sched.h 105 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908290929102911291229132914291529162917291829192920292129222923292429252926292729282929293029312932293329342935293629372938293929402941294229432944294529462947294829492950295129522953295429552956295729582959296029612962296329642965296629672968296929702971297229732974297529762977297829792980298129822983298429852986298729882989299029912992299329942995299629972998299930003001300230033004300530063007300830093010301130123013301430153016301730183019302030213022302330243025302630273028302930303031303230333034303530363037303830393040304130423043304430453046304730483049305030513052305330543055305630573058305930603061306230633064306530663067306830693070307130723073307430753076307730783079308030813082308330843085308630873088308930903091309230933094309530963097309830993100310131023103310431053106310731083109311031113112311331143115311631173118311931203121312231233124312531263127312831293130313131323133313431353136313731383139314031413142314331443145314631473148314931503151315231533154315531563157315831593160316131623163316431653166316731683169317031713172317331743175317631773178317931803181318231833184318531863187318831893190319131923193319431953196319731983199320032013202320332043205320632073208320932103211321232133214321532163217321832193220322132223223322432253226322732283229323032313232323332343235323632373238323932403241324232433244324532463247324832493250325132523253325432553256325732583259326032613262326332643265326632673268326932703271327232733274327532763277327832793280328132823283328432853286328732883289329032913292329332943295329632973298329933003301330233033304330533063307330833093310331133123313331433153316331733183319332033213322332333243325332633273328332933303331333233333334333533363337333833393340334133423343334433453346334733483349335033513352335333543355335633573358335933603361336233633364336533663367336833693370337133723373337433753376337733783379338033813382338333843385338633873388338933903391339233933394339533963397339833993400340134023403340434053406340734083409341034113412341334143415341634173418341934203421342234233424342534263427342834293430343134323433343434353436343734383439344034413442344334443445344634473448344934503451345234533454345534563457345834593460346134623463346434653466346734683469347034713472347334743475347634773478347934803481348234833484348534863487348834893490349134923493349434953496349734983499350035013502350335043505350635073508350935103511351235133514351535163517351835193520352135223523352435253526352735283529353035313532353335343535353635373538353935403541354235433544354535463547354835493550355135523553355435553556355735583559356035613562356335643565356635673568356935703571357235733574357535763577357835793580358135823583358435853586358735883589359035913592359335943595359635973598359936003601360236033604360536063607360836093610361136123613361436153616361736183619362036213622362336243625362636273628362936303631363236333634363536363637363836393640364136423643364436453646364736483649365036513652365336543655365636573658365936603661366236633664366536663667366836693670367136723673367436753676367736783679368036813682368336843685368636873688368936903691369236933694369536963697369836993700
  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/smp.h>
  26. #include <linux/sem.h>
  27. #include <linux/shm.h>
  28. #include <linux/signal.h>
  29. #include <linux/compiler.h>
  30. #include <linux/completion.h>
  31. #include <linux/pid.h>
  32. #include <linux/percpu.h>
  33. #include <linux/topology.h>
  34. #include <linux/seccomp.h>
  35. #include <linux/rcupdate.h>
  36. #include <linux/rculist.h>
  37. #include <linux/rtmutex.h>
  38. #include <linux/time.h>
  39. #include <linux/param.h>
  40. #include <linux/resource.h>
  41. #include <linux/timer.h>
  42. #include <linux/hrtimer.h>
  43. #include <linux/kcov.h>
  44. #include <linux/task_io_accounting.h>
  45. #include <linux/latencytop.h>
  46. #include <linux/cred.h>
  47. #include <linux/llist.h>
  48. #include <linux/uidgid.h>
  49. #include <linux/gfp.h>
  50. #include <linux/magic.h>
  51. #include <linux/cgroup-defs.h>
  52. #include <asm/processor.h>
  53. #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
  54. /*
  55. * Extended scheduling parameters data structure.
  56. *
  57. * This is needed because the original struct sched_param can not be
  58. * altered without introducing ABI issues with legacy applications
  59. * (e.g., in sched_getparam()).
  60. *
  61. * However, the possibility of specifying more than just a priority for
  62. * the tasks may be useful for a wide variety of application fields, e.g.,
  63. * multimedia, streaming, automation and control, and many others.
  64. *
  65. * This variant (sched_attr) is meant at describing a so-called
  66. * sporadic time-constrained task. In such model a task is specified by:
  67. * - the activation period or minimum instance inter-arrival time;
  68. * - the maximum (or average, depending on the actual scheduling
  69. * discipline) computation time of all instances, a.k.a. runtime;
  70. * - the deadline (relative to the actual activation time) of each
  71. * instance.
  72. * Very briefly, a periodic (sporadic) task asks for the execution of
  73. * some specific computation --which is typically called an instance--
  74. * (at most) every period. Moreover, each instance typically lasts no more
  75. * than the runtime and must be completed by time instant t equal to
  76. * the instance activation time + the deadline.
  77. *
  78. * This is reflected by the actual fields of the sched_attr structure:
  79. *
  80. * @size size of the structure, for fwd/bwd compat.
  81. *
  82. * @sched_policy task's scheduling policy
  83. * @sched_flags for customizing the scheduler behaviour
  84. * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
  85. * @sched_priority task's static priority (SCHED_FIFO/RR)
  86. * @sched_deadline representative of the task's deadline
  87. * @sched_runtime representative of the task's runtime
  88. * @sched_period representative of the task's period
  89. *
  90. * Given this task model, there are a multiplicity of scheduling algorithms
  91. * and policies, that can be used to ensure all the tasks will make their
  92. * timing constraints.
  93. *
  94. * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
  95. * only user of this new interface. More information about the algorithm
  96. * available in the scheduling class file or in Documentation/.
  97. */
  98. struct sched_attr {
  99. u32 size;
  100. u32 sched_policy;
  101. u64 sched_flags;
  102. /* SCHED_NORMAL, SCHED_BATCH */
  103. s32 sched_nice;
  104. /* SCHED_FIFO, SCHED_RR */
  105. u32 sched_priority;
  106. /* SCHED_DEADLINE */
  107. u64 sched_runtime;
  108. u64 sched_deadline;
  109. u64 sched_period;
  110. };
  111. struct futex_pi_state;
  112. struct robust_list_head;
  113. struct bio_list;
  114. struct fs_struct;
  115. struct perf_event_context;
  116. struct blk_plug;
  117. struct filename;
  118. struct nameidata;
  119. #define VMACACHE_BITS 2
  120. #define VMACACHE_SIZE (1U << VMACACHE_BITS)
  121. #define VMACACHE_MASK (VMACACHE_SIZE - 1)
  122. /*
  123. * These are the constant used to fake the fixed-point load-average
  124. * counting. Some notes:
  125. * - 11 bit fractions expand to 22 bits by the multiplies: this gives
  126. * a load-average precision of 10 bits integer + 11 bits fractional
  127. * - if you want to count load-averages more often, you need more
  128. * precision, or rounding will get you. With 2-second counting freq,
  129. * the EXP_n values would be 1981, 2034 and 2043 if still using only
  130. * 11 bit fractions.
  131. */
  132. extern unsigned long avenrun[]; /* Load averages */
  133. extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
  134. #define FSHIFT 11 /* nr of bits of precision */
  135. #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
  136. #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
  137. #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
  138. #define EXP_5 2014 /* 1/exp(5sec/5min) */
  139. #define EXP_15 2037 /* 1/exp(5sec/15min) */
  140. #define CALC_LOAD(load,exp,n) \
  141. load *= exp; \
  142. load += n*(FIXED_1-exp); \
  143. load >>= FSHIFT;
  144. extern unsigned long total_forks;
  145. extern int nr_threads;
  146. DECLARE_PER_CPU(unsigned long, process_counts);
  147. extern int nr_processes(void);
  148. extern unsigned long nr_running(void);
  149. extern bool single_task_running(void);
  150. extern unsigned long nr_iowait(void);
  151. extern unsigned long nr_iowait_cpu(int cpu);
  152. extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
  153. extern void calc_global_load(unsigned long ticks);
  154. #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
  155. extern void cpu_load_update_nohz_start(void);
  156. extern void cpu_load_update_nohz_stop(void);
  157. #else
  158. static inline void cpu_load_update_nohz_start(void) { }
  159. static inline void cpu_load_update_nohz_stop(void) { }
  160. #endif
  161. extern void dump_cpu_task(int cpu);
  162. struct seq_file;
  163. struct cfs_rq;
  164. struct task_group;
  165. #ifdef CONFIG_SCHED_DEBUG
  166. extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
  167. extern void proc_sched_set_task(struct task_struct *p);
  168. #endif
  169. /*
  170. * Task state bitmask. NOTE! These bits are also
  171. * encoded in fs/proc/array.c: get_task_state().
  172. *
  173. * We have two separate sets of flags: task->state
  174. * is about runnability, while task->exit_state are
  175. * about the task exiting. Confusing, but this way
  176. * modifying one set can't modify the other one by
  177. * mistake.
  178. */
  179. #define TASK_RUNNING 0
  180. #define TASK_INTERRUPTIBLE 1
  181. #define TASK_UNINTERRUPTIBLE 2
  182. #define __TASK_STOPPED 4
  183. #define __TASK_TRACED 8
  184. /* in tsk->exit_state */
  185. #define EXIT_DEAD 16
  186. #define EXIT_ZOMBIE 32
  187. #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
  188. /* in tsk->state again */
  189. #define TASK_DEAD 64
  190. #define TASK_WAKEKILL 128
  191. #define TASK_WAKING 256
  192. #define TASK_PARKED 512
  193. #define TASK_NOLOAD 1024
  194. #define TASK_NEW 2048
  195. #define TASK_STATE_MAX 4096
  196. #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
  197. extern char ___assert_task_state[1 - 2*!!(
  198. sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
  199. /* Convenience macros for the sake of set_current_state */
  200. #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
  201. #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
  202. #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
  203. #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
  204. /* Convenience macros for the sake of wake_up */
  205. #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
  206. #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
  207. /* get_task_state() */
  208. #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
  209. TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
  210. __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
  211. #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
  212. #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
  213. #define task_is_stopped_or_traced(task) \
  214. ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
  215. #define task_contributes_to_load(task) \
  216. ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
  217. (task->flags & PF_FROZEN) == 0 && \
  218. (task->state & TASK_NOLOAD) == 0)
  219. #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
  220. #define __set_current_state(state_value) \
  221. do { \
  222. current->task_state_change = _THIS_IP_; \
  223. current->state = (state_value); \
  224. } while (0)
  225. #define set_current_state(state_value) \
  226. do { \
  227. current->task_state_change = _THIS_IP_; \
  228. smp_store_mb(current->state, (state_value)); \
  229. } while (0)
  230. #else
  231. /*
  232. * set_current_state() includes a barrier so that the write of current->state
  233. * is correctly serialised wrt the caller's subsequent test of whether to
  234. * actually sleep:
  235. *
  236. * for (;;) {
  237. * set_current_state(TASK_UNINTERRUPTIBLE);
  238. * if (!need_sleep)
  239. * break;
  240. *
  241. * schedule();
  242. * }
  243. * __set_current_state(TASK_RUNNING);
  244. *
  245. * If the caller does not need such serialisation (because, for instance, the
  246. * condition test and condition change and wakeup are under the same lock) then
  247. * use __set_current_state().
  248. *
  249. * The above is typically ordered against the wakeup, which does:
  250. *
  251. * need_sleep = false;
  252. * wake_up_state(p, TASK_UNINTERRUPTIBLE);
  253. *
  254. * Where wake_up_state() (and all other wakeup primitives) imply enough
  255. * barriers to order the store of the variable against wakeup.
  256. *
  257. * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
  258. * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
  259. * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
  260. *
  261. * This is obviously fine, since they both store the exact same value.
  262. *
  263. * Also see the comments of try_to_wake_up().
  264. */
  265. #define __set_current_state(state_value) \
  266. do { current->state = (state_value); } while (0)
  267. #define set_current_state(state_value) \
  268. smp_store_mb(current->state, (state_value))
  269. #endif
  270. /* Task command name length */
  271. #define TASK_COMM_LEN 16
  272. #include <linux/spinlock.h>
  273. /*
  274. * This serializes "schedule()" and also protects
  275. * the run-queue from deletions/modifications (but
  276. * _adding_ to the beginning of the run-queue has
  277. * a separate lock).
  278. */
  279. extern rwlock_t tasklist_lock;
  280. extern spinlock_t mmlist_lock;
  281. struct task_struct;
  282. #ifdef CONFIG_PROVE_RCU
  283. extern int lockdep_tasklist_lock_is_held(void);
  284. #endif /* #ifdef CONFIG_PROVE_RCU */
  285. extern void sched_init(void);
  286. extern void sched_init_smp(void);
  287. extern asmlinkage void schedule_tail(struct task_struct *prev);
  288. extern void init_idle(struct task_struct *idle, int cpu);
  289. extern void init_idle_bootup_task(struct task_struct *idle);
  290. extern cpumask_var_t cpu_isolated_map;
  291. extern int runqueue_is_locked(int cpu);
  292. #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
  293. extern void nohz_balance_enter_idle(int cpu);
  294. extern void set_cpu_sd_state_idle(void);
  295. extern int get_nohz_timer_target(void);
  296. #else
  297. static inline void nohz_balance_enter_idle(int cpu) { }
  298. static inline void set_cpu_sd_state_idle(void) { }
  299. #endif
  300. /*
  301. * Only dump TASK_* tasks. (0 for all tasks)
  302. */
  303. extern void show_state_filter(unsigned long state_filter);
  304. static inline void show_state(void)
  305. {
  306. show_state_filter(0);
  307. }
  308. extern void show_regs(struct pt_regs *);
  309. /*
  310. * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
  311. * task), SP is the stack pointer of the first frame that should be shown in the back
  312. * trace (or NULL if the entire call-chain of the task should be shown).
  313. */
  314. extern void show_stack(struct task_struct *task, unsigned long *sp);
  315. extern void cpu_init (void);
  316. extern void trap_init(void);
  317. extern void update_process_times(int user);
  318. extern void scheduler_tick(void);
  319. extern int sched_cpu_starting(unsigned int cpu);
  320. extern int sched_cpu_activate(unsigned int cpu);
  321. extern int sched_cpu_deactivate(unsigned int cpu);
  322. #ifdef CONFIG_HOTPLUG_CPU
  323. extern int sched_cpu_dying(unsigned int cpu);
  324. #else
  325. # define sched_cpu_dying NULL
  326. #endif
  327. extern void sched_show_task(struct task_struct *p);
  328. #ifdef CONFIG_LOCKUP_DETECTOR
  329. extern void touch_softlockup_watchdog_sched(void);
  330. extern void touch_softlockup_watchdog(void);
  331. extern void touch_softlockup_watchdog_sync(void);
  332. extern void touch_all_softlockup_watchdogs(void);
  333. extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
  334. void __user *buffer,
  335. size_t *lenp, loff_t *ppos);
  336. extern unsigned int softlockup_panic;
  337. extern unsigned int hardlockup_panic;
  338. void lockup_detector_init(void);
  339. #else
  340. static inline void touch_softlockup_watchdog_sched(void)
  341. {
  342. }
  343. static inline void touch_softlockup_watchdog(void)
  344. {
  345. }
  346. static inline void touch_softlockup_watchdog_sync(void)
  347. {
  348. }
  349. static inline void touch_all_softlockup_watchdogs(void)
  350. {
  351. }
  352. static inline void lockup_detector_init(void)
  353. {
  354. }
  355. #endif
  356. #ifdef CONFIG_DETECT_HUNG_TASK
  357. void reset_hung_task_detector(void);
  358. #else
  359. static inline void reset_hung_task_detector(void)
  360. {
  361. }
  362. #endif
  363. /* Attach to any functions which should be ignored in wchan output. */
  364. #define __sched __attribute__((__section__(".sched.text")))
  365. /* Linker adds these: start and end of __sched functions */
  366. extern char __sched_text_start[], __sched_text_end[];
  367. /* Is this address in the __sched functions? */
  368. extern int in_sched_functions(unsigned long addr);
  369. #define MAX_SCHEDULE_TIMEOUT LONG_MAX
  370. extern signed long schedule_timeout(signed long timeout);
  371. extern signed long schedule_timeout_interruptible(signed long timeout);
  372. extern signed long schedule_timeout_killable(signed long timeout);
  373. extern signed long schedule_timeout_uninterruptible(signed long timeout);
  374. extern signed long schedule_timeout_idle(signed long timeout);
  375. asmlinkage void schedule(void);
  376. extern void schedule_preempt_disabled(void);
  377. extern int __must_check io_schedule_prepare(void);
  378. extern void io_schedule_finish(int token);
  379. extern long io_schedule_timeout(long timeout);
  380. extern void io_schedule(void);
  381. void __noreturn do_task_dead(void);
  382. struct nsproxy;
  383. struct user_namespace;
  384. #ifdef CONFIG_MMU
  385. extern void arch_pick_mmap_layout(struct mm_struct *mm);
  386. extern unsigned long
  387. arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
  388. unsigned long, unsigned long);
  389. extern unsigned long
  390. arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
  391. unsigned long len, unsigned long pgoff,
  392. unsigned long flags);
  393. #else
  394. static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
  395. #endif
  396. #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
  397. #define SUID_DUMP_USER 1 /* Dump as user of process */
  398. #define SUID_DUMP_ROOT 2 /* Dump as root */
  399. /* mm flags */
  400. /* for SUID_DUMP_* above */
  401. #define MMF_DUMPABLE_BITS 2
  402. #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
  403. extern void set_dumpable(struct mm_struct *mm, int value);
  404. /*
  405. * This returns the actual value of the suid_dumpable flag. For things
  406. * that are using this for checking for privilege transitions, it must
  407. * test against SUID_DUMP_USER rather than treating it as a boolean
  408. * value.
