;
  1060. #endif
  1061. #ifdef CONFIG_FAIR_GROUP_SCHED
  1062. int depth;
  1063. struct sched_entity *parent;
  1064. /* rq on which this entity is (to be) queued: */
  1065. struct cfs_rq *cfs_rq;
  1066. /* rq "owned" by this entity/group: */
  1067. struct cfs_rq *my_q;
  1068. #endif
  1069. #ifdef CONFIG_SMP
  1070. /* Per entity load average tracking */
  1071. struct sched_avg avg;
  1072. #endif
  1073. };
  1074. struct sched_rt_entity {
  1075. struct list_head run_list;
  1076. unsigned long timeout;
  1077. unsigned long watchdog_stamp;
  1078. unsigned int time_slice;
  1079. struct sched_rt_entity *back;
  1080. #ifdef CONFIG_RT_GROUP_SCHED
  1081. struct sched_rt_entity *parent;
  1082. /* rq on which this entity is (to be) queued: */
  1083. struct rt_rq *rt_rq;
  1084. /* rq "owned" by this entity/group: */
  1085. struct rt_rq *my_q;
  1086. #endif
  1087. };
  1088. struct sched_dl_entity {
  1089. struct rb_node rb_node;
  1090. /*
  1091. * Original scheduling parameters. Copied here from sched_attr
  1092. * during sched_setattr(), they will remain the same until
  1093. * the next sched_setattr().
  1094. */
  1095. u64 dl_runtime; /* maximum runtime for each instance */
  1096. u64 dl_deadline; /* relative deadline of each instance */
  1097. u64 dl_period; /* separation of two instances (period) */
  1098. u64 dl_bw; /* dl_runtime / dl_deadline */
  1099. /*
  1100. * Actual scheduling parameters. Initialized with the values above,
  1101. * they are continously updated during task execution. Note that
  1102. * the remaining runtime could be < 0 in case we are in overrun.
  1103. */
  1104. s64 runtime; /* remaining runtime for this instance */
  1105. u64 deadline; /* absolute deadline for this instance */
  1106. unsigned int flags; /* specifying the scheduler behaviour */
  1107. /*
  1108. * Some bool flags:
  1109. *
  1110. * @dl_throttled tells if we exhausted the runtime. If so, the
  1111. * task has to wait for a replenishment to be performed at the
  1112. * next firing of dl_timer.
  1113. *
  1114. * @dl_new tells if a new instance arrived. If so we must
  1115. * start executing it with full runtime and reset its absolute
  1116. * deadline;
  1117. *
  1118. * @dl_boosted tells if we are boosted due to DI. If so we are
  1119. * outside bandwidth enforcement mechanism (but only until we
  1120. * exit the critical section);
  1121. *
  1122. * @dl_yielded tells if task gave up the cpu before consuming
  1123. * all its available runtime during the last job.
  1124. */
  1125. int dl_throttled, dl_new, dl_boosted, dl_yielded;
  1126. /*
  1127. * Bandwidth enforcement timer. Each -deadline task has its
  1128. * own bandwidth to be enforced, thus we need one timer per task.
  1129. */
  1130. struct hrtimer dl_timer;
  1131. };
  1132. union rcu_special {
  1133. struct {
  1134. bool blocked;
  1135. bool need_qs;
  1136. } b;
  1137. short s;
  1138. };
  1139. struct rcu_node;
  1140. enum perf_event_task_context {
  1141. perf_invalid_context = -1,
  1142. perf_hw_context = 0,
  1143. perf_sw_context,
  1144. perf_nr_task_contexts,
  1145. };
  1146. struct task_struct {
  1147. volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
  1148. void *stack;
  1149. atomic_t usage;
  1150. unsigned int flags; /* per process flags, defined below */
  1151. unsigned int ptrace;
  1152. #ifdef CONFIG_SMP
  1153. struct llist_node wake_entry;
  1154. int on_cpu;
  1155. unsigned int wakee_flips;
  1156. unsigned long wakee_flip_decay_ts;
  1157. struct task_struct *last_wakee;
  1158. int wake_cpu;
  1159. #endif
  1160. int on_rq;
  1161. int prio, static_prio, normal_prio;
  1162. unsigned int rt_priority;
  1163. const struct sched_class *sched_class;
  1164. struct sched_entity se;
  1165. struct sched_rt_entity rt;
  1166. #ifdef CONFIG_CGROUP_SCHED
  1167. struct task_group *sched_task_group;
  1168. #endif
  1169. struct sched_dl_entity dl;
  1170. #ifdef CONFIG_PREEMPT_NOTIFIERS
  1171. /* list of struct preempt_notifier: */
  1172. struct hlist_head preempt_notifiers;
  1173. #endif
  1174. #ifdef CONFIG_BLK_DEV_IO_TRACE
  1175. unsigned int btrace_seq;
  1176. #endif
  1177. unsigned int policy;
  1178. int nr_cpus_allowed;
  1179. cpumask_t cpus_allowed;
  1180. #ifdef CONFIG_PREEMPT_RCU
  1181. int rcu_read_lock_nesting;
  1182. union rcu_special rcu_read_unlock_special;
  1183. struct list_head rcu_node_entry;
  1184. struct rcu_node *rcu_blocked_node;
  1185. #endif /* #ifdef CONFIG_PREEMPT_RCU */
  1186. #ifdef CONFIG_TASKS_RCU
  1187. unsigned long rcu_tasks_nvcsw;
  1188. bool rcu_tasks_holdout;
  1189. struct list_head rcu_tasks_holdout_list;
  1190. int rcu_tasks_idle_cpu;
  1191. #endif /* #ifdef CONFIG_TASKS_RCU */
  1192. #ifdef CONFIG_SCHED_INFO
  1193. struct sched_info sched_info;
  1194. #endif
  1195. struct list_head tasks;
  1196. #ifdef CONFIG_SMP
  1197. struct plist_node pushable_tasks;
  1198. struct rb_node pushable_dl_tasks;
  1199. #endif
  1200. struct mm_struct *mm, *active_mm;
  1201. /* per-thread vma caching */
  1202. u32 vmacache_seqnum;
  1203. struct vm_area_struct *vmacache[VMACACHE_SIZE];
  1204. #if defined(SPLIT_RSS_COUNTING)
  1205. struct task_rss_stat rss_stat;
  1206. #endif
  1207. /* task state */
  1208. int exit_state;
  1209. int exit_code, exit_signal;
  1210. int pdeath_signal; /* The signal sent when the parent dies */
  1211. unsigned long jobctl; /* JOBCTL_*, siglock protected */
  1212. /* Used for emulating ABI behavior of previous Linux versions */
  1213. unsigned int personality;
  1214. unsigned in_execve:1; /* Tell the LSMs that the process is doing an
  1215. * execve */
  1216. unsigned in_iowait:1;
  1217. /* Revert to default priority/policy when forking */
  1218. unsigned sched_reset_on_fork:1;
  1219. unsigned sched_contributes_to_load:1;
  1220. unsigned sched_migrated:1;
  1221. #ifdef CONFIG_MEMCG_KMEM
  1222. unsigned memcg_kmem_skip_account:1;
  1223. #endif
  1224. #ifdef CONFIG_COMPAT_BRK
  1225. unsigned brk_randomized:1;
  1226. #endif
  1227. unsigned long atomic_flags; /* Flags needing atomic access. */
  1228. struct restart_block restart_block;
  1229. pid_t pid;
  1230. pid_t tgid;
  1231. #ifdef CONFIG_CC_STACKPROTECTOR
  1232. /* Canary value for the -fstack-protector gcc feature */
  1233. unsigned long stack_canary;
  1234. #endif
  1235. /*
  1236. * pointers to (original) parent process, youngest child, younger sibling,
  1237. * older sibling, respectively. (p->father can be replaced with
  1238. * p->real_parent->pid)
  1239. */
  1240. struct task_struct __rcu *real_parent; /* real parent process */
  1241. struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
  1242. /*
  1243. * children/sibling forms the list of my natural children
  1244. */
  1245. struct list_head children; /* list of my children */
  1246. struct list_head sibling; /* linkage in my parent's children list */
  1247. struct task_struct *group_leader; /* threadgroup leader */
  1248. /*
  1249. * ptraced is the list of tasks this task is using ptrace on.
