bpf.h 146 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908290929102911291229132914291529162917291829192920292129222923292429252926292729282929293029312932293329342935293629372938293929402941294229432944294529462947294829492950295129522953295429552956295729582959296029612962296329642965296629672968296929702971297229732974297529762977297829792980298129822983298429852986298729882989299029912992299329942995299629972998299930003001300230033004300530063007300830093010301130123013301430153016301730183019302030213022302330243025302630273028302930303031303230333034303530363037303830393040304130423043304430453046304730483049305030513052305330543055305630573058305930603061306230633064306530663067306830693070307130723073307430753076307730783079308030813082308330843085308630873088308930903091309230933094309530963097309830993100310131023103310431053106310731083109311031113112311331143115311631173118311931203121312231233124312531263127312831293130313131323133313431353136313731383139314031413142314331443145314631473148314931503151315231533154315531563157315831593160316131623163316431653166316731683169317031713172317331743175317631773178317931803181318231833184318531863187318831893190319131923193319431953196319731983199320032013202320332043205320632073208320932103211321232133214321532163217321832193220322132223223322432253226322732283229323032313232323332343235323632373238323932403241324232433244324532463247324832493250325132523253325432553256325732583259326032613262326332643265326632673268326932703271327232733274327532763277327832793280328132823283328432853286328732883289329032913292329332943295329632973298329933003301330233033304330533063307330833093310331133123313331433153316331733183319332033213322332333243325332633273328332933303331333233333334333533363337333833393340334133423343334433453346334733483349335033513352335333543355335633573358335933603361336233633364336533663367336833693370337133723373337433753376337733783379338033813382338333843385338633873388338933903391339233933394339533963397339833993400340134023403340434053406340734083409341034113412341334143415341634173418341934203421342234233424342534263427342834293430343134323433343434353436343734383439344034413442344334443445344634473448344934503451345234533454345534563457345834593460346134623463346434653466346734683469347034713472347334743475347634773478347934803481348234833484348534863487348834893490349134923493349434953496349734983499350035013502350335043505350635073508350935103511351235133514351535163517351835193520352135223523352435253526352735283529353035313532353335343535353635373538353935403541354235433544354535463547354835493550355135523553355435553556355735583559356035613562356335643565356635673568356935703571357235733574357535763577357835793580358135823583358435853586358735883589359035913592359335943595359635973598359936003601360236033604360536063607360836093610361136123613361436153616361736183619362036213622362336243625362636273628362936303631363236333634363536363637363836393640364136423643364436453646364736483649365036513652365336543655365636573658365936603661366236633664366536663667366836693670367136723673367436753676367736783679368036813682368336843685368636873688368936903691369236933694369536963697369836993700370137023703370437053706370737083709371037113712371337143715371637173718371937203721372237233724372537263727372837293730373137323733373437353736373737383739374037413742374337443745374637473748374937503751375237533754375537563757375837593760376137623763376437653766376737683769377037713772377337743775377637773778377937803781378237833784378537863787378837893790379137923793379437953796379737983799380038013802380338043805380638073808380938103811381238133814381538163817381838193820382138223823382438253826382738283829383038313832383338343835383638373838383938403841384238433844384538463847384838493850385138523853385438553856385738583859386038613862386338643865386638673868386938703871387238733874387538763877387838793880388138823883388438853886388738883889389038913892389338943895389638973898389939003901390239033904390539063907390839093910391139123913391439153916391739183919392039213922392339243925392639273928392939303931393239333934393539363937393839393940394139423943394439453946394739483949395039513952
  1. /* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
  2. /* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com
  3. *
  4. * This program is free software; you can redistribute it and/or
  5. * modify it under the terms of version 2 of the GNU General Public
  6. * License as published by the Free Software Foundation.
  7. */
  8. #ifndef __LINUX_BPF_H__
  9. #define __LINUX_BPF_H__
  10. #include <linux/types.h>
  11. #include <linux/bpf_common.h>
  12. /* Extended instruction set based on top of classic BPF */
  13. /* instruction classes */
  14. #define BPF_JMP32 0x06 /* jmp mode in word width */
  15. #define BPF_ALU64 0x07 /* alu mode in double word width */
  16. /* ld/ldx fields */
  17. #define BPF_DW 0x18 /* double word (64-bit) */
  18. #define BPF_XADD 0xc0 /* exclusive add */
  19. /* alu/jmp fields */
  20. #define BPF_MOV 0xb0 /* mov reg to reg */
  21. #define BPF_ARSH 0xc0 /* sign extending arithmetic shift right */
  22. /* change endianness of a register */
  23. #define BPF_END 0xd0 /* flags for endianness conversion: */
  24. #define BPF_TO_LE 0x00 /* convert to little-endian */
  25. #define BPF_TO_BE 0x08 /* convert to big-endian */
  26. #define BPF_FROM_LE BPF_TO_LE
  27. #define BPF_FROM_BE BPF_TO_BE
  28. /* jmp encodings */
  29. #define BPF_JNE 0x50 /* jump != */
  30. #define BPF_JLT 0xa0 /* LT is unsigned, '<' */
  31. #define BPF_JLE 0xb0 /* LE is unsigned, '<=' */
  32. #define BPF_JSGT 0x60 /* SGT is signed '>', GT in x86 */
  33. #define BPF_JSGE 0x70 /* SGE is signed '>=', GE in x86 */
  34. #define BPF_JSLT 0xc0 /* SLT is signed, '<' */
  35. #define BPF_JSLE 0xd0 /* SLE is signed, '<=' */
  36. #define BPF_CALL 0x80 /* function call */
  37. #define BPF_EXIT 0x90 /* function return */
  38. /* Register numbers */
  39. enum {
  40. BPF_REG_0 = 0,
  41. BPF_REG_1,
  42. BPF_REG_2,
  43. BPF_REG_3,
  44. BPF_REG_4,
  45. BPF_REG_5,
  46. BPF_REG_6,
  47. BPF_REG_7,
  48. BPF_REG_8,
  49. BPF_REG_9,
  50. BPF_REG_10,
  51. __MAX_BPF_REG,
  52. };
  53. /* BPF has 10 general purpose 64-bit registers and stack frame. */
  54. #define MAX_BPF_REG __MAX_BPF_REG
  55. struct bpf_insn {
  56. __u8 code; /* opcode */
  57. __u8 dst_reg:4; /* dest register */
  58. __u8 src_reg:4; /* source register */
  59. __s16 off; /* signed offset */
  60. __s32 imm; /* signed immediate constant */
  61. };
  62. /* Key of an a BPF_MAP_TYPE_LPM_TRIE entry */
  63. struct bpf_lpm_trie_key {
  64. __u32 prefixlen; /* up to 32 for AF_INET, 128 for AF_INET6 */
  65. __u8 data[0]; /* Arbitrary size */
  66. };
  67. struct bpf_cgroup_storage_key {
  68. __u64 cgroup_inode_id; /* cgroup inode id */
  69. __u32 attach_type; /* program attach type */
  70. };
  71. /* BPF syscall commands, see bpf(2) man-page for details. */
  72. enum bpf_cmd {
  73. BPF_MAP_CREATE,
  74. BPF_MAP_LOOKUP_ELEM,
  75. BPF_MAP_UPDATE_ELEM,
  76. BPF_MAP_DELETE_ELEM,
  77. BPF_MAP_GET_NEXT_KEY,
  78. BPF_PROG_LOAD,
  79. BPF_OBJ_PIN,
  80. BPF_OBJ_GET,
  81. BPF_PROG_ATTACH,
  82. BPF_PROG_DETACH,
  83. BPF_PROG_TEST_RUN,
  84. BPF_PROG_GET_NEXT_ID,
  85. BPF_MAP_GET_NEXT_ID,
  86. BPF_PROG_GET_FD_BY_ID,
  87. BPF_MAP_GET_FD_BY_ID,
  88. BPF_OBJ_GET_INFO_BY_FD,
  89. BPF_PROG_QUERY,
  90. BPF_RAW_TRACEPOINT_OPEN,
  91. BPF_BTF_LOAD,
  92. BPF_BTF_GET_FD_BY_ID,
  93. BPF_TASK_FD_QUERY,
  94. BPF_MAP_LOOKUP_AND_DELETE_ELEM,
  95. BPF_MAP_FREEZE,
  96. BPF_BTF_GET_NEXT_ID,
  97. BPF_MAP_LOOKUP_BATCH,
  98. BPF_MAP_LOOKUP_AND_DELETE_BATCH,
  99. BPF_MAP_UPDATE_BATCH,
  100. BPF_MAP_DELETE_BATCH,
  101. BPF_LINK_CREATE,
  102. BPF_LINK_UPDATE,
  103. };
  104. enum bpf_map_type {
  105. BPF_MAP_TYPE_UNSPEC,
  106. BPF_MAP_TYPE_HASH,
  107. BPF_MAP_TYPE_ARRAY,
  108. BPF_MAP_TYPE_PROG_ARRAY,
  109. BPF_MAP_TYPE_PERF_EVENT_ARRAY,
  110. BPF_MAP_TYPE_PERCPU_HASH,
  111. BPF_MAP_TYPE_PERCPU_ARRAY,
  112. BPF_MAP_TYPE_STACK_TRACE,
  113. BPF_MAP_TYPE_CGROUP_ARRAY,
  114. BPF_MAP_TYPE_LRU_HASH,
  115. BPF_MAP_TYPE_LRU_PERCPU_HASH,
  116. BPF_MAP_TYPE_LPM_TRIE,
  117. BPF_MAP_TYPE_ARRAY_OF_MAPS,
  118. BPF_MAP_TYPE_HASH_OF_MAPS,
  119. BPF_MAP_TYPE_DEVMAP,
  120. BPF_MAP_TYPE_SOCKMAP,
  121. BPF_MAP_TYPE_CPUMAP,
  122. BPF_MAP_TYPE_XSKMAP,
  123. BPF_MAP_TYPE_SOCKHASH,
  124. BPF_MAP_TYPE_CGROUP_STORAGE,
  125. BPF_MAP_TYPE_REUSEPORT_SOCKARRAY,
  126. BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
  127. BPF_MAP_TYPE_QUEUE,
  128. BPF_MAP_TYPE_STACK,
  129. BPF_MAP_TYPE_SK_STORAGE,
  130. BPF_MAP_TYPE_DEVMAP_HASH,
  131. BPF_MAP_TYPE_STRUCT_OPS,
  132. };
  133. /* Note that tracing related programs such as
  134. * BPF_PROG_TYPE_{KPROBE,TRACEPOINT,PERF_EVENT,RAW_TRACEPOINT}
  135. * are not subject to a stable API since kernel internal data
  136. * structures can change from release to release and may
  137. * therefore break existing tracing BPF programs. Tracing BPF
  138. * programs correspond to /a/ specific kernel which is to be
  139. * analyzed, and not /a/ specific kernel /and/ all future ones.
  140. */
  141. enum bpf_prog_type {
  142. BPF_PROG_TYPE_UNSPEC,
  143. BPF_PROG_TYPE_SOCKET_FILTER,
  144. BPF_PROG_TYPE_KPROBE,
  145. BPF_PROG_TYPE_SCHED_CLS,
  146. BPF_PROG_TYPE_SCHED_ACT,
  147. BPF_PROG_TYPE_TRACEPOINT,
  148. BPF_PROG_TYPE_XDP,
  149. BPF_PROG_TYPE_PERF_EVENT,
  150. BPF_PROG_TYPE_CGROUP_SKB,
  151. BPF_PROG_TYPE_CGROUP_SOCK,
  152. BPF_PROG_TYPE_LWT_IN,
  153. BPF_PROG_TYPE_LWT_OUT,
  154. BPF_PROG_TYPE_LWT_XMIT,
  155. BPF_PROG_TYPE_SOCK_OPS,
  156. BPF_PROG_TYPE_SK_SKB,
  157. BPF_PROG_TYPE_CGROUP_DEVICE,
  158. BPF_PROG_TYPE_SK_MSG,
  159. BPF_PROG_TYPE_RAW_TRACEPOINT,
  160. BPF_PROG_TYPE_CGROUP_SOCK_ADDR,
  161. BPF_PROG_TYPE_LWT_SEG6LOCAL,
  162. BPF_PROG_TYPE_LIRC_MODE2,
  163. BPF_PROG_TYPE_SK_REUSEPORT,
  164. BPF_PROG_TYPE_FLOW_DISSECTOR,
  165. BPF_PROG_TYPE_CGROUP_SYSCTL,
  166. BPF_PROG_TYPE_RAW_TRACEPOINT_WRITABLE,
  167. BPF_PROG_TYPE_CGROUP_SOCKOPT,
  168. BPF_PROG_TYPE_TRACING,
  169. BPF_PROG_TYPE_STRUCT_OPS,
  170. BPF_PROG_TYPE_EXT,
  171. BPF_PROG_TYPE_LSM,
  172. };
  173. enum bpf_attach_type {
  174. BPF_CGROUP_INET_INGRESS,
  175. BPF_CGROUP_INET_EGRESS,
  176. BPF_CGROUP_INET_SOCK_CREATE,
  177. BPF_CGROUP_SOCK_OPS,
  178. BPF_SK_SKB_STREAM_PARSER,
  179. BPF_SK_SKB_STREAM_VERDICT,
  180. BPF_CGROUP_DEVICE,
  181. BPF_SK_MSG_VERDICT,
  182. BPF_CGROUP_INET4_BIND,
  183. BPF_CGROUP_INET6_BIND,
  184. BPF_CGROUP_INET4_CONNECT,
  185. BPF_CGROUP_INET6_CONNECT,
  186. BPF_CGROUP_INET4_POST_BIND,
  187. BPF_CGROUP_INET6_POST_BIND,
  188. BPF_CGROUP_UDP4_SENDMSG,
  189. BPF_CGROUP_UDP6_SENDMSG,
  190. BPF_LIRC_MODE2,
  191. BPF_FLOW_DISSECTOR,
  192. BPF_CGROUP_SYSCTL,
  193. BPF_CGROUP_UDP4_RECVMSG,
  194. BPF_CGROUP_UDP6_RECVMSG,
  195. BPF_CGROUP_GETSOCKOPT,
  196. BPF_CGROUP_SETSOCKOPT,
  197. BPF_TRACE_RAW_TP,
  198. BPF_TRACE_FENTRY,
  199. BPF_TRACE_FEXIT,
  200. BPF_MODIFY_RETURN,
  201. BPF_LSM_MAC,
  202. __MAX_BPF_ATTACH_TYPE
  203. };
  204. #define MAX_BPF_ATTACH_TYPE __MAX_BPF_ATTACH_TYPE
  205. /* cgroup-bpf attach flags used in BPF_PROG_ATTACH command
  206. *
  207. * NONE(default): No further bpf programs allowed in the subtree.
  208. *
  209. * BPF_F_ALLOW_OVERRIDE: If a sub-cgroup installs some bpf program,
  210. * the program in this cgroup yields to sub-cgroup program.
  211. *
  212. * BPF_F_ALLOW_MULTI: If a sub-cgroup installs some bpf program,
  213. * that cgroup program gets run in addition to the program in this cgroup.
  214. *
  215. * Only one program is allowed to be attached to a cgroup with
  216. * NONE or BPF_F_ALLOW_OVERRIDE flag.
  217. * Attaching another program on top of NONE or BPF_F_ALLOW_OVERRIDE will
  218. * release old program and attach the new one. Attach flags has to match.
  219. *
  220. * Multiple programs are allowed to be attached to a cgroup with
  221. * BPF_F_ALLOW_MULTI flag. They are executed in FIFO order
  222. * (those that were attached first, run first)
  223. * The programs of sub-cgroup are executed first, then programs of
  224. * this cgroup and then programs of parent cgroup.
  225. * When children program makes decision (like picking TCP CA or sock bind)
  226. * parent program has a chance to override it.
  227. *
  228. * With BPF_F_ALLOW_MULTI a new program is added to the end of the list of
  229. * programs for a cgroup. Though it's possible to replace an old program at
  230. * any position by also specifying BPF_F_REPLACE flag and position itself in
  231. * replace_bpf_fd attribute. Old program at this position will be released.
  232. *
  233. * A cgroup with MULTI or OVERRIDE flag allows any attach flags in sub-cgroups.
  234. * A cgroup with NONE doesn't allow any programs in sub-cgroups.
  235. * Ex1:
  236. * cgrp1 (MULTI progs A, B) ->
  237. * cgrp2 (OVERRIDE prog C) ->
  238. * cgrp3 (MULTI prog D) ->
  239. * cgrp4 (OVERRIDE prog E) ->
  240. * cgrp5 (NONE prog F)
  241. * the event in cgrp5 triggers execution of F,D,A,B in that order.
  242. * if prog F is detached, the execution is E,D,A,B
  243. * if prog F and D are detached, the execution is E,A,B
  244. * if prog F, E and D are detached, the execution is C,A,B
  245. *
  246. * All eligible programs are executed regardless of return code from
  247. * earlier programs.
  248. */
  249. #define BPF_F_ALLOW_OVERRIDE (1U << 0)
  250. #define BPF_F_ALLOW_MULTI (1U << 1)
  251. #define BPF_F_REPLACE (1U << 2)
  252. /* If BPF_F_STRICT_ALIGNMENT is used in BPF_PROG_LOAD command, the
  253. * verifier will perform strict alignment checking as if the kernel
  254. * has been built with CONFIG_EFFICIENT_UNALIGNED_ACCESS not set,
  255. * and NET_IP_ALIGN defined to 2.
  256. */
  257. #define BPF_F_STRICT_ALIGNMENT (1U << 0)
  258. /* If BPF_F_ANY_ALIGNMENT is used in BPF_PROF_LOAD command, the
  259. * verifier will allow any alignment whatsoever. On platforms
  260. * with strict alignment requirements for loads ands stores (such
  261. * as sparc and mips) the verifier validates that all loads and
  262. * stores provably follow this requirement. This flag turns that
  263. * checking and enforcement off.
  264. *
  265. * It is mostly used for testing when we want to validate the
  266. * context and memory access aspects of the verifier, but because
  267. * of an unaligned access the alignment check would trigger before
  268. * the one we are interested in.
  269. */
  270. #define BPF_F_ANY_ALIGNMENT (1U << 1)
  271. /* BPF_F_TEST_RND_HI32 is used in BPF_PROG_LOAD command for testing purpose.
  272. * Verifier does sub-register def/use analysis and identifies instructions whose
  273. * def only matters for low 32-bit, high 32-bit is never referenced later
  274. * through implicit zero extension. Therefore verifier notifies JIT back-ends
  275. * that it is safe to ignore clearing high 32-bit for these instructions. This
  276. * saves some back-ends a lot of code-gen. However such optimization is not
  277. * necessary on some arches, for example x86_64, arm64 etc, whose JIT back-ends
  278. * hence hasn't used verifier's analysis result. But, we really want to have a
  279. * way to be able to verify the correctness of the described optimization on
  280. * x86_64 on which testsuites are frequently exercised.
  281. *
  282. * So, this flag is introduced. Once it is set, verifier will randomize high
  283. * 32-bit for those instructions who has been identified as safe to ignore them.
  284. * Then, if verifier is not doing correct analysis, such randomization will
  285. * regress tests to expose bugs.
  286. */
  287. #define BPF_F_TEST_RND_HI32 (1U << 2)
  288. /* The verifier internal test flag. Behavior is undefined */
  289. #define BPF_F_TEST_STATE_FREQ (1U << 3)
  290. /* When BPF ldimm64's insn[0].src_reg != 0 then this can have
  291. * two extensions:
  292. *
  293. * insn[0].src_reg: BPF_PSEUDO_MAP_FD BPF_PSEUDO_MAP_VALUE
  294. * insn[0].imm: map fd map fd
  295. * insn[1].imm: 0 offset into value
  296. * insn[0].off: 0 0
  297. * insn[1].off: 0 0
  298. * ldimm64 rewrite: address of map address of map[0]+offset
  299. * verifier type: CONST_PTR_TO_MAP PTR_TO_MAP_VALUE
  300. */
  301. #define BPF_PSEUDO_MAP_FD 1
  302. #define BPF_PSEUDO_MAP_VALUE 2
  303. /* when bpf_call->src_reg == BPF_PSEUDO_CALL, bpf_call->imm == pc-relative
  304. * offset to another bpf function
  305. */
  306. #define BPF_PSEUDO_CALL 1
  307. /* flags for BPF_MAP_UPDATE_ELEM command */
  308. enum {
  309. BPF_ANY = 0, /* create new element or update existing */
  310. BPF_NOEXIST = 1, /* create new element if it didn't exist */
  311. BPF_EXIST = 2, /* update existing element */
  312. BPF_F_LOCK = 4, /* spin_lock-ed map_lookup/map_update */
  313. };
  314. /* flags for BPF_MAP_CREATE command */
  315. enum {
  316. BPF_F_NO_PREALLOC = (1U << 0),
  317. /* Instead of having one common LRU list in the
  318. * BPF_MAP_TYPE_LRU_[PERCPU_]HASH map, use a percpu LRU list
  319. * which can scale and perform better.