  409. */
  410. static inline int __get_dumpable(unsigned long mm_flags)
  411. {
  412. return mm_flags & MMF_DUMPABLE_MASK;
  413. }
  414. static inline int get_dumpable(struct mm_struct *mm)
  415. {
  416. return __get_dumpable(mm->flags);
  417. }
  418. /* coredump filter bits */
  419. #define MMF_DUMP_ANON_PRIVATE 2
  420. #define MMF_DUMP_ANON_SHARED 3
  421. #define MMF_DUMP_MAPPED_PRIVATE 4
  422. #define MMF_DUMP_MAPPED_SHARED 5
  423. #define MMF_DUMP_ELF_HEADERS 6
  424. #define MMF_DUMP_HUGETLB_PRIVATE 7
  425. #define MMF_DUMP_HUGETLB_SHARED 8
  426. #define MMF_DUMP_DAX_PRIVATE 9
  427. #define MMF_DUMP_DAX_SHARED 10
  428. #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
  429. #define MMF_DUMP_FILTER_BITS 9
  430. #define MMF_DUMP_FILTER_MASK \
  431. (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
  432. #define MMF_DUMP_FILTER_DEFAULT \
  433. ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
  434. (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
  435. #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
  436. # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
  437. #else
  438. # define MMF_DUMP_MASK_DEFAULT_ELF 0
  439. #endif
  440. /* leave room for more dump flags */
  441. #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
  442. #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
  443. /*
  444. * This one-shot flag is dropped due to necessity of changing exe once again
  445. * on NFS restore
  446. */
  447. //#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
  448. #define MMF_HAS_UPROBES 19 /* has uprobes */
  449. #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
  450. #define MMF_OOM_SKIP 21 /* mm is of no interest for the OOM killer */
  451. #define MMF_UNSTABLE 22 /* mm is unstable for copy_from_user */
  452. #define MMF_HUGE_ZERO_PAGE 23 /* mm has ever used the global huge zero page */
  453. #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
  454. struct sighand_struct {
  455. atomic_t count;
  456. struct k_sigaction action[_NSIG];
  457. spinlock_t siglock;
  458. wait_queue_head_t signalfd_wqh;
  459. };
  460. struct pacct_struct {
  461. int ac_flag;
  462. long ac_exitcode;
  463. unsigned long ac_mem;
  464. u64 ac_utime, ac_stime;
  465. unsigned long ac_minflt, ac_majflt;
  466. };
  467. struct cpu_itimer {
  468. u64 expires;
  469. u64 incr;
  470. };
  471. /**
  472. * struct prev_cputime - snaphsot of system and user cputime
  473. * @utime: time spent in user mode
  474. * @stime: time spent in system mode
  475. * @lock: protects the above two fields
  476. *
  477. * Stores previous user/system time values such that we can guarantee
  478. * monotonicity.
  479. */
  480. struct prev_cputime {
  481. #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
  482. u64 utime;
  483. u64 stime;
  484. raw_spinlock_t lock;
  485. #endif
  486. };
  487. static inline void prev_cputime_init(struct prev_cputime *prev)
  488. {
  489. #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
  490. prev->utime = prev->stime = 0;
  491. raw_spin_lock_init(&prev->lock);
  492. #endif
  493. }
  494. /**
  495. * struct task_cputime - collected CPU time counts
  496. * @utime: time spent in user mode, in nanoseconds
  497. * @stime: time spent in kernel mode, in nanoseconds
  498. * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
  499. *
  500. * This structure groups together three kinds of CPU time that are tracked for
  501. * threads and thread groups. Most things considering CPU time want to group
  502. * these counts together and treat all three of them in parallel.
  503. */
  504. struct task_cputime {
  505. u64 utime;
  506. u64 stime;
  507. unsigned long long sum_exec_runtime;
  508. };
  509. /* Alternate field names when used to cache expirations. */
  510. #define virt_exp utime
  511. #define prof_exp stime
  512. #define sched_exp sum_exec_runtime
  513. /*
  514. * This is the atomic variant of task_cputime, which can be used for
  515. * storing and updating task_cputime statistics without locking.
  516. */
  517. struct task_cputime_atomic {
  518. atomic64_t utime;
  519. atomic64_t stime;
  520. atomic64_t sum_exec_runtime;
  521. };
  522. #define INIT_CPUTIME_ATOMIC \
  523. (struct task_cputime_atomic) { \
  524. .utime = ATOMIC64_INIT(0), \
  525. .stime = ATOMIC64_INIT(0), \
  526. .sum_exec_runtime = ATOMIC64_INIT(0), \
  527. }
  528. #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
  529. /*
  530. * Disable preemption until the scheduler is running -- use an unconditional
  531. * value so that it also works on !PREEMPT_COUNT kernels.
  532. *
  533. * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
  534. */
  535. #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
  536. /*
  537. * Initial preempt_count value; reflects the preempt_count schedule invariant
  538. * which states that during context switches:
  539. *
  540. * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
  541. *
  542. * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
  543. * Note: See finish_task_switch().
  544. */
  545. #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
  546. /**
  547. * struct thread_group_cputimer - thread group interval timer counts
  548. * @cputime_atomic: atomic thread group interval timers.
  549. * @running: true when there are timers running and
  550. * @cputime_atomic receives updates.
  551. * @checking_timer: true when a thread in the group is in the
  552. * process of checking for thread group timers.
  553. *
  554. * This structure contains the version of task_cputime, above, that is
  555. * used for thread group CPU timer calculations.
  556. */
  557. struct thread_group_cputimer {
  558. struct task_cputime_atomic cputime_atomic;
  559. bool running;
  560. bool checking_timer;
  561. };
  562. #include <linux/rwsem.h>
  563. struct autogroup;
  564. /*
  565. * NOTE! "signal_struct" does not have its own
  566. * locking, because a shared signal_struct always
  567. * implies a shared sighand_struct, so locking
  568. * sighand_struct is always a proper superset of
  569. * the locking of signal_struct.
  570. */
  571. struct signal_struct {
  572. atomic_t sigcnt;
  573. atomic_t live;
  574. int nr_threads;
  575. struct list_head thread_head;
  576. wait_queue_head_t wait_chldexit; /* for wait4() */
  577. /* current thread group signal load-balancing target: */
  578. struct task_struct *curr_target;
  579. /* shared signal handling: */
  580. struct sigpending shared_pending;
  581. /* thread group exit support */
  582. int group_exit_code;
  583. /* overloaded:
  584. * - notify group_exit_task when ->count is equal to notify_count
  585. * - everyone except group_exit_task is stopped during signal delivery
  586. * of fatal signals, group_exit_task processes the signal.
  587. */
  588. int notify_count;
  589. struct task_struct *group_exit_task;
  590. /* thread group stop support, overloads group_exit_code too */
  591. int group_stop_count;
  592. unsigned int flags; /* see SIGNAL_* flags below */
  593. /*
  594. * PR_SET_CHILD_SUBREAPER marks a process, like a service
  595. * manager, to re-parent orphan (double-forking) child processes
  596. * to this process instead of 'init'. The service manager is
  597. * able to receive SIGCHLD signals and is able to investigate
  598. * the process until it calls wait(). All children of this
  599. * process will inherit a flag if they should look for a
  600. * child_subreaper process at exit.
  601. */
  602. unsigned int is_child_subreaper:1;
  603. unsigned int has_child_subreaper:1;
  604. #ifdef CONFIG_POSIX_TIMERS
  605. /* POSIX.1b Interval Timers */
  606. int posix_timer_id;
  607. struct list_head posix_timers;
  608. /* ITIMER_REAL timer for the process */
  609. struct hrtimer real_timer;
  610. ktime_t it_real_incr;
  611. /*
  612. * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
  613. * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
  614. * values are defined to 0 and 1 respectively
  615. */
  616. struct cpu_itimer it[2];
  617. /*
  618. * Thread group totals for process CPU timers.
  619. * See thread_group_cputimer(), et al, for details.
  620. */
  621. struct thread_group_cputimer cputimer;
  622. /* Earliest-expiration cache. */
  623. struct task_cputime cputime_expires;
  624. struct list_head cpu_timers[3];
  625. #endif
  626. struct pid *leader_pid;
  627. #ifdef CONFIG_NO_HZ_FULL
  628. atomic_t tick_dep_mask;
  629. #endif
  630. struct pid *tty_old_pgrp;
  631. /* boolean value for session group leader */
  632. int leader;
  633. struct tty_struct *tty; /* NULL if no tty */
  634. #ifdef CONFIG_SCHED_AUTOGROUP
  635. struct autogroup *autogroup;
  636. #endif
  637. /*
  638. * Cumulative resource counters for dead threads in the group,
  639. * and for reaped dead child processes forked by this group.
  640. * Live threads maintain their own counters and add to these
  641. * in __exit_signal, except for the group leader.
  642. */
  643. seqlock_t stats_lock;
  644. u64 utime, stime, cutime, cstime;
  645. u64 gtime;
  646. u64 cgtime;
  647. struct prev_cputime prev_cputime;
  648. unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
  649. unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
  650. unsigned long inblock, oublock, cinblock, coublock;
  651. unsigned long maxrss, cmaxrss;
  652. struct task_io_accounting ioac;
  653. /*
  654. * Cumulative ns of schedule CPU time fo dead threads in the
  655. * group, not including a zombie group leader, (This only differs
  656. * from jiffies_to_ns(utime + stime) if sched_clock uses something
  657. * other than jiffies.)
  658. */
  659. unsigned long long sum_sched_runtime;
  660. /*
  661. * We don't bother to synchronize most readers of this at all,
  662. * because there is no reader checking a limit that actually needs
  663. * to get both rlim_cur and rlim_max atomically, and either one
  664. * alone is a single word that can safely be read normally.
  665. * getrlimit/setrlimit use task_lock(current->group_leader) to
  666. * protect this instead of the siglock, because they really
  667. * have no need to disable irqs.
  668. */
  669. struct rlimit rlim[RLIM_NLIMITS];
  670. #ifdef CONFIG_BSD_PROCESS_ACCT
  671. struct pacct_struct pacct; /* per-process accounting information */
  672. #endif
  673. #ifdef CONFIG_TASKSTATS
  674. struct taskstats *stats;
  675. #endif
  676. #ifdef CONFIG_AUDIT
  677. unsigned audit_tty;
  678. struct tty_audit_buf *tty_audit_buf;
  679. #endif
  680. /*
  681. * Thread is the potential origin of an oom condition; kill first on
  682. * oom
  683. */
  684. bool oom_flag_origin;
  685. short oom_score_adj; /* OOM kill score adjustment */
  686. short oom_score_adj_min; /* OOM kill score adjustment min value.
  687. * Only settable by CAP_SYS_RESOURCE. */
  688. struct mm_struct *oom_mm; /* recorded mm when the thread group got
  689. * killed by the oom killer */
  690. struct mutex cred_guard_mutex; /* guard against foreign influences on
  691. * credential calculations
  692. * (notably. ptrace) */
  693. };
  694. /*
  695. * Bits in flags field of signal_struct.
  696. */
  697. #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
  698. #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
  699. #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
  700. #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
  701. /*
  702. * Pending notifications to parent.
  703. */
  704. #define SIGNAL_CLD_STOPPED 0x00000010
  705. #define SIGNAL_CLD_CONTINUED 0x00000020
  706. #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
  707. #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
  708. #define SIGNAL_STOP_MASK (SIGNAL_CLD_MASK | SIGNAL_STOP_STOPPED | \
  709. SIGNAL_STOP_CONTINUED)
  710. static inline void signal_set_stop_flags(struct signal_struct *sig,
  711. unsigned int flags)
  712. {
  713. WARN_ON(sig->flags & (SIGNAL_GROUP_EXIT|SIGNAL_GROUP_COREDUMP));
  714. sig->flags = (sig->flags & ~SIGNAL_STOP_MASK) | flags;
  715. }
  716. /* If true, all threads except ->group_exit_task have pending SIGKILL */
  717. static inline int signal_group_exit(const struct signal_struct *sig)
  718. {
  719. return (sig->flags & SIGNAL_GROUP_EXIT) ||
  720. (sig->group_exit_task != NULL);
  721. }
  722. /*
  723. * Some day this will be a full-fledged user tracking system..
  724. */
  725. struct user_struct {
  726. atomic_t __count; /* reference count */
  727. atomic_t processes; /* How many processes does this user have? */
  728. atomic_t sigpending; /* How many pending signals does this user have? */
  729. #ifdef CONFIG_FANOTIFY
  730. atomic_t fanotify_listeners;
  731. #endif
  732. #ifdef CONFIG_EPOLL
  733. atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
  734. #endif
  735. #ifdef CONFIG_POSIX_MQUEUE
  736. /* protected by mq_lock */
  737. unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
  738. #endif
  739. unsigned long locked_shm; /* How many pages of mlocked shm ? */
  740. unsigned long unix_inflight; /* How many files in flight in unix sockets */
  741. atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
  742. #ifdef CONFIG_KEYS
  743. struct key *uid_keyring; /* UID specific keyring */
  744. struct key *session_keyring; /* UID's default session keyring */
  745. #endif
  746. /* Hash table maintenance information */
  747. struct hlist_node uidhash_node;
  748. kuid_t uid;
  749. #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
  750. atomic_long_t locked_vm;
  751. #endif
  752. };
  753. extern int uids_sysfs_init(void);
  754. extern struct user_struct *find_user(kuid_t);
  755. extern struct user_struct root_user;
  756. #define INIT_USER (&root_user)
  757. struct backing_dev_info;
  758. struct reclaim_state;
  759. #ifdef CONFIG_SCHED_INFO
  760. struct sched_info {
  761. /* cumulative counters */
  762. unsigned long pcount; /* # of times run on this cpu */
  763. unsigned long long run_delay; /* time spent waiting on a runqueue */
  764. /* timestamps */
  765. unsigned long long last_arrival,/* when we last ran on a cpu */
  766. last_queued; /* when we were last queued to run */
  767. };
  768. #endif /* CONFIG_SCHED_INFO */
  769. #ifdef CONFIG_TASK_DELAY_ACCT
  770. struct task_delay_info {
  771. spinlock_t lock;
  772. unsigned int flags; /* Private per-task flags */
  773. /* For each stat XXX, add following, aligned appropriately
  774. *
  775. * struct timespec XXX_start, XXX_end;
  776. * u64 XXX_delay;
  777. * u32 XXX_count;
  778. *
  779. * Atomicity of updates to XXX_delay, XXX_count protected by
  780. * single lock above (split into XXX_lock if contention is an issue).