  1250. * This includes both natural children and PTRACE_ATTACH targets.
  1251. * p->ptrace_entry is p's link on the p->parent->ptraced list.
  1252. */
  1253. struct list_head ptraced;
  1254. struct list_head ptrace_entry;
  1255. /* PID/PID hash table linkage. */
  1256. struct pid_link pids[PIDTYPE_MAX];
  1257. struct list_head thread_group;
  1258. struct list_head thread_node;
  1259. struct completion *vfork_done; /* for vfork() */
  1260. int __user *set_child_tid; /* CLONE_CHILD_SETTID */
  1261. int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
  1262. cputime_t utime, stime, utimescaled, stimescaled;
  1263. cputime_t gtime;
  1264. struct prev_cputime prev_cputime;
  1265. #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
  1266. seqlock_t vtime_seqlock;
  1267. unsigned long long vtime_snap;
  1268. enum {
  1269. VTIME_SLEEPING = 0,
  1270. VTIME_USER,
  1271. VTIME_SYS,
  1272. } vtime_snap_whence;
  1273. #endif
  1274. unsigned long nvcsw, nivcsw; /* context switch counts */
  1275. u64 start_time; /* monotonic time in nsec */
  1276. u64 real_start_time; /* boot based time in nsec */
  1277. /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
  1278. unsigned long min_flt, maj_flt;
  1279. struct task_cputime cputime_expires;
  1280. struct list_head cpu_timers[3];
  1281. /* process credentials */
  1282. const struct cred __rcu *real_cred; /* objective and real subjective task
  1283. * credentials (COW) */
  1284. const struct cred __rcu *cred; /* effective (overridable) subjective task
  1285. * credentials (COW) */
  1286. char comm[TASK_COMM_LEN]; /* executable name excluding path
  1287. - access with [gs]et_task_comm (which lock
  1288. it with task_lock())
  1289. - initialized normally by setup_new_exec */
  1290. /* file system info */
  1291. struct nameidata *nameidata;
  1292. #ifdef CONFIG_SYSVIPC
  1293. /* ipc stuff */
  1294. struct sysv_sem sysvsem;
  1295. struct sysv_shm sysvshm;
  1296. #endif
  1297. #ifdef CONFIG_DETECT_HUNG_TASK
  1298. /* hung task detection */
  1299. unsigned long last_switch_count;
  1300. #endif
  1301. /* CPU-specific state of this task */
  1302. struct thread_struct thread;
  1303. /* filesystem information */
  1304. struct fs_struct *fs;
  1305. /* open file information */
  1306. struct files_struct *files;
  1307. /* namespaces */
  1308. struct nsproxy *nsproxy;
  1309. /* signal handlers */
  1310. struct signal_struct *signal;
  1311. struct sighand_struct *sighand;
  1312. sigset_t blocked, real_blocked;
  1313. sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
  1314. struct sigpending pending;
  1315. unsigned long sas_ss_sp;
  1316. size_t sas_ss_size;
  1317. int (*notifier)(void *priv);
  1318. void *notifier_data;
  1319. sigset_t *notifier_mask;
  1320. struct callback_head *task_works;
  1321. struct audit_context *audit_context;
  1322. #ifdef CONFIG_AUDITSYSCALL
  1323. kuid_t loginuid;
  1324. unsigned int sessionid;
  1325. #endif
  1326. struct seccomp seccomp;
  1327. /* Thread group tracking */
  1328. u32 parent_exec_id;
  1329. u32 self_exec_id;
  1330. /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
  1331. * mempolicy */
  1332. spinlock_t alloc_lock;
  1333. /* Protection of the PI data structures: */
  1334. raw_spinlock_t pi_lock;
  1335. struct wake_q_node wake_q;
  1336. #ifdef CONFIG_RT_MUTEXES
  1337. /* PI waiters blocked on a rt_mutex held by this task */
  1338. struct rb_root pi_waiters;
  1339. struct rb_node *pi_waiters_leftmost;
  1340. /* Deadlock detection and priority inheritance handling */
  1341. struct rt_mutex_waiter *pi_blocked_on;
  1342. #endif
  1343. #ifdef CONFIG_DEBUG_MUTEXES
  1344. /* mutex deadlock detection */
  1345. struct mutex_waiter *blocked_on;
  1346. #endif
  1347. #ifdef CONFIG_TRACE_IRQFLAGS
  1348. unsigned int irq_events;
  1349. unsigned long hardirq_enable_ip;
  1350. unsigned long hardirq_disable_ip;
  1351. unsigned int hardirq_enable_event;
  1352. unsigned int hardirq_disable_event;
  1353. int hardirqs_enabled;
  1354. int hardirq_context;
  1355. unsigned long softirq_disable_ip;
  1356. unsigned long softirq_enable_ip;
  1357. unsigned int softirq_disable_event;
  1358. unsigned int softirq_enable_event;
  1359. int softirqs_enabled;
  1360. int softirq_context;
  1361. #endif
  1362. #ifdef CONFIG_LOCKDEP
  1363. # define MAX_LOCK_DEPTH 48UL
  1364. u64 curr_chain_key;
  1365. int lockdep_depth;
  1366. unsigned int lockdep_recursion;
  1367. struct held_lock held_locks[MAX_LOCK_DEPTH];
  1368. gfp_t lockdep_reclaim_gfp;
  1369. #endif
  1370. /* journalling filesystem info */
  1371. void *journal_info;
  1372. /* stacked block device info */
  1373. struct bio_list *bio_list;
  1374. #ifdef CONFIG_BLOCK
  1375. /* stack plugging */
  1376. struct blk_plug *plug;
  1377. #endif
  1378. /* VM state */
  1379. struct reclaim_state *reclaim_state;
  1380. struct backing_dev_info *backing_dev_info;
  1381. struct io_context *io_context;
  1382. unsigned long ptrace_message;
  1383. siginfo_t *last_siginfo; /* For ptrace use. */
  1384. struct task_io_accounting ioac;
  1385. #if defined(CONFIG_TASK_XACCT)
  1386. u64 acct_rss_mem1; /* accumulated rss usage */
  1387. u64 acct_vm_mem1; /* accumulated virtual memory usage */
  1388. cputime_t acct_timexpd; /* stime + utime since last update */
  1389. #endif
  1390. #ifdef CONFIG_CPUSETS
  1391. nodemask_t mems_allowed; /* Protected by alloc_lock */
  1392. seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
  1393. int cpuset_mem_spread_rotor;
  1394. int cpuset_slab_spread_rotor;
  1395. #endif
  1396. #ifdef CONFIG_CGROUPS
  1397. /* Control Group info protected by css_set_lock */
  1398. struct css_set __rcu *cgroups;
  1399. /* cg_list protected by css_set_lock and tsk->alloc_lock */
  1400. struct list_head cg_list;
  1401. #endif
  1402. #ifdef CONFIG_FUTEX
  1403. struct robust_list_head __user *robust_list;
  1404. #ifdef CONFIG_COMPAT
  1405. struct compat_robust_list_head __user *compat_robust_list;
  1406. #endif
  1407. struct list_head pi_state_list;
  1408. struct futex_pi_state *pi_state_cache;
  1409. #endif
  1410. #ifdef CONFIG_PERF_EVENTS
  1411. struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
  1412. struct mutex perf_event_mutex;
  1413. struct list_head perf_event_list;
  1414. #endif
  1415. #ifdef CONFIG_DEBUG_PREEMPT
  1416. unsigned long preempt_disable_ip;
  1417. #endif
  1418. #ifdef CONFIG_NUMA
  1419. struct mempolicy *mempolicy; /* Protected by alloc_lock */
  1420. short il_next;
  1421. short pref_node_fork;
  1422. #endif
  1423. #ifdef CONFIG_NUMA_BALANCING
  1424. int numa_scan_seq;
  1425. unsigned int numa_scan_period;
  1426. unsigned int numa_scan_period_max;
  1427. int numa_preferred_nid;
  1428. unsigned long numa_migrate_retry;
  1429. u64 node_stamp; /* migration stamp */
  1430. u64 last_task_numa_placement;
  1431. u64 last_sum_exec_runtime;
  1432. struct callback_head numa_work;
  1433. struct list_head numa_entry;
  1434. struct numa_group *numa_group;
  1435. /*
  1436. * numa_faults is an array split into four regions:
  1437. * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
  1438. * in this precise order.