  320. * Note, the LRU nodes (including free nodes) cannot be moved
  321. * across different LRU lists.
  322. */
  323. BPF_F_NO_COMMON_LRU = (1U << 1),
  324. /* Specify numa node during map creation */
  325. BPF_F_NUMA_NODE = (1U << 2),
  326. /* Flags for accessing BPF object from syscall side. */
  327. BPF_F_RDONLY = (1U << 3),
  328. BPF_F_WRONLY = (1U << 4),
  329. /* Flag for stack_map, store build_id+offset instead of pointer */
  330. BPF_F_STACK_BUILD_ID = (1U << 5),
  331. /* Zero-initialize hash function seed. This should only be used for testing. */
  332. BPF_F_ZERO_SEED = (1U << 6),
  333. /* Flags for accessing BPF object from program side. */
  334. BPF_F_RDONLY_PROG = (1U << 7),
  335. BPF_F_WRONLY_PROG = (1U << 8),
  336. /* Clone map from listener for newly accepted socket */
  337. BPF_F_CLONE = (1U << 9),
  338. /* Enable memory-mapping BPF map */
  339. BPF_F_MMAPABLE = (1U << 10),
  340. };
  341. /* Flags for BPF_PROG_QUERY. */
  342. /* Query effective (directly attached + inherited from ancestor cgroups)
  343. * programs that will be executed for events within a cgroup.
  344. * attach_flags with this flag are returned only for directly attached programs.
  345. */
  346. #define BPF_F_QUERY_EFFECTIVE (1U << 0)
  347. enum bpf_stack_build_id_status {
  348. /* user space need an empty entry to identify end of a trace */
  349. BPF_STACK_BUILD_ID_EMPTY = 0,
  350. /* with valid build_id and offset */
  351. BPF_STACK_BUILD_ID_VALID = 1,
  352. /* couldn't get build_id, fallback to ip */
  353. BPF_STACK_BUILD_ID_IP = 2,
  354. };
  355. #define BPF_BUILD_ID_SIZE 20
  356. struct bpf_stack_build_id {
  357. __s32 status;
  358. unsigned char build_id[BPF_BUILD_ID_SIZE];
  359. union {
  360. __u64 offset;
  361. __u64 ip;
  362. };
  363. };
  364. #define BPF_OBJ_NAME_LEN 16U
  365. union bpf_attr {
  366. struct { /* anonymous struct used by BPF_MAP_CREATE command */
  367. __u32 map_type; /* one of enum bpf_map_type */
  368. __u32 key_size; /* size of key in bytes */
  369. __u32 value_size; /* size of value in bytes */
  370. __u32 max_entries; /* max number of entries in a map */
  371. __u32 map_flags; /* BPF_MAP_CREATE related
  372. * flags defined above.
  373. */
  374. __u32 inner_map_fd; /* fd pointing to the inner map */
  375. __u32 numa_node; /* numa node (effective only if
  376. * BPF_F_NUMA_NODE is set).
  377. */
  378. char map_name[BPF_OBJ_NAME_LEN];
  379. __u32 map_ifindex; /* ifindex of netdev to create on */
  380. __u32 btf_fd; /* fd pointing to a BTF type data */
  381. __u32 btf_key_type_id; /* BTF type_id of the key */
  382. __u32 btf_value_type_id; /* BTF type_id of the value */
  383. __u32 btf_vmlinux_value_type_id;/* BTF type_id of a kernel-
  384. * struct stored as the
  385. * map value
  386. */
  387. };
  388. struct { /* anonymous struct used by BPF_MAP_*_ELEM commands */
  389. __u32 map_fd;
  390. __aligned_u64 key;
  391. union {
  392. __aligned_u64 value;
  393. __aligned_u64 next_key;
  394. };
  395. __u64 flags;
  396. };
  397. struct { /* struct used by BPF_MAP_*_BATCH commands */
  398. __aligned_u64 in_batch; /* start batch,
  399. * NULL to start from beginning
  400. */
  401. __aligned_u64 out_batch; /* output: next start batch */
  402. __aligned_u64 keys;
  403. __aligned_u64 values;
  404. __u32 count; /* input/output:
  405. * input: # of key/value
  406. * elements
  407. * output: # of filled elements
  408. */
  409. __u32 map_fd;
  410. __u64 elem_flags;
  411. __u64 flags;
  412. } batch;
  413. struct { /* anonymous struct used by BPF_PROG_LOAD command */
  414. __u32 prog_type; /* one of enum bpf_prog_type */
  415. __u32 insn_cnt;
  416. __aligned_u64 insns;
  417. __aligned_u64 license;
  418. __u32 log_level; /* verbosity level of verifier */
  419. __u32 log_size; /* size of user buffer */
  420. __aligned_u64 log_buf; /* user supplied buffer */
  421. __u32 kern_version; /* not used */
  422. __u32 prog_flags;
  423. char prog_name[BPF_OBJ_NAME_LEN];
  424. __u32 prog_ifindex; /* ifindex of netdev to prep for */
  425. /* For some prog types expected attach type must be known at
  426. * load time to verify attach type specific parts of prog
  427. * (context accesses, allowed helpers, etc).
  428. */
  429. __u32 expected_attach_type;
  430. __u32 prog_btf_fd; /* fd pointing to BTF type data */
  431. __u32 func_info_rec_size; /* userspace bpf_func_info size */
  432. __aligned_u64 func_info; /* func info */
  433. __u32 func_info_cnt; /* number of bpf_func_info records */
  434. __u32 line_info_rec_size; /* userspace bpf_line_info size */
  435. __aligned_u64 line_info; /* line info */
  436. __u32 line_info_cnt; /* number of bpf_line_info records */
  437. __u32 attach_btf_id; /* in-kernel BTF type id to attach to */
  438. __u32 attach_prog_fd; /* 0 to attach to vmlinux */
  439. };
  440. struct { /* anonymous struct used by BPF_OBJ_* commands */
  441. __aligned_u64 pathname;
  442. __u32 bpf_fd;
  443. __u32 file_flags;
  444. };
  445. struct { /* anonymous struct used by BPF_PROG_ATTACH/DETACH commands */
  446. __u32 target_fd; /* container object to attach to */
  447. __u32 attach_bpf_fd; /* eBPF program to attach */
  448. __u32 attach_type;
  449. __u32 attach_flags;
  450. __u32 replace_bpf_fd; /* previously attached eBPF
  451. * program to replace if
  452. * BPF_F_REPLACE is used
  453. */
  454. };
  455. struct { /* anonymous struct used by BPF_PROG_TEST_RUN command */
  456. __u32 prog_fd;
  457. __u32 retval;
  458. __u32 data_size_in; /* input: len of data_in */
  459. __u32 data_size_out; /* input/output: len of data_out
  460. * returns ENOSPC if data_out
  461. * is too small.
  462. */
  463. __aligned_u64 data_in;
  464. __aligned_u64 data_out;
  465. __u32 repeat;
  466. __u32 duration;
  467. __u32 ctx_size_in; /* input: len of ctx_in */
  468. __u32 ctx_size_out; /* input/output: len of ctx_out
  469. * returns ENOSPC if ctx_out
  470. * is too small.
  471. */
  472. __aligned_u64 ctx_in;
  473. __aligned_u64 ctx_out;
  474. } test;
  475. struct { /* anonymous struct used by BPF_*_GET_*_ID */
  476. union {
  477. __u32 start_id;
  478. __u32 prog_id;
  479. __u32 map_id;
  480. __u32 btf_id;
  481. };
  482. __u32 next_id;
  483. __u32 open_flags;
  484. };
  485. struct { /* anonymous struct used by BPF_OBJ_GET_INFO_BY_FD */
  486. __u32 bpf_fd;
  487. __u32 info_len;
  488. __aligned_u64 info;
  489. } info;
  490. struct { /* anonymous struct used by BPF_PROG_QUERY command */
  491. __u32 target_fd; /* container object to query */
  492. __u32 attach_type;
  493. __u32 query_flags;
  494. __u32 attach_flags;
  495. __aligned_u64 prog_ids;
  496. __u32 prog_cnt;
  497. } query;
  498. struct { /* anonymous struct used by BPF_RAW_TRACEPOINT_OPEN command */
  499. __u64 name;
  500. __u32 prog_fd;
  501. } raw_tracepoint;
  502. struct { /* anonymous struct for BPF_BTF_LOAD */
  503. __aligned_u64 btf;
  504. __aligned_u64 btf_log_buf;
  505. __u32 btf_size;
  506. __u32 btf_log_size;
  507. __u32 btf_log_level;
  508. };
  509. struct {
  510. __u32 pid; /* input: pid */
  511. __u32 fd; /* input: fd */
  512. __u32 flags; /* input: flags */
  513. __u32 buf_len; /* input/output: buf len */
  514. __aligned_u64 buf; /* input/output:
  515. * tp_name for tracepoint
  516. * symbol for kprobe
  517. * filename for uprobe
  518. */
  519. __u32 prog_id; /* output: prod_id */
  520. __u32 fd_type; /* output: BPF_FD_TYPE_* */
  521. __u64 probe_offset; /* output: probe_offset */
  522. __u64 probe_addr; /* output: probe_addr */
  523. } task_fd_query;
  524. struct { /* struct used by BPF_LINK_CREATE command */
  525. __u32 prog_fd; /* eBPF program to attach */
  526. __u32 target_fd; /* object to attach to */
  527. __u32 attach_type; /* attach type */
  528. __u32 flags; /* extra flags */
  529. } link_create;
  530. struct { /* struct used by BPF_LINK_UPDATE command */
  531. __u32 link_fd; /* link fd */
  532. /* new program fd to update link with */
  533. __u32 new_prog_fd;
  534. __u32 flags; /* extra flags */
  535. /* expected link's program fd; is specified only if
  536. * BPF_F_REPLACE flag is set in flags */
  537. __u32 old_prog_fd;
  538. } link_update;
  539. } __attribute__((aligned(8)));
  540. /* The description below is an attempt at providing documentation to eBPF
  541. * developers about the multiple available eBPF helper functions. It can be
  542. * parsed and used to produce a manual page. The workflow is the following,
  543. * and requires the rst2man utility:
  544. *
  545. * $ ./scripts/bpf_helpers_doc.py \
  546. * --filename include/uapi/linux/bpf.h > /tmp/bpf-helpers.rst
  547. * $ rst2man /tmp/bpf-helpers.rst > /tmp/bpf-helpers.7
  548. * $ man /tmp/bpf-helpers.7
  549. *
  550. * Note that in order to produce this external documentation, some RST
  551. * formatting is used in the descriptions to get "bold" and "italics" in
  552. * manual pages. Also note that the few trailing white spaces are
  553. * intentional, removing them would break paragraphs for rst2man.
  554. *
  555. * Start of BPF helper function descriptions:
  556. *
  557. * void *bpf_map_lookup_elem(struct bpf_map *map, const void *key)
  558. * Description
  559. * Perform a lookup in *map* for an entry associated to *key*.
  560. * Return
  561. * Map value associated to *key*, or **NULL** if no entry was
  562. * found.
  563. *
  564. * int bpf_map_update_elem(struct bpf_map *map, const void *key, const void *value, u64 flags)
  565. * Description
  566. * Add or update the value of the entry associated to *key* in
  567. * *map* with *value*. *flags* is one of:
  568. *
  569. * **BPF_NOEXIST**
  570. * The entry for *key* must not exist in the map.
  571. * **BPF_EXIST**
  572. * The entry for *key* must already exist in the map.
  573. * **BPF_ANY**
  574. * No condition on the existence of the entry for *key*.
  575. *
  576. * Flag value **BPF_NOEXIST** cannot be used for maps of types
  577. * **BPF_MAP_TYPE_ARRAY** or **BPF_MAP_TYPE_PERCPU_ARRAY** (all
  578. * elements always exist), the helper would return an error.
  579. * Return
  580. * 0 on success, or a negative error in case of failure.
  581. *
  582. * int bpf_map_delete_elem(struct bpf_map *map, const void *key)
  583. * Description
  584. * Delete entry with *key* from *map*.
  585. * Return
  586. * 0 on success, or a negative error in case of failure.
  587. *
  588. * int bpf_probe_read(void *dst, u32 size, const void *unsafe_ptr)
  589. * Description
  590. * For tracing programs, safely attempt to read *size* bytes from
  591. * kernel space address *unsafe_ptr* and store the data in *dst*.
  592. *
  593. * Generally, use bpf_probe_read_user() or bpf_probe_read_kernel()
  594. * instead.
  595. * Return
  596. * 0 on success, or a negative error in case of failure.
  597. *
  598. * u64 bpf_ktime_get_ns(void)
  599. * Description
  600. * Return the time elapsed since system boot, in nanoseconds.
  601. * Return
  602. * Current *ktime*.
  603. *
  604. * int bpf_trace_printk(const char *fmt, u32 fmt_size, ...)
  605. * Description
  606. * This helper is a "printk()-like" facility for debugging. It
  607. * prints a message defined by format *fmt* (of size *fmt_size*)
  608. * to file *\/sys/kernel/debug/tracing/trace* from DebugFS, if
  609. * available. It can take up to three additional **u64**
  610. * arguments (as an eBPF helpers, the total number of arguments is
  611. * limited to five).
  612. *
  613. * Each time the helper is called, it appends a line to the trace.
  614. * Lines are discarded while *\/sys/kernel/debug/tracing/trace* is
  615. * open, use *\/sys/kernel/debug/tracing/trace_pipe* to avoid this.
  616. * The format of the trace is customizable, and the exact output
  617. * one will get depends on the options set in
  618. * *\/sys/kernel/debug/tracing/trace_options* (see also the
  619. * *README* file under the same directory). However, it usually
  620. * defaults to something like:
  621. *
  622. * ::
  623. *
  624. * telnet-470 [001] .N.. 419421.045894: 0x00000001: <formatted msg>
  625. *
  626. * In the above:
  627. *
  628. * * ``telnet`` is the name of the current task.
  629. * * ``470`` is the PID of the current task.
  630. * * ``001`` is the CPU number on which the task is
  631. * running.
  632. * * In ``.N..``, each character refers to a set of
  633. * options (whether irqs are enabled, scheduling
  634. * options, whether hard/softirqs are running, level of
  635. * preempt_disabled respectively). **N** means that
  636. * **TIF_NEED_RESCHED** and **PREEMPT_NEED_RESCHED**
  637. * are set.
  638. * * ``419421.045894`` is a timestamp.
  639. * * ``0x00000001`` is a fake value used by BPF for the
  640. * instruction pointer register.
  641. * * ``<formatted msg>`` is the message formatted with
  642. * *fmt*.
  643. *
  644. * The conversion specifiers supported by *fmt* are similar, but
  645. * more limited than for printk(). They are **%d**, **%i**,
  646. * **%u**, **%x**, **%ld**, **%li**, **%lu**, **%lx**, **%lld**,
  647. * **%lli**, **%llu**, **%llx**, **%p**, **%s**. No modifier (size
  648. * of field, padding with zeroes, etc.) is available, and the
  649. * helper will return **-EINVAL** (but print nothing) if it
  650. * encounters an unknown specifier.
  651. *
  652. * Also, note that **bpf_trace_printk**\ () is slow, and should
  653. * only be used for debugging purposes. For this reason, a notice
  654. * bloc (spanning several lines) is printed to kernel logs and
  655. * states that the helper should not be used "for production use"
  656. * the first time this helper is used (or more precisely, when
  657. * **trace_printk**\ () buffers are allocated). For passing values
  658. * to user space, perf events should be preferred.
  659. * Return
  660. * The number of bytes written to the buffer, or a negative error
  661. * in case of failure.
  662. *
  663. * u32 bpf_get_prandom_u32(void)
  664. * Description
  665. * Get a pseudo-random number.
  666. *
  667. * From a security point of view, this helper uses its own
  668. * pseudo-random internal state, and cannot be used to infer the
  669. * seed of other random functions in the kernel. However, it is
  670. * essential to note that the generator used by the helper is not
  671. * cryptographically secure.
  672. * Return
  673. * A random 32-bit unsigned value.
  674. *
  675. * u32 bpf_get_smp_processor_id(void)
  676. * Description
  677. * Get the SMP (symmetric multiprocessing) processor id. Note that
  678. * all programs run with preemption disabled, which means that the
  679. * SMP processor id is stable during all the execution of the
  680. * program.
  681. * Return
  682. * The SMP id of the processor running the program.
  683. *
  684. * int bpf_skb_store_bytes(struct sk_buff *skb, u32 offset, const void *from, u32 len, u64 flags)
  685. * Description
  686. * Store *len* bytes from address *from* into the packet
  687. * associated to *skb*, at *offset*. *flags* are a combination of
  688. * **BPF_F_RECOMPUTE_CSUM** (automatically recompute the
  689. * checksum for the packet after storing the bytes) and
  690. * **BPF_F_INVALIDATE_HASH** (set *skb*\ **->hash**, *skb*\
  691. * **->swhash** and *skb*\ **->l4hash** to 0).
  692. *
  693. * A call to this helper is susceptible to change the underlying
  694. * packet buffer. Therefore, at load time, all checks on pointers
  695. * previously done by the verifier are invalidated and must be
  696. * performed again, if the helper is used in combination with
  697. * direct packet access.
  698. * Return
  699. * 0 on success, or a negative error in case of failure.
  700. *
  701. * int bpf_l3_csum_replace(struct sk_buff *skb, u32 offset, u64 from, u64 to, u64 size)
  702. * Description
  703. * Recompute the layer 3 (e.g. IP) checksum for the packet
  704. * associated to *skb*. Computation is incremental, so the helper
  705. * must know the former value of the header field that was
  706. * modified (*from*), the new value of this field (*to*), and the
  707. * number of bytes (2 or 4) for this field, stored in *size*.
  708. * Alternatively, it is possible to store the difference between
  709. * the previous and the new values of the header field in *to*, by
  710. * setting *from* and *size* to 0. For both methods, *offset*
  711. * indicates the location of the IP checksum within the packet.
  712. *
  713. * This helper works in combination with **bpf_csum_diff**\ (),
  714. * which does not update the checksum in-place, but offers more
  715. * flexibility and can handle sizes larger than 2 or 4 for the
  716. * checksum to update.
  717. *
  718. * A call to this helper is susceptible to change the underlying
  719. * packet buffer. Therefore, at load time, all checks on pointers
  720. * previously done by the verifier are invalidated and must be
  721. * performed again, if the helper is used in combination with
  722. * direct packet access.
  723. * Return
  724. * 0 on success, or a negative error in case of failure.
  725. *
  726. * int bpf_l4_csum_replace(struct sk_buff *skb, u32 offset, u64 from, u64 to, u64 flags)
  727. * Description
  728. * Recompute the layer 4 (e.g. TCP, UDP or ICMP) checksum for the
  729. * packet associated to *skb*. Computation is incremental, so the
  730. * helper must know the former value of the header field that was
  731. * modified (*from*), the new value of this field (*to*), and the
  732. * number of bytes (2 or 4) for this field, stored on the lowest
  733. * four bits of *flags*. Alternatively, it is possible to store
  734. * the difference between the previous and the new values of the
  735. * header field in *to*, by setting *from* and the four lowest
  736. * bits of *flags* to 0. For both methods, *offset* indicates the
  737. * location of the IP checksum within the packet. In addition to
  738. * the size of the field, *flags* can be added (bitwise OR) actual
  739. * flags. With **BPF_F_MARK_MANGLED_0**, a null checksum is left
  740. * untouched (unless **BPF_F_MARK_ENFORCE** is added as well), and
  741. * for updates resulting in a null checksum the value is set to
  742. * **CSUM_MANGLED_0** instead. Flag **BPF_F_PSEUDO_HDR** indicates
  743. * the checksum is to be computed against a pseudo-header.