  781. */
  782. /*
  783. * XXX_count is incremented on every XXX operation, the delay
  784. * associated with the operation is added to XXX_delay.
  785. * XXX_delay contains the accumulated delay time in nanoseconds.
  786. */
  787. u64 blkio_start; /* Shared by blkio, swapin */
  788. u64 blkio_delay; /* wait for sync block io completion */
  789. u64 swapin_delay; /* wait for swapin block io completion */
  790. u32 blkio_count; /* total count of the number of sync block */
  791. /* io operations performed */
  792. u32 swapin_count; /* total count of the number of swapin block */
  793. /* io operations performed */
  794. u64 freepages_start;
  795. u64 freepages_delay; /* wait for memory reclaim */
  796. u32 freepages_count; /* total count of memory reclaim */
  797. };
  798. #endif /* CONFIG_TASK_DELAY_ACCT */
  799. static inline int sched_info_on(void)
  800. {
  801. #ifdef CONFIG_SCHEDSTATS
  802. return 1;
  803. #elif defined(CONFIG_TASK_DELAY_ACCT)
  804. extern int delayacct_on;
  805. return delayacct_on;
  806. #else
  807. return 0;
  808. #endif
  809. }
  810. #ifdef CONFIG_SCHEDSTATS
  811. void force_schedstat_enabled(void);
  812. #endif
  813. enum cpu_idle_type {
  814. CPU_IDLE,
  815. CPU_NOT_IDLE,
  816. CPU_NEWLY_IDLE,
  817. CPU_MAX_IDLE_TYPES
  818. };
  819. /*
  820. * Integer metrics need fixed point arithmetic, e.g., sched/fair
  821. * has a few: load, load_avg, util_avg, freq, and capacity.
  822. *
  823. * We define a basic fixed point arithmetic range, and then formalize
  824. * all these metrics based on that basic range.
  825. */
  826. # define SCHED_FIXEDPOINT_SHIFT 10
  827. # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
  828. /*
  829. * Increase resolution of cpu_capacity calculations
  830. */
  831. #define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
  832. #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
  833. /*
  834. * Wake-queues are lists of tasks with a pending wakeup, whose
  835. * callers have already marked the task as woken internally,
  836. * and can thus carry on. A common use case is being able to
  837. * do the wakeups once the corresponding user lock as been
  838. * released.
  839. *
  840. * We hold reference to each task in the list across the wakeup,
  841. * thus guaranteeing that the memory is still valid by the time
  842. * the actual wakeups are performed in wake_up_q().
  843. *
  844. * One per task suffices, because there's never a need for a task to be
  845. * in two wake queues simultaneously; it is forbidden to abandon a task
  846. * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
  847. * already in a wake queue, the wakeup will happen soon and the second
  848. * waker can just skip it.
  849. *
  850. * The DEFINE_WAKE_Q macro declares and initializes the list head.
  851. * wake_up_q() does NOT reinitialize the list; it's expected to be
  852. * called near the end of a function. Otherwise, the list can be
  853. * re-initialized for later re-use by wake_q_init().
  854. *
  855. * Note that this can cause spurious wakeups. schedule() callers
  856. * must ensure the call is done inside a loop, confirming that the
  857. * wakeup condition has in fact occurred.
  858. */
  859. struct wake_q_node {
  860. struct wake_q_node *next;
  861. };
  862. struct wake_q_head {
  863. struct wake_q_node *first;
  864. struct wake_q_node **lastp;
  865. };
  866. #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
  867. #define DEFINE_WAKE_Q(name) \
  868. struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
  869. static inline void wake_q_init(struct wake_q_head *head)
  870. {
  871. head->first = WAKE_Q_TAIL;
  872. head->lastp = &head->first;
  873. }
  874. extern void wake_q_add(struct wake_q_head *head,
  875. struct task_struct *task);
  876. extern void wake_up_q(struct wake_q_head *head);
  877. /*
  878. * sched-domains (multiprocessor balancing) declarations:
  879. */
  880. #ifdef CONFIG_SMP
  881. #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
  882. #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
  883. #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
  884. #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
  885. #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
  886. #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
  887. #define SD_ASYM_CPUCAPACITY 0x0040 /* Groups have different max cpu capacities */
  888. #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu capacity */
  889. #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
  890. #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
  891. #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
  892. #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
  893. #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
  894. #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
  895. #define SD_NUMA 0x4000 /* cross-node balancing */
  896. #ifdef CONFIG_SCHED_SMT
  897. static inline int cpu_smt_flags(void)
  898. {
  899. return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
  900. }
  901. #endif
  902. #ifdef CONFIG_SCHED_MC
  903. static inline int cpu_core_flags(void)
  904. {
  905. return SD_SHARE_PKG_RESOURCES;
  906. }
  907. #endif
  908. #ifdef CONFIG_NUMA
  909. static inline int cpu_numa_flags(void)
  910. {
  911. return SD_NUMA;
  912. }
  913. #endif
  914. extern int arch_asym_cpu_priority(int cpu);
  915. struct sched_domain_attr {
  916. int relax_domain_level;
  917. };
  918. #define SD_ATTR_INIT (struct sched_domain_attr) { \
  919. .relax_domain_level = -1, \
  920. }
  921. extern int sched_domain_level_max;
  922. struct sched_group;
  923. struct sched_domain_shared {
  924. atomic_t ref;
  925. atomic_t nr_busy_cpus;
  926. int has_idle_cores;
  927. };
  928. struct sched_domain {
  929. /* These fields must be setup */
  930. struct sched_domain *parent; /* top domain must be null terminated */
  931. struct sched_domain *child; /* bottom domain must be null terminated */
  932. struct sched_group *groups; /* the balancing groups of the domain */
  933. unsigned long min_interval; /* Minimum balance interval ms */
  934. unsigned long max_interval; /* Maximum balance interval ms */
  935. unsigned int busy_factor; /* less balancing by factor if busy */
  936. unsigned int imbalance_pct; /* No balance until over watermark */
  937. unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
  938. unsigned int busy_idx;
  939. unsigned int idle_idx;
  940. unsigned int newidle_idx;
  941. unsigned int wake_idx;
  942. unsigned int forkexec_idx;
  943. unsigned int smt_gain;
  944. int nohz_idle; /* NOHZ IDLE status */
  945. int flags; /* See SD_* */
  946. int level;
  947. /* Runtime fields. */
  948. unsigned long last_balance; /* init to jiffies. units in jiffies */
  949. unsigned int balance_interval; /* initialise to 1. units in ms. */
  950. unsigned int nr_balance_failed; /* initialise to 0 */
  951. /* idle_balance() stats */
  952. u64 max_newidle_lb_cost;
  953. unsigned long next_decay_max_lb_cost;
  954. u64 avg_scan_cost; /* select_idle_sibling */
  955. #ifdef CONFIG_SCHEDSTATS
  956. /* load_balance() stats */
  957. unsigned int lb_count[CPU_MAX_IDLE_TYPES];
  958. unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
  959. unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
  960. unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
  961. unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
  962. unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
  963. unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
  964. unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
  965. /* Active load balancing */
  966. unsigned int alb_count;
  967. unsigned int alb_failed;
  968. unsigned int alb_pushed;
  969. /* SD_BALANCE_EXEC stats */
  970. unsigned int sbe_count;
  971. unsigned int sbe_balanced;
  972. unsigned int sbe_pushed;
  973. /* SD_BALANCE_FORK stats */
  974. unsigned int sbf_count;
  975. unsigned int sbf_balanced;
  976. unsigned int sbf_pushed;
  977. /* try_to_wake_up() stats */
  978. unsigned int ttwu_wake_remote;
  979. unsigned int ttwu_move_affine;
  980. unsigned int ttwu_move_balance;
  981. #endif
  982. #ifdef CONFIG_SCHED_DEBUG
  983. char *name;
  984. #endif
  985. union {
  986. void *private; /* used during construction */
  987. struct rcu_head rcu; /* used during destruction */
  988. };
  989. struct sched_domain_shared *shared;
  990. unsigned int span_weight;
  991. /*
  992. * Span of all CPUs in this domain.
  993. *
  994. * NOTE: this field is variable length. (Allocated dynamically
  995. * by attaching extra space to the end of the structure,
  996. * depending on how many CPUs the kernel has booted up with)
  997. */
  998. unsigned long span[0];
  999. };
  1000. static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
  1001. {
  1002. return to_cpumask(sd->span);
  1003. }
  1004. extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  1005. struct sched_domain_attr *dattr_new);
  1006. /* Allocate an array of sched domains, for partition_sched_domains(). */
  1007. cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
  1008. void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
  1009. bool cpus_share_cache(int this_cpu, int that_cpu);
  1010. typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
  1011. typedef int (*sched_domain_flags_f)(void);
  1012. #define SDTL_OVERLAP 0x01
  1013. struct sd_data {
  1014. struct sched_domain **__percpu sd;
  1015. struct sched_domain_shared **__percpu sds;
  1016. struct sched_group **__percpu sg;
  1017. struct sched_group_capacity **__percpu sgc;
  1018. };
  1019. struct sched_domain_topology_level {
  1020. sched_domain_mask_f mask;
  1021. sched_domain_flags_f sd_flags;
  1022. int flags;
  1023. int numa_level;
  1024. struct sd_data data;
  1025. #ifdef CONFIG_SCHED_DEBUG
  1026. char *name;
  1027. #endif
  1028. };
  1029. extern void set_sched_topology(struct sched_domain_topology_level *tl);
  1030. extern void wake_up_if_idle(int cpu);
  1031. #ifdef CONFIG_SCHED_DEBUG
  1032. # define SD_INIT_NAME(type) .name = #type
  1033. #else
  1034. # define SD_INIT_NAME(type)
  1035. #endif
  1036. #else /* CONFIG_SMP */
  1037. struct sched_domain_attr;
  1038. static inline void
  1039. partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
  1040. struct sched_domain_attr *dattr_new)
  1041. {
  1042. }
  1043. static inline bool cpus_share_cache(int this_cpu, int that_cpu)
  1044. {
  1045. return true;
  1046. }
  1047. #endif /* !CONFIG_SMP */
  1048. struct io_context; /* See blkdev.h */
  1049. #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
  1050. extern void prefetch_stack(struct task_struct *t);
  1051. #else
  1052. static inline void prefetch_stack(struct task_struct *t) { }
  1053. #endif
  1054. struct audit_context; /* See audit.c */
  1055. struct mempolicy;
  1056. struct pipe_inode_info;
  1057. struct uts_namespace;
  1058. struct load_weight {
  1059. unsigned long weight;
  1060. u32 inv_weight;
  1061. };
  1062. /*
  1063. * The load_avg/util_avg accumulates an infinite geometric series
  1064. * (see __update_load_avg() in kernel/sched/fair.c).
  1065. *
  1066. * [load_avg definition]
  1067. *
  1068. * load_avg = runnable% * scale_load_down(load)
  1069. *
  1070. * where runnable% is the time ratio that a sched_entity is runnable.
  1071. * For cfs_rq, it is the aggregated load_avg of all runnable and
  1072. * blocked sched_entities.
  1073. *
  1074. * load_avg may also take frequency scaling into account:
  1075. *
  1076. * load_avg = runnable% * scale_load_down(load) * freq%
  1077. *
  1078. * where freq% is the CPU frequency normalized to the highest frequency.
  1079. *
  1080. * [util_avg definition]
  1081. *
  1082. * util_avg = running% * SCHED_CAPACITY_SCALE
  1083. *
  1084. * where running% is the time ratio that a sched_entity is running on
  1085. * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
  1086. * and blocked sched_entities.
  1087. *
  1088. * util_avg may also factor frequency scaling and CPU capacity scaling:
  1089. *
  1090. * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
  1091. *
  1092. * where freq% is the same as above, and capacity% is the CPU capacity
  1093. * normalized to the greatest capacity (due to uarch differences, etc).
  1094. *
  1095. * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
  1096. * themselves are in the range of [0, 1]. To do fixed point arithmetics,
  1097. * we therefore scale them to as large a range as necessary. This is for
  1098. * example reflected by util_avg's SCHED_CAPACITY_SCALE.
  1099. *
  1100. * [Overflow issue]
  1101. *
  1102. * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
  1103. * with the highest load (=88761), always runnable on a single cfs_rq,
  1104. * and should not overflow as the number already hits PID_MAX_LIMIT.
  1105. *
  1106. * For all other cases (including 32-bit kernels), struct load_weight's
  1107. * weight will overflow first before we do, because:
  1108. *
  1109. * Max(load_avg) <= Max(load.weight)
  1110. *
  1111. * Then it is the load_weight's responsibility to consider overflow
  1112. * issues.
  1113. */
  1114. struct sched_avg {
  1115. u64 last_update_time, load_sum;
  1116. u32 util_sum, period_contrib;
  1117. unsigned long load_avg, util_avg;
  1118. };
  1119. #ifdef CONFIG_SCHEDSTATS
  1120. struct sched_statistics {
  1121. u64 wait_start;
  1122. u64 wait_max;
  1123. u64 wait_count;
  1124. u64 wait_sum;
  1125. u64 iowait_count;
  1126. u64 iowait_sum;
  1127. u64 sleep_start;
  1128. u64 sleep_max;
  1129. s64 sum_sleep_runtime;
  1130. u64 block_start;
  1131. u64 block_max;
  1132. u64 exec_max;
  1133. u64 slice_max;
  1134. u64 nr_migrations_cold;
  1135. u64 nr_failed_migrations_affine;
  1136. u64 nr_failed_migrations_running;
  1137. u64 nr_failed_migrations_hot;
  1138. u64 nr_forced_migrations;
  1139. u64 nr_wakeups;
  1140. u64 nr_wakeups_sync;
  1141. u64 nr_wakeups_migrate;
  1142. u64 nr_wakeups_local;
  1143. u64 nr_wakeups_remote;
  1144. u64 nr_wakeups_affine;
  1145. u64 nr_wakeups_affine_attempts;
  1146. u64 nr_wakeups_passive;
  1147. u64 nr_wakeups_idle;
  1148. };
  1149. #endif
  1150. struct sched_entity {
  1151. struct load_weight load; /* for load-balancing */
  1152. struct rb_node run_node;
  1153. struct list_head group_node;
  1154. unsigned int on_rq;
  1155. u64 exec_start;
  1156. u64 sum_exec_runtime;
  1157. u64 vruntime;
  1158. u64 prev_sum_exec_runtime;
  1159. u64 nr_migrations;
  1160. #ifdef CONFIG_SCHEDSTATS
  1161. struct sched_statistics statistics;
  1162. #endif
  1163. #ifdef CONFIG_FAIR_GROUP_SCHED
  1164. int depth;
  1165. struct sched_entity *parent;
  1166. /* rq on which this entity is (to be) queued: */
  1167. struct cfs_rq *cfs_rq;
  1168. /* rq "owned" by this entity/group: */
  1169. struct cfs_rq *my_q;
  1170. #endif
  1171. #ifdef CONFIG_SMP
  1172. /*
  1173. * Per entity load average tracking.
  1174. *
  1175. * Put into separate cache line so it does not
  1176. * collide with read-mostly values above.