  1439. *
  1440. * faults_memory: Exponential decaying average of faults on a per-node
  1441. * basis. Scheduling placement decisions are made based on these
  1442. * counts. The values remain static for the duration of a PTE scan.
  1443. * faults_cpu: Track the nodes the process was running on when a NUMA
  1444. * hinting fault was incurred.
  1445. * faults_memory_buffer and faults_cpu_buffer: Record faults per node
  1446. * during the current scan window. When the scan completes, the counts
  1447. * in faults_memory and faults_cpu decay and these values are copied.
  1448. */
  1449. unsigned long *numa_faults;
  1450. unsigned long total_numa_faults;
  1451. /*
  1452. * numa_faults_locality tracks if faults recorded during the last
  1453. * scan window were remote/local or failed to migrate. The task scan
  1454. * period is adapted based on the locality of the faults with different
  1455. * weights depending on whether they were shared or private faults
  1456. */
  1457. unsigned long numa_faults_locality[3];
  1458. unsigned long numa_pages_migrated;
  1459. #endif /* CONFIG_NUMA_BALANCING */
  1460. struct rcu_head rcu;
  1461. /*
  1462. * cache last used pipe for splice
  1463. */
  1464. struct pipe_inode_info *splice_pipe;
  1465. struct page_frag task_frag;
  1466. #ifdef CONFIG_TASK_DELAY_ACCT
  1467. struct task_delay_info *delays;
  1468. #endif
  1469. #ifdef CONFIG_FAULT_INJECTION
  1470. int make_it_fail;
  1471. #endif
  1472. /*
  1473. * when (nr_dirtied >= nr_dirtied_pause), it's time to call
  1474. * balance_dirty_pages() for some dirty throttling pause
  1475. */
  1476. int nr_dirtied;
  1477. int nr_dirtied_pause;
  1478. unsigned long dirty_paused_when; /* start of a write-and-pause period */
  1479. #ifdef CONFIG_LATENCYTOP
  1480. int latency_record_count;
  1481. struct latency_record latency_record[LT_SAVECOUNT];
  1482. #endif
  1483. /*
  1484. * time slack values; these are used to round up poll() and
  1485. * select() etc timeout values. These are in nanoseconds.
  1486. */
  1487. unsigned long timer_slack_ns;
  1488. unsigned long default_timer_slack_ns;
  1489. #ifdef CONFIG_KASAN
  1490. unsigned int kasan_depth;
  1491. #endif
  1492. #ifdef CONFIG_FUNCTION_GRAPH_TRACER
  1493. /* Index of current stored address in ret_stack */
  1494. int curr_ret_stack;
  1495. /* Stack of return addresses for return function tracing */
  1496. struct ftrace_ret_stack *ret_stack;
  1497. /* time stamp for last schedule */
  1498. unsigned long long ftrace_timestamp;
  1499. /*
  1500. * Number of functions that haven't been traced
  1501. * because of depth overrun.
  1502. */
  1503. atomic_t trace_overrun;
  1504. /* Pause for the tracing */
  1505. atomic_t tracing_graph_pause;
  1506. #endif
  1507. #ifdef CONFIG_TRACING
  1508. /* state flags for use by tracers */
  1509. unsigned long trace;
  1510. /* bitmask and counter of trace recursion */
  1511. unsigned long trace_recursion;
  1512. #endif /* CONFIG_TRACING */
  1513. #ifdef CONFIG_MEMCG
  1514. struct memcg_oom_info {
  1515. struct mem_cgroup *memcg;
  1516. gfp_t gfp_mask;
  1517. int order;
  1518. unsigned int may_oom:1;
  1519. } memcg_oom;
  1520. #endif
  1521. #ifdef CONFIG_UPROBES
  1522. struct uprobe_task *utask;
  1523. #endif
  1524. #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
  1525. unsigned int sequential_io;
  1526. unsigned int sequential_io_avg;
  1527. #endif
  1528. #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
  1529. unsigned long task_state_change;
  1530. #endif
  1531. int pagefault_disabled;
  1532. };
  1533. /* Future-safe accessor for struct task_struct's cpus_allowed. */
  1534. #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
  1535. #define TNF_MIGRATED 0x01
  1536. #define TNF_NO_GROUP 0x02
  1537. #define TNF_SHARED 0x04
  1538. #define TNF_FAULT_LOCAL 0x08
  1539. #define TNF_MIGRATE_FAIL 0x10
  1540. #ifdef CONFIG_NUMA_BALANCING
  1541. extern void task_numa_fault(int last_node, int node, int pages, int flags);
  1542. extern pid_t task_numa_group_id(struct task_struct *p);
  1543. extern void set_numabalancing_state(bool enabled);
  1544. extern void task_numa_free(struct task_struct *p);
  1545. extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
  1546. int src_nid, int dst_cpu);
  1547. #else
  1548. static inline void task_numa_fault(int last_node, int node, int pages,
  1549. int flags)
  1550. {
  1551. }
  1552. static inline pid_t task_numa_group_id(struct task_struct *p)
  1553. {
  1554. return 0;
  1555. }
  1556. static inline void set_numabalancing_state(bool enabled)
  1557. {
  1558. }
  1559. static inline void task_numa_free(struct task_struct *p)
  1560. {
  1561. }
  1562. static inline bool should_numa_migrate_memory(struct task_struct *p,
  1563. struct page *page, int src_nid, int dst_cpu)
  1564. {
  1565. return true;
  1566. }
  1567. #endif
  1568. static inline struct pid *task_pid(struct task_struct *task)
  1569. {
  1570. return task->pids[PIDTYPE_PID].pid;
  1571. }
  1572. static inline struct pid *task_tgid(struct task_struct *task)
  1573. {
  1574. return task->group_leader->pids[PIDTYPE_PID].pid;
  1575. }
  1576. /*
  1577. * Without tasklist or rcu lock it is not safe to dereference
  1578. * the result of task_pgrp/task_session even if task == current,
  1579. * we can race with another thread doing sys_setsid/sys_setpgid.
  1580. */
  1581. static inline struct pid *task_pgrp(struct task_struct *task)
  1582. {
  1583. return task->group_leader->pids[PIDTYPE_PGID].pid;
  1584. }
  1585. static inline struct pid *task_session(struct task_struct *task)
  1586. {
  1587. return task->group_leader->pids[PIDTYPE_SID].pid;
  1588. }
  1589. struct pid_namespace;
  1590. /*
  1591. * the helpers to get the task's different pids as they are seen
  1592. * from various namespaces
  1593. *
  1594. * task_xid_nr() : global id, i.e. the id seen from the init namespace;
  1595. * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
  1596. * current.
  1597. * task_xid_nr_ns() : id seen from the ns specified;
  1598. *
  1599. * set_task_vxid() : assigns a virtual id to a task;
  1600. *
  1601. * see also pid_nr() etc in include/linux/pid.h
  1602. */
  1603. pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
  1604. struct pid_namespace *ns);
  1605. static inline pid_t task_pid_nr(struct task_struct *tsk)
  1606. {
  1607. return tsk->pid;
  1608. }
  1609. static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
  1610. struct pid_namespace *ns)
  1611. {
  1612. return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
  1613. }
  1614. static inline pid_t task_pid_vnr(struct task_struct *tsk)
  1615. {
  1616. return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
  1617. }
  1618. static inline pid_t task_tgid_nr(struct task_struct *tsk)
  1619. {
  1620. return tsk->tgid;
  1621. }
  1622. pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
  1623. static inline pid_t task_tgid_vnr(struct task_struct *tsk)
  1624. {
  1625. return pid_vnr(task_tgid(tsk));
  1626. }
  1627. static inline int pid_alive(const struct task_struct *p);
  1628. static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
  1629. {
  1630. pid_t pid = 0;
  1631. rcu_read_lock();
  1632. if (pid_alive(tsk))
  1633. pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
  1634. rcu_read_unlock();
  1635. return pid;
  1636. }
  1637. static inline pid_t task_ppid_nr(const struct task_struct *tsk)
  1638. {
  1639. return task_ppid_nr_ns(tsk, &init_pid_ns);
  1640. }
  1641. static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
  1642. struct pid_namespace *ns)
  1643. {
  1644. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
  1645. }
  1646. static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
  1647. {
  1648. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
  1649. }
  1650. static inline pid_t task_session_nr_ns(struct task_struct *tsk,
  1651. struct pid_namespace *ns)
  1652. {
  1653. return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
  1654. }
  1655. static inline pid_t task_session_vnr(struct task_struct *tsk)
  1656. {
  1657. return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
  1658. }
  1659. /* obsolete, do not use */
  1660. static inline pid_t task_pgrp_nr(struct task_struct *tsk)
  1661. {
  1662. return task_pgrp_nr_ns(tsk, &init_pid_ns);
  1663. }
  1664. /**
  1665. * pid_alive - check that a task structure is not stale
  1666. * @p: Task structure to be checked.