  744. *
  745. * This helper works in combination with **bpf_csum_diff**\ (),
  746. * which does not update the checksum in-place, but offers more
  747. * flexibility and can handle sizes larger than 2 or 4 for the
  748. * checksum to update.
  749. *
  750. * A call to this helper is susceptible to change the underlying
  751. * packet buffer. Therefore, at load time, all checks on pointers
  752. * previously done by the verifier are invalidated and must be
  753. * performed again, if the helper is used in combination with
  754. * direct packet access.
  755. * Return
  756. * 0 on success, or a negative error in case of failure.
  757. *
  758. * int bpf_tail_call(void *ctx, struct bpf_map *prog_array_map, u32 index)
  759. * Description
  760. * This special helper is used to trigger a "tail call", or in
  761. * other words, to jump into another eBPF program. The same stack
  762. * frame is used (but values on stack and in registers for the
  763. * caller are not accessible to the callee). This mechanism allows
  764. * for program chaining, either for raising the maximum number of
  765. * available eBPF instructions, or to execute given programs in
  766. * conditional blocks. For security reasons, there is an upper
  767. * limit to the number of successive tail calls that can be
  768. * performed.
  769. *
  770. * Upon call of this helper, the program attempts to jump into a
  771. * program referenced at index *index* in *prog_array_map*, a
  772. * special map of type **BPF_MAP_TYPE_PROG_ARRAY**, and passes
  773. * *ctx*, a pointer to the context.
  774. *
  775. * If the call succeeds, the kernel immediately runs the first
  776. * instruction of the new program. This is not a function call,
  777. * and it never returns to the previous program. If the call
  778. * fails, then the helper has no effect, and the caller continues
  779. * to run its subsequent instructions. A call can fail if the
  780. * destination program for the jump does not exist (i.e. *index*
  781. * is superior to the number of entries in *prog_array_map*), or
  782. * if the maximum number of tail calls has been reached for this
  783. * chain of programs. This limit is defined in the kernel by the
  784. * macro **MAX_TAIL_CALL_CNT** (not accessible to user space),
  785. * which is currently set to 32.
  786. * Return
  787. * 0 on success, or a negative error in case of failure.
  788. *
  789. * int bpf_clone_redirect(struct sk_buff *skb, u32 ifindex, u64 flags)
  790. * Description
  791. * Clone and redirect the packet associated to *skb* to another
  792. * net device of index *ifindex*. Both ingress and egress
  793. * interfaces can be used for redirection. The **BPF_F_INGRESS**
  794. * value in *flags* is used to make the distinction (ingress path
  795. * is selected if the flag is present, egress path otherwise).
  796. * This is the only flag supported for now.
  797. *
  798. * In comparison with **bpf_redirect**\ () helper,
  799. * **bpf_clone_redirect**\ () has the associated cost of
  800. * duplicating the packet buffer, but this can be executed out of
  801. * the eBPF program. Conversely, **bpf_redirect**\ () is more
  802. * efficient, but it is handled through an action code where the
  803. * redirection happens only after the eBPF program has returned.
  804. *
  805. * A call to this helper is susceptible to change the underlying
  806. * packet buffer. Therefore, at load time, all checks on pointers
  807. * previously done by the verifier are invalidated and must be
  808. * performed again, if the helper is used in combination with
  809. * direct packet access.
  810. * Return
  811. * 0 on success, or a negative error in case of failure.
  812. *
  813. * u64 bpf_get_current_pid_tgid(void)
  814. * Return
  815. * A 64-bit integer containing the current tgid and pid, and
  816. * created as such:
  817. * *current_task*\ **->tgid << 32 \|**
  818. * *current_task*\ **->pid**.
  819. *
  820. * u64 bpf_get_current_uid_gid(void)
  821. * Return
  822. * A 64-bit integer containing the current GID and UID, and
  823. * created as such: *current_gid* **<< 32 \|** *current_uid*.
  824. *
  825. * int bpf_get_current_comm(void *buf, u32 size_of_buf)
  826. * Description
  827. * Copy the **comm** attribute of the current task into *buf* of
  828. * *size_of_buf*. The **comm** attribute contains the name of
  829. * the executable (excluding the path) for the current task. The
  830. * *size_of_buf* must be strictly positive. On success, the
  831. * helper makes sure that the *buf* is NUL-terminated. On failure,
  832. * it is filled with zeroes.
  833. * Return
  834. * 0 on success, or a negative error in case of failure.
  835. *
  836. * u32 bpf_get_cgroup_classid(struct sk_buff *skb)
  837. * Description
  838. * Retrieve the classid for the current task, i.e. for the net_cls
  839. * cgroup to which *skb* belongs.
  840. *
  841. * This helper can be used on TC egress path, but not on ingress.
  842. *
  843. * The net_cls cgroup provides an interface to tag network packets
  844. * based on a user-provided identifier for all traffic coming from
  845. * the tasks belonging to the related cgroup. See also the related
  846. * kernel documentation, available from the Linux sources in file
  847. * *Documentation/admin-guide/cgroup-v1/net_cls.rst*.
  848. *
  849. * The Linux kernel has two versions for cgroups: there are
  850. * cgroups v1 and cgroups v2. Both are available to users, who can
  851. * use a mixture of them, but note that the net_cls cgroup is for
  852. * cgroup v1 only. This makes it incompatible with BPF programs
  853. * run on cgroups, which is a cgroup-v2-only feature (a socket can
  854. * only hold data for one version of cgroups at a time).
  855. *
  856. * This helper is only available is the kernel was compiled with
  857. * the **CONFIG_CGROUP_NET_CLASSID** configuration option set to
  858. * "**y**" or to "**m**".
  859. * Return
  860. * The classid, or 0 for the default unconfigured classid.
  861. *
  862. * int bpf_skb_vlan_push(struct sk_buff *skb, __be16 vlan_proto, u16 vlan_tci)
  863. * Description
  864. * Push a *vlan_tci* (VLAN tag control information) of protocol
  865. * *vlan_proto* to the packet associated to *skb*, then update
  866. * the checksum. Note that if *vlan_proto* is different from
  867. * **ETH_P_8021Q** and **ETH_P_8021AD**, it is considered to
  868. * be **ETH_P_8021Q**.
  869. *
  870. * A call to this helper is susceptible to change the underlying
  871. * packet buffer. Therefore, at load time, all checks on pointers
  872. * previously done by the verifier are invalidated and must be
  873. * performed again, if the helper is used in combination with
  874. * direct packet access.
  875. * Return
  876. * 0 on success, or a negative error in case of failure.
  877. *
  878. * int bpf_skb_vlan_pop(struct sk_buff *skb)
  879. * Description
  880. * Pop a VLAN header from the packet associated to *skb*.
  881. *
  882. * A call to this helper is susceptible to change the underlying
  883. * packet buffer. Therefore, at load time, all checks on pointers
  884. * previously done by the verifier are invalidated and must be
  885. * performed again, if the helper is used in combination with
  886. * direct packet access.
  887. * Return
  888. * 0 on success, or a negative error in case of failure.
  889. *
  890. * int bpf_skb_get_tunnel_key(struct sk_buff *skb, struct bpf_tunnel_key *key, u32 size, u64 flags)
  891. * Description
  892. * Get tunnel metadata. This helper takes a pointer *key* to an
  893. * empty **struct bpf_tunnel_key** of **size**, that will be
  894. * filled with tunnel metadata for the packet associated to *skb*.
  895. * The *flags* can be set to **BPF_F_TUNINFO_IPV6**, which
  896. * indicates that the tunnel is based on IPv6 protocol instead of
  897. * IPv4.
  898. *
  899. * The **struct bpf_tunnel_key** is an object that generalizes the
  900. * principal parameters used by various tunneling protocols into a
  901. * single struct. This way, it can be used to easily make a
  902. * decision based on the contents of the encapsulation header,
  903. * "summarized" in this struct. In particular, it holds the IP
  904. * address of the remote end (IPv4 or IPv6, depending on the case)
  905. * in *key*\ **->remote_ipv4** or *key*\ **->remote_ipv6**. Also,
  906. * this struct exposes the *key*\ **->tunnel_id**, which is
  907. * generally mapped to a VNI (Virtual Network Identifier), making
  908. * it programmable together with the **bpf_skb_set_tunnel_key**\
  909. * () helper.
  910. *
  911. * Let's imagine that the following code is part of a program
  912. * attached to the TC ingress interface, on one end of a GRE
  913. * tunnel, and is supposed to filter out all messages coming from
  914. * remote ends with IPv4 address other than 10.0.0.1:
  915. *
  916. * ::
  917. *
  918. * int ret;
  919. * struct bpf_tunnel_key key = {};
  920. *
  921. * ret = bpf_skb_get_tunnel_key(skb, &key, sizeof(key), 0);
  922. * if (ret < 0)
  923. * return TC_ACT_SHOT; // drop packet
  924. *
  925. * if (key.remote_ipv4 != 0x0a000001)
  926. * return TC_ACT_SHOT; // drop packet
  927. *
  928. * return TC_ACT_OK; // accept packet
  929. *
  930. * This interface can also be used with all encapsulation devices
  931. * that can operate in "collect metadata" mode: instead of having
  932. * one network device per specific configuration, the "collect
  933. * metadata" mode only requires a single device where the
  934. * configuration can be extracted from this helper.
  935. *
  936. * This can be used together with various tunnels such as VXLan,
  937. * Geneve, GRE or IP in IP (IPIP).
  938. * Return
  939. * 0 on success, or a negative error in case of failure.
  940. *
  941. * int bpf_skb_set_tunnel_key(struct sk_buff *skb, struct bpf_tunnel_key *key, u32 size, u64 flags)
  942. * Description
  943. * Populate tunnel metadata for packet associated to *skb.* The
  944. * tunnel metadata is set to the contents of *key*, of *size*. The
  945. * *flags* can be set to a combination of the following values:
  946. *
  947. * **BPF_F_TUNINFO_IPV6**
  948. * Indicate that the tunnel is based on IPv6 protocol
  949. * instead of IPv4.
  950. * **BPF_F_ZERO_CSUM_TX**
  951. * For IPv4 packets, add a flag to tunnel metadata
  952. * indicating that checksum computation should be skipped
  953. * and checksum set to zeroes.
  954. * **BPF_F_DONT_FRAGMENT**
  955. * Add a flag to tunnel metadata indicating that the
  956. * packet should not be fragmented.
  957. * **BPF_F_SEQ_NUMBER**
  958. * Add a flag to tunnel metadata indicating that a
  959. * sequence number should be added to tunnel header before
  960. * sending the packet. This flag was added for GRE
  961. * encapsulation, but might be used with other protocols
  962. * as well in the future.
  963. *
  964. * Here is a typical usage on the transmit path:
  965. *
  966. * ::
  967. *
  968. * struct bpf_tunnel_key key;
  969. * populate key ...
  970. * bpf_skb_set_tunnel_key(skb, &key, sizeof(key), 0);
  971. * bpf_clone_redirect(skb, vxlan_dev_ifindex, 0);
  972. *
  973. * See also the description of the **bpf_skb_get_tunnel_key**\ ()
  974. * helper for additional information.
  975. * Return
  976. * 0 on success, or a negative error in case of failure.
  977. *
  978. * u64 bpf_perf_event_read(struct bpf_map *map, u64 flags)
  979. * Description
  980. * Read the value of a perf event counter. This helper relies on a
  981. * *map* of type **BPF_MAP_TYPE_PERF_EVENT_ARRAY**. The nature of
  982. * the perf event counter is selected when *map* is updated with
  983. * perf event file descriptors. The *map* is an array whose size
  984. * is the number of available CPUs, and each cell contains a value
  985. * relative to one CPU. The value to retrieve is indicated by
  986. * *flags*, that contains the index of the CPU to look up, masked
  987. * with **BPF_F_INDEX_MASK**. Alternatively, *flags* can be set to
  988. * **BPF_F_CURRENT_CPU** to indicate that the value for the
  989. * current CPU should be retrieved.
  990. *
  991. * Note that before Linux 4.13, only hardware perf event can be
  992. * retrieved.
  993. *
  994. * Also, be aware that the newer helper
  995. * **bpf_perf_event_read_value**\ () is recommended over
  996. * **bpf_perf_event_read**\ () in general. The latter has some ABI
  997. * quirks where error and counter value are used as a return code
  998. * (which is wrong to do since ranges may overlap). This issue is
  999. * fixed with **bpf_perf_event_read_value**\ (), which at the same
  1000. * time provides more features over the **bpf_perf_event_read**\
  1001. * () interface. Please refer to the description of
  1002. * **bpf_perf_event_read_value**\ () for details.
  1003. * Return
  1004. * The value of the perf event counter read from the map, or a
  1005. * negative error code in case of failure.
  1006. *
  1007. * int bpf_redirect(u32 ifindex, u64 flags)
  1008. * Description
  1009. * Redirect the packet to another net device of index *ifindex*.
  1010. * This helper is somewhat similar to **bpf_clone_redirect**\
  1011. * (), except that the packet is not cloned, which provides
  1012. * increased performance.
  1013. *
  1014. * Except for XDP, both ingress and egress interfaces can be used
  1015. * for redirection. The **BPF_F_INGRESS** value in *flags* is used
  1016. * to make the distinction (ingress path is selected if the flag
  1017. * is present, egress path otherwise). Currently, XDP only
  1018. * supports redirection to the egress interface, and accepts no
  1019. * flag at all.
  1020. *
  1021. * The same effect can also be attained with the more generic
  1022. * **bpf_redirect_map**\ (), which uses a BPF map to store the
  1023. * redirect target instead of providing it directly to the helper.
  1024. * Return
  1025. * For XDP, the helper returns **XDP_REDIRECT** on success or
  1026. * **XDP_ABORTED** on error. For other program types, the values
  1027. * are **TC_ACT_REDIRECT** on success or **TC_ACT_SHOT** on
  1028. * error.
  1029. *
  1030. * u32 bpf_get_route_realm(struct sk_buff *skb)
  1031. * Description
  1032. * Retrieve the realm or the route, that is to say the
  1033. * **tclassid** field of the destination for the *skb*. The
  1034. * indentifier retrieved is a user-provided tag, similar to the
  1035. * one used with the net_cls cgroup (see description for
  1036. * **bpf_get_cgroup_classid**\ () helper), but here this tag is
  1037. * held by a route (a destination entry), not by a task.
  1038. *
  1039. * Retrieving this identifier works with the clsact TC egress hook
  1040. * (see also **tc-bpf(8)**), or alternatively on conventional
  1041. * classful egress qdiscs, but not on TC ingress path. In case of
  1042. * clsact TC egress hook, this has the advantage that, internally,
  1043. * the destination entry has not been dropped yet in the transmit
  1044. * path. Therefore, the destination entry does not need to be
  1045. * artificially held via **netif_keep_dst**\ () for a classful
  1046. * qdisc until the *skb* is freed.
  1047. *
  1048. * This helper is available only if the kernel was compiled with
  1049. * **CONFIG_IP_ROUTE_CLASSID** configuration option.
  1050. * Return
  1051. * The realm of the route for the packet associated to *skb*, or 0
  1052. * if none was found.
  1053. *
  1054. * int bpf_perf_event_output(void *ctx, struct bpf_map *map, u64 flags, void *data, u64 size)
  1055. * Description
  1056. * Write raw *data* blob into a special BPF perf event held by
  1057. * *map* of type **BPF_MAP_TYPE_PERF_EVENT_ARRAY**. This perf
  1058. * event must have the following attributes: **PERF_SAMPLE_RAW**
  1059. * as **sample_type**, **PERF_TYPE_SOFTWARE** as **type**, and
  1060. * **PERF_COUNT_SW_BPF_OUTPUT** as **config**.
  1061. *
  1062. * The *flags* are used to indicate the index in *map* for which
  1063. * the value must be put, masked with **BPF_F_INDEX_MASK**.
  1064. * Alternatively, *flags* can be set to **BPF_F_CURRENT_CPU**
  1065. * to indicate that the index of the current CPU core should be
  1066. * used.
  1067. *
  1068. * The value to write, of *size*, is passed through eBPF stack and
  1069. * pointed by *data*.
  1070. *
  1071. * The context of the program *ctx* needs also be passed to the
  1072. * helper.
  1073. *
  1074. * On user space, a program willing to read the values needs to
  1075. * call **perf_event_open**\ () on the perf event (either for
  1076. * one or for all CPUs) and to store the file descriptor into the
  1077. * *map*. This must be done before the eBPF program can send data
  1078. * into it. An example is available in file
  1079. * *samples/bpf/trace_output_user.c* in the Linux kernel source
  1080. * tree (the eBPF program counterpart is in
  1081. * *samples/bpf/trace_output_kern.c*).
  1082. *
  1083. * **bpf_perf_event_output**\ () achieves better performance
  1084. * than **bpf_trace_printk**\ () for sharing data with user
  1085. * space, and is much better suitable for streaming data from eBPF
  1086. * programs.
  1087. *
  1088. * Note that this helper is not restricted to tracing use cases
  1089. * and can be used with programs attached to TC or XDP as well,
  1090. * where it allows for passing data to user space listeners. Data
  1091. * can be:
  1092. *
  1093. * * Only custom structs,
  1094. * * Only the packet payload, or
  1095. * * A combination of both.
  1096. * Return
  1097. * 0 on success, or a negative error in case of failure.
  1098. *
  1099. * int bpf_skb_load_bytes(const void *skb, u32 offset, void *to, u32 len)
  1100. * Description
  1101. * This helper was provided as an easy way to load data from a
  1102. * packet. It can be used to load *len* bytes from *offset* from
  1103. * the packet associated to *skb*, into the buffer pointed by
  1104. * *to*.
  1105. *
  1106. * Since Linux 4.7, usage of this helper has mostly been replaced
  1107. * by "direct packet access", enabling packet data to be
  1108. * manipulated with *skb*\ **->data** and *skb*\ **->data_end**
  1109. * pointing respectively to the first byte of packet data and to
  1110. * the byte after the last byte of packet data. However, it
  1111. * remains useful if one wishes to read large quantities of data
  1112. * at once from a packet into the eBPF stack.
  1113. * Return
  1114. * 0 on success, or a negative error in case of failure.
  1115. *
  1116. * int bpf_get_stackid(void *ctx, struct bpf_map *map, u64 flags)
  1117. * Description
  1118. * Walk a user or a kernel stack and return its id. To achieve
  1119. * this, the helper needs *ctx*, which is a pointer to the context
  1120. * on which the tracing program is executed, and a pointer to a
  1121. * *map* of type **BPF_MAP_TYPE_STACK_TRACE**.
  1122. *
  1123. * The last argument, *flags*, holds the number of stack frames to
  1124. * skip (from 0 to 255), masked with
  1125. * **BPF_F_SKIP_FIELD_MASK**. The next bits can be used to set
  1126. * a combination of the following flags:
  1127. *
  1128. * **BPF_F_USER_STACK**
  1129. * Collect a user space stack instead of a kernel stack.
  1130. * **BPF_F_FAST_STACK_CMP**
  1131. * Compare stacks by hash only.
  1132. * **BPF_F_REUSE_STACKID**
  1133. * If two different stacks hash into the same *stackid*,
  1134. * discard the old one.
  1135. *
  1136. * The stack id retrieved is a 32 bit long integer handle which
  1137. * can be further combined with other data (including other stack
  1138. * ids) and used as a key into maps. This can be useful for
  1139. * generating a variety of graphs (such as flame graphs or off-cpu
  1140. * graphs).
  1141. *
  1142. * For walking a stack, this helper is an improvement over
  1143. * **bpf_probe_read**\ (), which can be used with unrolled loops
  1144. * but is not efficient and consumes a lot of eBPF instructions.
  1145. * Instead, **bpf_get_stackid**\ () can collect up to
  1146. * **PERF_MAX_STACK_DEPTH** both kernel and user frames. Note that
  1147. * this limit can be controlled with the **sysctl** program, and
  1148. * that it should be manually increased in order to profile long
  1149. * user stacks (such as stacks for Java programs). To do so, use:
  1150. *
  1151. * ::
  1152. *
  1153. * # sysctl kernel.perf_event_max_stack=<new value>
  1154. * Return
  1155. * The positive or null stack id on success, or a negative error
  1156. * in case of failure.