  1177. */
  1178. struct sched_avg avg ____cacheline_aligned_in_smp;
  1179. #endif
  1180. };
  1181. struct sched_rt_entity {
  1182. struct list_head run_list;
  1183. unsigned long timeout;
  1184. unsigned long watchdog_stamp;
  1185. unsigned int time_slice;
  1186. unsigned short on_rq;
  1187. unsigned short on_list;
  1188. struct sched_rt_entity *back;
  1189. #ifdef CONFIG_RT_GROUP_SCHED
  1190. struct sched_rt_entity *parent;
  1191. /* rq on which this entity is (to be) queued: */
  1192. struct rt_rq *rt_rq;
  1193. /* rq "owned" by this entity/group: */
  1194. struct rt_rq *my_q;
  1195. #endif
  1196. };
  1197. struct sched_dl_entity {
  1198. struct rb_node rb_node;
  1199. /*
  1200. * Original scheduling parameters. Copied here from sched_attr
  1201. * during sched_setattr(), they will remain the same until
  1202. * the next sched_setattr().
  1203. */
  1204. u64 dl_runtime; /* maximum runtime for each instance */
  1205. u64 dl_deadline; /* relative deadline of each instance */
  1206. u64 dl_period; /* separation of two instances (period) */
  1207. u64 dl_bw; /* dl_runtime / dl_deadline */
  1208. /*
  1209. * Actual scheduling parameters. Initialized with the values above,
  1210. * they are continously updated during task execution. Note that
  1211. * the remaining runtime could be < 0 in case we are in overrun.
  1212. */
  1213. s64 runtime; /* remaining runtime for this instance */
  1214. u64 deadline; /* absolute deadline for this instance */
  1215. unsigned int flags; /* specifying the scheduler behaviour */
  1216. /*
  1217. * Some bool flags:
  1218. *
  1219. * @dl_throttled tells if we exhausted the runtime. If so, the
  1220. * task has to wait for a replenishment to be performed at the
  1221. * next firing of dl_timer.
  1222. *
  1223. * @dl_boosted tells if we are boosted due to DI. If so we are
  1224. * outside bandwidth enforcement mechanism (but only until we
  1225. * exit the critical section);
  1226. *
  1227. * @dl_yielded tells if task gave up the cpu before consuming
  1228. * all its available runtime during the last job.
  1229. */
  1230. int dl_throttled, dl_boosted, dl_yielded;
  1231. /*
  1232. * Bandwidth enforcement timer. Each -deadline task has its
  1233. * own bandwidth to be enforced, thus we need one timer per task.
  1234. */
  1235. struct hrtimer dl_timer;
  1236. };
  1237. union rcu_special {
  1238. struct {
  1239. u8 blocked;
  1240. u8 need_qs;
  1241. u8 exp_need_qs;
  1242. u8 pad; /* Otherwise the compiler can store garbage here. */
  1243. } b; /* Bits. */
  1244. u32 s; /* Set of bits. */
  1245. };
  1246. struct rcu_node;
  1247. enum perf_event_task_context {
  1248. perf_invalid_context = -1,
  1249. perf_hw_context = 0,
  1250. perf_sw_context,
  1251. perf_nr_task_contexts,
  1252. };
  1253. /* Track pages that require TLB flushes */
  1254. struct tlbflush_unmap_batch {
  1255. /*
  1256. * Each bit set is a CPU that potentially has a TLB entry for one of
  1257. * the PFNs being flushed. See set_tlb_ubc_flush_pending().
  1258. */
  1259. struct cpumask cpumask;
  1260. /* True if any bit in cpumask is set */
  1261. bool flush_required;
  1262. /*
  1263. * If true then the PTE was dirty when unmapped. The entry must be
  1264. * flushed before IO is initiated or a stale TLB entry potentially
  1265. * allows an update without redirtying the page.
  1266. */
  1267. bool writable;
  1268. };
  1269. struct task_struct {
  1270. #ifdef CONFIG_THREAD_INFO_IN_TASK
  1271. /*
  1272. * For reasons of header soup (see current_thread_info()), this
  1273. * must be the first element of task_struct.
  1274. */
  1275. struct thread_info thread_info;
  1276. #endif
  1277. volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
  1278. void *stack;
  1279. atomic_t usage;
  1280. unsigned int flags; /* per process flags, defined below */
  1281. unsigned int ptrace;
  1282. #ifdef CONFIG_SMP
  1283. struct llist_node wake_entry;
  1284. int on_cpu;
  1285. #ifdef CONFIG_THREAD_INFO_IN_TASK
  1286. unsigned int cpu; /* current CPU */
  1287. #endif
  1288. unsigned int wakee_flips;
  1289. unsigned long wakee_flip_decay_ts;
  1290. struct task_struct *last_wakee;
  1291. int wake_cpu;
  1292. #endif
  1293. int on_rq;
  1294. int prio, static_prio, normal_prio;
  1295. unsigned int rt_priority;
  1296. const struct sched_class *sched_class;
  1297. struct sched_entity se;
  1298. struct sched_rt_entity rt;
  1299. #ifdef CONFIG_CGROUP_SCHED
  1300. struct task_group *sched_task_group;
  1301. #endif
  1302. struct sched_dl_entity dl;
  1303. #ifdef CONFIG_PREEMPT_NOTIFIERS
  1304. /* list of struct preempt_notifier: */
  1305. struct hlist_head preempt_notifiers;
  1306. #endif
  1307. #ifdef CONFIG_BLK_DEV_IO_TRACE
  1308. unsigned int btrace_seq;
  1309. #endif
  1310. unsigned int policy;
  1311. int nr_cpus_allowed;
  1312. cpumask_t cpus_allowed;
  1313. #ifdef CONFIG_PREEMPT_RCU
  1314. int rcu_read_lock_nesting;
  1315. union rcu_special rcu_read_unlock_special;
  1316. struct list_head rcu_node_entry;
  1317. struct rcu_node *rcu_blocked_node;
  1318. #endif /* #ifdef CONFIG_PREEMPT_RCU */
  1319. #ifdef CONFIG_TASKS_RCU
  1320. unsigned long rcu_tasks_nvcsw;
  1321. bool rcu_tasks_holdout;
  1322. struct list_head rcu_tasks_holdout_list;
  1323. int rcu_tasks_idle_cpu;
  1324. #endif /* #ifdef CONFIG_TASKS_RCU */
  1325. #ifdef CONFIG_SCHED_INFO
  1326. struct sched_info sched_info;
  1327. #endif
  1328. struct list_head tasks;
  1329. #ifdef CONFIG_SMP
  1330. struct plist_node pushable_tasks;
  1331. struct rb_node pushable_dl_tasks;
  1332. #endif
  1333. struct mm_struct *mm, *active_mm;
  1334. /* per-thread vma caching */
  1335. u32 vmacache_seqnum;
  1336. struct vm_area_struct *vmacache[VMACACHE_SIZE];
  1337. #if defined(SPLIT_RSS_COUNTING)
  1338. struct task_rss_stat rss_stat;
  1339. #endif
  1340. /* task state */
  1341. int exit_state;
  1342. int exit_code, exit_signal;
  1343. int pdeath_signal; /* The signal sent when the parent dies */
  1344. unsigned long jobctl; /* JOBCTL_*, siglock protected */
  1345. /* Used for emulating ABI behavior of previous Linux versions */
  1346. unsigned int personality;
  1347. /* scheduler bits, serialized by scheduler locks */
  1348. unsigned sched_reset_on_fork:1;
  1349. unsigned sched_contributes_to_load:1;
  1350. unsigned sched_migrated:1;
  1351. unsigned sched_remote_wakeup:1;
  1352. unsigned :0; /* force alignment to the next boundary */
  1353. /* unserialized, strictly 'current' */
  1354. unsigned in_execve:1; /* bit to tell LSMs we're in execve */
  1355. unsigned in_iowait:1;
  1356. #if !defined(TIF_RESTORE_SIGMASK)
  1357. unsigned restore_sigmask:1;
  1358. #endif
  1359. #ifdef CONFIG_MEMCG
  1360. unsigned memcg_may_oom:1;
  1361. #ifndef CONFIG_SLOB
  1362. unsigned memcg_kmem_skip_account:1;
  1363. #endif
  1364. #endif
  1365. #ifdef CONFIG_COMPAT_BRK
  1366. unsigned brk_randomized:1;
  1367. #endif
  1368. unsigned long atomic_flags; /* Flags needing atomic access. */
  1369. struct restart_block restart_block;
  1370. pid_t pid;
  1371. pid_t tgid;
  1372. #ifdef CONFIG_CC_STACKPROTECTOR
  1373. /* Canary value for the -fstack-protector gcc feature */
  1374. unsigned long stack_canary;
  1375. #endif
  1376. /*
  1377. * pointers to (original) parent process, youngest child, younger sibling,
  1378. * older sibling, respectively. (p->father can be replaced with
  1379. * p->real_parent->pid)
  1380. */
  1381. struct task_struct __rcu *real_parent; /* real parent process */
  1382. struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
  1383. /*
  1384. * children/sibling forms the list of my natural children
  1385. */
  1386. struct list_head children; /* list of my children */
  1387. struct list_head sibling; /* linkage in my parent's children list */
  1388. struct task_struct *group_leader; /* threadgroup leader */
  1389. /*
  1390. * ptraced is the list of tasks this task is using ptrace on.
  1391. * This includes both natural children and PTRACE_ATTACH targets.
  1392. * p->ptrace_entry is p's link on the p->parent->ptraced list.
  1393. */
  1394. struct list_head ptraced;
  1395. struct list_head ptrace_entry;
  1396. /* PID/PID hash table linkage. */
  1397. struct pid_link pids[PIDTYPE_MAX];
  1398. struct list_head thread_group;
  1399. struct list_head thread_node;
  1400. struct completion *vfork_done; /* for vfork() */
  1401. int __user *set_child_tid; /* CLONE_CHILD_SETTID */
  1402. int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
  1403. u64 utime, stime;
  1404. #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
  1405. u64 utimescaled, stimescaled;
  1406. #endif
  1407. u64 gtime;
  1408. struct prev_cputime prev_cputime;
  1409. #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
  1410. seqcount_t vtime_seqcount;
  1411. unsigned long long vtime_snap;
  1412. enum {
  1413. /* Task is sleeping or running in a CPU with VTIME inactive */
  1414. VTIME_INACTIVE = 0,
  1415. /* Task runs in userspace in a CPU with VTIME active */
  1416. VTIME_USER,
  1417. /* Task runs in kernelspace in a CPU with VTIME active */
  1418. VTIME_SYS,
  1419. } vtime_snap_whence;
  1420. #endif
  1421. #ifdef CONFIG_NO_HZ_FULL
  1422. atomic_t tick_dep_mask;
  1423. #endif
  1424. unsigned long nvcsw, nivcsw; /* context switch counts */
  1425. u64 start_time; /* monotonic time in nsec */
  1426. u64 real_start_time; /* boot based time in nsec */
  1427. /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
  1428. unsigned long min_flt, maj_flt;
  1429. #ifdef CONFIG_POSIX_TIMERS
  1430. struct task_cputime cputime_expires;
  1431. struct list_head cpu_timers[3];
  1432. #endif
  1433. /* process credentials */
  1434. const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
  1435. const struct cred __rcu *real_cred; /* objective and real subjective task
  1436. * credentials (COW) */
  1437. const struct cred __rcu *cred; /* effective (overridable) subjective task
  1438. * credentials (COW) */
  1439. char comm[TASK_COMM_LEN]; /* executable name excluding path
  1440. - access with [gs]et_task_comm (which lock
  1441. it with task_lock())
  1442. - initialized normally by setup_new_exec */
  1443. /* file system info */
  1444. struct nameidata *nameidata;
  1445. #ifdef CONFIG_SYSVIPC
  1446. /* ipc stuff */
  1447. struct sysv_sem sysvsem;
  1448. struct sysv_shm sysvshm;
  1449. #endif
  1450. #ifdef CONFIG_DETECT_HUNG_TASK
  1451. /* hung task detection */
  1452. unsigned long last_switch_count;
  1453. #endif
  1454. /* filesystem information */
  1455. struct fs_struct *fs;
  1456. /* open file information */
  1457. struct files_struct *files;
  1458. /* namespaces */
  1459. struct nsproxy *nsproxy;
  1460. /* signal handlers */
  1461. struct signal_struct *signal;
  1462. struct sighand_struct *sighand;
  1463. sigset_t blocked, real_blocked;
  1464. sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
  1465. struct sigpending pending;
  1466. unsigned long sas_ss_sp;
  1467. size_t sas_ss_size;
  1468. unsigned sas_ss_flags;
  1469. struct callback_head *task_works;
  1470. struct audit_context *audit_context;
  1471. #ifdef CONFIG_AUDITSYSCALL
  1472. kuid_t loginuid;
  1473. unsigned int sessionid;
  1474. #endif
  1475. struct seccomp seccomp;
  1476. /* Thread group tracking */
  1477. u32 parent_exec_id;
  1478. u32 self_exec_id;
  1479. /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
  1480. * mempolicy */
  1481. spinlock_t alloc_lock;
  1482. /* Protection of the PI data structures: */
  1483. raw_spinlock_t pi_lock;
  1484. struct wake_q_node wake_q;
  1485. #ifdef CONFIG_RT_MUTEXES
  1486. /* PI waiters blocked on a rt_mutex held by this task */
  1487. struct rb_root pi_waiters;
  1488. struct rb_node *pi_waiters_leftmost;
  1489. /* Deadlock detection and priority inheritance handling */
  1490. struct rt_mutex_waiter *pi_blocked_on;
  1491. #endif
  1492. #ifdef CONFIG_DEBUG_MUTEXES
  1493. /* mutex deadlock detection */
  1494. struct mutex_waiter *blocked_on;
  1495. #endif
  1496. #ifdef CONFIG_TRACE_IRQFLAGS
  1497. unsigned int irq_events;
  1498. unsigned long hardirq_enable_ip;
  1499. unsigned long hardirq_disable_ip;
  1500. unsigned int hardirq_enable_event;
  1501. unsigned int hardirq_disable_event;
  1502. int hardirqs_enabled;
  1503. int hardirq_context;
  1504. unsigned long softirq_disable_ip;
  1505. unsigned long softirq_enable_ip;
  1506. unsigned int softirq_disable_event;
  1507. unsigned int softirq_enable_event;
  1508. int softirqs_enabled;
  1509. int softirq_context;
  1510. #endif
  1511. #ifdef CONFIG_LOCKDEP
  1512. # define MAX_LOCK_DEPTH 48UL
  1513. u64 curr_chain_key;
  1514. int lockdep_depth;
  1515. unsigned int lockdep_recursion;
  1516. struct held_lock held_locks[MAX_LOCK_DEPTH];
  1517. gfp_t lockdep_reclaim_gfp;
  1518. #endif
  1519. #ifdef CONFIG_UBSAN
  1520. unsigned int in_ubsan;
  1521. #endif
  1522. /* journalling filesystem info */
  1523. void *journal_info;
  1524. /* stacked block device info */
  1525. struct bio_list *bio_list;
  1526. #ifdef CONFIG_BLOCK
  1527. /* stack plugging */
  1528. struct blk_plug *plug;
  1529. #endif
  1530. /* VM state */
  1531. struct reclaim_state *reclaim_state;
  1532. struct backing_dev_info *backing_dev_info;
  1533. struct io_context *io_context;
  1534. unsigned long ptrace_message;
  1535. siginfo_t *last_siginfo; /* For ptrace use. */
  1536. struct task_io_accounting ioac;
  1537. #if defined(CONFIG_TASK_XACCT)
  1538. u64 acct_rss_mem1; /* accumulated rss usage */
  1539. u64 acct_vm_mem1; /* accumulated virtual memory usage */
  1540. u64 acct_timexpd; /* stime + utime since last update */
  1541. #endif
  1542. #ifdef CONFIG_CPUSETS
  1543. nodemask_t mems_allowed; /* Protected by alloc_lock */
  1544. seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
  1545. int cpuset_mem_spread_rotor;
  1546. int cpuset_slab_spread_rotor;
  1547. #endif
  1548. #ifdef CONFIG_CGROUPS
  1549. /* Control Group info protected by css_set_lock */
  1550. struct css_set __rcu *cgroups;
  1551. /* cg_list protected by css_set_lock and tsk->alloc_lock */
  1552. struct list_head cg_list;
  1553. #endif
  1554. #ifdef CONFIG_INTEL_RDT_A
  1555. int closid;
  1556. #endif
  1557. #ifdef CONFIG_FUTEX
  1558. struct robust_list_head __user *robust_list;
  1559. #ifdef CONFIG_COMPAT
  1560. struct compat_robust_list_head __user *compat_robust_list;
  1561. #endif
  1562. struct list_head pi_state_list;
  1563. struct futex_pi_state *pi_state_cache;
  1564. #endif
  1565. #ifdef CONFIG_PERF_EVENTS
  1566. struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
  1567. struct mutex perf_event_mutex;
  1568. struct list_head perf_event_list;
  1569. #endif
  1570. #ifdef CONFIG_DEBUG_PREEMPT
  1571. unsigned long preempt_disable_ip;
  1572. #endif
  1573. #ifdef CONFIG_NUMA
  1574. struct mempolicy *mempolicy; /* Protected by alloc_lock */
  1575. short il_next;
  1576. short pref_node_fork;
  1577. #endif
  1578. #ifdef CONFIG_NUMA_BALANCING
  1579. int numa_scan_seq;
  1580. unsigned int numa_scan_period;
  1581. unsigned int numa_scan_period_max;
  1582. int numa_preferred_nid;
  1583. unsigned long numa_migrate_retry;
  1584. u64 node_stamp; /* migration stamp */
  1585. u64 last_task_numa_placement;
  1586. u64 last_sum_exec_runtime;
  1587. struct callback_head numa_work;
  1588. struct list_head numa_entry;
  1589. struct numa_group *numa_group;
  1590. /*
  1591. * numa_faults is an array split into four regions:
  1592. * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
  1593. * in this precise order.