  1667. *
  1668. * Test if a process is not yet dead (at most zombie state)
  1669. * If pid_alive fails, then pointers within the task structure
  1670. * can be stale and must not be dereferenced.
  1671. *
  1672. * Return: 1 if the process is alive. 0 otherwise.
  1673. */
  1674. static inline int pid_alive(const struct task_struct *p)
  1675. {
  1676. return p->pids[PIDTYPE_PID].pid != NULL;
  1677. }
  1678. /**
  1679. * is_global_init - check if a task structure is init
  1680. * @tsk: Task structure to be checked.
  1681. *
  1682. * Check if a task structure is the first user space task the kernel created.
  1683. *
  1684. * Return: 1 if the task structure is init. 0 otherwise.
  1685. */
  1686. static inline int is_global_init(struct task_struct *tsk)
  1687. {
  1688. return tsk->pid == 1;
  1689. }
  1690. extern struct pid *cad_pid;
  1691. extern void free_task(struct task_struct *tsk);
  1692. #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
  1693. extern void __put_task_struct(struct task_struct *t);
  1694. static inline void put_task_struct(struct task_struct *t)
  1695. {
  1696. if (atomic_dec_and_test(&t->usage))
  1697. __put_task_struct(t);
  1698. }
  1699. #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
  1700. extern void task_cputime(struct task_struct *t,
  1701. cputime_t *utime, cputime_t *stime);
  1702. extern void task_cputime_scaled(struct task_struct *t,
  1703. cputime_t *utimescaled, cputime_t *stimescaled);
  1704. extern cputime_t task_gtime(struct task_struct *t);
  1705. #else
  1706. static inline void task_cputime(struct task_struct *t,
  1707. cputime_t *utime, cputime_t *stime)
  1708. {
  1709. if (utime)
  1710. *utime = t->utime;
  1711. if (stime)
  1712. *stime = t->stime;
  1713. }
  1714. static inline void task_cputime_scaled(struct task_struct *t,
  1715. cputime_t *utimescaled,
  1716. cputime_t *stimescaled)
  1717. {
  1718. if (utimescaled)
  1719. *utimescaled = t->utimescaled;
  1720. if (stimescaled)
  1721. *stimescaled = t->stimescaled;
  1722. }
  1723. static inline cputime_t task_gtime(struct task_struct *t)
  1724. {
  1725. return t->gtime;
  1726. }
  1727. #endif
  1728. extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
  1729. extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
  1730. /*
  1731. * Per process flags
  1732. */
  1733. #define PF_EXITING 0x00000004 /* getting shut down */
  1734. #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
  1735. #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
  1736. #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
  1737. #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
  1738. #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
  1739. #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
  1740. #define PF_DUMPCORE 0x00000200 /* dumped core */
  1741. #define PF_SIGNALED 0x00000400 /* killed by a signal */
  1742. #define PF_MEMALLOC 0x00000800 /* Allocating memory */
  1743. #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
  1744. #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
  1745. #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
  1746. #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
  1747. #define PF_FROZEN 0x00010000 /* frozen for system suspend */
  1748. #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
  1749. #define PF_KSWAPD 0x00040000 /* I am kswapd */
  1750. #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
  1751. #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
  1752. #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
  1753. #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
  1754. #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
  1755. #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
  1756. #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
  1757. #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
  1758. #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
  1759. #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
  1760. /*
  1761. * Only the _current_ task can read/write to tsk->flags, but other
  1762. * tasks can access tsk->flags in readonly mode for example
  1763. * with tsk_used_math (like during threaded core dumping).
  1764. * There is however an exception to this rule during ptrace
  1765. * or during fork: the ptracer task is allowed to write to the
  1766. * child->flags of its traced child (same goes for fork, the parent
  1767. * can write to the child->flags), because we're guaranteed the
  1768. * child is not running and in turn not changing child->flags
  1769. * at the same time the parent does it.
  1770. */
  1771. #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
  1772. #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
  1773. #define clear_used_math() clear_stopped_child_used_math(current)
  1774. #define set_used_math() set_stopped_child_used_math(current)
  1775. #define conditional_stopped_child_used_math(condition, child) \
  1776. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
  1777. #define conditional_used_math(condition) \
  1778. conditional_stopped_child_used_math(condition, current)
  1779. #define copy_to_stopped_child_used_math(child) \
  1780. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
  1781. /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
  1782. #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
  1783. #define used_math() tsk_used_math(current)
  1784. /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
  1785. * __GFP_FS is also cleared as it implies __GFP_IO.
  1786. */
  1787. static inline gfp_t memalloc_noio_flags(gfp_t flags)
  1788. {
  1789. if (unlikely(current->flags & PF_MEMALLOC_NOIO))
  1790. flags &= ~(__GFP_IO | __GFP_FS);
  1791. return flags;
  1792. }
  1793. static inline unsigned int memalloc_noio_save(void)
  1794. {
  1795. unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
  1796. current->flags |= PF_MEMALLOC_NOIO;
  1797. return flags;
  1798. }
  1799. static inline void memalloc_noio_restore(unsigned int flags)
  1800. {
  1801. current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
  1802. }
  1803. /* Per-process atomic flags. */
  1804. #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
  1805. #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
  1806. #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
  1807. #define TASK_PFA_TEST(name, func) \
  1808. static inline bool task_##func(struct task_struct *p) \
  1809. { return test_bit(PFA_##name, &p->atomic_flags); }
  1810. #define TASK_PFA_SET(name, func) \
  1811. static inline void task_set_##func(struct task_struct *p) \
  1812. { set_bit(PFA_##name, &p->atomic_flags); }
  1813. #define TASK_PFA_CLEAR(name, func) \
  1814. static inline void task_clear_##func(struct task_struct *p) \
  1815. { clear_bit(PFA_##name, &p->atomic_flags); }
  1816. TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
  1817. TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
  1818. TASK_PFA_TEST(SPREAD_PAGE, spread_page)
  1819. TASK_PFA_SET(SPREAD_PAGE, spread_page)
  1820. TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
  1821. TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
  1822. TASK_PFA_SET(SPREAD_SLAB, spread_slab)
  1823. TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
  1824. /*
  1825. * task->jobctl flags
  1826. */
  1827. #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
  1828. #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
  1829. #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
  1830. #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
  1831. #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
  1832. #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
  1833. #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
  1834. #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
  1835. #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
  1836. #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
  1837. #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
  1838. #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
  1839. #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
  1840. #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
  1841. #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
  1842. #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
  1843. #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
  1844. extern bool task_set_jobctl_pending(struct task_struct *task,
  1845. unsigned long mask);
  1846. extern void task_clear_jobctl_trapping(struct task_struct *task);
  1847. extern void task_clear_jobctl_pending(struct task_struct *task,
  1848. unsigned long mask);
  1849. static inline void rcu_copy_process(struct task_struct *p)
  1850. {
  1851. #ifdef CONFIG_PREEMPT_RCU
  1852. p->rcu_read_lock_nesting = 0;
  1853. p->rcu_read_unlock_special.s = 0;
  1854. p->rcu_blocked_node = NULL;
  1855. INIT_LIST_HEAD(&p->rcu_node_entry);
  1856. #endif /* #ifdef CONFIG_PREEMPT_RCU */
  1857. #ifdef CONFIG_TASKS_RCU
  1858. p->rcu_tasks_holdout = false;
  1859. INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
  1860. p->rcu_tasks_idle_cpu = -1;
  1861. #endif /* #ifdef CONFIG_TASKS_RCU */
  1862. }
  1863. static inline void tsk_restore_flags(struct task_struct *task,
  1864. unsigned long orig_flags, unsigned long flags)
  1865. {
  1866. task->flags &= ~flags;
  1867. task->flags |= orig_flags & flags;
  1868. }
  1869. extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
  1870. const struct cpumask *trial);
  1871. extern int task_can_attach(struct task_struct *p,
  1872. const struct cpumask *cs_cpus_allowed);
  1873. #ifdef CONFIG_SMP
  1874. extern void do_set_cpus_allowed(struct task_struct *p,
  1875. const struct cpumask *new_mask);
  1876. extern int set_cpus_allowed_ptr(struct task_struct *p,
  1877. const struct cpumask *new_mask);
  1878. #else
  1879. static inline void do_set_cpus_allowed(struct task_struct *p,
  1880. const struct cpumask *new_mask)
  1881. {
  1882. }
  1883. static inline int set_cpus_allowed_ptr(struct task_struct *p,
  1884. const struct cpumask *new_mask)
  1885. {
  1886. if (!cpumask_test_cpu(0, new_mask))
  1887. return -EINVAL;
  1888. return 0;
  1889. }
  1890. #endif
  1891. #ifdef CONFIG_NO_HZ_COMMON
  1892. void calc_load_enter_idle(void);
  1893. void calc_load_exit_idle(void);
  1894. #else
  1895. static inline void calc_load_enter_idle(void) { }
  1896. static inline void calc_load_exit_idle(void) { }
  1897. #endif /* CONFIG_NO_HZ_COMMON */
  1898. #ifndef CONFIG_CPUMASK_OFFSTACK
  1899. static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
  1900. {
  1901. return set_cpus_allowed_ptr(p, &new_mask);
  1902. }
  1903. #endif
  1904. /*
  1905. * Do not use outside of architecture code which knows its limitations.