  1157. *
  1158. * s64 bpf_csum_diff(__be32 *from, u32 from_size, __be32 *to, u32 to_size, __wsum seed)
  1159. * Description
  1160. * Compute a checksum difference, from the raw buffer pointed by
  1161. * *from*, of length *from_size* (that must be a multiple of 4),
  1162. * towards the raw buffer pointed by *to*, of size *to_size*
  1163. * (same remark). An optional *seed* can be added to the value
  1164. * (this can be cascaded, the seed may come from a previous call
  1165. * to the helper).
  1166. *
  1167. * This is flexible enough to be used in several ways:
  1168. *
  1169. * * With *from_size* == 0, *to_size* > 0 and *seed* set to
  1170. * checksum, it can be used when pushing new data.
  1171. * * With *from_size* > 0, *to_size* == 0 and *seed* set to
  1172. * checksum, it can be used when removing data from a packet.
  1173. * * With *from_size* > 0, *to_size* > 0 and *seed* set to 0, it
  1174. * can be used to compute a diff. Note that *from_size* and
  1175. * *to_size* do not need to be equal.
  1176. *
  1177. * This helper can be used in combination with
  1178. * **bpf_l3_csum_replace**\ () and **bpf_l4_csum_replace**\ (), to
  1179. * which one can feed in the difference computed with
  1180. * **bpf_csum_diff**\ ().
  1181. * Return
  1182. * The checksum result, or a negative error code in case of
  1183. * failure.
  1184. *
  1185. * int bpf_skb_get_tunnel_opt(struct sk_buff *skb, void *opt, u32 size)
  1186. * Description
  1187. * Retrieve tunnel options metadata for the packet associated to
  1188. * *skb*, and store the raw tunnel option data to the buffer *opt*
  1189. * of *size*.
  1190. *
  1191. * This helper can be used with encapsulation devices that can
  1192. * operate in "collect metadata" mode (please refer to the related
  1193. * note in the description of **bpf_skb_get_tunnel_key**\ () for
  1194. * more details). A particular example where this can be used is
  1195. * in combination with the Geneve encapsulation protocol, where it
  1196. * allows for pushing (with **bpf_skb_get_tunnel_opt**\ () helper)
  1197. * and retrieving arbitrary TLVs (Type-Length-Value headers) from
  1198. * the eBPF program. This allows for full customization of these
  1199. * headers.
  1200. * Return
  1201. * The size of the option data retrieved.
  1202. *
  1203. * int bpf_skb_set_tunnel_opt(struct sk_buff *skb, void *opt, u32 size)
  1204. * Description
  1205. * Set tunnel options metadata for the packet associated to *skb*
  1206. * to the option data contained in the raw buffer *opt* of *size*.
  1207. *
  1208. * See also the description of the **bpf_skb_get_tunnel_opt**\ ()
  1209. * helper for additional information.
  1210. * Return
  1211. * 0 on success, or a negative error in case of failure.
  1212. *
  1213. * int bpf_skb_change_proto(struct sk_buff *skb, __be16 proto, u64 flags)
  1214. * Description
  1215. * Change the protocol of the *skb* to *proto*. Currently
  1216. * supported are transition from IPv4 to IPv6, and from IPv6 to
  1217. * IPv4. The helper takes care of the groundwork for the
  1218. * transition, including resizing the socket buffer. The eBPF
  1219. * program is expected to fill the new headers, if any, via
  1220. * **skb_store_bytes**\ () and to recompute the checksums with
  1221. * **bpf_l3_csum_replace**\ () and **bpf_l4_csum_replace**\
  1222. * (). The main case for this helper is to perform NAT64
  1223. * operations out of an eBPF program.
  1224. *
  1225. * Internally, the GSO type is marked as dodgy so that headers are
  1226. * checked and segments are recalculated by the GSO/GRO engine.
  1227. * The size for GSO target is adapted as well.
  1228. *
  1229. * All values for *flags* are reserved for future usage, and must
  1230. * be left at zero.
  1231. *
  1232. * A call to this helper is susceptible to change the underlying
  1233. * packet buffer. Therefore, at load time, all checks on pointers
  1234. * previously done by the verifier are invalidated and must be
  1235. * performed again, if the helper is used in combination with
  1236. * direct packet access.
  1237. * Return
  1238. * 0 on success, or a negative error in case of failure.
  1239. *
  1240. * int bpf_skb_change_type(struct sk_buff *skb, u32 type)
  1241. * Description
  1242. * Change the packet type for the packet associated to *skb*. This
  1243. * comes down to setting *skb*\ **->pkt_type** to *type*, except
  1244. * the eBPF program does not have a write access to *skb*\
  1245. * **->pkt_type** beside this helper. Using a helper here allows
  1246. * for graceful handling of errors.
  1247. *
  1248. * The major use case is to change incoming *skb*s to
  1249. * **PACKET_HOST** in a programmatic way instead of having to
  1250. * recirculate via **redirect**\ (..., **BPF_F_INGRESS**), for
  1251. * example.
  1252. *
  1253. * Note that *type* only allows certain values. At this time, they
  1254. * are:
  1255. *
  1256. * **PACKET_HOST**
  1257. * Packet is for us.
  1258. * **PACKET_BROADCAST**
  1259. * Send packet to all.
  1260. * **PACKET_MULTICAST**
  1261. * Send packet to group.
  1262. * **PACKET_OTHERHOST**
  1263. * Send packet to someone else.
  1264. * Return
  1265. * 0 on success, or a negative error in case of failure.
  1266. *
  1267. * int bpf_skb_under_cgroup(struct sk_buff *skb, struct bpf_map *map, u32 index)
  1268. * Description
  1269. * Check whether *skb* is a descendant of the cgroup2 held by
  1270. * *map* of type **BPF_MAP_TYPE_CGROUP_ARRAY**, at *index*.
  1271. * Return
  1272. * The return value depends on the result of the test, and can be:
  1273. *
  1274. * * 0, if the *skb* failed the cgroup2 descendant test.
  1275. * * 1, if the *skb* succeeded the cgroup2 descendant test.
  1276. * * A negative error code, if an error occurred.
  1277. *
  1278. * u32 bpf_get_hash_recalc(struct sk_buff *skb)
  1279. * Description
  1280. * Retrieve the hash of the packet, *skb*\ **->hash**. If it is
  1281. * not set, in particular if the hash was cleared due to mangling,
  1282. * recompute this hash. Later accesses to the hash can be done
  1283. * directly with *skb*\ **->hash**.
  1284. *
  1285. * Calling **bpf_set_hash_invalid**\ (), changing a packet
  1286. * prototype with **bpf_skb_change_proto**\ (), or calling
  1287. * **bpf_skb_store_bytes**\ () with the
  1288. * **BPF_F_INVALIDATE_HASH** are actions susceptible to clear
  1289. * the hash and to trigger a new computation for the next call to
  1290. * **bpf_get_hash_recalc**\ ().
  1291. * Return
  1292. * The 32-bit hash.
  1293. *
  1294. * u64 bpf_get_current_task(void)
  1295. * Return
  1296. * A pointer to the current task struct.
  1297. *
  1298. * int bpf_probe_write_user(void *dst, const void *src, u32 len)
  1299. * Description
  1300. * Attempt in a safe way to write *len* bytes from the buffer
  1301. * *src* to *dst* in memory. It only works for threads that are in
  1302. * user context, and *dst* must be a valid user space address.
  1303. *
  1304. * This helper should not be used to implement any kind of
  1305. * security mechanism because of TOC-TOU attacks, but rather to
  1306. * debug, divert, and manipulate execution of semi-cooperative
  1307. * processes.
  1308. *
  1309. * Keep in mind that this feature is meant for experiments, and it
  1310. * has a risk of crashing the system and running programs.
  1311. * Therefore, when an eBPF program using this helper is attached,
  1312. * a warning including PID and process name is printed to kernel
  1313. * logs.
  1314. * Return
  1315. * 0 on success, or a negative error in case of failure.
  1316. *
  1317. * int bpf_current_task_under_cgroup(struct bpf_map *map, u32 index)
  1318. * Description
  1319. * Check whether the probe is being run is the context of a given
  1320. * subset of the cgroup2 hierarchy. The cgroup2 to test is held by
  1321. * *map* of type **BPF_MAP_TYPE_CGROUP_ARRAY**, at *index*.
  1322. * Return
  1323. * The return value depends on the result of the test, and can be:
  1324. *
  1325. * * 0, if the *skb* task belongs to the cgroup2.
  1326. * * 1, if the *skb* task does not belong to the cgroup2.
  1327. * * A negative error code, if an error occurred.
  1328. *
  1329. * int bpf_skb_change_tail(struct sk_buff *skb, u32 len, u64 flags)
  1330. * Description
  1331. * Resize (trim or grow) the packet associated to *skb* to the
  1332. * new *len*. The *flags* are reserved for future usage, and must
  1333. * be left at zero.
  1334. *
  1335. * The basic idea is that the helper performs the needed work to
  1336. * change the size of the packet, then the eBPF program rewrites
  1337. * the rest via helpers like **bpf_skb_store_bytes**\ (),
  1338. * **bpf_l3_csum_replace**\ (), **bpf_l3_csum_replace**\ ()
  1339. * and others. This helper is a slow path utility intended for
  1340. * replies with control messages. And because it is targeted for
  1341. * slow path, the helper itself can afford to be slow: it
  1342. * implicitly linearizes, unclones and drops offloads from the
  1343. * *skb*.
  1344. *
  1345. * A call to this helper is susceptible to change the underlying
  1346. * packet buffer. Therefore, at load time, all checks on pointers
  1347. * previously done by the verifier are invalidated and must be
  1348. * performed again, if the helper is used in combination with
  1349. * direct packet access.
  1350. * Return
  1351. * 0 on success, or a negative error in case of failure.
  1352. *
  1353. * int bpf_skb_pull_data(struct sk_buff *skb, u32 len)
  1354. * Description
  1355. * Pull in non-linear data in case the *skb* is non-linear and not
  1356. * all of *len* are part of the linear section. Make *len* bytes
  1357. * from *skb* readable and writable. If a zero value is passed for
  1358. * *len*, then the whole length of the *skb* is pulled.
  1359. *
  1360. * This helper is only needed for reading and writing with direct
  1361. * packet access.
  1362. *
  1363. * For direct packet access, testing that offsets to access
  1364. * are within packet boundaries (test on *skb*\ **->data_end**) is
  1365. * susceptible to fail if offsets are invalid, or if the requested
  1366. * data is in non-linear parts of the *skb*. On failure the
  1367. * program can just bail out, or in the case of a non-linear
  1368. * buffer, use a helper to make the data available. The
  1369. * **bpf_skb_load_bytes**\ () helper is a first solution to access
  1370. * the data. Another one consists in using **bpf_skb_pull_data**
  1371. * to pull in once the non-linear parts, then retesting and
  1372. * eventually access the data.
  1373. *
  1374. * At the same time, this also makes sure the *skb* is uncloned,
  1375. * which is a necessary condition for direct write. As this needs
  1376. * to be an invariant for the write part only, the verifier
  1377. * detects writes and adds a prologue that is calling
  1378. * **bpf_skb_pull_data()** to effectively unclone the *skb* from
  1379. * the very beginning in case it is indeed cloned.
  1380. *
  1381. * A call to this helper is susceptible to change the underlying
  1382. * packet buffer. Therefore, at load time, all checks on pointers
  1383. * previously done by the verifier are invalidated and must be
  1384. * performed again, if the helper is used in combination with
  1385. * direct packet access.
  1386. * Return
  1387. * 0 on success, or a negative error in case of failure.
  1388. *
  1389. * s64 bpf_csum_update(struct sk_buff *skb, __wsum csum)
  1390. * Description
  1391. * Add the checksum *csum* into *skb*\ **->csum** in case the
  1392. * driver has supplied a checksum for the entire packet into that
  1393. * field. Return an error otherwise. This helper is intended to be
  1394. * used in combination with **bpf_csum_diff**\ (), in particular
  1395. * when the checksum needs to be updated after data has been
  1396. * written into the packet through direct packet access.
  1397. * Return
  1398. * The checksum on success, or a negative error code in case of
  1399. * failure.
  1400. *
  1401. * void bpf_set_hash_invalid(struct sk_buff *skb)
  1402. * Description
  1403. * Invalidate the current *skb*\ **->hash**. It can be used after
  1404. * mangling on headers through direct packet access, in order to
  1405. * indicate that the hash is outdated and to trigger a
  1406. * recalculation the next time the kernel tries to access this
  1407. * hash or when the **bpf_get_hash_recalc**\ () helper is called.
  1408. *
  1409. * int bpf_get_numa_node_id(void)
  1410. * Description
  1411. * Return the id of the current NUMA node. The primary use case
  1412. * for this helper is the selection of sockets for the local NUMA
  1413. * node, when the program is attached to sockets using the
  1414. * **SO_ATTACH_REUSEPORT_EBPF** option (see also **socket(7)**),
  1415. * but the helper is also available to other eBPF program types,
  1416. * similarly to **bpf_get_smp_processor_id**\ ().
  1417. * Return
  1418. * The id of current NUMA node.
  1419. *
  1420. * int bpf_skb_change_head(struct sk_buff *skb, u32 len, u64 flags)
  1421. * Description
  1422. * Grows headroom of packet associated to *skb* and adjusts the
  1423. * offset of the MAC header accordingly, adding *len* bytes of
  1424. * space. It automatically extends and reallocates memory as
  1425. * required.
  1426. *
  1427. * This helper can be used on a layer 3 *skb* to push a MAC header
  1428. * for redirection into a layer 2 device.
  1429. *
  1430. * All values for *flags* are reserved for future usage, and must
  1431. * be left at zero.
  1432. *
  1433. * A call to this helper is susceptible to change the underlying
  1434. * packet buffer. Therefore, at load time, all checks on pointers
  1435. * previously done by the verifier are invalidated and must be
  1436. * performed again, if the helper is used in combination with
  1437. * direct packet access.
  1438. * Return
  1439. * 0 on success, or a negative error in case of failure.
  1440. *
  1441. * int bpf_xdp_adjust_head(struct xdp_buff *xdp_md, int delta)
  1442. * Description
  1443. * Adjust (move) *xdp_md*\ **->data** by *delta* bytes. Note that
  1444. * it is possible to use a negative value for *delta*. This helper
  1445. * can be used to prepare the packet for pushing or popping
  1446. * headers.
  1447. *
  1448. * A call to this helper is susceptible to change the underlying
  1449. * packet buffer. Therefore, at load time, all checks on pointers
  1450. * previously done by the verifier are invalidated and must be
  1451. * performed again, if the helper is used in combination with
  1452. * direct packet access.
  1453. * Return
  1454. * 0 on success, or a negative error in case of failure.
  1455. *
  1456. * int bpf_probe_read_str(void *dst, u32 size, const void *unsafe_ptr)
  1457. * Description
  1458. * Copy a NUL terminated string from an unsafe kernel address
  1459. * *unsafe_ptr* to *dst*. See bpf_probe_read_kernel_str() for
  1460. * more details.
  1461. *
  1462. * Generally, use bpf_probe_read_user_str() or bpf_probe_read_kernel_str()
  1463. * instead.
  1464. * Return
  1465. * On success, the strictly positive length of the string,
  1466. * including the trailing NUL character. On error, a negative
  1467. * value.
  1468. *
  1469. * u64 bpf_get_socket_cookie(struct sk_buff *skb)
  1470. * Description
  1471. * If the **struct sk_buff** pointed by *skb* has a known socket,
  1472. * retrieve the cookie (generated by the kernel) of this socket.
  1473. * If no cookie has been set yet, generate a new cookie. Once
  1474. * generated, the socket cookie remains stable for the life of the
  1475. * socket. This helper can be useful for monitoring per socket
  1476. * networking traffic statistics as it provides a global socket
  1477. * identifier that can be assumed unique.
  1478. * Return
  1479. * A 8-byte long non-decreasing number on success, or 0 if the
  1480. * socket field is missing inside *skb*.
  1481. *
  1482. * u64 bpf_get_socket_cookie(struct bpf_sock_addr *ctx)
  1483. * Description
  1484. * Equivalent to bpf_get_socket_cookie() helper that accepts
  1485. * *skb*, but gets socket from **struct bpf_sock_addr** context.
  1486. * Return
  1487. * A 8-byte long non-decreasing number.
  1488. *
  1489. * u64 bpf_get_socket_cookie(struct bpf_sock_ops *ctx)
  1490. * Description
  1491. * Equivalent to bpf_get_socket_cookie() helper that accepts
  1492. * *skb*, but gets socket from **struct bpf_sock_ops** context.
  1493. * Return
  1494. * A 8-byte long non-decreasing number.
  1495. *
  1496. * u32 bpf_get_socket_uid(struct sk_buff *skb)
  1497. * Return
  1498. * The owner UID of the socket associated to *skb*. If the socket
  1499. * is **NULL**, or if it is not a full socket (i.e. if it is a
  1500. * time-wait or a request socket instead), **overflowuid** value
  1501. * is returned (note that **overflowuid** might also be the actual
  1502. * UID value for the socket).
  1503. *
  1504. * u32 bpf_set_hash(struct sk_buff *skb, u32 hash)
  1505. * Description
  1506. * Set the full hash for *skb* (set the field *skb*\ **->hash**)
  1507. * to value *hash*.
  1508. * Return
  1509. * 0
  1510. *
  1511. * int bpf_setsockopt(struct bpf_sock_ops *bpf_socket, int level, int optname, void *optval, int optlen)
  1512. * Description
  1513. * Emulate a call to **setsockopt()** on the socket associated to
  1514. * *bpf_socket*, which must be a full socket. The *level* at
  1515. * which the option resides and the name *optname* of the option
  1516. * must be specified, see **setsockopt(2)** for more information.
  1517. * The option value of length *optlen* is pointed by *optval*.
  1518. *
  1519. * This helper actually implements a subset of **setsockopt()**.
  1520. * It supports the following *level*\ s:
  1521. *
  1522. * * **SOL_SOCKET**, which supports the following *optname*\ s:
  1523. * **SO_RCVBUF**, **SO_SNDBUF**, **SO_MAX_PACING_RATE**,
  1524. * **SO_PRIORITY**, **SO_RCVLOWAT**, **SO_MARK**.
  1525. * * **IPPROTO_TCP**, which supports the following *optname*\ s:
  1526. * **TCP_CONGESTION**, **TCP_BPF_IW**,
  1527. * **TCP_BPF_SNDCWND_CLAMP**.
  1528. * * **IPPROTO_IP**, which supports *optname* **IP_TOS**.
  1529. * * **IPPROTO_IPV6**, which supports *optname* **IPV6_TCLASS**.
  1530. * Return
  1531. * 0 on success, or a negative error in case of failure.
  1532. *
  1533. * int bpf_skb_adjust_room(struct sk_buff *skb, s32 len_diff, u32 mode, u64 flags)
  1534. * Description
  1535. * Grow or shrink the room for data in the packet associated to
  1536. * *skb* by *len_diff*, and according to the selected *mode*.
  1537. *
  1538. * There are two supported modes at this time:
  1539. *
  1540. * * **BPF_ADJ_ROOM_MAC**: Adjust room at the mac layer
  1541. * (room space is added or removed below the layer 2 header).
  1542. *
  1543. * * **BPF_ADJ_ROOM_NET**: Adjust room at the network layer
  1544. * (room space is added or removed below the layer 3 header).
  1545. *
  1546. * The following flags are supported at this time:
  1547. *
  1548. * * **BPF_F_ADJ_ROOM_FIXED_GSO**: Do not adjust gso_size.
  1549. * Adjusting mss in this way is not allowed for datagrams.
  1550. *
  1551. * * **BPF_F_ADJ_ROOM_ENCAP_L3_IPV4**,
  1552. * **BPF_F_ADJ_ROOM_ENCAP_L3_IPV6**:
  1553. * Any new space is reserved to hold a tunnel header.
  1554. * Configure skb offsets and other fields accordingly.
  1555. *
  1556. * * **BPF_F_ADJ_ROOM_ENCAP_L4_GRE**,
  1557. * **BPF_F_ADJ_ROOM_ENCAP_L4_UDP**:
  1558. * Use with ENCAP_L3 flags to further specify the tunnel type.