  1594. *
  1595. * faults_memory: Exponential decaying average of faults on a per-node
  1596. * basis. Scheduling placement decisions are made based on these
  1597. * counts. The values remain static for the duration of a PTE scan.
  1598. * faults_cpu: Track the nodes the process was running on when a NUMA
  1599. * hinting fault was incurred.
  1600. * faults_memory_buffer and faults_cpu_buffer: Record faults per node
  1601. * during the current scan window. When the scan completes, the counts
  1602. * in faults_memory and faults_cpu decay and these values are copied.
  1603. */
  1604. unsigned long *numa_faults;
  1605. unsigned long total_numa_faults;
  1606. /*
  1607. * numa_faults_locality tracks if faults recorded during the last
  1608. * scan window were remote/local or failed to migrate. The task scan
  1609. * period is adapted based on the locality of the faults with different
  1610. * weights depending on whether they were shared or private faults
  1611. */
  1612. unsigned long numa_faults_locality[3];
  1613. unsigned long numa_pages_migrated;
  1614. #endif /* CONFIG_NUMA_BALANCING */
  1615. #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
  1616. struct tlbflush_unmap_batch tlb_ubc;
  1617. #endif
  1618. struct rcu_head rcu;
  1619. /*
  1620. * cache last used pipe for splice
  1621. */
  1622. struct pipe_inode_info *splice_pipe;
  1623. struct page_frag task_frag;
  1624. #ifdef CONFIG_TASK_DELAY_ACCT
  1625. struct task_delay_info *delays;
  1626. #endif
  1627. #ifdef CONFIG_FAULT_INJECTION
  1628. int make_it_fail;
  1629. #endif
  1630. /*
  1631. * when (nr_dirtied >= nr_dirtied_pause), it's time to call
  1632. * balance_dirty_pages() for some dirty throttling pause
  1633. */
  1634. int nr_dirtied;
  1635. int nr_dirtied_pause;
  1636. unsigned long dirty_paused_when; /* start of a write-and-pause period */
  1637. #ifdef CONFIG_LATENCYTOP
  1638. int latency_record_count;
  1639. struct latency_record latency_record[LT_SAVECOUNT];
  1640. #endif
  1641. /*
  1642. * time slack values; these are used to round up poll() and
  1643. * select() etc timeout values. These are in nanoseconds.
  1644. */
  1645. u64 timer_slack_ns;
  1646. u64 default_timer_slack_ns;
  1647. #ifdef CONFIG_KASAN
  1648. unsigned int kasan_depth;
  1649. #endif
  1650. #ifdef CONFIG_FUNCTION_GRAPH_TRACER
  1651. /* Index of current stored address in ret_stack */
  1652. int curr_ret_stack;
  1653. /* Stack of return addresses for return function tracing */
  1654. struct ftrace_ret_stack *ret_stack;
  1655. /* time stamp for last schedule */
  1656. unsigned long long ftrace_timestamp;
  1657. /*
  1658. * Number of functions that haven't been traced
  1659. * because of depth overrun.
  1660. */
  1661. atomic_t trace_overrun;
  1662. /* Pause for the tracing */
  1663. atomic_t tracing_graph_pause;
  1664. #endif
  1665. #ifdef CONFIG_TRACING
  1666. /* state flags for use by tracers */
  1667. unsigned long trace;
  1668. /* bitmask and counter of trace recursion */
  1669. unsigned long trace_recursion;
  1670. #endif /* CONFIG_TRACING */
  1671. #ifdef CONFIG_KCOV
  1672. /* Coverage collection mode enabled for this task (0 if disabled). */
  1673. enum kcov_mode kcov_mode;
  1674. /* Size of the kcov_area. */
  1675. unsigned kcov_size;
  1676. /* Buffer for coverage collection. */
  1677. void *kcov_area;
  1678. /* kcov desciptor wired with this task or NULL. */
  1679. struct kcov *kcov;
  1680. #endif
  1681. #ifdef CONFIG_MEMCG
  1682. struct mem_cgroup *memcg_in_oom;
  1683. gfp_t memcg_oom_gfp_mask;
  1684. int memcg_oom_order;
  1685. /* number of pages to reclaim on returning to userland */
  1686. unsigned int memcg_nr_pages_over_high;
  1687. #endif
  1688. #ifdef CONFIG_UPROBES
  1689. struct uprobe_task *utask;
  1690. #endif
  1691. #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
  1692. unsigned int sequential_io;
  1693. unsigned int sequential_io_avg;
  1694. #endif
  1695. #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
  1696. unsigned long task_state_change;
  1697. #endif
  1698. int pagefault_disabled;
  1699. #ifdef CONFIG_MMU
  1700. struct task_struct *oom_reaper_list;
  1701. #endif
  1702. #ifdef CONFIG_VMAP_STACK
  1703. struct vm_struct *stack_vm_area;
  1704. #endif
  1705. #ifdef CONFIG_THREAD_INFO_IN_TASK
  1706. /* A live task holds one reference. */
  1707. atomic_t stack_refcount;
  1708. #endif
  1709. /* CPU-specific state of this task */
  1710. struct thread_struct thread;
  1711. /*
  1712. * WARNING: on x86, 'thread_struct' contains a variable-sized
  1713. * structure. It *MUST* be at the end of 'task_struct'.
  1714. *
  1715. * Do not put anything below here!
  1716. */
  1717. };
  1718. #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
  1719. extern int arch_task_struct_size __read_mostly;
  1720. #else
  1721. # define arch_task_struct_size (sizeof(struct task_struct))
  1722. #endif
  1723. #ifdef CONFIG_VMAP_STACK
  1724. static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
  1725. {
  1726. return t->stack_vm_area;
  1727. }
  1728. #else
  1729. static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
  1730. {
  1731. return NULL;
  1732. }
  1733. #endif
  1734. /* Future-safe accessor for struct task_struct's cpus_allowed. */
  1735. #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
  1736. static inline int tsk_nr_cpus_allowed(struct task_struct *p)
  1737. {
  1738. return p->nr_cpus_allowed;
  1739. }
  1740. #define TNF_MIGRATED 0x01
  1741. #define TNF_NO_GROUP 0x02
  1742. #define TNF_SHARED 0x04
  1743. #define TNF_FAULT_LOCAL 0x08
  1744. #define TNF_MIGRATE_FAIL 0x10
  1745. static inline bool in_vfork(struct task_struct *tsk)
  1746. {
  1747. bool ret;
  1748. /*
  1749. * need RCU to access ->real_parent if CLONE_VM was used along with
  1750. * CLONE_PARENT.
  1751. *
  1752. * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
  1753. * imply CLONE_VM
  1754. *
  1755. * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
  1756. * ->real_parent is not necessarily the task doing vfork(), so in
  1757. * theory we can't rely on task_lock() if we want to dereference it.
  1758. *
  1759. * And in this case we can't trust the real_parent->mm == tsk->mm
  1760. * check, it can be false negative. But we do not care, if init or
  1761. * another oom-unkillable task does this it should blame itself.
  1762. */
  1763. rcu_read_lock();
  1764. ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm;
  1765. rcu_read_unlock();
  1766. return ret;
  1767. }
  1768. #ifdef CONFIG_NUMA_BALANCING
  1769. extern void task_numa_fault(int last_node, int node, int pages, int flags);
  1770. extern pid_t task_numa_group_id(struct task_struct *p);
  1771. extern void set_numabalancing_state(bool enabled);
  1772. extern void task_numa_free(struct task_struct *p);
  1773. extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
  1774. int src_nid, int dst_cpu);
  1775. #else
  1776. static inline void task_numa_fault(int last_node, int node, int pages,
  1777. int flags)
  1778. {
  1779. }
  1780. static inline pid_t task_numa_group_id(struct task_struct *p)
  1781. {
  1782. return 0;
  1783. }
  1784. static inline void set_numabalancing_state(bool enabled)
  1785. {
  1786. }
  1787. static inline void task_numa_free(struct task_struct *p)
  1788. {
  1789. }
  1790. static inline bool should_numa_migrate_memory(struct task_struct *p,
  1791. struct page *page, int src_nid, int dst_cpu)
  1792. {
  1793. return true;
  1794. }
  1795. #endif
  1796. static inline struct pid *task_pid(struct task_struct *task)
  1797. {
  1798. return task->pids[PIDTYPE_PID].pid;
  1799. }
  1800. static inline struct pid *task_tgid(struct task_struct *task)
  1801. {
  1802. return task->group_leader->pids[PIDTYPE_PID].pid;
  1803. }
  1804. /*
  1805. * Without tasklist or rcu lock it is not safe to dereference
  1806. * the result of task_pgrp/task_session even if task == current,
  1807. * we can race with another thread doing sys_setsid/sys_setpgid.
  1808. */
  1809. static inline struct pid *task_pgrp(struct task_struct *task)
  1810. {
  1811. return task->group_leader->pids[PIDTYPE_PGID].pid;
  1812. }
  1813. static inline struct pid *task_session(struct task_struct *task)
  1814. {
  1815. return task->group_leader->pids[PIDTYPE_SID].pid;
  1816. }
  1817. struct pid_namespace;
  1818. /*
  1819. * the helpers to get the task's different pids as they are seen
  1820. * from various namespaces
  1821. *
  1822. * task_xid_nr() : global id, i.e. the id seen from the init namespace;
  1823. * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
  1824. * current.
  1825. * task_xid_nr_ns() : id seen from the ns specified;
  1826. *
  1827. * set_task_vxid() : assigns a virtual id to a task;
  1828. *
  1829. * see also pid_nr() etc in include/linux/pid.h
  1830. */
  1831. pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
  1832. struct pid_namespace *ns);
  1833. static inline pid_t task_pid_nr(struct task_struct *tsk)
  1834. {
  1835. return tsk->pid;
  1836. }
  1837. static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
  1838. struct pid_namespace *ns)
  1839. {
  1840. return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
  1841. }
  1842. static inline pid_t task_pid_vnr(struct task_struct *tsk)
  1843. {
  1844. return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
  1845. }
  1846. static inline pid_t task_tgid_nr(struct task_struct *tsk)
  1847. {
  1848. return tsk->tgid;
  1849. }
  1850. pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
  1851. static inline pid_t task_tgid_vnr(struct task_struct *tsk)
  1852. {
  1853. return pid_vnr(task_tgid(tsk));
  1854. }
  1855. static inline int pid_alive(const struct task_struct *p);
  1856. static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
  1857. {
  1858. pid_t pid = 0;
  1859. rcu_read_lock();
  1860. if (pid_alive(tsk))
  1861. pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
  1862. rcu_read_unlock();
  1863. return pid;
  1864. }
  1865. static inline pid_t task_ppid_nr(const struct task_struct *tsk)
  1866. {
  1867. return task_ppid_nr_ns(tsk, &init_pid_ns);
  1868. }
  1869. static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
  1870. struct pid_namespace *ns)
  1871. {
  1872. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
  1873. }
  1874. static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
  1875. {
  1876. return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
  1877. }
  1878. static inline pid_t task_session_nr_ns(struct task_struct *tsk,
  1879. struct pid_namespace *ns)
  1880. {
  1881. return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
  1882. }
  1883. static inline pid_t task_session_vnr(struct task_struct *tsk)
  1884. {
  1885. return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
  1886. }
  1887. /* obsolete, do not use */
  1888. static inline pid_t task_pgrp_nr(struct task_struct *tsk)
  1889. {
  1890. return task_pgrp_nr_ns(tsk, &init_pid_ns);
  1891. }
  1892. /**
  1893. * pid_alive - check that a task structure is not stale
  1894. * @p: Task structure to be checked.
  1895. *
  1896. * Test if a process is not yet dead (at most zombie state)
  1897. * If pid_alive fails, then pointers within the task structure
  1898. * can be stale and must not be dereferenced.
  1899. *
  1900. * Return: 1 if the process is alive. 0 otherwise.
  1901. */
  1902. static inline int pid_alive(const struct task_struct *p)
  1903. {
  1904. return p->pids[PIDTYPE_PID].pid != NULL;
  1905. }
  1906. /**
  1907. * is_global_init - check if a task structure is init. Since init
  1908. * is free to have sub-threads we need to check tgid.
  1909. * @tsk: Task structure to be checked.
  1910. *
  1911. * Check if a task structure is the first user space task the kernel created.
  1912. *
  1913. * Return: 1 if the task structure is init. 0 otherwise.
  1914. */
  1915. static inline int is_global_init(struct task_struct *tsk)
  1916. {
  1917. return task_tgid_nr(tsk) == 1;
  1918. }
  1919. extern struct pid *cad_pid;
  1920. extern void free_task(struct task_struct *tsk);
  1921. #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
  1922. extern void __put_task_struct(struct task_struct *t);
  1923. static inline void put_task_struct(struct task_struct *t)
  1924. {
  1925. if (atomic_dec_and_test(&t->usage))
  1926. __put_task_struct(t);
  1927. }
  1928. struct task_struct *task_rcu_dereference(struct task_struct **ptask);
  1929. struct task_struct *try_get_task_struct(struct task_struct **ptask);
  1930. #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
  1931. extern void task_cputime(struct task_struct *t,
  1932. u64 *utime, u64 *stime);
  1933. extern u64 task_gtime(struct task_struct *t);
  1934. #else
  1935. static inline void task_cputime(struct task_struct *t,
  1936. u64 *utime, u64 *stime)
  1937. {
  1938. *utime = t->utime;
  1939. *stime = t->stime;
  1940. }
  1941. static inline u64 task_gtime(struct task_struct *t)
  1942. {
  1943. return t->gtime;
  1944. }
  1945. #endif
  1946. #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
  1947. static inline void task_cputime_scaled(struct task_struct *t,
  1948. u64 *utimescaled,
  1949. u64 *stimescaled)
  1950. {
  1951. *utimescaled = t->utimescaled;
  1952. *stimescaled = t->stimescaled;
  1953. }
  1954. #else
  1955. static inline void task_cputime_scaled(struct task_struct *t,
  1956. u64 *utimescaled,
  1957. u64 *stimescaled)
  1958. {
  1959. task_cputime(t, utimescaled, stimescaled);
  1960. }
  1961. #endif
  1962. extern void task_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st);
  1963. extern void thread_group_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st);
  1964. /*
  1965. * Per process flags
  1966. */
  1967. #define PF_IDLE 0x00000002 /* I am an IDLE thread */
  1968. #define PF_EXITING 0x00000004 /* getting shut down */
  1969. #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
  1970. #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
  1971. #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
  1972. #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
  1973. #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
  1974. #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
  1975. #define PF_DUMPCORE 0x00000200 /* dumped core */
  1976. #define PF_SIGNALED 0x00000400 /* killed by a signal */
  1977. #define PF_MEMALLOC 0x00000800 /* Allocating memory */
  1978. #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
  1979. #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
  1980. #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
  1981. #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
  1982. #define PF_FROZEN 0x00010000 /* frozen for system suspend */
  1983. #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
  1984. #define PF_KSWAPD 0x00040000 /* I am kswapd */
  1985. #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
  1986. #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
  1987. #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
  1988. #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
  1989. #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
  1990. #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
  1991. #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
  1992. #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
  1993. #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
  1994. #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
  1995. /*
  1996. * Only the _current_ task can read/write to tsk->flags, but other
  1997. * tasks can access tsk->flags in readonly mode for example
  1998. * with tsk_used_math (like during threaded core dumping).