  1906. *
  1907. * sched_clock() has no promise of monotonicity or bounded drift between
  1908. * CPUs, use (which you should not) requires disabling IRQs.
  1909. *
  1910. * Please use one of the three interfaces below.
  1911. */
  1912. extern unsigned long long notrace sched_clock(void);
  1913. /*
  1914. * See the comment in kernel/sched/clock.c
  1915. */
  1916. extern u64 cpu_clock(int cpu);
  1917. extern u64 local_clock(void);
  1918. extern u64 running_clock(void);
  1919. extern u64 sched_clock_cpu(int cpu);
  1920. extern void sched_clock_init(void);
  1921. #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
  1922. static inline void sched_clock_tick(void)
  1923. {
  1924. }
  1925. static inline void sched_clock_idle_sleep_event(void)
  1926. {
  1927. }
  1928. static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
  1929. {
  1930. }
  1931. #else
  1932. /*
  1933. * Architectures can set this to 1 if they have specified
  1934. * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
  1935. * but then during bootup it turns out that sched_clock()
  1936. * is reliable after all:
  1937. */
  1938. extern int sched_clock_stable(void);
  1939. extern void set_sched_clock_stable(void);
  1940. extern void clear_sched_clock_stable(void);
  1941. extern void sched_clock_tick(void);
  1942. extern void sched_clock_idle_sleep_event(void);
  1943. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  1944. #endif
  1945. #ifdef CONFIG_IRQ_TIME_ACCOUNTING
  1946. /*
  1947. * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
  1948. * The reason for this explicit opt-in is not to have perf penalty with
  1949. * slow sched_clocks.
  1950. */
  1951. extern void enable_sched_clock_irqtime(void);
  1952. extern void disable_sched_clock_irqtime(void);
  1953. #else
  1954. static inline void enable_sched_clock_irqtime(void) {}
  1955. static inline void disable_sched_clock_irqtime(void) {}
  1956. #endif
  1957. extern unsigned long long
  1958. task_sched_runtime(struct task_struct *task);
  1959. /* sched_exec is called by processes performing an exec */
  1960. #ifdef CONFIG_SMP
  1961. extern void sched_exec(void);
  1962. #else
  1963. #define sched_exec() {}
  1964. #endif
  1965. extern void sched_clock_idle_sleep_event(void);
  1966. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  1967. #ifdef CONFIG_HOTPLUG_CPU
  1968. extern void idle_task_exit(void);
  1969. #else
  1970. static inline void idle_task_exit(void) {}
  1971. #endif
  1972. #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
  1973. extern void wake_up_nohz_cpu(int cpu);
  1974. #else
  1975. static inline void wake_up_nohz_cpu(int cpu) { }
  1976. #endif
  1977. #ifdef CONFIG_NO_HZ_FULL
  1978. extern bool sched_can_stop_tick(void);
  1979. extern u64 scheduler_tick_max_deferment(void);
  1980. #else
  1981. static inline bool sched_can_stop_tick(void) { return false; }
  1982. #endif
  1983. #ifdef CONFIG_SCHED_AUTOGROUP
  1984. extern void sched_autogroup_create_attach(struct task_struct *p);
  1985. extern void sched_autogroup_detach(struct task_struct *p);
  1986. extern void sched_autogroup_fork(struct signal_struct *sig);
  1987. extern void sched_autogroup_exit(struct signal_struct *sig);
  1988. #ifdef CONFIG_PROC_FS
  1989. extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
  1990. extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
  1991. #endif
  1992. #else
  1993. static inline void sched_autogroup_create_attach(struct task_struct *p) { }
  1994. static inline void sched_autogroup_detach(struct task_struct *p) { }
  1995. static inline void sched_autogroup_fork(struct signal_struct *sig) { }
  1996. static inline void sched_autogroup_exit(struct signal_struct *sig) { }
  1997. #endif
  1998. extern int yield_to(struct task_struct *p, bool preempt);
  1999. extern void set_user_nice(struct task_struct *p, long nice);
  2000. extern int task_prio(const struct task_struct *p);
  2001. /**
  2002. * task_nice - return the nice value of a given task.
  2003. * @p: the task in question.
  2004. *
  2005. * Return: The nice value [ -20 ... 0 ... 19 ].
  2006. */
  2007. static inline int task_nice(const struct task_struct *p)
  2008. {
  2009. return PRIO_TO_NICE((p)->static_prio);
  2010. }
  2011. extern int can_nice(const struct task_struct *p, const int nice);
  2012. extern int task_curr(const struct task_struct *p);
  2013. extern int idle_cpu(int cpu);
  2014. extern int sched_setscheduler(struct task_struct *, int,
  2015. const struct sched_param *);
  2016. extern int sched_setscheduler_nocheck(struct task_struct *, int,
  2017. const struct sched_param *);
  2018. extern int sched_setattr(struct task_struct *,
  2019. const struct sched_attr *);
  2020. extern struct task_struct *idle_task(int cpu);
  2021. /**
  2022. * is_idle_task - is the specified task an idle task?
  2023. * @p: the task in question.
  2024. *
  2025. * Return: 1 if @p is an idle task. 0 otherwise.