  1559. *
  1560. * * **BPF_F_ADJ_ROOM_ENCAP_L2**\ (*len*):
  1561. * Use with ENCAP_L3/L4 flags to further specify the tunnel
  1562. * type; *len* is the length of the inner MAC header.
  1563. *
  1564. * A call to this helper is susceptible to change the underlying
  1565. * packet buffer. Therefore, at load time, all checks on pointers
  1566. * previously done by the verifier are invalidated and must be
  1567. * performed again, if the helper is used in combination with
  1568. * direct packet access.
  1569. * Return
  1570. * 0 on success, or a negative error in case of failure.
  1571. *
  1572. * int bpf_redirect_map(struct bpf_map *map, u32 key, u64 flags)
  1573. * Description
  1574. * Redirect the packet to the endpoint referenced by *map* at
  1575. * index *key*. Depending on its type, this *map* can contain
  1576. * references to net devices (for forwarding packets through other
  1577. * ports), or to CPUs (for redirecting XDP frames to another CPU;
  1578. * but this is only implemented for native XDP (with driver
  1579. * support) as of this writing).
  1580. *
  1581. * The lower two bits of *flags* are used as the return code if
  1582. * the map lookup fails. This is so that the return value can be
  1583. * one of the XDP program return codes up to XDP_TX, as chosen by
  1584. * the caller. Any higher bits in the *flags* argument must be
  1585. * unset.
  1586. *
  1587. * See also bpf_redirect(), which only supports redirecting to an
  1588. * ifindex, but doesn't require a map to do so.
  1589. * Return
  1590. * **XDP_REDIRECT** on success, or the value of the two lower bits
  1591. * of the *flags* argument on error.
  1592. *
  1593. * int bpf_sk_redirect_map(struct sk_buff *skb, struct bpf_map *map, u32 key, u64 flags)
  1594. * Description
  1595. * Redirect the packet to the socket referenced by *map* (of type
  1596. * **BPF_MAP_TYPE_SOCKMAP**) at index *key*. Both ingress and
  1597. * egress interfaces can be used for redirection. The
  1598. * **BPF_F_INGRESS** value in *flags* is used to make the
  1599. * distinction (ingress path is selected if the flag is present,
  1600. * egress path otherwise). This is the only flag supported for now.
  1601. * Return
  1602. * **SK_PASS** on success, or **SK_DROP** on error.
  1603. *
  1604. * int bpf_sock_map_update(struct bpf_sock_ops *skops, struct bpf_map *map, void *key, u64 flags)
  1605. * Description
  1606. * Add an entry to, or update a *map* referencing sockets. The
  1607. * *skops* is used as a new value for the entry associated to
  1608. * *key*. *flags* is one of:
  1609. *
  1610. * **BPF_NOEXIST**
  1611. * The entry for *key* must not exist in the map.
  1612. * **BPF_EXIST**
  1613. * The entry for *key* must already exist in the map.
  1614. * **BPF_ANY**
  1615. * No condition on the existence of the entry for *key*.
  1616. *
  1617. * If the *map* has eBPF programs (parser and verdict), those will
  1618. * be inherited by the socket being added. If the socket is
  1619. * already attached to eBPF programs, this results in an error.
  1620. * Return
  1621. * 0 on success, or a negative error in case of failure.
  1622. *
  1623. * int bpf_xdp_adjust_meta(struct xdp_buff *xdp_md, int delta)
  1624. * Description
  1625. * Adjust the address pointed by *xdp_md*\ **->data_meta** by
  1626. * *delta* (which can be positive or negative). Note that this
  1627. * operation modifies the address stored in *xdp_md*\ **->data**,
  1628. * so the latter must be loaded only after the helper has been
  1629. * called.
  1630. *
  1631. * The use of *xdp_md*\ **->data_meta** is optional and programs
  1632. * are not required to use it. The rationale is that when the
  1633. * packet is processed with XDP (e.g. as DoS filter), it is
  1634. * possible to push further meta data along with it before passing
  1635. * to the stack, and to give the guarantee that an ingress eBPF
  1636. * program attached as a TC classifier on the same device can pick
  1637. * this up for further post-processing. Since TC works with socket
  1638. * buffers, it remains possible to set from XDP the **mark** or
  1639. * **priority** pointers, or other pointers for the socket buffer.
  1640. * Having this scratch space generic and programmable allows for
  1641. * more flexibility as the user is free to store whatever meta
  1642. * data they need.
  1643. *
  1644. * A call to this helper is susceptible to change the underlying
  1645. * packet buffer. Therefore, at load time, all checks on pointers
  1646. * previously done by the verifier are invalidated and must be
  1647. * performed again, if the helper is used in combination with
  1648. * direct packet access.
  1649. * Return
  1650. * 0 on success, or a negative error in case of failure.
  1651. *
  1652. * int bpf_perf_event_read_value(struct bpf_map *map, u64 flags, struct bpf_perf_event_value *buf, u32 buf_size)
  1653. * Description
  1654. * Read the value of a perf event counter, and store it into *buf*
  1655. * of size *buf_size*. This helper relies on a *map* of type
  1656. * **BPF_MAP_TYPE_PERF_EVENT_ARRAY**. The nature of the perf event
  1657. * counter is selected when *map* is updated with perf event file
  1658. * descriptors. The *map* is an array whose size is the number of
  1659. * available CPUs, and each cell contains a value relative to one
  1660. * CPU. The value to retrieve is indicated by *flags*, that
  1661. * contains the index of the CPU to look up, masked with
  1662. * **BPF_F_INDEX_MASK**. Alternatively, *flags* can be set to
  1663. * **BPF_F_CURRENT_CPU** to indicate that the value for the
  1664. * current CPU should be retrieved.
  1665. *
  1666. * This helper behaves in a way close to
  1667. * **bpf_perf_event_read**\ () helper, save that instead of
  1668. * just returning the value observed, it fills the *buf*
  1669. * structure. This allows for additional data to be retrieved: in
  1670. * particular, the enabled and running times (in *buf*\
  1671. * **->enabled** and *buf*\ **->running**, respectively) are
  1672. * copied. In general, **bpf_perf_event_read_value**\ () is
  1673. * recommended over **bpf_perf_event_read**\ (), which has some
  1674. * ABI issues and provides fewer functionalities.
  1675. *
  1676. * These values are interesting, because hardware PMU (Performance
  1677. * Monitoring Unit) counters are limited resources. When there are
  1678. * more PMU based perf events opened than available counters,
  1679. * kernel will multiplex these events so each event gets certain
  1680. * percentage (but not all) of the PMU time. In case that
  1681. * multiplexing happens, the number of samples or counter value
  1682. * will not reflect the case compared to when no multiplexing
  1683. * occurs. This makes comparison between different runs difficult.
  1684. * Typically, the counter value should be normalized before
  1685. * comparing to other experiments. The usual normalization is done
  1686. * as follows.
  1687. *
  1688. * ::
  1689. *
  1690. * normalized_counter = counter * t_enabled / t_running
  1691. *
  1692. * Where t_enabled is the time enabled for event and t_running is
  1693. * the time running for event since last normalization. The
  1694. * enabled and running times are accumulated since the perf event
  1695. * open. To achieve scaling factor between two invocations of an
  1696. * eBPF program, users can can use CPU id as the key (which is
  1697. * typical for perf array usage model) to remember the previous
  1698. * value and do the calculation inside the eBPF program.
  1699. * Return
  1700. * 0 on success, or a negative error in case of failure.
  1701. *
  1702. * int bpf_perf_prog_read_value(struct bpf_perf_event_data *ctx, struct bpf_perf_event_value *buf, u32 buf_size)
  1703. * Description
  1704. * For en eBPF program attached to a perf event, retrieve the
  1705. * value of the event counter associated to *ctx* and store it in
  1706. * the structure pointed by *buf* and of size *buf_size*. Enabled
  1707. * and running times are also stored in the structure (see
  1708. * description of helper **bpf_perf_event_read_value**\ () for
  1709. * more details).
  1710. * Return
  1711. * 0 on success, or a negative error in case of failure.
  1712. *
  1713. * int bpf_getsockopt(struct bpf_sock_ops *bpf_socket, int level, int optname, void *optval, int optlen)
  1714. * Description
  1715. * Emulate a call to **getsockopt()** on the socket associated to
  1716. * *bpf_socket*, which must be a full socket. The *level* at
  1717. * which the option resides and the name *optname* of the option
  1718. * must be specified, see **getsockopt(2)** for more information.
  1719. * The retrieved value is stored in the structure pointed by
  1720. * *opval* and of length *optlen*.
  1721. *
  1722. * This helper actually implements a subset of **getsockopt()**.
  1723. * It supports the following *level*\ s:
  1724. *
  1725. * * **IPPROTO_TCP**, which supports *optname*
  1726. * **TCP_CONGESTION**.
  1727. * * **IPPROTO_IP**, which supports *optname* **IP_TOS**.
  1728. * * **IPPROTO_IPV6**, which supports *optname* **IPV6_TCLASS**.
  1729. * Return
  1730. * 0 on success, or a negative error in case of failure.
  1731. *
  1732. * int bpf_override_return(struct pt_regs *regs, u64 rc)
  1733. * Description
  1734. * Used for error injection, this helper uses kprobes to override
  1735. * the return value of the probed function, and to set it to *rc*.
  1736. * The first argument is the context *regs* on which the kprobe
  1737. * works.
  1738. *
  1739. * This helper works by setting setting the PC (program counter)
  1740. * to an override function which is run in place of the original
  1741. * probed function. This means the probed function is not run at
  1742. * all. The replacement function just returns with the required
  1743. * value.
  1744. *
  1745. * This helper has security implications, and thus is subject to
  1746. * restrictions. It is only available if the kernel was compiled
  1747. * with the **CONFIG_BPF_KPROBE_OVERRIDE** configuration
  1748. * option, and in this case it only works on functions tagged with
  1749. * **ALLOW_ERROR_INJECTION** in the kernel code.
  1750. *
  1751. * Also, the helper is only available for the architectures having
  1752. * the CONFIG_FUNCTION_ERROR_INJECTION option. As of this writing,
  1753. * x86 architecture is the only one to support this feature.
  1754. * Return
  1755. * 0
  1756. *
  1757. * int bpf_sock_ops_cb_flags_set(struct bpf_sock_ops *bpf_sock, int argval)
  1758. * Description
  1759. * Attempt to set the value of the **bpf_sock_ops_cb_flags** field
  1760. * for the full TCP socket associated to *bpf_sock_ops* to
  1761. * *argval*.
  1762. *
  1763. * The primary use of this field is to determine if there should
  1764. * be calls to eBPF programs of type
  1765. * **BPF_PROG_TYPE_SOCK_OPS** at various points in the TCP
  1766. * code. A program of the same type can change its value, per
  1767. * connection and as necessary, when the connection is
  1768. * established. This field is directly accessible for reading, but
  1769. * this helper must be used for updates in order to return an
  1770. * error if an eBPF program tries to set a callback that is not
  1771. * supported in the current kernel.
  1772. *
  1773. * *argval* is a flag array which can combine these flags:
  1774. *
  1775. * * **BPF_SOCK_OPS_RTO_CB_FLAG** (retransmission time out)
  1776. * * **BPF_SOCK_OPS_RETRANS_CB_FLAG** (retransmission)
  1777. * * **BPF_SOCK_OPS_STATE_CB_FLAG** (TCP state change)
  1778. * * **BPF_SOCK_OPS_RTT_CB_FLAG** (every RTT)
  1779. *
  1780. * Therefore, this function can be used to clear a callback flag by
  1781. * setting the appropriate bit to zero. e.g. to disable the RTO
  1782. * callback:
  1783. *
  1784. * **bpf_sock_ops_cb_flags_set(bpf_sock,**
  1785. * **bpf_sock->bpf_sock_ops_cb_flags & ~BPF_SOCK_OPS_RTO_CB_FLAG)**
  1786. *
  1787. * Here are some examples of where one could call such eBPF
  1788. * program:
  1789. *
  1790. * * When RTO fires.
  1791. * * When a packet is retransmitted.
  1792. * * When the connection terminates.
  1793. * * When a packet is sent.
  1794. * * When a packet is received.
  1795. * Return
  1796. * Code **-EINVAL** if the socket is not a full TCP socket;
  1797. * otherwise, a positive number containing the bits that could not
  1798. * be set is returned (which comes down to 0 if all bits were set
  1799. * as required).
  1800. *
  1801. * int bpf_msg_redirect_map(struct sk_msg_buff *msg, struct bpf_map *map, u32 key, u64 flags)
  1802. * Description
  1803. * This helper is used in programs implementing policies at the
  1804. * socket level. If the message *msg* is allowed to pass (i.e. if
  1805. * the verdict eBPF program returns **SK_PASS**), redirect it to
  1806. * the socket referenced by *map* (of type
  1807. * **BPF_MAP_TYPE_SOCKMAP**) at index *key*. Both ingress and
  1808. * egress interfaces can be used for redirection. The
  1809. * **BPF_F_INGRESS** value in *flags* is used to make the
  1810. * distinction (ingress path is selected if the flag is present,
  1811. * egress path otherwise). This is the only flag supported for now.
  1812. * Return
  1813. * **SK_PASS** on success, or **SK_DROP** on error.
  1814. *
  1815. * int bpf_msg_apply_bytes(struct sk_msg_buff *msg, u32 bytes)
  1816. * Description
  1817. * For socket policies, apply the verdict of the eBPF program to
  1818. * the next *bytes* (number of bytes) of message *msg*.
  1819. *
  1820. * For example, this helper can be used in the following cases:
  1821. *
  1822. * * A single **sendmsg**\ () or **sendfile**\ () system call
  1823. * contains multiple logical messages that the eBPF program is
  1824. * supposed to read and for which it should apply a verdict.
  1825. * * An eBPF program only cares to read the first *bytes* of a
  1826. * *msg*. If the message has a large payload, then setting up
  1827. * and calling the eBPF program repeatedly for all bytes, even
  1828. * though the verdict is already known, would create unnecessary
  1829. * overhead.
  1830. *
  1831. * When called from within an eBPF program, the helper sets a
  1832. * counter internal to the BPF infrastructure, that is used to
  1833. * apply the last verdict to the next *bytes*. If *bytes* is
  1834. * smaller than the current data being processed from a
  1835. * **sendmsg**\ () or **sendfile**\ () system call, the first
  1836. * *bytes* will be sent and the eBPF program will be re-run with
  1837. * the pointer for start of data pointing to byte number *bytes*
  1838. * **+ 1**. If *bytes* is larger than the current data being
  1839. * processed, then the eBPF verdict will be applied to multiple
  1840. * **sendmsg**\ () or **sendfile**\ () calls until *bytes* are
  1841. * consumed.
  1842. *
  1843. * Note that if a socket closes with the internal counter holding
  1844. * a non-zero value, this is not a problem because data is not
  1845. * being buffered for *bytes* and is sent as it is received.
  1846. * Return
  1847. * 0
  1848. *
  1849. * int bpf_msg_cork_bytes(struct sk_msg_buff *msg, u32 bytes)
  1850. * Description
  1851. * For socket policies, prevent the execution of the verdict eBPF
  1852. * program for message *msg* until *bytes* (byte number) have been
  1853. * accumulated.
  1854. *
  1855. * This can be used when one needs a specific number of bytes
  1856. * before a verdict can be assigned, even if the data spans
  1857. * multiple **sendmsg**\ () or **sendfile**\ () calls. The extreme
  1858. * case would be a user calling **sendmsg**\ () repeatedly with
  1859. * 1-byte long message segments. Obviously, this is bad for
  1860. * performance, but it is still valid. If the eBPF program needs
  1861. * *bytes* bytes to validate a header, this helper can be used to
  1862. * prevent the eBPF program to be called again until *bytes* have
  1863. * been accumulated.
  1864. * Return
  1865. * 0
  1866. *
  1867. * int bpf_msg_pull_data(struct sk_msg_buff *msg, u32 start, u32 end, u64 flags)
  1868. * Description
  1869. * For socket policies, pull in non-linear data from user space
  1870. * for *msg* and set pointers *msg*\ **->data** and *msg*\
  1871. * **->data_end** to *start* and *end* bytes offsets into *msg*,
  1872. * respectively.
  1873. *
  1874. * If a program of type **BPF_PROG_TYPE_SK_MSG** is run on a
  1875. * *msg* it can only parse data that the (**data**, **data_end**)
  1876. * pointers have already consumed. For **sendmsg**\ () hooks this
  1877. * is likely the first scatterlist element. But for calls relying
  1878. * on the **sendpage** handler (e.g. **sendfile**\ ()) this will
  1879. * be the range (**0**, **0**) because the data is shared with
  1880. * user space and by default the objective is to avoid allowing
  1881. * user space to modify data while (or after) eBPF verdict is
  1882. * being decided. This helper can be used to pull in data and to
  1883. * set the start and end pointer to given values. Data will be
  1884. * copied if necessary (i.e. if data was not linear and if start
  1885. * and end pointers do not point to the same chunk).
  1886. *
  1887. * A call to this helper is susceptible to change the underlying
  1888. * packet buffer. Therefore, at load time, all checks on pointers
  1889. * previously done by the verifier are invalidated and must be
  1890. * performed again, if the helper is used in combination with
  1891. * direct packet access.
  1892. *
  1893. * All values for *flags* are reserved for future usage, and must
  1894. * be left at zero.
  1895. * Return
  1896. * 0 on success, or a negative error in case of failure.
  1897. *
  1898. * int bpf_bind(struct bpf_sock_addr *ctx, struct sockaddr *addr, int addr_len)
  1899. * Description
  1900. * Bind the socket associated to *ctx* to the address pointed by
  1901. * *addr*, of length *addr_len*. This allows for making outgoing
  1902. * connection from the desired IP address, which can be useful for
  1903. * example when all processes inside a cgroup should use one
  1904. * single IP address on a host that has multiple IP configured.
  1905. *
  1906. * This helper works for IPv4 and IPv6, TCP and UDP sockets. The
  1907. * domain (*addr*\ **->sa_family**) must be **AF_INET** (or
  1908. * **AF_INET6**). Looking for a free port to bind to can be
  1909. * expensive, therefore binding to port is not permitted by the
  1910. * helper: *addr*\ **->sin_port** (or **sin6_port**, respectively)
  1911. * must be set to zero.
  1912. * Return
  1913. * 0 on success, or a negative error in case of failure.
  1914. *
  1915. * int bpf_xdp_adjust_tail(struct xdp_buff *xdp_md, int delta)
  1916. * Description
  1917. * Adjust (move) *xdp_md*\ **->data_end** by *delta* bytes. It is
  1918. * only possible to shrink the packet as of this writing,
  1919. * therefore *delta* must be a negative integer.
  1920. *
  1921. * A call to this helper is susceptible to change the underlying
  1922. * packet buffer. Therefore, at load time, all checks on pointers
  1923. * previously done by the verifier are invalidated and must be
  1924. * performed again, if the helper is used in combination with
  1925. * direct packet access.
  1926. * Return
  1927. * 0 on success, or a negative error in case of failure.
  1928. *
  1929. * int bpf_skb_get_xfrm_state(struct sk_buff *skb, u32 index, struct bpf_xfrm_state *xfrm_state, u32 size, u64 flags)
  1930. * Description
  1931. * Retrieve the XFRM state (IP transform framework, see also
  1932. * **ip-xfrm(8)**) at *index* in XFRM "security path" for *skb*.
  1933. *
  1934. * The retrieved value is stored in the **struct bpf_xfrm_state**
  1935. * pointed by *xfrm_state* and of length *size*.
  1936. *
  1937. * All values for *flags* are reserved for future usage, and must
  1938. * be left at zero.
  1939. *
  1940. * This helper is available only if the kernel was compiled with
  1941. * **CONFIG_XFRM** configuration option.
  1942. * Return
  1943. * 0 on success, or a negative error in case of failure.
  1944. *
  1945. * int bpf_get_stack(void *ctx, void *buf, u32 size, u64 flags)
  1946. * Description
  1947. * Return a user or a kernel stack in bpf program provided buffer.
  1948. * To achieve this, the helper needs *ctx*, which is a pointer
  1949. * to the context on which the tracing program is executed.
  1950. * To store the stacktrace, the bpf program provides *buf* with
  1951. * a nonnegative *size*.