  1999. * There is however an exception to this rule during ptrace
  2000. * or during fork: the ptracer task is allowed to write to the
  2001. * child->flags of its traced child (same goes for fork, the parent
  2002. * can write to the child->flags), because we're guaranteed the
  2003. * child is not running and in turn not changing child->flags
  2004. * at the same time the parent does it.
  2005. */
  2006. #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
  2007. #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
  2008. #define clear_used_math() clear_stopped_child_used_math(current)
  2009. #define set_used_math() set_stopped_child_used_math(current)
  2010. #define conditional_stopped_child_used_math(condition, child) \
  2011. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
  2012. #define conditional_used_math(condition) \
  2013. conditional_stopped_child_used_math(condition, current)
  2014. #define copy_to_stopped_child_used_math(child) \
  2015. do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
  2016. /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
  2017. #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
  2018. #define used_math() tsk_used_math(current)
  2019. /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
  2020. * __GFP_FS is also cleared as it implies __GFP_IO.
  2021. */
  2022. static inline gfp_t memalloc_noio_flags(gfp_t flags)
  2023. {
  2024. if (unlikely(current->flags & PF_MEMALLOC_NOIO))
  2025. flags &= ~(__GFP_IO | __GFP_FS);
  2026. return flags;
  2027. }
  2028. static inline unsigned int memalloc_noio_save(void)
  2029. {
  2030. unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
  2031. current->flags |= PF_MEMALLOC_NOIO;
  2032. return flags;
  2033. }
  2034. static inline void memalloc_noio_restore(unsigned int flags)
  2035. {
  2036. current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
  2037. }
  2038. /* Per-process atomic flags. */
  2039. #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
  2040. #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
  2041. #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
  2042. #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
  2043. #define TASK_PFA_TEST(name, func) \
  2044. static inline bool task_##func(struct task_struct *p) \
  2045. { return test_bit(PFA_##name, &p->atomic_flags); }
  2046. #define TASK_PFA_SET(name, func) \
  2047. static inline void task_set_##func(struct task_struct *p) \
  2048. { set_bit(PFA_##name, &p->atomic_flags); }
  2049. #define TASK_PFA_CLEAR(name, func) \
  2050. static inline void task_clear_##func(struct task_struct *p) \
  2051. { clear_bit(PFA_##name, &p->atomic_flags); }
  2052. TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
  2053. TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
  2054. TASK_PFA_TEST(SPREAD_PAGE, spread_page)
  2055. TASK_PFA_SET(SPREAD_PAGE, spread_page)
  2056. TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
  2057. TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
  2058. TASK_PFA_SET(SPREAD_SLAB, spread_slab)
  2059. TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
  2060. TASK_PFA_TEST(LMK_WAITING, lmk_waiting)
  2061. TASK_PFA_SET(LMK_WAITING, lmk_waiting)
  2062. /*
  2063. * task->jobctl flags
  2064. */
  2065. #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
  2066. #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
  2067. #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
  2068. #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
  2069. #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
  2070. #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
  2071. #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
  2072. #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
  2073. #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
  2074. #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
  2075. #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
  2076. #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
  2077. #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
  2078. #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
  2079. #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
  2080. #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
  2081. #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
  2082. extern bool task_set_jobctl_pending(struct task_struct *task,
  2083. unsigned long mask);
  2084. extern void task_clear_jobctl_trapping(struct task_struct *task);
  2085. extern void task_clear_jobctl_pending(struct task_struct *task,
  2086. unsigned long mask);
  2087. static inline void rcu_copy_process(struct task_struct *p)
  2088. {
  2089. #ifdef CONFIG_PREEMPT_RCU
  2090. p->rcu_read_lock_nesting = 0;
  2091. p->rcu_read_unlock_special.s = 0;
  2092. p->rcu_blocked_node = NULL;
  2093. INIT_LIST_HEAD(&p->rcu_node_entry);
  2094. #endif /* #ifdef CONFIG_PREEMPT_RCU */
  2095. #ifdef CONFIG_TASKS_RCU
  2096. p->rcu_tasks_holdout = false;
  2097. INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
  2098. p->rcu_tasks_idle_cpu = -1;
  2099. #endif /* #ifdef CONFIG_TASKS_RCU */
  2100. }
  2101. static inline void tsk_restore_flags(struct task_struct *task,
  2102. unsigned long orig_flags, unsigned long flags)
  2103. {
  2104. task->flags &= ~flags;
  2105. task->flags |= orig_flags & flags;
  2106. }
  2107. extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
  2108. const struct cpumask *trial);
  2109. extern int task_can_attach(struct task_struct *p,
  2110. const struct cpumask *cs_cpus_allowed);
  2111. #ifdef CONFIG_SMP
  2112. extern void do_set_cpus_allowed(struct task_struct *p,
  2113. const struct cpumask *new_mask);
  2114. extern int set_cpus_allowed_ptr(struct task_struct *p,
  2115. const struct cpumask *new_mask);
  2116. #else
  2117. static inline void do_set_cpus_allowed(struct task_struct *p,
  2118. const struct cpumask *new_mask)
  2119. {
  2120. }
  2121. static inline int set_cpus_allowed_ptr(struct task_struct *p,
  2122. const struct cpumask *new_mask)
  2123. {
  2124. if (!cpumask_test_cpu(0, new_mask))
  2125. return -EINVAL;
  2126. return 0;
  2127. }
  2128. #endif
  2129. #ifdef CONFIG_NO_HZ_COMMON
  2130. void calc_load_enter_idle(void);
  2131. void calc_load_exit_idle(void);
  2132. #else
  2133. static inline void calc_load_enter_idle(void) { }
  2134. static inline void calc_load_exit_idle(void) { }
  2135. #endif /* CONFIG_NO_HZ_COMMON */
  2136. #ifndef cpu_relax_yield
  2137. #define cpu_relax_yield() cpu_relax()
  2138. #endif
  2139. /*
  2140. * Do not use outside of architecture code which knows its limitations.
  2141. *
  2142. * sched_clock() has no promise of monotonicity or bounded drift between
  2143. * CPUs, use (which you should not) requires disabling IRQs.
  2144. *
  2145. * Please use one of the three interfaces below.
  2146. */
  2147. extern unsigned long long notrace sched_clock(void);
  2148. /*
  2149. * See the comment in kernel/sched/clock.c
  2150. */
  2151. extern u64 running_clock(void);
  2152. extern u64 sched_clock_cpu(int cpu);
  2153. extern void sched_clock_init(void);
  2154. #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
  2155. static inline void sched_clock_init_late(void)
  2156. {
  2157. }
  2158. static inline void sched_clock_tick(void)
  2159. {
  2160. }
  2161. static inline void clear_sched_clock_stable(void)
  2162. {
  2163. }
  2164. static inline void sched_clock_idle_sleep_event(void)
  2165. {
  2166. }
  2167. static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
  2168. {
  2169. }
  2170. static inline u64 cpu_clock(int cpu)
  2171. {
  2172. return sched_clock();
  2173. }
  2174. static inline u64 local_clock(void)
  2175. {
  2176. return sched_clock();
  2177. }
  2178. #else
  2179. extern void sched_clock_init_late(void);
  2180. /*
  2181. * Architectures can set this to 1 if they have specified
  2182. * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
  2183. * but then during bootup it turns out that sched_clock()
  2184. * is reliable after all:
  2185. */
  2186. extern int sched_clock_stable(void);
  2187. extern void clear_sched_clock_stable(void);
  2188. extern void sched_clock_tick(void);
  2189. extern void sched_clock_idle_sleep_event(void);
  2190. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  2191. /*
  2192. * As outlined in clock.c, provides a fast, high resolution, nanosecond
  2193. * time source that is monotonic per cpu argument and has bounded drift
  2194. * between cpus.
  2195. *
  2196. * ######################### BIG FAT WARNING ##########################
  2197. * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
  2198. * # go backwards !! #
  2199. * ####################################################################
  2200. */
  2201. static inline u64 cpu_clock(int cpu)
  2202. {
  2203. return sched_clock_cpu(cpu);
  2204. }
  2205. static inline u64 local_clock(void)
  2206. {
  2207. return sched_clock_cpu(raw_smp_processor_id());
  2208. }
  2209. #endif
  2210. #ifdef CONFIG_IRQ_TIME_ACCOUNTING
  2211. /*
  2212. * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
  2213. * The reason for this explicit opt-in is not to have perf penalty with
  2214. * slow sched_clocks.
  2215. */
  2216. extern void enable_sched_clock_irqtime(void);
  2217. extern void disable_sched_clock_irqtime(void);
  2218. #else
  2219. static inline void enable_sched_clock_irqtime(void) {}
  2220. static inline void disable_sched_clock_irqtime(void) {}
  2221. #endif
  2222. extern unsigned long long
  2223. task_sched_runtime(struct task_struct *task);
  2224. /* sched_exec is called by processes performing an exec */
  2225. #ifdef CONFIG_SMP
  2226. extern void sched_exec(void);
  2227. #else
  2228. #define sched_exec() {}
  2229. #endif
  2230. extern void sched_clock_idle_sleep_event(void);
  2231. extern void sched_clock_idle_wakeup_event(u64 delta_ns);
  2232. #ifdef CONFIG_HOTPLUG_CPU
  2233. extern void idle_task_exit(void);
  2234. #else
  2235. static inline void idle_task_exit(void) {}
  2236. #endif
  2237. #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
  2238. extern void wake_up_nohz_cpu(int cpu);
  2239. #else
  2240. static inline void wake_up_nohz_cpu(int cpu) { }
  2241. #endif
  2242. #ifdef CONFIG_NO_HZ_FULL
  2243. extern u64 scheduler_tick_max_deferment(void);
  2244. #endif
  2245. #ifdef CONFIG_SCHED_AUTOGROUP
  2246. extern void sched_autogroup_create_attach(struct task_struct *p);
  2247. extern void sched_autogroup_detach(struct task_struct *p);
  2248. extern void sched_autogroup_fork(struct signal_struct *sig);
  2249. extern void sched_autogroup_exit(struct signal_struct *sig);
  2250. extern void sched_autogroup_exit_task(struct task_struct *p);
  2251. #ifdef CONFIG_PROC_FS
  2252. extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
  2253. extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
  2254. #endif
  2255. #else
  2256. static inline void sched_autogroup_create_attach(struct task_struct *p) { }
  2257. static inline void sched_autogroup_detach(struct task_struct *p) { }
  2258. static inline void sched_autogroup_fork(struct signal_struct *sig) { }
  2259. static inline void sched_autogroup_exit(struct signal_struct *sig) { }
  2260. static inline void sched_autogroup_exit_task(struct task_struct *p) { }
  2261. #endif
  2262. extern int yield_to(struct task_struct *p, bool preempt);
  2263. extern void set_user_nice(struct task_struct *p, long nice);
  2264. extern int task_prio(const struct task_struct *p);
  2265. /**
  2266. * task_nice - return the nice value of a given task.
  2267. * @p: the task in question.
  2268. *
  2269. * Return: The nice value [ -20 ... 0 ... 19 ].
  2270. */
  2271. static inline int task_nice(const struct task_struct *p)
  2272. {
  2273. return PRIO_TO_NICE((p)->static_prio);
  2274. }
  2275. extern int can_nice(const struct task_struct *p, const int nice);
  2276. extern int task_curr(const struct task_struct *p);
  2277. extern int idle_cpu(int cpu);
  2278. extern int sched_setscheduler(struct task_struct *, int,
  2279. const struct sched_param *);
  2280. extern int sched_setscheduler_nocheck(struct task_struct *, int,
  2281. const struct sched_param *);
  2282. extern int sched_setattr(struct task_struct *,
  2283. const struct sched_attr *);
  2284. extern struct task_struct *idle_task(int cpu);
  2285. /**
  2286. * is_idle_task - is the specified task an idle task?
  2287. * @p: the task in question.
  2288. *
  2289. * Return: 1 if @p is an idle task. 0 otherwise.
  2290. */
  2291. static inline bool is_idle_task(const struct task_struct *p)
  2292. {
  2293. return !!(p->flags & PF_IDLE);
  2294. }
  2295. extern struct task_struct *curr_task(int cpu);
  2296. extern void ia64_set_curr_task(int cpu, struct task_struct *p);
  2297. void yield(void);
  2298. union thread_union {
  2299. #ifndef CONFIG_THREAD_INFO_IN_TASK
  2300. struct thread_info thread_info;
  2301. #endif
  2302. unsigned long stack[THREAD_SIZE/sizeof(long)];
  2303. };
  2304. #ifndef __HAVE_ARCH_KSTACK_END
  2305. static inline int kstack_end(void *addr)
  2306. {
  2307. /* Reliable end of stack detection:
  2308. * Some APM bios versions misalign the stack
  2309. */
  2310. return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
  2311. }
  2312. #endif
  2313. extern union thread_union init_thread_union;
  2314. extern struct task_struct init_task;
  2315. extern struct mm_struct init_mm;
  2316. extern struct pid_namespace init_pid_ns;
  2317. /*
  2318. * find a task by one of its numerical ids
  2319. *
  2320. * find_task_by_pid_ns():
  2321. * finds a task by its pid in the specified namespace
  2322. * find_task_by_vpid():
  2323. * finds a task by its virtual pid
  2324. *
  2325. * see also find_vpid() etc in include/linux/pid.h
  2326. */
  2327. extern struct task_struct *find_task_by_vpid(pid_t nr);
  2328. extern struct task_struct *find_task_by_pid_ns(pid_t nr,
  2329. struct pid_namespace *ns);
  2330. /* per-UID process charging. */
  2331. extern struct user_struct * alloc_uid(kuid_t);
  2332. static inline struct user_struct *get_uid(struct user_struct *u)
  2333. {
  2334. atomic_inc(&u->__count);
  2335. return u;
  2336. }
  2337. extern void free_uid(struct user_struct *);
  2338. #include <asm/current.h>
  2339. extern void xtime_update(unsigned long ticks);
  2340. extern int wake_up_state(struct task_struct *tsk, unsigned int state);
  2341. extern int wake_up_process(struct task_struct *tsk);
  2342. extern void wake_up_new_task(struct task_struct *tsk);
  2343. #ifdef CONFIG_SMP
  2344. extern void kick_process(struct task_struct *tsk);
  2345. #else
  2346. static inline void kick_process(struct task_struct *tsk) { }
  2347. #endif
  2348. extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
  2349. extern void sched_dead(struct task_struct *p);
  2350. extern void proc_caches_init(void);
  2351. extern void flush_signals(struct task_struct *);
  2352. extern void ignore_signals(struct task_struct *);
  2353. extern void flush_signal_handlers(struct task_struct *, int force_default);
  2354. extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
  2355. static inline int kernel_dequeue_signal(siginfo_t *info)
  2356. {
  2357. struct task_struct *tsk = current;
  2358. siginfo_t __info;
  2359. int ret;
  2360. spin_lock_irq(&tsk->sighand->siglock);
  2361. ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
  2362. spin_unlock_irq(&tsk->sighand->siglock);
  2363. return ret;
  2364. }
  2365. static inline void kernel_signal_stop(void)
  2366. {
  2367. spin_lock_irq(&current->sighand->siglock);
  2368. if (current->jobctl & JOBCTL_STOP_DEQUEUED)
  2369. __set_current_state(TASK_STOPPED);
  2370. spin_unlock_irq(&current->sighand->siglock);
  2371. schedule();
  2372. }
  2373. extern void release_task(struct task_struct * p);
  2374. extern int send_sig_info(int, struct siginfo *, struct task_struct *);
  2375. extern int force_sigsegv(int, struct task_struct *);
  2376. extern int force_sig_info(int, struct siginfo *, struct task_struct *);
  2377. extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
  2378. extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
  2379. extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
  2380. const struct cred *, u32);
  2381. extern int kill_pgrp(struct pid *pid, int sig, int priv);
  2382. extern int kill_pid(struct pid *pid, int sig, int priv);
  2383. extern int kill_proc_info(int, struct siginfo *, pid_t);
  2384. extern __must_check bool do_notify_parent(struct task_struct *, int);
  2385. extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
  2386. extern void force_sig(int, struct task_struct *);
  2387. extern int send_sig(int, struct task_struct *, int);
  2388. extern int zap_other_threads(struct task_struct *p);
  2389. extern struct sigqueue *sigqueue_alloc(void);
  2390. extern void sigqueue_free(struct sigqueue *);
  2391. extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
  2392. extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
  2393. #ifdef TIF_RESTORE_SIGMASK
  2394. /*
  2395. * Legacy restore_sigmask accessors. These are inefficient on
  2396. * SMP architectures because they require atomic operations.