  2026. */
  2027. static inline bool is_idle_task(const struct task_struct *p)
  2028. {
  2029. return p->pid == 0;
  2030. }
  2031. extern struct task_struct *curr_task(int cpu);
  2032. extern void set_curr_task(int cpu, struct task_struct *p);
  2033. void yield(void);
  2034. union thread_union {
  2035. struct thread_info thread_info;
  2036. unsigned long stack[THREAD_SIZE/sizeof(long)];
  2037. };
  2038. #ifndef __HAVE_ARCH_KSTACK_END
  2039. static inline int kstack_end(void *addr)
  2040. {
  2041. /* Reliable end of stack detection:
  2042. * Some APM bios versions misalign the stack
  2043. */
  2044. return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
  2045. }
  2046. #endif
  2047. extern union thread_union init_thread_union;
  2048. extern struct task_struct init_task;
  2049. extern struct mm_struct init_mm;
  2050. extern struct pid_namespace init_pid_ns;
  2051. /*
  2052. * find a task by one of its numerical ids
  2053. *
  2054. * find_task_by_pid_ns():
  2055. * finds a task by its pid in the specified namespace
  2056. * find_task_by_vpid():
  2057. * finds a task by its virtual pid
  2058. *
  2059. * see also find_vpid() etc in include/linux/pid.h
  2060. */
  2061. extern struct task_struct *find_task_by_vpid(pid_t nr);
  2062. extern struct task_struct *find_task_by_pid_ns(pid_t nr,
  2063. struct pid_namespace *ns);
  2064. /* per-UID process charging. */
  2065. extern struct user_struct * alloc_uid(kuid_t);
  2066. static inline struct user_struct *get_uid(struct user_struct *u)
  2067. {
  2068. atomic_inc(&u->__count);
  2069. return u;
  2070. }
  2071. extern void free_uid(struct user_struct *);
  2072. #include <asm/current.h>
  2073. extern void xtime_update(unsigned long ticks);
  2074. extern int wake_up_state(struct task_struct *tsk, unsigned int state);
  2075. extern int wake_up_process(struct task_struct *tsk);
  2076. extern void wake_up_new_task(struct task_struct *tsk);
  2077. #ifdef CONFIG_SMP
  2078. extern void kick_process(struct task_struct *tsk);
  2079. #else
  2080. static inline void kick_process(struct task_struct *tsk) { }
  2081. #endif
  2082. extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
  2083. extern void sched_dead(struct task_struct *p);
  2084. extern void proc_caches_init(void);
  2085. extern void flush_signals(struct task_struct *);
  2086. extern void ignore_signals(struct task_struct *);
  2087. extern void flush_signal_handlers(struct task_struct *, int force_default);
  2088. extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
  2089. static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
  2090. {
  2091. unsigned long flags;
  2092. int ret;
  2093. spin_lock_irqsave(&tsk->sighand->siglock, flags);
  2094. ret = dequeue_signal(tsk, mask, info);
  2095. spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
  2096. return ret;
  2097. }
  2098. extern void block_all_signals(int (*notifier)(void *priv), void *priv,
  2099. sigset_t *mask);
  2100. extern void unblock_all_signals(void);
  2101. extern void release_task(struct task_struct * p);
  2102. extern int send_sig_info(int, struct siginfo *, struct task_struct *);
  2103. extern int force_sigsegv(int, struct task_struct *);
  2104. extern int force_sig_info(int, struct siginfo *, struct task_struct *);
  2105. extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
  2106. extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
  2107. extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
  2108. const struct cred *, u32);
  2109. extern int kill_pgrp(struct pid *pid, int sig, int priv);
  2110. extern int kill_pid(struct pid *pid, int sig, int priv);
  2111. extern int kill_proc_info(int, struct siginfo *, pid_t);
  2112. extern __must_check bool do_notify_parent(struct task_struct *, int);
  2113. extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
  2114. extern void force_sig(int, struct task_struct *);
  2115. extern int send_sig(int, struct task_struct *, int);
  2116. extern int zap_other_threads(struct task_struct *p);
  2117. extern struct sigqueue *sigqueue_alloc(void);
  2118. extern void sigqueue_free(struct sigqueue *);
  2119. extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
  2120. extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
  2121. static inline void restore_saved_sigmask(void)
  2122. {
  2123. if (test_and_clear_restore_sigmask())
  2124. __set_current_blocked(&current->saved_sigmask);
  2125. }
  2126. static inline sigset_t *sigmask_to_save(void)
  2127. {
  2128. sigset_t *res = &current->blocked;
  2129. if (unlikely(test_restore_sigmask()))
  2130. res = &current->saved_sigmask;
  2131. return res;
  2132. }
  2133. static inline int kill_cad_pid(int sig, int priv)
  2134. {
  2135. return kill_pid(cad_pid, sig, priv);
  2136. }
  2137. /* These can be the second arg to send_sig_info/send_group_sig_info. */
  2138. #define SEND_SIG_NOINFO ((struct siginfo *) 0)
  2139. #define SEND_SIG_PRIV ((struct siginfo *) 1)
  2140. #define SEND_SIG_FORCED ((struct siginfo *) 2)
  2141. /*
  2142. * True if we are on the alternate signal stack.
  2143. */
  2144. static inline int on_sig_stack(unsigned long sp)
  2145. {
  2146. #ifdef CONFIG_STACK_GROWSUP
  2147. return sp >= current->sas_ss_sp &&
  2148. sp - current->sas_ss_sp < current->sas_ss_size;
  2149. #else
  2150. return sp > current->sas_ss_sp &&
  2151. sp - current->sas_ss_sp <= current->sas_ss_size;
  2152. #endif
  2153. }
  2154. static inline int sas_ss_flags(unsigned long sp)
  2155. {
  2156. if (!current->sas_ss_size)
  2157. return SS_DISABLE;
  2158. return on_sig_stack(sp) ? SS_ONSTACK : 0;
  2159. }
  2160. static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
  2161. {
  2162. if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
  2163. #ifdef CONFIG_STACK_GROWSUP
  2164. return current->sas_ss_sp;
  2165. #else
  2166. return current->sas_ss_sp + current->sas_ss_size;
  2167. #endif
  2168. return sp;
  2169. }
  2170. /*
  2171. * Routines for handling mm_structs
  2172. */
  2173. extern struct mm_struct * mm_alloc(void);
  2174. /* mmdrop drops the mm and the page tables */
  2175. extern void __mmdrop(struct mm_struct *);
  2176. static inline void mmdrop(struct mm_struct * mm)
  2177. {
  2178. if (unlikely(atomic_dec_and_test(&mm->mm_count)))
  2179. __mmdrop(mm);
  2180. }
  2181. /* mmput gets rid of the mappings and all user-space */
  2182. extern void mmput(struct mm_struct *);
  2183. /* Grab a reference to a task's mm, if it is not already going away */
  2184. extern struct mm_struct *get_task_mm(struct task_struct *task);
  2185. /*
  2186. * Grab a reference to a task's mm, if it is not already going away
  2187. * and ptrace_may_access with the mode parameter passed to it
  2188. * succeeds.
  2189. */
  2190. extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
  2191. /* Remove the current tasks stale references to the old mm_struct */
  2192. extern void mm_release(struct task_struct *, struct mm_struct *);
  2193. #ifdef CONFIG_HAVE_COPY_THREAD_TLS
  2194. extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
  2195. struct task_struct *, unsigned long);
  2196. #else
  2197. extern int copy_thread(unsigned long, unsigned long, unsigned long,
  2198. struct task_struct *);
  2199. /* Architectures that haven't opted into copy_thread_tls get the tls argument
  2200. * via pt_regs, so ignore the tls argument passed via C. */
  2201. static inline int copy_thread_tls(
  2202. unsigned long clone_flags, unsigned long sp, unsigned long arg,
  2203. struct task_struct *p, unsigned long tls)
  2204. {
  2205. return copy_thread(clone_flags, sp, arg, p);
  2206. }
  2207. #endif
  2208. extern void flush_thread(void);
  2209. extern void exit_thread(void);
  2210. extern void exit_files(struct task_struct *);
  2211. extern void __cleanup_sighand(struct sighand_struct *);
  2212. extern void exit_itimers(struct signal_struct *);
  2213. extern void flush_itimer_signals(void);
  2214. extern void do_group_exit(int);
  2215. extern int do_execve(struct filename *,
  2216. const char __user * const __user *,
  2217. const char __user * const __user *);
  2218. extern int do_execveat(int, struct filename *,
  2219. const char __user * const __user *,
  2220. const char __user * const __user *,
  2221. int);
  2222. extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
  2223. extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
  2224. struct task_struct *fork_idle(int);
  2225. extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
  2226. extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
  2227. static inline void set_task_comm(struct task_struct *tsk, const char *from)
  2228. {
  2229. __set_task_comm(tsk, from, false);
  2230. }
  2231. extern char *get_task_comm(char *to, struct task_struct *tsk);
  2232. #ifdef CONFIG_SMP
  2233. void scheduler_ipi(void);
  2234. extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
  2235. #else
  2236. static inline void scheduler_ipi(void) { }
  2237. static inline unsigned long wait_task_inactive(struct task_struct *p,
  2238. long match_state)
  2239. {
  2240. return 1;
  2241. }
  2242. #endif
  2243. #define tasklist_empty() \
  2244. list_empty(&init_task.tasks)
  2245. #define next_task(p) \
  2246. list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
  2247. #define for_each_process(p) \
  2248. for (p = &init_task ; (p = next_task(p)) != &init_task ; )
  2249. extern bool current_is_single_threaded(void);
  2250. /*
  2251. * Careful: do_each_thread/while_each_thread is a double loop so
  2252. * 'break' will not work as expected - use goto instead.