  1952. *
  1953. * The last argument, *flags*, holds the number of stack frames to
  1954. * skip (from 0 to 255), masked with
  1955. * **BPF_F_SKIP_FIELD_MASK**. The next bits can be used to set
  1956. * the following flags:
  1957. *
  1958. * **BPF_F_USER_STACK**
  1959. * Collect a user space stack instead of a kernel stack.
  1960. * **BPF_F_USER_BUILD_ID**
  1961. * Collect buildid+offset instead of ips for user stack,
  1962. * only valid if **BPF_F_USER_STACK** is also specified.
  1963. *
  1964. * **bpf_get_stack**\ () can collect up to
  1965. * **PERF_MAX_STACK_DEPTH** both kernel and user frames, subject
  1966. * to sufficient large buffer size. Note that
  1967. * this limit can be controlled with the **sysctl** program, and
  1968. * that it should be manually increased in order to profile long
  1969. * user stacks (such as stacks for Java programs). To do so, use:
  1970. *
  1971. * ::
  1972. *
  1973. * # sysctl kernel.perf_event_max_stack=<new value>
  1974. * Return
  1975. * A non-negative value equal to or less than *size* on success,
  1976. * or a negative error in case of failure.
  1977. *
  1978. * int bpf_skb_load_bytes_relative(const void *skb, u32 offset, void *to, u32 len, u32 start_header)
  1979. * Description
  1980. * This helper is similar to **bpf_skb_load_bytes**\ () in that
  1981. * it provides an easy way to load *len* bytes from *offset*
  1982. * from the packet associated to *skb*, into the buffer pointed
  1983. * by *to*. The difference to **bpf_skb_load_bytes**\ () is that
  1984. * a fifth argument *start_header* exists in order to select a
  1985. * base offset to start from. *start_header* can be one of:
  1986. *
  1987. * **BPF_HDR_START_MAC**
  1988. * Base offset to load data from is *skb*'s mac header.
  1989. * **BPF_HDR_START_NET**
  1990. * Base offset to load data from is *skb*'s network header.
  1991. *
  1992. * In general, "direct packet access" is the preferred method to
  1993. * access packet data, however, this helper is in particular useful
  1994. * in socket filters where *skb*\ **->data** does not always point
  1995. * to the start of the mac header and where "direct packet access"
  1996. * is not available.
  1997. * Return
  1998. * 0 on success, or a negative error in case of failure.
  1999. *
  2000. * int bpf_fib_lookup(void *ctx, struct bpf_fib_lookup *params, int plen, u32 flags)
  2001. * Description
  2002. * Do FIB lookup in kernel tables using parameters in *params*.
  2003. * If lookup is successful and result shows packet is to be
  2004. * forwarded, the neighbor tables are searched for the nexthop.
  2005. * If successful (ie., FIB lookup shows forwarding and nexthop
  2006. * is resolved), the nexthop address is returned in ipv4_dst
  2007. * or ipv6_dst based on family, smac is set to mac address of
  2008. * egress device, dmac is set to nexthop mac address, rt_metric
  2009. * is set to metric from route (IPv4/IPv6 only), and ifindex
  2010. * is set to the device index of the nexthop from the FIB lookup.
  2011. *
  2012. * *plen* argument is the size of the passed in struct.
  2013. * *flags* argument can be a combination of one or more of the
  2014. * following values:
  2015. *
  2016. * **BPF_FIB_LOOKUP_DIRECT**
  2017. * Do a direct table lookup vs full lookup using FIB
  2018. * rules.
  2019. * **BPF_FIB_LOOKUP_OUTPUT**
  2020. * Perform lookup from an egress perspective (default is
  2021. * ingress).
  2022. *
  2023. * *ctx* is either **struct xdp_md** for XDP programs or
  2024. * **struct sk_buff** tc cls_act programs.
  2025. * Return
  2026. * * < 0 if any input argument is invalid
  2027. * * 0 on success (packet is forwarded, nexthop neighbor exists)
  2028. * * > 0 one of **BPF_FIB_LKUP_RET_** codes explaining why the
  2029. * packet is not forwarded or needs assist from full stack
  2030. *
  2031. * int bpf_sock_hash_update(struct bpf_sock_ops *skops, struct bpf_map *map, void *key, u64 flags)
  2032. * Description
  2033. * Add an entry to, or update a sockhash *map* referencing sockets.
  2034. * The *skops* is used as a new value for the entry associated to
  2035. * *key*. *flags* is one of:
  2036. *
  2037. * **BPF_NOEXIST**
  2038. * The entry for *key* must not exist in the map.
  2039. * **BPF_EXIST**
  2040. * The entry for *key* must already exist in the map.
  2041. * **BPF_ANY**
  2042. * No condition on the existence of the entry for *key*.
  2043. *
  2044. * If the *map* has eBPF programs (parser and verdict), those will
  2045. * be inherited by the socket being added. If the socket is
  2046. * already attached to eBPF programs, this results in an error.
  2047. * Return
  2048. * 0 on success, or a negative error in case of failure.
  2049. *
  2050. * int bpf_msg_redirect_hash(struct sk_msg_buff *msg, struct bpf_map *map, void *key, u64 flags)
  2051. * Description
  2052. * This helper is used in programs implementing policies at the
  2053. * socket level. If the message *msg* is allowed to pass (i.e. if
  2054. * the verdict eBPF program returns **SK_PASS**), redirect it to
  2055. * the socket referenced by *map* (of type
  2056. * **BPF_MAP_TYPE_SOCKHASH**) using hash *key*. Both ingress and
  2057. * egress interfaces can be used for redirection. The
  2058. * **BPF_F_INGRESS** value in *flags* is used to make the
  2059. * distinction (ingress path is selected if the flag is present,
  2060. * egress path otherwise). This is the only flag supported for now.
  2061. * Return
  2062. * **SK_PASS** on success, or **SK_DROP** on error.
  2063. *
  2064. * int bpf_sk_redirect_hash(struct sk_buff *skb, struct bpf_map *map, void *key, u64 flags)
  2065. * Description
  2066. * This helper is used in programs implementing policies at the
  2067. * skb socket level. If the sk_buff *skb* is allowed to pass (i.e.
  2068. * if the verdeict eBPF program returns **SK_PASS**), redirect it
  2069. * to the socket referenced by *map* (of type
  2070. * **BPF_MAP_TYPE_SOCKHASH**) using hash *key*. Both ingress and
  2071. * egress interfaces can be used for redirection. The
  2072. * **BPF_F_INGRESS** value in *flags* is used to make the
  2073. * distinction (ingress path is selected if the flag is present,
  2074. * egress otherwise). This is the only flag supported for now.
  2075. * Return
  2076. * **SK_PASS** on success, or **SK_DROP** on error.
  2077. *
  2078. * int bpf_lwt_push_encap(struct sk_buff *skb, u32 type, void *hdr, u32 len)
  2079. * Description
  2080. * Encapsulate the packet associated to *skb* within a Layer 3
  2081. * protocol header. This header is provided in the buffer at
  2082. * address *hdr*, with *len* its size in bytes. *type* indicates
  2083. * the protocol of the header and can be one of:
  2084. *
  2085. * **BPF_LWT_ENCAP_SEG6**
  2086. * IPv6 encapsulation with Segment Routing Header
  2087. * (**struct ipv6_sr_hdr**). *hdr* only contains the SRH,
  2088. * the IPv6 header is computed by the kernel.
  2089. * **BPF_LWT_ENCAP_SEG6_INLINE**
  2090. * Only works if *skb* contains an IPv6 packet. Insert a
  2091. * Segment Routing Header (**struct ipv6_sr_hdr**) inside
  2092. * the IPv6 header.
  2093. * **BPF_LWT_ENCAP_IP**
  2094. * IP encapsulation (GRE/GUE/IPIP/etc). The outer header
  2095. * must be IPv4 or IPv6, followed by zero or more
  2096. * additional headers, up to **LWT_BPF_MAX_HEADROOM**
  2097. * total bytes in all prepended headers. Please note that
  2098. * if **skb_is_gso**\ (*skb*) is true, no more than two
  2099. * headers can be prepended, and the inner header, if
  2100. * present, should be either GRE or UDP/GUE.
  2101. *
  2102. * **BPF_LWT_ENCAP_SEG6**\ \* types can be called by BPF programs
  2103. * of type **BPF_PROG_TYPE_LWT_IN**; **BPF_LWT_ENCAP_IP** type can
  2104. * be called by bpf programs of types **BPF_PROG_TYPE_LWT_IN** and
  2105. * **BPF_PROG_TYPE_LWT_XMIT**.
  2106. *
  2107. * A call to this helper is susceptible to change the underlying
  2108. * packet buffer. Therefore, at load time, all checks on pointers
  2109. * previously done by the verifier are invalidated and must be
  2110. * performed again, if the helper is used in combination with
  2111. * direct packet access.
  2112. * Return
  2113. * 0 on success, or a negative error in case of failure.
  2114. *
  2115. * int bpf_lwt_seg6_store_bytes(struct sk_buff *skb, u32 offset, const void *from, u32 len)
  2116. * Description
  2117. * Store *len* bytes from address *from* into the packet
  2118. * associated to *skb*, at *offset*. Only the flags, tag and TLVs
  2119. * inside the outermost IPv6 Segment Routing Header can be
  2120. * modified through this helper.
  2121. *
  2122. * A call to this helper is susceptible to change the underlying
  2123. * packet buffer. Therefore, at load time, all checks on pointers
  2124. * previously done by the verifier are invalidated and must be
  2125. * performed again, if the helper is used in combination with
  2126. * direct packet access.
  2127. * Return
  2128. * 0 on success, or a negative error in case of failure.
  2129. *
  2130. * int bpf_lwt_seg6_adjust_srh(struct sk_buff *skb, u32 offset, s32 delta)
  2131. * Description
  2132. * Adjust the size allocated to TLVs in the outermost IPv6
  2133. * Segment Routing Header contained in the packet associated to
  2134. * *skb*, at position *offset* by *delta* bytes. Only offsets
  2135. * after the segments are accepted. *delta* can be as well
  2136. * positive (growing) as negative (shrinking).
  2137. *
  2138. * A call to this helper is susceptible to change the underlying
  2139. * packet buffer. Therefore, at load time, all checks on pointers
  2140. * previously done by the verifier are invalidated and must be
  2141. * performed again, if the helper is used in combination with
  2142. * direct packet access.
  2143. * Return
  2144. * 0 on success, or a negative error in case of failure.
  2145. *
  2146. * int bpf_lwt_seg6_action(struct sk_buff *skb, u32 action, void *param, u32 param_len)
  2147. * Description
  2148. * Apply an IPv6 Segment Routing action of type *action* to the
  2149. * packet associated to *skb*. Each action takes a parameter
  2150. * contained at address *param*, and of length *param_len* bytes.
  2151. * *action* can be one of:
  2152. *
  2153. * **SEG6_LOCAL_ACTION_END_X**
  2154. * End.X action: Endpoint with Layer-3 cross-connect.
  2155. * Type of *param*: **struct in6_addr**.
  2156. * **SEG6_LOCAL_ACTION_END_T**
  2157. * End.T action: Endpoint with specific IPv6 table lookup.
  2158. * Type of *param*: **int**.
  2159. * **SEG6_LOCAL_ACTION_END_B6**
  2160. * End.B6 action: Endpoint bound to an SRv6 policy.
  2161. * Type of *param*: **struct ipv6_sr_hdr**.
  2162. * **SEG6_LOCAL_ACTION_END_B6_ENCAP**
  2163. * End.B6.Encap action: Endpoint bound to an SRv6
  2164. * encapsulation policy.
  2165. * Type of *param*: **struct ipv6_sr_hdr**.
  2166. *
  2167. * A call to this helper is susceptible to change the underlying
  2168. * packet buffer. Therefore, at load time, all checks on pointers
  2169. * previously done by the verifier are invalidated and must be
  2170. * performed again, if the helper is used in combination with
  2171. * direct packet access.
  2172. * Return
  2173. * 0 on success, or a negative error in case of failure.
  2174. *
  2175. * int bpf_rc_repeat(void *ctx)
  2176. * Description
  2177. * This helper is used in programs implementing IR decoding, to
  2178. * report a successfully decoded repeat key message. This delays
  2179. * the generation of a key up event for previously generated
  2180. * key down event.
  2181. *
  2182. * Some IR protocols like NEC have a special IR message for
  2183. * repeating last button, for when a button is held down.
  2184. *
  2185. * The *ctx* should point to the lirc sample as passed into
  2186. * the program.
  2187. *
  2188. * This helper is only available is the kernel was compiled with
  2189. * the **CONFIG_BPF_LIRC_MODE2** configuration option set to
  2190. * "**y**".
  2191. * Return
  2192. * 0
  2193. *
  2194. * int bpf_rc_keydown(void *ctx, u32 protocol, u64 scancode, u32 toggle)
  2195. * Description
  2196. * This helper is used in programs implementing IR decoding, to
  2197. * report a successfully decoded key press with *scancode*,
  2198. * *toggle* value in the given *protocol*. The scancode will be
  2199. * translated to a keycode using the rc keymap, and reported as
  2200. * an input key down event. After a period a key up event is
  2201. * generated. This period can be extended by calling either
  2202. * **bpf_rc_keydown**\ () again with the same values, or calling
  2203. * **bpf_rc_repeat**\ ().
  2204. *
  2205. * Some protocols include a toggle bit, in case the button was
  2206. * released and pressed again between consecutive scancodes.
  2207. *
  2208. * The *ctx* should point to the lirc sample as passed into
  2209. * the program.
  2210. *
  2211. * The *protocol* is the decoded protocol number (see
  2212. * **enum rc_proto** for some predefined values).
  2213. *
  2214. * This helper is only available is the kernel was compiled with
  2215. * the **CONFIG_BPF_LIRC_MODE2** configuration option set to
  2216. * "**y**".
  2217. * Return
  2218. * 0
  2219. *
  2220. * u64 bpf_skb_cgroup_id(struct sk_buff *skb)
  2221. * Description
  2222. * Return the cgroup v2 id of the socket associated with the *skb*.
  2223. * This is roughly similar to the **bpf_get_cgroup_classid**\ ()
  2224. * helper for cgroup v1 by providing a tag resp. identifier that
  2225. * can be matched on or used for map lookups e.g. to implement
  2226. * policy. The cgroup v2 id of a given path in the hierarchy is
  2227. * exposed in user space through the f_handle API in order to get
  2228. * to the same 64-bit id.
  2229. *
  2230. * This helper can be used on TC egress path, but not on ingress,
  2231. * and is available only if the kernel was compiled with the
  2232. * **CONFIG_SOCK_CGROUP_DATA** configuration option.
  2233. * Return
  2234. * The id is returned or 0 in case the id could not be retrieved.
  2235. *
  2236. * u64 bpf_get_current_cgroup_id(void)
  2237. * Return
  2238. * A 64-bit integer containing the current cgroup id based
  2239. * on the cgroup within which the current task is running.
  2240. *
  2241. * void *bpf_get_local_storage(void *map, u64 flags)
  2242. * Description
  2243. * Get the pointer to the local storage area.
  2244. * The type and the size of the local storage is defined
  2245. * by the *map* argument.
  2246. * The *flags* meaning is specific for each map type,
  2247. * and has to be 0 for cgroup local storage.
  2248. *
  2249. * Depending on the BPF program type, a local storage area
  2250. * can be shared between multiple instances of the BPF program,
  2251. * running simultaneously.
  2252. *
  2253. * A user should care about the synchronization by himself.
  2254. * For example, by using the **BPF_STX_XADD** instruction to alter
  2255. * the shared data.
  2256. * Return
  2257. * A pointer to the local storage area.
  2258. *
  2259. * int bpf_sk_select_reuseport(struct sk_reuseport_md *reuse, struct bpf_map *map, void *key, u64 flags)
  2260. * Description
  2261. * Select a **SO_REUSEPORT** socket from a
  2262. * **BPF_MAP_TYPE_REUSEPORT_ARRAY** *map*.
  2263. * It checks the selected socket is matching the incoming
  2264. * request in the socket buffer.
  2265. * Return
  2266. * 0 on success, or a negative error in case of failure.
  2267. *
  2268. * u64 bpf_skb_ancestor_cgroup_id(struct sk_buff *skb, int ancestor_level)
  2269. * Description
  2270. * Return id of cgroup v2 that is ancestor of cgroup associated
  2271. * with the *skb* at the *ancestor_level*. The root cgroup is at
  2272. * *ancestor_level* zero and each step down the hierarchy
  2273. * increments the level. If *ancestor_level* == level of cgroup
  2274. * associated with *skb*, then return value will be same as that
  2275. * of **bpf_skb_cgroup_id**\ ().
  2276. *
  2277. * The helper is useful to implement policies based on cgroups
  2278. * that are upper in hierarchy than immediate cgroup associated
  2279. * with *skb*.
  2280. *
  2281. * The format of returned id and helper limitations are same as in
  2282. * **bpf_skb_cgroup_id**\ ().
  2283. * Return
  2284. * The id is returned or 0 in case the id could not be retrieved.
  2285. *
  2286. * struct bpf_sock *bpf_sk_lookup_tcp(void *ctx, struct bpf_sock_tuple *tuple, u32 tuple_size, u64 netns, u64 flags)
  2287. * Description
  2288. * Look for TCP socket matching *tuple*, optionally in a child
  2289. * network namespace *netns*. The return value must be checked,
  2290. * and if non-**NULL**, released via **bpf_sk_release**\ ().
  2291. *
  2292. * The *ctx* should point to the context of the program, such as
  2293. * the skb or socket (depending on the hook in use). This is used
  2294. * to determine the base network namespace for the lookup.
  2295. *
  2296. * *tuple_size* must be one of:
  2297. *
  2298. * **sizeof**\ (*tuple*\ **->ipv4**)
  2299. * Look for an IPv4 socket.
  2300. * **sizeof**\ (*tuple*\ **->ipv6**)
  2301. * Look for an IPv6 socket.
  2302. *
  2303. * If the *netns* is a negative signed 32-bit integer, then the
  2304. * socket lookup table in the netns associated with the *ctx* will
  2305. * will be used. For the TC hooks, this is the netns of the device
  2306. * in the skb. For socket hooks, this is the netns of the socket.
  2307. * If *netns* is any other signed 32-bit value greater than or
  2308. * equal to zero then it specifies the ID of the netns relative to
  2309. * the netns associated with the *ctx*. *netns* values beyond the
  2310. * range of 32-bit integers are reserved for future use.
  2311. *
  2312. * All values for *flags* are reserved for future usage, and must
  2313. * be left at zero.
  2314. *
  2315. * This helper is available only if the kernel was compiled with
  2316. * **CONFIG_NET** configuration option.
  2317. * Return
  2318. * Pointer to **struct bpf_sock**, or **NULL** in case of failure.
  2319. * For sockets with reuseport option, the **struct bpf_sock**
  2320. * result is from *reuse*\ **->socks**\ [] using the hash of the
  2321. * tuple.
  2322. *
  2323. * struct bpf_sock *bpf_sk_lookup_udp(void *ctx, struct bpf_sock_tuple *tuple, u32 tuple_size, u64 netns, u64 flags)
  2324. * Description
  2325. * Look for UDP socket matching *tuple*, optionally in a child
  2326. * network namespace *netns*. The return value must be checked,
  2327. * and if non-**NULL**, released via **bpf_sk_release**\ ().
  2328. *
  2329. * The *ctx* should point to the context of the program, such as
  2330. * the skb or socket (depending on the hook in use). This is used
  2331. * to determine the base network namespace for the lookup.
  2332. *
  2333. * *tuple_size* must be one of:
  2334. *
  2335. * **sizeof**\ (*tuple*\ **->ipv4**)
  2336. * Look for an IPv4 socket.
  2337. * **sizeof**\ (*tuple*\ **->ipv6**)
  2338. * Look for an IPv6 socket.
  2339. *
  2340. * If the *netns* is a negative signed 32-bit integer, then the
  2341. * socket lookup table in the netns associated with the *ctx* will
  2342. * will be used. For the TC hooks, this is the netns of the device
  2343. * in the skb. For socket hooks, this is the netns of the socket.
  2344. * If *netns* is any other signed 32-bit value greater than or
  2345. * equal to zero then it specifies the ID of the netns relative to
  2346. * the netns associated with the *ctx*. *netns* values beyond the
  2347. * range of 32-bit integers are reserved for future use.