  2397. */
  2398. /**
  2399. * set_restore_sigmask() - make sure saved_sigmask processing gets done
  2400. *
  2401. * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code
  2402. * will run before returning to user mode, to process the flag. For
  2403. * all callers, TIF_SIGPENDING is already set or it's no harm to set
  2404. * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the
  2405. * arch code will notice on return to user mode, in case those bits
  2406. * are scarce. We set TIF_SIGPENDING here to ensure that the arch
  2407. * signal code always gets run when TIF_RESTORE_SIGMASK is set.
  2408. */
  2409. static inline void set_restore_sigmask(void)
  2410. {
  2411. set_thread_flag(TIF_RESTORE_SIGMASK);
  2412. WARN_ON(!test_thread_flag(TIF_SIGPENDING));
  2413. }
  2414. static inline void clear_restore_sigmask(void)
  2415. {
  2416. clear_thread_flag(TIF_RESTORE_SIGMASK);
  2417. }
  2418. static inline bool test_restore_sigmask(void)
  2419. {
  2420. return test_thread_flag(TIF_RESTORE_SIGMASK);
  2421. }
  2422. static inline bool test_and_clear_restore_sigmask(void)
  2423. {
  2424. return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK);
  2425. }
  2426. #else /* TIF_RESTORE_SIGMASK */
  2427. /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */
  2428. static inline void set_restore_sigmask(void)
  2429. {
  2430. current->restore_sigmask = true;
  2431. WARN_ON(!test_thread_flag(TIF_SIGPENDING));
  2432. }
  2433. static inline void clear_restore_sigmask(void)
  2434. {
  2435. current->restore_sigmask = false;
  2436. }
  2437. static inline bool test_restore_sigmask(void)
  2438. {
  2439. return current->restore_sigmask;
  2440. }
  2441. static inline bool test_and_clear_restore_sigmask(void)
  2442. {
  2443. if (!current->restore_sigmask)
  2444. return false;
  2445. current->restore_sigmask = false;
  2446. return true;
  2447. }
  2448. #endif
  2449. static inline void restore_saved_sigmask(void)
  2450. {
  2451. if (test_and_clear_restore_sigmask())
  2452. __set_current_blocked(&current->saved_sigmask);
  2453. }
  2454. static inline sigset_t *sigmask_to_save(void)
  2455. {
  2456. sigset_t *res = &current->blocked;
  2457. if (unlikely(test_restore_sigmask()))
  2458. res = &current->saved_sigmask;
  2459. return res;
  2460. }
  2461. static inline int kill_cad_pid(int sig, int priv)
  2462. {
  2463. return kill_pid(cad_pid, sig, priv);
  2464. }
  2465. /* These can be the second arg to send_sig_info/send_group_sig_info. */
  2466. #define SEND_SIG_NOINFO ((struct siginfo *) 0)
  2467. #define SEND_SIG_PRIV ((struct siginfo *) 1)
  2468. #define SEND_SIG_FORCED ((struct siginfo *) 2)
  2469. /*
  2470. * True if we are on the alternate signal stack.
  2471. */
  2472. static inline int on_sig_stack(unsigned long sp)
  2473. {
  2474. /*
  2475. * If the signal stack is SS_AUTODISARM then, by construction, we
  2476. * can't be on the signal stack unless user code deliberately set
  2477. * SS_AUTODISARM when we were already on it.
  2478. *
  2479. * This improves reliability: if user state gets corrupted such that
  2480. * the stack pointer points very close to the end of the signal stack,
  2481. * then this check will enable the signal to be handled anyway.
  2482. */
  2483. if (current->sas_ss_flags & SS_AUTODISARM)
  2484. return 0;
  2485. #ifdef CONFIG_STACK_GROWSUP
  2486. return sp >= current->sas_ss_sp &&
  2487. sp - current->sas_ss_sp < current->sas_ss_size;
  2488. #else
  2489. return sp > current->sas_ss_sp &&
  2490. sp - current->sas_ss_sp <= current->sas_ss_size;
  2491. #endif
  2492. }
  2493. static inline int sas_ss_flags(unsigned long sp)
  2494. {
  2495. if (!current->sas_ss_size)
  2496. return SS_DISABLE;
  2497. return on_sig_stack(sp) ? SS_ONSTACK : 0;
  2498. }
  2499. static inline void sas_ss_reset(struct task_struct *p)
  2500. {
  2501. p->sas_ss_sp = 0;
  2502. p->sas_ss_size = 0;
  2503. p->sas_ss_flags = SS_DISABLE;
  2504. }
  2505. static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
  2506. {
  2507. if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
  2508. #ifdef CONFIG_STACK_GROWSUP
  2509. return current->sas_ss_sp;
  2510. #else
  2511. return current->sas_ss_sp + current->sas_ss_size;
  2512. #endif
  2513. return sp;
  2514. }
  2515. /*
  2516. * Routines for handling mm_structs
  2517. */
  2518. extern struct mm_struct * mm_alloc(void);
  2519. /**
  2520. * mmgrab() - Pin a &struct mm_struct.
  2521. * @mm: The &struct mm_struct to pin.
  2522. *
  2523. * Make sure that @mm will not get freed even after the owning task
  2524. * exits. This doesn't guarantee that the associated address space
  2525. * will still exist later on and mmget_not_zero() has to be used before
  2526. * accessing it.
  2527. *
  2528. * This is a preferred way to to pin @mm for a longer/unbounded amount
  2529. * of time.
  2530. *
  2531. * Use mmdrop() to release the reference acquired by mmgrab().
  2532. *
  2533. * See also <Documentation/vm/active_mm.txt> for an in-depth explanation
  2534. * of &mm_struct.mm_count vs &mm_struct.mm_users.
  2535. */
  2536. static inline void mmgrab(struct mm_struct *mm)
  2537. {
  2538. atomic_inc(&mm->mm_count);
  2539. }
  2540. /* mmdrop drops the mm and the page tables */
  2541. extern void __mmdrop(struct mm_struct *);
  2542. static inline void mmdrop(struct mm_struct *mm)
  2543. {
  2544. if (unlikely(atomic_dec_and_test(&mm->mm_count)))
  2545. __mmdrop(mm);
  2546. }
  2547. static inline void mmdrop_async_fn(struct work_struct *work)
  2548. {
  2549. struct mm_struct *mm = container_of(work, struct mm_struct, async_put_work);
  2550. __mmdrop(mm);
  2551. }
  2552. static inline void mmdrop_async(struct mm_struct *mm)
  2553. {
  2554. if (unlikely(atomic_dec_and_test(&mm->mm_count))) {
  2555. INIT_WORK(&mm->async_put_work, mmdrop_async_fn);
  2556. schedule_work(&mm->async_put_work);
  2557. }
  2558. }
  2559. /**
  2560. * mmget() - Pin the address space associated with a &struct mm_struct.
  2561. * @mm: The address space to pin.
  2562. *
  2563. * Make sure that the address space of the given &struct mm_struct doesn't
  2564. * go away. This does not protect against parts of the address space being
  2565. * modified or freed, however.
  2566. *
  2567. * Never use this function to pin this address space for an
  2568. * unbounded/indefinite amount of time.
  2569. *
  2570. * Use mmput() to release the reference acquired by mmget().
  2571. *
  2572. * See also <Documentation/vm/active_mm.txt> for an in-depth explanation
  2573. * of &mm_struct.mm_count vs &mm_struct.mm_users.
  2574. */
  2575. static inline void mmget(struct mm_struct *mm)
  2576. {
  2577. atomic_inc(&mm->mm_users);
  2578. }
  2579. static inline bool mmget_not_zero(struct mm_struct *mm)
  2580. {
  2581. return atomic_inc_not_zero(&mm->mm_users);
  2582. }
  2583. /* mmput gets rid of the mappings and all user-space */
  2584. extern void mmput(struct mm_struct *);
  2585. #ifdef CONFIG_MMU
  2586. /* same as above but performs the slow path from the async context. Can
  2587. * be called from the atomic context as well
  2588. */
  2589. extern void mmput_async(struct mm_struct *);
  2590. #endif
  2591. /* Grab a reference to a task's mm, if it is not already going away */
  2592. extern struct mm_struct *get_task_mm(struct task_struct *task);
  2593. /*
  2594. * Grab a reference to a task's mm, if it is not already going away
  2595. * and ptrace_may_access with the mode parameter passed to it
  2596. * succeeds.
  2597. */
  2598. extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
  2599. /* Remove the current tasks stale references to the old mm_struct */
  2600. extern void mm_release(struct task_struct *, struct mm_struct *);
  2601. #ifdef CONFIG_HAVE_COPY_THREAD_TLS
  2602. extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
  2603. struct task_struct *, unsigned long);
  2604. #else
  2605. extern int copy_thread(unsigned long, unsigned long, unsigned long,
  2606. struct task_struct *);
  2607. /* Architectures that haven't opted into copy_thread_tls get the tls argument
  2608. * via pt_regs, so ignore the tls argument passed via C. */
  2609. static inline int copy_thread_tls(
  2610. unsigned long clone_flags, unsigned long sp, unsigned long arg,
  2611. struct task_struct *p, unsigned long tls)
  2612. {
  2613. return copy_thread(clone_flags, sp, arg, p);
  2614. }
  2615. #endif
  2616. extern void flush_thread(void);
  2617. #ifdef CONFIG_HAVE_EXIT_THREAD
  2618. extern void exit_thread(struct task_struct *tsk);
  2619. #else
  2620. static inline void exit_thread(struct task_struct *tsk)
  2621. {
  2622. }
  2623. #endif
  2624. extern void exit_files(struct task_struct *);
  2625. extern void __cleanup_sighand(struct sighand_struct *);
  2626. extern void exit_itimers(struct signal_struct *);
  2627. extern void flush_itimer_signals(void);
  2628. extern void do_group_exit(int);
  2629. extern int do_execve(struct filename *,
  2630. const char __user * const __user *,
  2631. const char __user * const __user *);
  2632. extern int do_execveat(int, struct filename *,
  2633. const char __user * const __user *,
  2634. const char __user * const __user *,
  2635. int);
  2636. extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
  2637. extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
  2638. struct task_struct *fork_idle(int);
  2639. extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
  2640. extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
  2641. static inline void set_task_comm(struct task_struct *tsk, const char *from)
  2642. {
  2643. __set_task_comm(tsk, from, false);
  2644. }
  2645. extern char *get_task_comm(char *to, struct task_struct *tsk);
  2646. #ifdef CONFIG_SMP
  2647. void scheduler_ipi(void);
  2648. extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
  2649. #else
  2650. static inline void scheduler_ipi(void) { }
  2651. static inline unsigned long wait_task_inactive(struct task_struct *p,
  2652. long match_state)
  2653. {
  2654. return 1;
  2655. }
  2656. #endif
  2657. #define tasklist_empty() \
  2658. list_empty(&init_task.tasks)
  2659. #define next_task(p) \
  2660. list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
  2661. #define for_each_process(p) \
  2662. for (p = &init_task ; (p = next_task(p)) != &init_task ; )
  2663. extern bool current_is_single_threaded(void);
  2664. /*
  2665. * Careful: do_each_thread/while_each_thread is a double loop so
  2666. * 'break' will not work as expected - use goto instead.
  2667. */
  2668. #define do_each_thread(g, t) \
  2669. for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
  2670. #define while_each_thread(g, t) \
  2671. while ((t = next_thread(t)) != g)
  2672. #define __for_each_thread(signal, t) \
  2673. list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
  2674. #define for_each_thread(p, t) \
  2675. __for_each_thread((p)->signal, t)
  2676. /* Careful: this is a double loop, 'break' won't work as expected. */
  2677. #define for_each_process_thread(p, t) \
  2678. for_each_process(p) for_each_thread(p, t)
  2679. typedef int (*proc_visitor)(struct task_struct *p, void *data);
  2680. void walk_process_tree(struct task_struct *top, proc_visitor, void *);
  2681. static inline int get_nr_threads(struct task_struct *tsk)
  2682. {
  2683. return tsk->signal->nr_threads;
  2684. }
  2685. static inline bool thread_group_leader(struct task_struct *p)
  2686. {
  2687. return p->exit_signal >= 0;
  2688. }
  2689. /* Do to the insanities of de_thread it is possible for a process
  2690. * to have the pid of the thread group leader without actually being
  2691. * the thread group leader. For iteration through the pids in proc
  2692. * all we care about is that we have a task with the appropriate
  2693. * pid, we don't actually care if we have the right task.
  2694. */
  2695. static inline bool has_group_leader_pid(struct task_struct *p)
  2696. {
  2697. return task_pid(p) == p->signal->leader_pid;
  2698. }
  2699. static inline
  2700. bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
  2701. {
  2702. return p1->signal == p2->signal;
  2703. }
  2704. static inline struct task_struct *next_thread(const struct task_struct *p)
  2705. {
  2706. return list_entry_rcu(p->thread_group.next,
  2707. struct task_struct, thread_group);
  2708. }
  2709. static inline int thread_group_empty(struct task_struct *p)
  2710. {
  2711. return list_empty(&p->thread_group);
  2712. }
  2713. #define delay_group_leader(p) \
  2714. (thread_group_leader(p) && !thread_group_empty(p))
  2715. /*
  2716. * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
  2717. * subscriptions and synchronises with wait4(). Also used in procfs. Also
  2718. * pins the final release of task.io_context. Also protects ->cpuset and
  2719. * ->cgroup.subsys[]. And ->vfork_done.
  2720. *
  2721. * Nests both inside and outside of read_lock(&tasklist_lock).
  2722. * It must not be nested with write_lock_irq(&tasklist_lock),
  2723. * neither inside nor outside.