  2253. */
  2254. #define do_each_thread(g, t) \
  2255. for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
  2256. #define while_each_thread(g, t) \
  2257. while ((t = next_thread(t)) != g)
  2258. #define __for_each_thread(signal, t) \
  2259. list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
  2260. #define for_each_thread(p, t) \
  2261. __for_each_thread((p)->signal, t)
  2262. /* Careful: this is a double loop, 'break' won't work as expected. */
  2263. #define for_each_process_thread(p, t) \
  2264. for_each_process(p) for_each_thread(p, t)
  2265. static inline int get_nr_threads(struct task_struct *tsk)
  2266. {
  2267. return tsk->signal->nr_threads;
  2268. }
  2269. static inline bool thread_group_leader(struct task_struct *p)
  2270. {
  2271. return p->exit_signal >= 0;
  2272. }
  2273. /* Do to the insanities of de_thread it is possible for a process
  2274. * to have the pid of the thread group leader without actually being
  2275. * the thread group leader. For iteration through the pids in proc
  2276. * all we care about is that we have a task with the appropriate
  2277. * pid, we don't actually care if we have the right task.
  2278. */
  2279. static inline bool has_group_leader_pid(struct task_struct *p)
  2280. {
  2281. return task_pid(p) == p->signal->leader_pid;
  2282. }
  2283. static inline
  2284. bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
  2285. {
  2286. return p1->signal == p2->signal;
  2287. }
  2288. static inline struct task_struct *next_thread(const struct task_struct *p)
  2289. {
  2290. return list_entry_rcu(p->thread_group.next,
  2291. struct task_struct, thread_group);
  2292. }
  2293. static inline int thread_group_empty(struct task_struct *p)
  2294. {
  2295. return list_empty(&p->thread_group);
  2296. }
  2297. #define delay_group_leader(p) \
  2298. (thread_group_leader(p) && !thread_group_empty(p))
  2299. /*
  2300. * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
  2301. * subscriptions and synchronises with wait4(). Also used in procfs. Also
  2302. * pins the final release of task.io_context. Also protects ->cpuset and
  2303. * ->cgroup.subsys[]. And ->vfork_done.
  2304. *
  2305. * Nests both inside and outside of read_lock(&tasklist_lock).
  2306. * It must not be nested with write_lock_irq(&tasklist_lock),
  2307. * neither inside nor outside.
  2308. */
  2309. static inline void task_lock(struct task_struct *p)
  2310. {
  2311. spin_lock(&p->alloc_lock);
  2312. }
  2313. static inline void task_unlock(struct task_struct *p)
  2314. {
  2315. spin_unlock(&p->alloc_lock);
  2316. }
  2317. extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
  2318. unsigned long *flags);
  2319. static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
  2320. unsigned long *flags)
  2321. {
  2322. struct sighand_struct *ret;
  2323. ret = __lock_task_sighand(tsk, flags);
  2324. (void)__cond_lock(&tsk->sighand->siglock, ret);
  2325. return ret;
  2326. }
  2327. static inline void unlock_task_sighand(struct task_struct *tsk,
  2328. unsigned long *flags)
  2329. {
  2330. spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
  2331. }
  2332. /**
  2333. * threadgroup_change_begin - mark the beginning of changes to a threadgroup
  2334. * @tsk: task causing the changes
  2335. *
  2336. * All operations which modify a threadgroup - a new thread joining the
  2337. * group, death of a member thread (the assertion of PF_EXITING) and
  2338. * exec(2) dethreading the process and replacing the leader - are wrapped
  2339. * by threadgroup_change_{begin|end}(). This is to provide a place which
  2340. * subsystems needing threadgroup stability can hook into for
  2341. * synchronization.
  2342. */
  2343. static inline void threadgroup_change_begin(struct task_struct *tsk)
  2344. {
  2345. might_sleep();
  2346. cgroup_threadgroup_change_begin(tsk);
  2347. }
  2348. /**
  2349. * threadgroup_change_end - mark the end of changes to a threadgroup
  2350. * @tsk: task causing the changes
  2351. *
  2352. * See threadgroup_change_begin().
  2353. */
  2354. static inline void threadgroup_change_end(struct task_struct *tsk)
  2355. {
  2356. cgroup_threadgroup_change_end(tsk);
  2357. }
  2358. #ifndef __HAVE_THREAD_FUNCTIONS
  2359. #define task_thread_info(task) ((struct thread_info *)(task)->stack)
  2360. #define task_stack_page(task) ((task)->stack)
  2361. static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
  2362. {
  2363. *task_thread_info(p) = *task_thread_info(org);
  2364. task_thread_info(p)->task = p;
  2365. }
  2366. /*
  2367. * Return the address of the last usable long on the stack.
  2368. *
  2369. * When the stack grows down, this is just above the thread
  2370. * info struct. Going any lower will corrupt the threadinfo.
  2371. *
  2372. * When the stack grows up, this is the highest address.
  2373. * Beyond that position, we corrupt data on the next page.
  2374. */
  2375. static inline unsigned long *end_of_stack(struct task_struct *p)
  2376. {
  2377. #ifdef CONFIG_STACK_GROWSUP
  2378. return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
  2379. #else
  2380. return (unsigned long *)(task_thread_info(p) + 1);
  2381. #endif
  2382. }
  2383. #endif
  2384. #define task_stack_end_corrupted(task) \
  2385. (*(end_of_stack(task)) != STACK_END_MAGIC)
  2386. static inline int object_is_on_stack(void *obj)
  2387. {
  2388. void *stack = task_stack_page(current);
  2389. return (obj >= stack) && (obj < (stack + THREAD_SIZE));
  2390. }
  2391. extern void thread_info_cache_init(void);
  2392. #ifdef CONFIG_DEBUG_STACK_USAGE
  2393. static inline unsigned long stack_not_used(struct task_struct *p)
  2394. {
  2395. unsigned long *n = end_of_stack(p);
  2396. do { /* Skip over canary */
  2397. n++;
  2398. } while (!*n);
  2399. return (unsigned long)n - (unsigned long)end_of_stack(p);
  2400. }
  2401. #endif
  2402. extern void set_task_stack_end_magic(struct task_struct *tsk);
  2403. /* set thread flags in other task's structures
  2404. * - see asm/thread_info.h for TIF_xxxx flags available
  2405. */
  2406. static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2407. {
  2408. set_ti_thread_flag(task_thread_info(tsk), flag);
  2409. }
  2410. static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2411. {
  2412. clear_ti_thread_flag(task_thread_info(tsk), flag);
  2413. }
  2414. static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2415. {
  2416. return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
  2417. }
  2418. static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2419. {
  2420. return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
  2421. }
  2422. static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
  2423. {
  2424. return test_ti_thread_flag(task_thread_info(tsk), flag);
  2425. }
  2426. static inline void set_tsk_need_resched(struct task_struct *tsk)
  2427. {
  2428. set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2429. }
  2430. static inline void clear_tsk_need_resched(struct task_struct *tsk)
  2431. {
  2432. clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2433. }
  2434. static inline int test_tsk_need_resched(struct task_struct *tsk)
  2435. {
  2436. return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
  2437. }
  2438. static inline int restart_syscall(void)
  2439. {
  2440. set_tsk_thread_flag(current, TIF_SIGPENDING);
  2441. return -ERESTARTNOINTR;
  2442. }
  2443. static inline int signal_pending(struct task_struct *p)
  2444. {
  2445. return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
  2446. }
  2447. static inline int __fatal_signal_pending(struct task_struct *p)
  2448. {
  2449. return unlikely(sigismember(&p->pending.signal, SIGKILL));
  2450. }
  2451. static inline int fatal_signal_pending(struct task_struct *p)
  2452. {
  2453. return signal_pending(p) && __fatal_signal_pending(p);
  2454. }
  2455. static inline int signal_pending_state(long state, struct task_struct *p)
  2456. {
  2457. if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
  2458. return 0;
  2459. if (!signal_pending(p))
  2460. return 0;
  2461. return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
  2462. }
  2463. /*
  2464. * cond_resched() and cond_resched_lock(): latency reduction via
  2465. * explicit rescheduling in places that are safe. The return
  2466. * value indicates whether a reschedule was done in fact.