  2348. *
  2349. * All values for *flags* are reserved for future usage, and must
  2350. * be left at zero.
  2351. *
  2352. * This helper is available only if the kernel was compiled with
  2353. * **CONFIG_NET** configuration option.
  2354. * Return
  2355. * Pointer to **struct bpf_sock**, or **NULL** in case of failure.
  2356. * For sockets with reuseport option, the **struct bpf_sock**
  2357. * result is from *reuse*\ **->socks**\ [] using the hash of the
  2358. * tuple.
  2359. *
  2360. * int bpf_sk_release(struct bpf_sock *sock)
  2361. * Description
  2362. * Release the reference held by *sock*. *sock* must be a
  2363. * non-**NULL** pointer that was returned from
  2364. * **bpf_sk_lookup_xxx**\ ().
  2365. * Return
  2366. * 0 on success, or a negative error in case of failure.
  2367. *
  2368. * int bpf_map_push_elem(struct bpf_map *map, const void *value, u64 flags)
  2369. * Description
  2370. * Push an element *value* in *map*. *flags* is one of:
  2371. *
  2372. * **BPF_EXIST**
  2373. * If the queue/stack is full, the oldest element is
  2374. * removed to make room for this.
  2375. * Return
  2376. * 0 on success, or a negative error in case of failure.
  2377. *
  2378. * int bpf_map_pop_elem(struct bpf_map *map, void *value)
  2379. * Description
  2380. * Pop an element from *map*.
  2381. * Return
  2382. * 0 on success, or a negative error in case of failure.
  2383. *
  2384. * int bpf_map_peek_elem(struct bpf_map *map, void *value)
  2385. * Description
  2386. * Get an element from *map* without removing it.
  2387. * Return
  2388. * 0 on success, or a negative error in case of failure.
  2389. *
  2390. * int bpf_msg_push_data(struct sk_msg_buff *msg, u32 start, u32 len, u64 flags)
  2391. * Description
  2392. * For socket policies, insert *len* bytes into *msg* at offset
  2393. * *start*.
  2394. *
  2395. * If a program of type **BPF_PROG_TYPE_SK_MSG** is run on a
  2396. * *msg* it may want to insert metadata or options into the *msg*.
  2397. * This can later be read and used by any of the lower layer BPF
  2398. * hooks.
  2399. *
  2400. * This helper may fail if under memory pressure (a malloc
  2401. * fails) in these cases BPF programs will get an appropriate
  2402. * error and BPF programs will need to handle them.
  2403. * Return
  2404. * 0 on success, or a negative error in case of failure.
  2405. *
  2406. * int bpf_msg_pop_data(struct sk_msg_buff *msg, u32 start, u32 len, u64 flags)
  2407. * Description
  2408. * Will remove *len* bytes from a *msg* starting at byte *start*.
  2409. * This may result in **ENOMEM** errors under certain situations if
  2410. * an allocation and copy are required due to a full ring buffer.
  2411. * However, the helper will try to avoid doing the allocation
  2412. * if possible. Other errors can occur if input parameters are
  2413. * invalid either due to *start* byte not being valid part of *msg*
  2414. * payload and/or *pop* value being to large.
  2415. * Return
  2416. * 0 on success, or a negative error in case of failure.
  2417. *
  2418. * int bpf_rc_pointer_rel(void *ctx, s32 rel_x, s32 rel_y)
  2419. * Description
  2420. * This helper is used in programs implementing IR decoding, to
  2421. * report a successfully decoded pointer movement.
  2422. *
  2423. * The *ctx* should point to the lirc sample as passed into
  2424. * the program.
  2425. *
  2426. * This helper is only available is the kernel was compiled with
  2427. * the **CONFIG_BPF_LIRC_MODE2** configuration option set to
  2428. * "**y**".
  2429. * Return
  2430. * 0
  2431. *
  2432. * int bpf_spin_lock(struct bpf_spin_lock *lock)
  2433. * Description
  2434. * Acquire a spinlock represented by the pointer *lock*, which is
  2435. * stored as part of a value of a map. Taking the lock allows to
  2436. * safely update the rest of the fields in that value. The
  2437. * spinlock can (and must) later be released with a call to
  2438. * **bpf_spin_unlock**\ (\ *lock*\ ).
  2439. *
  2440. * Spinlocks in BPF programs come with a number of restrictions
  2441. * and constraints:
  2442. *
  2443. * * **bpf_spin_lock** objects are only allowed inside maps of
  2444. * types **BPF_MAP_TYPE_HASH** and **BPF_MAP_TYPE_ARRAY** (this
  2445. * list could be extended in the future).
  2446. * * BTF description of the map is mandatory.
  2447. * * The BPF program can take ONE lock at a time, since taking two
  2448. * or more could cause dead locks.
  2449. * * Only one **struct bpf_spin_lock** is allowed per map element.
  2450. * * When the lock is taken, calls (either BPF to BPF or helpers)
  2451. * are not allowed.
  2452. * * The **BPF_LD_ABS** and **BPF_LD_IND** instructions are not
  2453. * allowed inside a spinlock-ed region.
  2454. * * The BPF program MUST call **bpf_spin_unlock**\ () to release
  2455. * the lock, on all execution paths, before it returns.
  2456. * * The BPF program can access **struct bpf_spin_lock** only via
  2457. * the **bpf_spin_lock**\ () and **bpf_spin_unlock**\ ()
  2458. * helpers. Loading or storing data into the **struct
  2459. * bpf_spin_lock** *lock*\ **;** field of a map is not allowed.
  2460. * * To use the **bpf_spin_lock**\ () helper, the BTF description
  2461. * of the map value must be a struct and have **struct
  2462. * bpf_spin_lock** *anyname*\ **;** field at the top level.
  2463. * Nested lock inside another struct is not allowed.
  2464. * * The **struct bpf_spin_lock** *lock* field in a map value must
  2465. * be aligned on a multiple of 4 bytes in that value.
  2466. * * Syscall with command **BPF_MAP_LOOKUP_ELEM** does not copy
  2467. * the **bpf_spin_lock** field to user space.
  2468. * * Syscall with command **BPF_MAP_UPDATE_ELEM**, or update from
  2469. * a BPF program, do not update the **bpf_spin_lock** field.
  2470. * * **bpf_spin_lock** cannot be on the stack or inside a
  2471. * networking packet (it can only be inside of a map values).
  2472. * * **bpf_spin_lock** is available to root only.
  2473. * * Tracing programs and socket filter programs cannot use
  2474. * **bpf_spin_lock**\ () due to insufficient preemption checks
  2475. * (but this may change in the future).
  2476. * * **bpf_spin_lock** is not allowed in inner maps of map-in-map.
  2477. * Return
  2478. * 0
  2479. *
  2480. * int bpf_spin_unlock(struct bpf_spin_lock *lock)
  2481. * Description
  2482. * Release the *lock* previously locked by a call to
  2483. * **bpf_spin_lock**\ (\ *lock*\ ).
  2484. * Return
  2485. * 0
  2486. *
  2487. * struct bpf_sock *bpf_sk_fullsock(struct bpf_sock *sk)
  2488. * Description
  2489. * This helper gets a **struct bpf_sock** pointer such
  2490. * that all the fields in this **bpf_sock** can be accessed.
  2491. * Return
  2492. * A **struct bpf_sock** pointer on success, or **NULL** in
  2493. * case of failure.
  2494. *
  2495. * struct bpf_tcp_sock *bpf_tcp_sock(struct bpf_sock *sk)
  2496. * Description
  2497. * This helper gets a **struct bpf_tcp_sock** pointer from a
  2498. * **struct bpf_sock** pointer.
  2499. * Return
  2500. * A **struct bpf_tcp_sock** pointer on success, or **NULL** in
  2501. * case of failure.
  2502. *
  2503. * int bpf_skb_ecn_set_ce(struct sk_buff *skb)
  2504. * Description
  2505. * Set ECN (Explicit Congestion Notification) field of IP header
  2506. * to **CE** (Congestion Encountered) if current value is **ECT**
  2507. * (ECN Capable Transport). Otherwise, do nothing. Works with IPv6
  2508. * and IPv4.
  2509. * Return
  2510. * 1 if the **CE** flag is set (either by the current helper call
  2511. * or because it was already present), 0 if it is not set.
  2512. *
  2513. * struct bpf_sock *bpf_get_listener_sock(struct bpf_sock *sk)
  2514. * Description
  2515. * Return a **struct bpf_sock** pointer in **TCP_LISTEN** state.
  2516. * **bpf_sk_release**\ () is unnecessary and not allowed.
  2517. * Return
  2518. * A **struct bpf_sock** pointer on success, or **NULL** in
  2519. * case of failure.
  2520. *
  2521. * struct bpf_sock *bpf_skc_lookup_tcp(void *ctx, struct bpf_sock_tuple *tuple, u32 tuple_size, u64 netns, u64 flags)
  2522. * Description
  2523. * Look for TCP socket matching *tuple*, optionally in a child
  2524. * network namespace *netns*. The return value must be checked,
  2525. * and if non-**NULL**, released via **bpf_sk_release**\ ().
  2526. *
  2527. * This function is identical to **bpf_sk_lookup_tcp**\ (), except
  2528. * that it also returns timewait or request sockets. Use
  2529. * **bpf_sk_fullsock**\ () or **bpf_tcp_sock**\ () to access the
  2530. * full structure.
  2531. *
  2532. * This helper is available only if the kernel was compiled with
  2533. * **CONFIG_NET** configuration option.
  2534. * Return
  2535. * Pointer to **struct bpf_sock**, or **NULL** in case of failure.
  2536. * For sockets with reuseport option, the **struct bpf_sock**
  2537. * result is from *reuse*\ **->socks**\ [] using the hash of the
  2538. * tuple.
  2539. *
  2540. * int bpf_tcp_check_syncookie(struct bpf_sock *sk, void *iph, u32 iph_len, struct tcphdr *th, u32 th_len)
  2541. * Description
  2542. * Check whether *iph* and *th* contain a valid SYN cookie ACK for
  2543. * the listening socket in *sk*.
  2544. *
  2545. * *iph* points to the start of the IPv4 or IPv6 header, while
  2546. * *iph_len* contains **sizeof**\ (**struct iphdr**) or
  2547. * **sizeof**\ (**struct ip6hdr**).
  2548. *
  2549. * *th* points to the start of the TCP header, while *th_len*
  2550. * contains **sizeof**\ (**struct tcphdr**).
  2551. *
  2552. * Return
  2553. * 0 if *iph* and *th* are a valid SYN cookie ACK, or a negative
  2554. * error otherwise.
  2555. *
  2556. * int bpf_sysctl_get_name(struct bpf_sysctl *ctx, char *buf, size_t buf_len, u64 flags)
  2557. * Description
  2558. * Get name of sysctl in /proc/sys/ and copy it into provided by
  2559. * program buffer *buf* of size *buf_len*.
  2560. *
  2561. * The buffer is always NUL terminated, unless it's zero-sized.
  2562. *
  2563. * If *flags* is zero, full name (e.g. "net/ipv4/tcp_mem") is
  2564. * copied. Use **BPF_F_SYSCTL_BASE_NAME** flag to copy base name
  2565. * only (e.g. "tcp_mem").
  2566. * Return
  2567. * Number of character copied (not including the trailing NUL).
  2568. *
  2569. * **-E2BIG** if the buffer wasn't big enough (*buf* will contain
  2570. * truncated name in this case).
  2571. *
  2572. * int bpf_sysctl_get_current_value(struct bpf_sysctl *ctx, char *buf, size_t buf_len)
  2573. * Description
  2574. * Get current value of sysctl as it is presented in /proc/sys
  2575. * (incl. newline, etc), and copy it as a string into provided
  2576. * by program buffer *buf* of size *buf_len*.
  2577. *
  2578. * The whole value is copied, no matter what file position user
  2579. * space issued e.g. sys_read at.
  2580. *
  2581. * The buffer is always NUL terminated, unless it's zero-sized.
  2582. * Return
  2583. * Number of character copied (not including the trailing NUL).
  2584. *
  2585. * **-E2BIG** if the buffer wasn't big enough (*buf* will contain
  2586. * truncated name in this case).
  2587. *
  2588. * **-EINVAL** if current value was unavailable, e.g. because
  2589. * sysctl is uninitialized and read returns -EIO for it.
  2590. *
  2591. * int bpf_sysctl_get_new_value(struct bpf_sysctl *ctx, char *buf, size_t buf_len)
  2592. * Description
  2593. * Get new value being written by user space to sysctl (before
  2594. * the actual write happens) and copy it as a string into
  2595. * provided by program buffer *buf* of size *buf_len*.
  2596. *
  2597. * User space may write new value at file position > 0.
  2598. *
  2599. * The buffer is always NUL terminated, unless it's zero-sized.
  2600. * Return
  2601. * Number of character copied (not including the trailing NUL).
  2602. *
  2603. * **-E2BIG** if the buffer wasn't big enough (*buf* will contain
  2604. * truncated name in this case).
  2605. *
  2606. * **-EINVAL** if sysctl is being read.
  2607. *
  2608. * int bpf_sysctl_set_new_value(struct bpf_sysctl *ctx, const char *buf, size_t buf_len)
  2609. * Description
  2610. * Override new value being written by user space to sysctl with
  2611. * value provided by program in buffer *buf* of size *buf_len*.
  2612. *
  2613. * *buf* should contain a string in same form as provided by user
  2614. * space on sysctl write.
  2615. *
  2616. * User space may write new value at file position > 0. To override
  2617. * the whole sysctl value file position should be set to zero.
  2618. * Return
  2619. * 0 on success.
  2620. *
  2621. * **-E2BIG** if the *buf_len* is too big.
  2622. *
  2623. * **-EINVAL** if sysctl is being read.
  2624. *
  2625. * int bpf_strtol(const char *buf, size_t buf_len, u64 flags, long *res)
  2626. * Description
  2627. * Convert the initial part of the string from buffer *buf* of
  2628. * size *buf_len* to a long integer according to the given base
  2629. * and save the result in *res*.
  2630. *
  2631. * The string may begin with an arbitrary amount of white space
  2632. * (as determined by **isspace**\ (3)) followed by a single
  2633. * optional '**-**' sign.
  2634. *
  2635. * Five least significant bits of *flags* encode base, other bits
  2636. * are currently unused.
  2637. *
  2638. * Base must be either 8, 10, 16 or 0 to detect it automatically
  2639. * similar to user space **strtol**\ (3).
  2640. * Return
  2641. * Number of characters consumed on success. Must be positive but
  2642. * no more than *buf_len*.
  2643. *
  2644. * **-EINVAL** if no valid digits were found or unsupported base
  2645. * was provided.
  2646. *
  2647. * **-ERANGE** if resulting value was out of range.
  2648. *
  2649. * int bpf_strtoul(const char *buf, size_t buf_len, u64 flags, unsigned long *res)
  2650. * Description
  2651. * Convert the initial part of the string from buffer *buf* of
  2652. * size *buf_len* to an unsigned long integer according to the
  2653. * given base and save the result in *res*.
  2654. *
  2655. * The string may begin with an arbitrary amount of white space
  2656. * (as determined by **isspace**\ (3)).
  2657. *
  2658. * Five least significant bits of *flags* encode base, other bits
  2659. * are currently unused.
  2660. *
  2661. * Base must be either 8, 10, 16 or 0 to detect it automatically
  2662. * similar to user space **strtoul**\ (3).
  2663. * Return
  2664. * Number of characters consumed on success. Must be positive but
  2665. * no more than *buf_len*.
  2666. *
  2667. * **-EINVAL** if no valid digits were found or unsupported base
  2668. * was provided.
  2669. *
  2670. * **-ERANGE** if resulting value was out of range.
  2671. *
  2672. * void *bpf_sk_storage_get(struct bpf_map *map, struct bpf_sock *sk, void *value, u64 flags)
  2673. * Description
  2674. * Get a bpf-local-storage from a *sk*.
  2675. *
  2676. * Logically, it could be thought of getting the value from
  2677. * a *map* with *sk* as the **key**. From this
  2678. * perspective, the usage is not much different from
  2679. * **bpf_map_lookup_elem**\ (*map*, **&**\ *sk*) except this
  2680. * helper enforces the key must be a full socket and the map must
  2681. * be a **BPF_MAP_TYPE_SK_STORAGE** also.
  2682. *
  2683. * Underneath, the value is stored locally at *sk* instead of
  2684. * the *map*. The *map* is used as the bpf-local-storage
  2685. * "type". The bpf-local-storage "type" (i.e. the *map*) is
  2686. * searched against all bpf-local-storages residing at *sk*.
  2687. *
  2688. * An optional *flags* (**BPF_SK_STORAGE_GET_F_CREATE**) can be
  2689. * used such that a new bpf-local-storage will be
  2690. * created if one does not exist. *value* can be used
  2691. * together with **BPF_SK_STORAGE_GET_F_CREATE** to specify
  2692. * the initial value of a bpf-local-storage. If *value* is
  2693. * **NULL**, the new bpf-local-storage will be zero initialized.
  2694. * Return
  2695. * A bpf-local-storage pointer is returned on success.
  2696. *
  2697. * **NULL** if not found or there was an error in adding
  2698. * a new bpf-local-storage.
  2699. *
  2700. * int bpf_sk_storage_delete(struct bpf_map *map, struct bpf_sock *sk)
  2701. * Description
  2702. * Delete a bpf-local-storage from a *sk*.
  2703. * Return
  2704. * 0 on success.
  2705. *
  2706. * **-ENOENT** if the bpf-local-storage cannot be found.
  2707. *
  2708. * int bpf_send_signal(u32 sig)
  2709. * Description
  2710. * Send signal *sig* to the process of the current task.
  2711. * The signal may be delivered to any of this process's threads.
  2712. * Return
  2713. * 0 on success or successfully queued.
  2714. *
  2715. * **-EBUSY** if work queue under nmi is full.
  2716. *
  2717. * **-EINVAL** if *sig* is invalid.
  2718. *
  2719. * **-EPERM** if no permission to send the *sig*.
  2720. *
  2721. * **-EAGAIN** if bpf program can try again.
  2722. *
  2723. * s64 bpf_tcp_gen_syncookie(struct bpf_sock *sk, void *iph, u32 iph_len, struct tcphdr *th, u32 th_len)
  2724. * Description
  2725. * Try to issue a SYN cookie for the packet with corresponding
  2726. * IP/TCP headers, *iph* and *th*, on the listening socket in *sk*.
  2727. *
  2728. * *iph* points to the start of the IPv4 or IPv6 header, while
  2729. * *iph_len* contains **sizeof**\ (**struct iphdr**) or
  2730. * **sizeof**\ (**struct ip6hdr**).
  2731. *
  2732. * *th* points to the start of the TCP header, while *th_len*
  2733. * contains the length of the TCP header.
  2734. *
  2735. * Return
  2736. * On success, lower 32 bits hold the generated SYN cookie in
  2737. * followed by 16 bits which hold the MSS value for that cookie,
  2738. * and the top 16 bits are unused.
  2739. *
  2740. * On failure, the returned value is one of the following:
  2741. *
  2742. * **-EINVAL** SYN cookie cannot be issued due to error
  2743. *
  2744. * **-ENOENT** SYN cookie should not be issued (no SYN flood)
  2745. *
  2746. * **-EOPNOTSUPP** kernel configuration does not enable SYN cookies
  2747. *
  2748. * **-EPROTONOSUPPORT** IP packet version is not 4 or 6
  2749. *
  2750. * int bpf_skb_output(void *ctx, struct bpf_map *map, u64 flags, void *data, u64 size)
  2751. * Description
  2752. * Write raw *data* blob into a special BPF perf event held by
  2753. * *map* of type **BPF_MAP_TYPE_PERF_EVENT_ARRAY**. This perf
  2754. * event must have the following attributes: **PERF_SAMPLE_RAW**
  2755. * as **sample_type**, **PERF_TYPE_SOFTWARE** as **type**, and
  2756. * **PERF_COUNT_SW_BPF_OUTPUT** as **config**.
  2757. *
  2758. * The *flags* are used to indicate the index in *map* for which
  2759. * the value must be put, masked with **BPF_F_INDEX_MASK**.