  2724. */
  2725. static inline void task_lock(struct task_struct *p)
  2726. {
  2727. spin_lock(&p->alloc_lock);
  2728. }
  2729. static inline void task_unlock(struct task_struct *p)
  2730. {
  2731. spin_unlock(&p->alloc_lock);
  2732. }
  2733. extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
  2734. unsigned long *flags);
  2735. static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
  2736. unsigned long *flags)
  2737. {
  2738. struct sighand_struct *ret;
  2739. ret = __lock_task_sighand(tsk, flags);
  2740. (void)__cond_lock(&tsk->sighand->siglock, ret);
  2741. return ret;
  2742. }
  2743. static inline void unlock_task_sighand(struct task_struct *tsk,
  2744. unsigned long *flags)
  2745. {
  2746. spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
  2747. }
  2748. /**
  2749. * threadgroup_change_begin - mark the beginning of changes to a threadgroup
  2750. * @tsk: task causing the changes
  2751. *
  2752. * All operations which modify a threadgroup - a new thread joining the
  2753. * group, death of a member thread (the assertion of PF_EXITING) and
  2754. * exec(2) dethreading the process and replacing the leader - are wrapped
  2755. * by threadgroup_change_{begin|end}(). This is to provide a place which
  2756. * subsystems needing threadgroup stability can hook into for
  2757. * synchronization.
  2758. */
  2759. static inline void threadgroup_change_begin(struct task_struct *tsk)
  2760. {
  2761. might_sleep();
  2762. cgroup_threadgroup_change_begin(tsk);
  2763. }
  2764. /**
  2765. * threadgroup_change_end - mark the end of changes to a threadgroup
  2766. * @tsk: task causing the changes
  2767. *
  2768. * See threadgroup_change_begin().
  2769. */
  2770. static inline void threadgroup_change_end(struct task_struct *tsk)
  2771. {
  2772. cgroup_threadgroup_change_end(tsk);
  2773. }
  2774. #ifdef CONFIG_THREAD_INFO_IN_TASK
  2775. static inline struct thread_info *task_thread_info(struct task_struct *task)
  2776. {
  2777. return &task->thread_info;
  2778. }
  2779. /*
  2780. * When accessing the stack of a non-current task that might exit, use
  2781. * try_get_task_stack() instead. task_stack_page will return a pointer
  2782. * that could get freed out from under you.
  2783. */
  2784. static inline void *task_stack_page(const struct task_struct *task)
  2785. {
  2786. return task->stack;
  2787. }
  2788. #define setup_thread_stack(new,old) do { } while(0)
  2789. static inline unsigned long *end_of_stack(const struct task_struct *task)
  2790. {
  2791. return task->stack;
  2792. }
  2793. #elif !defined(__HAVE_THREAD_FUNCTIONS)
  2794. #define task_thread_info(task) ((struct thread_info *)(task)->stack)
  2795. #define task_stack_page(task) ((void *)(task)->stack)
  2796. static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
  2797. {
  2798. *task_thread_info(p) = *task_thread_info(org);
  2799. task_thread_info(p)->task = p;
  2800. }
  2801. /*
  2802. * Return the address of the last usable long on the stack.
  2803. *
  2804. * When the stack grows down, this is just above the thread
  2805. * info struct. Going any lower will corrupt the threadinfo.
  2806. *
  2807. * When the stack grows up, this is the highest address.
  2808. * Beyond that position, we corrupt data on the next page.
  2809. */
  2810. static inline unsigned long *end_of_stack(struct task_struct *p)
  2811. {
  2812. #ifdef CONFIG_STACK_GROWSUP
  2813. return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
  2814. #else
  2815. return (unsigned long *)(task_thread_info(p) + 1);
  2816. #endif
  2817. }
  2818. #endif
  2819. #ifdef CONFIG_THREAD_INFO_IN_TASK
  2820. static inline void *try_get_task_stack(struct task_struct *tsk)
  2821. {
  2822. return atomic_inc_not_zero(&tsk->stack_refcount) ?
  2823. task_stack_page(tsk) : NULL;
  2824. }
  2825. extern void put_task_stack(struct task_struct *tsk);
  2826. #else
  2827. static inline void *try_get_task_stack(struct task_struct *tsk)
  2828. {
  2829. return task_stack_page(tsk);
  2830. }
  2831. static inline void put_task_stack(struct task_struct *tsk) {}
  2832. #endif
  2833. #define task_stack_end_corrupted(task) \
  2834. (*(end_of_stack(task)) != STACK_END_MAGIC)
  2835. static inline int object_is_on_stack(void *obj)
  2836. {
  2837. void *stack = task_stack_page(current);
  2838. return (obj >= stack) && (obj < (stack + THREAD_SIZE));
  2839. }
  2840. extern void thread_stack_cache_init(void);
  2841. #ifdef CONFIG_DEBUG_STACK_USAGE
  2842. static inline unsigned long stack_not_used(struct task_struct *p)
  2843. {
  2844. unsigned long *n = end_of_stack(p);
  2845. do { /* Skip over canary */
  2846. # ifdef CONFIG_STACK_GROWSUP
  2847. n--;
  2848. # else
  2849. n++;
  2850. # endif
  2851. } while (!*n);
  2852. # ifdef CONFIG_STACK_GROWSUP
  2853. return (unsigned long)end_of_stack(p) - (unsigned long)n;
  2854. # else
  2855. return (unsigned long)n - (unsigned long)end_of_stack(p);
  2856. # endif
  2857. }
  2858. #endif
  2859. extern void set_task_stack_end_magic(struct task_struct *tsk);
  2860. /* set thread flags in other task's structures
  2861. * - see asm/thread_info.h for TIF_xxxx flags available
  2862. */
  2863. static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2864. {
  2865. set_ti_thread_flag(task_thread_info(tsk), flag);
  2866. }
  2867. static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2868. {
  2869. clear_ti_thread_flag(task_thread_info(tsk), flag);
  2870. }
  2871. static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
  2872. {
  2873. return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
  2874. }
  2875. static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
  2876. {
  2877. return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
  2878. }
  2879. static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
  2880. {
  2881. return test_ti_thread_flag(task_thread_info(tsk), flag);
  2882. }
  2883. static inline void set_tsk_need_resched(struct task_struct *tsk)
  2884. {
  2885. set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2886. }
  2887. static inline void clear_tsk_need_resched(struct task_struct *tsk)
  2888. {
  2889. clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
  2890. }
  2891. static inline int test_tsk_need_resched(struct task_struct *tsk)
  2892. {
  2893. return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
  2894. }
  2895. static inline int restart_syscall(void)
  2896. {
  2897. set_tsk_thread_flag(current, TIF_SIGPENDING);
  2898. return -ERESTARTNOINTR;
  2899. }
  2900. static inline int signal_pending(struct task_struct *p)
  2901. {
  2902. return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
  2903. }
  2904. static inline int __fatal_signal_pending(struct task_struct *p)
  2905. {
  2906. return unlikely(sigismember(&p->pending.signal, SIGKILL));
  2907. }
  2908. static inline int fatal_signal_pending(struct task_struct *p)
  2909. {
  2910. return signal_pending(p) && __fatal_signal_pending(p);
  2911. }
  2912. static inline int signal_pending_state(long state, struct task_struct *p)
  2913. {
  2914. if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
  2915. return 0;
  2916. if (!signal_pending(p))
  2917. return 0;
  2918. return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
  2919. }
  2920. /*
  2921. * cond_resched() and cond_resched_lock(): latency reduction via
  2922. * explicit rescheduling in places that are safe. The return
  2923. * value indicates whether a reschedule was done in fact.
  2924. * cond_resched_lock() will drop the spinlock before scheduling,
  2925. * cond_resched_softirq() will enable bhs before scheduling.
  2926. */
  2927. #ifndef CONFIG_PREEMPT
  2928. extern int _cond_resched(void);
  2929. #else
  2930. static inline int _cond_resched(void) { return 0; }
  2931. #endif
  2932. #define cond_resched() ({ \
  2933. ___might_sleep(__FILE__, __LINE__, 0); \
  2934. _cond_resched(); \
  2935. })
  2936. extern int __cond_resched_lock(spinlock_t *lock);
  2937. #define cond_resched_lock(lock) ({ \
  2938. ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
  2939. __cond_resched_lock(lock); \
  2940. })
  2941. extern int __cond_resched_softirq(void);
  2942. #define cond_resched_softirq() ({ \
  2943. ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
  2944. __cond_resched_softirq(); \
  2945. })
  2946. static inline void cond_resched_rcu(void)
  2947. {
  2948. #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
  2949. rcu_read_unlock();
  2950. cond_resched();
  2951. rcu_read_lock();
  2952. #endif
  2953. }
  2954. static inline unsigned long get_preempt_disable_ip(struct task_struct *p)
  2955. {
  2956. #ifdef CONFIG_DEBUG_PREEMPT
  2957. return p->preempt_disable_ip;
  2958. #else
  2959. return 0;
  2960. #endif
  2961. }
  2962. /*
  2963. * Does a critical section need to be broken due to another
  2964. * task waiting?: (technically does not depend on CONFIG_PREEMPT,
  2965. * but a general need for low latency)
  2966. */
  2967. static inline int spin_needbreak(spinlock_t *lock)
  2968. {
  2969. #ifdef CONFIG_PREEMPT
  2970. return spin_is_contended(lock);
  2971. #else
  2972. return 0;
  2973. #endif
  2974. }
  2975. /*
  2976. * Idle thread specific functions to determine the need_resched
  2977. * polling state.
  2978. */
  2979. #ifdef TIF_POLLING_NRFLAG
  2980. static inline int tsk_is_polling(struct task_struct *p)
  2981. {
  2982. return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
  2983. }
  2984. static inline void __current_set_polling(void)
  2985. {
  2986. set_thread_flag(TIF_POLLING_NRFLAG);
  2987. }
  2988. static inline bool __must_check current_set_polling_and_test(void)
  2989. {
  2990. __current_set_polling();
  2991. /*
  2992. * Polling state must be visible before we test NEED_RESCHED,
  2993. * paired by resched_curr()
  2994. */
  2995. smp_mb__after_atomic();
  2996. return unlikely(tif_need_resched());
  2997. }
  2998. static inline void __current_clr_polling(void)
  2999. {
  3000. clear_thread_flag(TIF_POLLING_NRFLAG);
  3001. }
  3002. static inline bool __must_check current_clr_polling_and_test(void)
  3003. {
  3004. __current_clr_polling();
  3005. /*
  3006. * Polling state must be visible before we test NEED_RESCHED,
  3007. * paired by resched_curr()
  3008. */
  3009. smp_mb__after_atomic();
  3010. return unlikely(tif_need_resched());
  3011. }
  3012. #else
  3013. static inline int tsk_is_polling(struct task_struct *p) { return 0; }
  3014. static inline void __current_set_polling(void) { }
  3015. static inline void __current_clr_polling(void) { }
  3016. static inline bool __must_check current_set_polling_and_test(void)
  3017. {
  3018. return unlikely(tif_need_resched());
  3019. }
  3020. static inline bool __must_check current_clr_polling_and_test(void)
  3021. {
  3022. return unlikely(tif_need_resched());
  3023. }
  3024. #endif
  3025. static inline void current_clr_polling(void)
  3026. {
  3027. __current_clr_polling();
  3028. /*
  3029. * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
  3030. * Once the bit is cleared, we'll get IPIs with every new
  3031. * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
  3032. * fold.
  3033. */
  3034. smp_mb(); /* paired with resched_curr() */
  3035. preempt_fold_need_resched();
  3036. }
  3037. static __always_inline bool need_resched(void)
  3038. {
  3039. return unlikely(tif_need_resched());
  3040. }
  3041. /*
  3042. * Thread group CPU time accounting.
  3043. */
  3044. void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
  3045. void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
  3046. /*
  3047. * Reevaluate whether the task has signals pending delivery.
  3048. * Wake the task if so.
  3049. * This is required every time the blocked sigset_t changes.
  3050. * callers must hold sighand->siglock.
  3051. */
  3052. extern void recalc_sigpending_and_wake(struct task_struct *t);
  3053. extern void recalc_sigpending(void);
  3054. extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
  3055. static inline void signal_wake_up(struct task_struct *t, bool resume)
  3056. {
  3057. signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
  3058. }
  3059. static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
  3060. {
  3061. signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
  3062. }
  3063. /*
  3064. * Wrappers for p->thread_info->cpu access. No-op on UP.
  3065. */
  3066. #ifdef CONFIG_SMP
  3067. static inline unsigned int task_cpu(const struct task_struct *p)
  3068. {
  3069. #ifdef CONFIG_THREAD_INFO_IN_TASK
  3070. return p->cpu;
  3071. #else
  3072. return task_thread_info(p)->cpu;
  3073. #endif
  3074. }
  3075. static inline int task_node(const struct task_struct *p)
  3076. {
  3077. return cpu_to_node(task_cpu(p));
  3078. }
  3079. extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
  3080. #else
  3081. static inline unsigned int task_cpu(const struct task_struct *p)
  3082. {
  3083. return 0;
  3084. }
  3085. static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
  3086. {
  3087. }
  3088. #endif /* CONFIG_SMP */
  3089. /*
  3090. * In order to reduce various lock holder preemption latencies provide an
  3091. * interface to see if a vCPU is currently running or not.
  3092. *
  3093. * This allows us to terminate optimistic spin loops and block, analogous to
  3094. * the native optimistic spin heuristic of testing if the lock owner task is
  3095. * running or not.
  3096. */
  3097. #ifndef vcpu_is_preempted
  3098. # define vcpu_is_preempted(cpu) false
  3099. #endif
  3100. extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
  3101. extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
  3102. #ifdef CONFIG_CGROUP_SCHED
  3103. extern struct task_group root_task_group;
  3104. #endif /* CONFIG_CGROUP_SCHED */
  3105. extern int task_can_switch_user(struct user_struct *up,
  3106. struct task_struct *tsk);
  3107. #ifdef CONFIG_TASK_XACCT
  3108. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  3109. {
  3110. tsk->ioac.rchar += amt;
  3111. }
  3112. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  3113. {
  3114. tsk->ioac.wchar += amt;
  3115. }
  3116. static inline void inc_syscr(struct task_struct *tsk)
  3117. {
  3118. tsk->ioac.syscr++;
  3119. }
  3120. static inline void inc_syscw(struct task_struct *tsk)
  3121. {
  3122. tsk->ioac.syscw++;
  3123. }
  3124. #else
  3125. static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
  3126. {
  3127. }
  3128. static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
  3129. {
  3130. }
  3131. static inline void inc_syscr(struct task_struct *tsk)
  3132. {
  3133. }
  3134. static inline void inc_syscw(struct task_struct *tsk)
  3135. {
  3136. }
  3137. #endif
  3138. #ifndef TASK_SIZE_OF
  3139. #define TASK_SIZE_OF(tsk) TASK_SIZE
  3140. #endif
  3141. #ifdef CONFIG_MEMCG
  3142. extern void mm_update_next_owner(struct mm_struct *mm);
  3143. #else
  3144. static inline void mm_update_next_owner(struct mm_struct *mm)
  3145. {
  3146. }
  3147. #endif /* CONFIG_MEMCG */
  3148. static inline unsigned long task_rlimit(const struct task_struct *tsk,
  3149. unsigned int limit)
  3150. {
  3151. return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
  3152. }
  3153. static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
  3154. unsigned int limit)
  3155. {
  3156. return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
  3157. }
  3158. static inline unsigned long rlimit(unsigned int limit)
  3159. {
  3160. return task_rlimit(current, limit);
  3161. }
  3162. static inline unsigned long rlimit_max(unsigned int limit)
  3163. {
  3164. return task_rlimit_max(current, limit);
  3165. }
  3166. #define SCHED_CPUFREQ_RT (1U << 0)
  3167. #define SCHED_CPUFREQ_DL (1U << 1)
  3168. #define SCHED_CPUFREQ_IOWAIT (1U << 2)
  3169. #define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL)
  3170. #ifdef CONFIG_CPU_FREQ
  3171. struct update_util_data {
  3172. void (*func)(struct update_util_data *data, u64 time, unsigned int flags);
  3173. };
  3174. void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data,
  3175. void (*func)(struct update_util_data *data, u64 time,
  3176. unsigned int flags));
  3177. void cpufreq_remove_update_util_hook(int cpu);
  3178. #endif /* CONFIG_CPU_FREQ */
  3179. #endif