  2467. * cond_resched_lock() will drop the spinlock before scheduling,
  2468. * cond_resched_softirq() will enable bhs before scheduling.
  2469. */
  2470. extern int _cond_resched(void);
  2471. #define cond_resched() ({ \
  2472. ___might_sleep(__FILE__, __LINE__, 0); \
  2473. _cond_resched(); \
  2474. })
  2475. extern int __cond_resched_lock(spinlock_t *lock);
  2476. #define cond_resched_lock(lock) ({ \
  2477. ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
  2478. __cond_resched_lock(lock); \
  2479. })
  2480. extern int __cond_resched_softirq(void);
  2481. #define cond_resched_softirq() ({ \
  2482. ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
  2483. __cond_resched_softirq(); \
  2484. })
  2485. static inline void cond_resched_rcu(void)
  2486. {
  2487. #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
  2488. rcu_read_unlock();
  2489. cond_resched();
  2490. rcu_read_lock();
  2491. #endif
  2492. }
  2493. /*
  2494. * Does a critical section need to be broken due to another
  2495. * task waiting?: (technically does not depend on CONFIG_PREEMPT,
  2496. * but a general need for low latency)
  2497. */
  2498. static inline int spin_needbreak(spinlock_t *lock)
  2499. {
  2500. #ifdef CONFIG_PREEMPT
  2501. return spin_is_contended(lock);
  2502. #else
  2503. return 0;
  2504. #endif
  2505. }
  2506. /*
  2507. * Idle thread specific functions to determine the need_resched
  2508. * polling state.
  2509. */
  2510. #ifdef TIF_POLLING_NRFLAG
  2511. static inline int tsk_is_polling(struct task_struct *p)
  2512. {
  2513. return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
  2514. }
  2515. static inline void __current_set_polling(void)
  2516. {
  2517. set_thread_flag(TIF_POLLING_NRFLAG);
  2518. }
  2519. static inline bool __must_check current_set_polling_and_test(void)
  2520. {
  2521. __current_set_polling();
  2522. /*
  2523. * Polling state must be visible before we test NEED_RESCHED,
  2524. * paired by resched_curr()
  2525. */
  2526. smp_mb__after_atomic();
  2527. return unlikely(tif_need_resched());
  2528. }
  2529. static inline void __current_clr_polling(void)
  2530. {
  2531. clear_thread_flag(TIF_POLLING_NRFLAG);
  2532. }
  2533. static inline bool __must_check current_clr_polling_and_test(void)
  2534. {
  2535. __current_clr_polling();
  2536. /*
  2537. * Polling state must be visible before we test NEED_RESCHED,
  2538. * paired by resched_curr()
  2539. */
  2540. smp_mb__after_atomic();
  2541. return unlikely(tif_need_resched());
  2542. }
  2543. #else
  2544. static inline int tsk_is_polling(struct task_struct *p) { return 0; }
  2545. static inline void __current_set_polling(void) { }
  2546. static inline void __current_clr_polling(void) { }
  2547. static inline bool __must_check current_set_polling_and_test(void)
  2548. {
  2549. return unlikely(tif_need_resched());
  2550. }
  2551. static inline bool __must_check current_clr_polling_and_test(void)
  2552. {
  2553. return unlikely(tif_need_resched());
  2554. }
  2555. #endif
  2556. static inline void current_clr_polling(void)
  2557. {
  2558. __current_clr_polling();
  2559. /*
  2560. * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
  2561. * Once the bit is cleared, we'll get IPIs with every new
  2562. * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
  2563. * fold.
  2564. */
  2565. smp_mb(); /* paired with resched_curr() */
  2566. preempt_fold_need_resched();
  2567. }
  2568. static __always_inline bool need_resched(void)
  2569. {
  2570. return unlikely(tif_need_resched());
  2571. }
  2572. /*
  2573. * Thread group CPU time accounting.
  2574. */
  2575. void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
  2576. void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
  2577. /*
  2578. * Reevaluate whether the task has signals pending delivery.
  2579. * Wake the task if so.
  2580. * This is required every time the blocked sigset_t changes.
  2581. * callers must hold sighand->siglock.
  2582. */
  2583. extern void recalc_sigpending_and_wake(struct task_struct *t);
  2584. extern void recalc_sigpending(void);
  2585. extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
  2586. static inline void signal_wake_up(struct task_struct *t, bool resume)
  2587. {
  2588. signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
  2589. }
  2590. static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
  2591. {
  2592. signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
  2593. }
  2594. /*
  2595. * Wrappers for p->thread_info->cpu access. No-op on UP.
  2596. */
  2597. #ifdef CONFIG_SMP
  2598. static inline unsigned int task_cpu(const struct task_struct *p)
  2599. {
  2600. return task_thread_info(p)->cpu;
  2601. }
  2602. static inline int task_node(const struct task_struct *p)
  2603. {
  2604. return cpu_to_node(task_cpu(p));
  2605. }
  2606. extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
  2607. #else
  2608. static inline unsigned int task_cpu(const struct task_struct *p)
  2609. {
  2610. return 0;
  2611. }
  2612. static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
  2613. {
  2614. }
  2615. #endif /* CONFIG_SMP */
  2616. extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
  2617. extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
  2618. #ifdef CONFIG_CGROUP_SCHED
  2619. extern struct task_group root_task_group;
  2620. #endif /* CONFIG_CGROUP_SCHED */
  2621. extern int task_can_switch_user(struct user_struct *up,
  2622. struct task_struct *tsk);
  2623. #ifdef CONFIG_TASK_XACCT
  2624. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  2625. {
  2626. tsk->ioac.rchar += amt;
  2627. }
  2628. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  2629. {
  2630. tsk->ioac.wchar += amt;
  2631. }
  2632. static inline void inc_syscr(struct task_struct *tsk)
  2633. {
  2634. tsk->ioac.syscr++;
  2635. }
  2636. static inline void inc_syscw(struct task_struct *tsk)
  2637. {
  2638. tsk->ioac.syscw++;
  2639. }
  2640. #else
  2641. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  2642. {
  2643. }
  2644. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  2645. {
  2646. }
  2647. static inline void inc_syscr(struct task_struct *tsk)
  2648. {
  2649. }
  2650. static inline void inc_syscw(struct task_struct *tsk)
  2651. {
  2652. }
  2653. #endif
  2654. #ifndef TASK_SIZE_OF
  2655. #define TASK_SIZE_OF(tsk) TASK_SIZE
  2656. #endif
  2657. #ifdef CONFIG_MEMCG
  2658. extern void mm_update_next_owner(struct mm_struct *mm);
  2659. #else
  2660. static inline void mm_update_next_owner(struct mm_struct *mm)
  2661. {
  2662. }
  2663. #endif /* CONFIG_MEMCG */
  2664. static inline unsigned long task_rlimit(const struct task_struct *tsk,
  2665. unsigned int limit)
  2666. {
  2667. return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
  2668. }
  2669. static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
  2670. unsigned int limit)
  2671. {
  2672. return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
  2673. }
  2674. static inline unsigned long rlimit(unsigned int limit)
  2675. {
  2676. return task_rlimit(current, limit);
  2677. }
  2678. static inline unsigned long rlimit_max(unsigned int limit)
  2679. {
  2680. return task_rlimit_max(current, limit);
  2681. }
  2682. #endif