  2760. * Alternatively, *flags* can be set to **BPF_F_CURRENT_CPU**
  2761. * to indicate that the index of the current CPU core should be
  2762. * used.
  2763. *
  2764. * The value to write, of *size*, is passed through eBPF stack and
  2765. * pointed by *data*.
  2766. *
  2767. * *ctx* is a pointer to in-kernel struct sk_buff.
  2768. *
  2769. * This helper is similar to **bpf_perf_event_output**\ () but
  2770. * restricted to raw_tracepoint bpf programs.
  2771. * Return
  2772. * 0 on success, or a negative error in case of failure.
  2773. *
  2774. * int bpf_probe_read_user(void *dst, u32 size, const void *unsafe_ptr)
  2775. * Description
  2776. * Safely attempt to read *size* bytes from user space address
  2777. * *unsafe_ptr* and store the data in *dst*.
  2778. * Return
  2779. * 0 on success, or a negative error in case of failure.
  2780. *
  2781. * int bpf_probe_read_kernel(void *dst, u32 size, const void *unsafe_ptr)
  2782. * Description
  2783. * Safely attempt to read *size* bytes from kernel space address
  2784. * *unsafe_ptr* and store the data in *dst*.
  2785. * Return
  2786. * 0 on success, or a negative error in case of failure.
  2787. *
  2788. * int bpf_probe_read_user_str(void *dst, u32 size, const void *unsafe_ptr)
  2789. * Description
  2790. * Copy a NUL terminated string from an unsafe user address
  2791. * *unsafe_ptr* to *dst*. The *size* should include the
  2792. * terminating NUL byte. In case the string length is smaller than
  2793. * *size*, the target is not padded with further NUL bytes. If the
  2794. * string length is larger than *size*, just *size*-1 bytes are
  2795. * copied and the last byte is set to NUL.
  2796. *
  2797. * On success, the length of the copied string is returned. This
  2798. * makes this helper useful in tracing programs for reading
  2799. * strings, and more importantly to get its length at runtime. See
  2800. * the following snippet:
  2801. *
  2802. * ::
  2803. *
  2804. * SEC("kprobe/sys_open")
  2805. * void bpf_sys_open(struct pt_regs *ctx)
  2806. * {
  2807. * char buf[PATHLEN]; // PATHLEN is defined to 256
  2808. * int res = bpf_probe_read_user_str(buf, sizeof(buf),
  2809. * ctx->di);
  2810. *
  2811. * // Consume buf, for example push it to
  2812. * // userspace via bpf_perf_event_output(); we
  2813. * // can use res (the string length) as event
  2814. * // size, after checking its boundaries.
  2815. * }
  2816. *
  2817. * In comparison, using **bpf_probe_read_user()** helper here
  2818. * instead to read the string would require to estimate the length
  2819. * at compile time, and would often result in copying more memory
  2820. * than necessary.
  2821. *
  2822. * Another useful use case is when parsing individual process
  2823. * arguments or individual environment variables navigating
  2824. * *current*\ **->mm->arg_start** and *current*\
  2825. * **->mm->env_start**: using this helper and the return value,
  2826. * one can quickly iterate at the right offset of the memory area.
  2827. * Return
  2828. * On success, the strictly positive length of the string,
  2829. * including the trailing NUL character. On error, a negative
  2830. * value.
  2831. *
  2832. * int bpf_probe_read_kernel_str(void *dst, u32 size, const void *unsafe_ptr)
  2833. * Description
  2834. * Copy a NUL terminated string from an unsafe kernel address *unsafe_ptr*
  2835. * to *dst*. Same semantics as with bpf_probe_read_user_str() apply.
  2836. * Return
  2837. * On success, the strictly positive length of the string, including
  2838. * the trailing NUL character. On error, a negative value.
  2839. *
  2840. * int bpf_tcp_send_ack(void *tp, u32 rcv_nxt)
  2841. * Description
  2842. * Send out a tcp-ack. *tp* is the in-kernel struct tcp_sock.
  2843. * *rcv_nxt* is the ack_seq to be sent out.
  2844. * Return
  2845. * 0 on success, or a negative error in case of failure.
  2846. *
  2847. * int bpf_send_signal_thread(u32 sig)
  2848. * Description
  2849. * Send signal *sig* to the thread corresponding to the current task.
  2850. * Return
  2851. * 0 on success or successfully queued.
  2852. *
  2853. * **-EBUSY** if work queue under nmi is full.
  2854. *
  2855. * **-EINVAL** if *sig* is invalid.
  2856. *
  2857. * **-EPERM** if no permission to send the *sig*.
  2858. *
  2859. * **-EAGAIN** if bpf program can try again.
  2860. *
  2861. * u64 bpf_jiffies64(void)
  2862. * Description
  2863. * Obtain the 64bit jiffies
  2864. * Return
  2865. * The 64 bit jiffies
  2866. *
  2867. * int bpf_read_branch_records(struct bpf_perf_event_data *ctx, void *buf, u32 size, u64 flags)
  2868. * Description
  2869. * For an eBPF program attached to a perf event, retrieve the
  2870. * branch records (struct perf_branch_entry) associated to *ctx*
  2871. * and store it in the buffer pointed by *buf* up to size
  2872. * *size* bytes.
  2873. * Return
  2874. * On success, number of bytes written to *buf*. On error, a
  2875. * negative value.
  2876. *
  2877. * The *flags* can be set to **BPF_F_GET_BRANCH_RECORDS_SIZE** to
  2878. * instead return the number of bytes required to store all the
  2879. * branch entries. If this flag is set, *buf* may be NULL.
  2880. *
  2881. * **-EINVAL** if arguments invalid or **size** not a multiple
  2882. * of sizeof(struct perf_branch_entry).
  2883. *
  2884. * **-ENOENT** if architecture does not support branch records.
  2885. *
  2886. * int bpf_get_ns_current_pid_tgid(u64 dev, u64 ino, struct bpf_pidns_info *nsdata, u32 size)
  2887. * Description
  2888. * Returns 0 on success, values for *pid* and *tgid* as seen from the current
  2889. * *namespace* will be returned in *nsdata*.
  2890. *
  2891. * On failure, the returned value is one of the following:
  2892. *
  2893. * **-EINVAL** if dev and inum supplied don't match dev_t and inode number
  2894. * with nsfs of current task, or if dev conversion to dev_t lost high bits.
  2895. *
  2896. * **-ENOENT** if pidns does not exists for the current task.
  2897. *
  2898. * int bpf_xdp_output(void *ctx, struct bpf_map *map, u64 flags, void *data, u64 size)
  2899. * Description
  2900. * Write raw *data* blob into a special BPF perf event held by
  2901. * *map* of type **BPF_MAP_TYPE_PERF_EVENT_ARRAY**. This perf
  2902. * event must have the following attributes: **PERF_SAMPLE_RAW**
  2903. * as **sample_type**, **PERF_TYPE_SOFTWARE** as **type**, and
  2904. * **PERF_COUNT_SW_BPF_OUTPUT** as **config**.
  2905. *
  2906. * The *flags* are used to indicate the index in *map* for which
  2907. * the value must be put, masked with **BPF_F_INDEX_MASK**.
  2908. * Alternatively, *flags* can be set to **BPF_F_CURRENT_CPU**
  2909. * to indicate that the index of the current CPU core should be
  2910. * used.
  2911. *
  2912. * The value to write, of *size*, is passed through eBPF stack and
  2913. * pointed by *data*.
  2914. *
  2915. * *ctx* is a pointer to in-kernel struct xdp_buff.
  2916. *
  2917. * This helper is similar to **bpf_perf_eventoutput**\ () but
  2918. * restricted to raw_tracepoint bpf programs.
  2919. * Return
  2920. * 0 on success, or a negative error in case of failure.
  2921. *
  2922. * u64 bpf_get_netns_cookie(void *ctx)
  2923. * Description
  2924. * Retrieve the cookie (generated by the kernel) of the network
  2925. * namespace the input *ctx* is associated with. The network
  2926. * namespace cookie remains stable for its lifetime and provides
  2927. * a global identifier that can be assumed unique. If *ctx* is
  2928. * NULL, then the helper returns the cookie for the initial
  2929. * network namespace. The cookie itself is very similar to that
  2930. * of bpf_get_socket_cookie() helper, but for network namespaces
  2931. * instead of sockets.
  2932. * Return
  2933. * A 8-byte long opaque number.
  2934. *
  2935. * u64 bpf_get_current_ancestor_cgroup_id(int ancestor_level)
  2936. * Description
  2937. * Return id of cgroup v2 that is ancestor of the cgroup associated
  2938. * with the current task at the *ancestor_level*. The root cgroup
  2939. * is at *ancestor_level* zero and each step down the hierarchy
  2940. * increments the level. If *ancestor_level* == level of cgroup
  2941. * associated with the current task, then return value will be the
  2942. * same as that of **bpf_get_current_cgroup_id**\ ().
  2943. *
  2944. * The helper is useful to implement policies based on cgroups
  2945. * that are upper in hierarchy than immediate cgroup associated
  2946. * with the current task.
  2947. *
  2948. * The format of returned id and helper limitations are same as in
  2949. * **bpf_get_current_cgroup_id**\ ().
  2950. * Return
  2951. * The id is returned or 0 in case the id could not be retrieved.
  2952. *
  2953. * int bpf_sk_assign(struct sk_buff *skb, struct bpf_sock *sk, u64 flags)
  2954. * Description
  2955. * Assign the *sk* to the *skb*. When combined with appropriate
  2956. * routing configuration to receive the packet towards the socket,
  2957. * will cause *skb* to be delivered to the specified socket.
  2958. * Subsequent redirection of *skb* via **bpf_redirect**\ (),
  2959. * **bpf_clone_redirect**\ () or other methods outside of BPF may
  2960. * interfere with successful delivery to the socket.
  2961. *
  2962. * This operation is only valid from TC ingress path.
  2963. *
  2964. * The *flags* argument must be zero.
  2965. * Return
  2966. * 0 on success, or a negative errno in case of failure.
  2967. *
  2968. * * **-EINVAL** Unsupported flags specified.
  2969. * * **-ENOENT** Socket is unavailable for assignment.
  2970. * * **-ENETUNREACH** Socket is unreachable (wrong netns).
  2971. * * **-EOPNOTSUPP** Unsupported operation, for example a
  2972. * call from outside of TC ingress.
  2973. * * **-ESOCKTNOSUPPORT** Socket type not supported (reuseport).
  2974. */
  2975. #define __BPF_FUNC_MAPPER(FN) \
  2976. FN(unspec), \
  2977. FN(map_lookup_elem), \
  2978. FN(map_update_elem), \
  2979. FN(map_delete_elem), \
  2980. FN(probe_read), \
  2981. FN(ktime_get_ns), \
  2982. FN(trace_printk), \
  2983. FN(get_prandom_u32), \
  2984. FN(get_smp_processor_id), \
  2985. FN(skb_store_bytes), \
  2986. FN(l3_csum_replace), \
  2987. FN(l4_csum_replace), \
  2988. FN(tail_call), \
  2989. FN(clone_redirect), \
  2990. FN(get_current_pid_tgid), \
  2991. FN(get_current_uid_gid), \
  2992. FN(get_current_comm), \
  2993. FN(get_cgroup_classid), \
  2994. FN(skb_vlan_push), \
  2995. FN(skb_vlan_pop), \
  2996. FN(skb_get_tunnel_key), \
  2997. FN(skb_set_tunnel_key), \
  2998. FN(perf_event_read), \
  2999. FN(redirect), \
  3000. FN(get_route_realm), \
  3001. FN(perf_event_output), \
  3002. FN(skb_load_bytes), \
  3003. FN(get_stackid), \
  3004. FN(csum_diff), \
  3005. FN(skb_get_tunnel_opt), \
  3006. FN(skb_set_tunnel_opt), \
  3007. FN(skb_change_proto), \
  3008. FN(skb_change_type), \
  3009. FN(skb_under_cgroup), \
  3010. FN(get_hash_recalc), \
  3011. FN(get_current_task), \
  3012. FN(probe_write_user), \
  3013. FN(current_task_under_cgroup), \
  3014. FN(skb_change_tail), \
  3015. FN(skb_pull_data), \
  3016. FN(csum_update), \
  3017. FN(set_hash_invalid), \
  3018. FN(get_numa_node_id), \
  3019. FN(skb_change_head), \
  3020. FN(xdp_adjust_head), \
  3021. FN(probe_read_str), \
  3022. FN(get_socket_cookie), \
  3023. FN(get_socket_uid), \
  3024. FN(set_hash), \
  3025. FN(setsockopt), \
  3026. FN(skb_adjust_room), \
  3027. FN(redirect_map), \
  3028. FN(sk_redirect_map), \
  3029. FN(sock_map_update), \
  3030. FN(xdp_adjust_meta), \
  3031. FN(perf_event_read_value), \
  3032. FN(perf_prog_read_value), \
  3033. FN(getsockopt), \
  3034. FN(override_return), \
  3035. FN(sock_ops_cb_flags_set), \
  3036. FN(msg_redirect_map), \
  3037. FN(msg_apply_bytes), \
  3038. FN(msg_cork_bytes), \
  3039. FN(msg_pull_data), \
  3040. FN(bind), \
  3041. FN(xdp_adjust_tail), \
  3042. FN(skb_get_xfrm_state), \
  3043. FN(get_stack), \
  3044. FN(skb_load_bytes_relative), \
  3045. FN(fib_lookup), \
  3046. FN(sock_hash_update), \
  3047. FN(msg_redirect_hash), \
  3048. FN(sk_redirect_hash), \
  3049. FN(lwt_push_encap), \
  3050. FN(lwt_seg6_store_bytes), \
  3051. FN(lwt_seg6_adjust_srh), \
  3052. FN(lwt_seg6_action), \
  3053. FN(rc_repeat), \
  3054. FN(rc_keydown), \
  3055. FN(skb_cgroup_id), \
  3056. FN(get_current_cgroup_id), \
  3057. FN(get_local_storage), \
  3058. FN(sk_select_reuseport), \
  3059. FN(skb_ancestor_cgroup_id), \
  3060. FN(sk_lookup_tcp), \
  3061. FN(sk_lookup_udp), \
  3062. FN(sk_release), \
  3063. FN(map_push_elem), \
  3064. FN(map_pop_elem), \
  3065. FN(map_peek_elem), \
  3066. FN(msg_push_data), \
  3067. FN(msg_pop_data), \
  3068. FN(rc_pointer_rel), \
  3069. FN(spin_lock), \
  3070. FN(spin_unlock), \
  3071. FN(sk_fullsock), \
  3072. FN(tcp_sock), \
  3073. FN(skb_ecn_set_ce), \
  3074. FN(get_listener_sock), \
  3075. FN(skc_lookup_tcp), \
  3076. FN(tcp_check_syncookie), \
  3077. FN(sysctl_get_name), \
  3078. FN(sysctl_get_current_value), \
  3079. FN(sysctl_get_new_value), \
  3080. FN(sysctl_set_new_value), \
  3081. FN(strtol), \
  3082. FN(strtoul), \
  3083. FN(sk_storage_get), \
  3084. FN(sk_storage_delete), \
  3085. FN(send_signal), \
  3086. FN(tcp_gen_syncookie), \
  3087. FN(skb_output), \
  3088. FN(probe_read_user), \
  3089. FN(probe_read_kernel), \
  3090. FN(probe_read_user_str), \
  3091. FN(probe_read_kernel_str), \
  3092. FN(tcp_send_ack), \
  3093. FN(send_signal_thread), \
  3094. FN(jiffies64), \
  3095. FN(read_branch_records), \
  3096. FN(get_ns_current_pid_tgid), \
  3097. FN(xdp_output), \
  3098. FN(get_netns_cookie), \
  3099. FN(get_current_ancestor_cgroup_id), \
  3100. FN(sk_assign),
  3101. /* integer value in 'imm' field of BPF_CALL instruction selects which helper
  3102. * function eBPF program intends to call
  3103. */
  3104. #define __BPF_ENUM_FN(x) BPF_FUNC_ ## x
  3105. enum bpf_func_id {
  3106. __BPF_FUNC_MAPPER(__BPF_ENUM_FN)
  3107. __BPF_FUNC_MAX_ID,
  3108. };
  3109. #undef __BPF_ENUM_FN
  3110. /* All flags used by eBPF helper functions, placed here. */
  3111. /* BPF_FUNC_skb_store_bytes flags. */
  3112. enum {
  3113. BPF_F_RECOMPUTE_CSUM = (1ULL << 0),
  3114. BPF_F_INVALIDATE_HASH = (1ULL << 1),
  3115. };
  3116. /* BPF_FUNC_l3_csum_replace and BPF_FUNC_l4_csum_replace flags.
  3117. * First 4 bits are for passing the header field size.
  3118. */
  3119. enum {
  3120. BPF_F_HDR_FIELD_MASK = 0xfULL,
  3121. };
  3122. /* BPF_FUNC_l4_csum_replace flags. */
  3123. enum {
  3124. BPF_F_PSEUDO_HDR = (1ULL << 4),
  3125. BPF_F_MARK_MANGLED_0 = (1ULL << 5),
  3126. BPF_F_MARK_ENFORCE = (1ULL << 6),
  3127. };
  3128. /* BPF_FUNC_clone_redirect and BPF_FUNC_redirect flags. */
  3129. enum {
  3130. BPF_F_INGRESS = (1ULL << 0),
  3131. };
  3132. /* BPF_FUNC_skb_set_tunnel_key and BPF_FUNC_skb_get_tunnel_key flags. */
  3133. enum {
  3134. BPF_F_TUNINFO_IPV6 = (1ULL << 0),
  3135. };
  3136. /* flags for both BPF_FUNC_get_stackid and BPF_FUNC_get_stack. */
  3137. enum {
  3138. BPF_F_SKIP_FIELD_MASK = 0xffULL,
  3139. BPF_F_USER_STACK = (1ULL << 8),
  3140. /* flags used by BPF_FUNC_get_stackid only. */
  3141. BPF_F_FAST_STACK_CMP = (1ULL << 9),
  3142. BPF_F_REUSE_STACKID = (1ULL << 10),
  3143. /* flags used by BPF_FUNC_get_stack only. */
  3144. BPF_F_USER_BUILD_ID = (1ULL << 11),
  3145. };
  3146. /* BPF_FUNC_skb_set_tunnel_key flags. */
  3147. enum {
  3148. BPF_F_ZERO_CSUM_TX = (1ULL << 1),
  3149. BPF_F_DONT_FRAGMENT = (1ULL << 2),
  3150. BPF_F_SEQ_NUMBER = (1ULL << 3),
  3151. };
  3152. /* BPF_FUNC_perf_event_output, BPF_FUNC_perf_event_read and
  3153. * BPF_FUNC_perf_event_read_value flags.
  3154. */
  3155. enum {
  3156. BPF_F_INDEX_MASK = 0xffffffffULL,
  3157. BPF_F_CURRENT_CPU = BPF_F_INDEX_MASK,
  3158. /* BPF_FUNC_perf_event_output for sk_buff input context. */
  3159. BPF_F_CTXLEN_MASK = (0xfffffULL << 32),
  3160. };
  3161. /* Current network namespace */
  3162. enum {
  3163. BPF_F_CURRENT_NETNS = (-1L),
  3164. };
  3165. /* BPF_FUNC_skb_adjust_room flags. */
  3166. enum {
  3167. BPF_F_ADJ_ROOM_FIXED_GSO = (1ULL << 0),
  3168. BPF_F_ADJ_ROOM_ENCAP_L3_IPV4 = (1ULL << 1),
  3169. BPF_F_ADJ_ROOM_ENCAP_L3_IPV6 = (1ULL << 2),
  3170. BPF_F_ADJ_ROOM_ENCAP_L4_GRE = (1ULL << 3),
  3171. BPF_F_ADJ_ROOM_ENCAP_L4_UDP = (1ULL << 4),
  3172. };
  3173. enum {
  3174. BPF_ADJ_ROOM_ENCAP_L2_MASK = 0xff,
  3175. BPF_ADJ_ROOM_ENCAP_L2_SHIFT = 56,
  3176. };
  3177. #define BPF_F_ADJ_ROOM_ENCAP_L2(len) (((__u64)len & \
  3178. BPF_ADJ_ROOM_ENCAP_L2_MASK) \
  3179. << BPF_ADJ_ROOM_ENCAP_L2_SHIFT)
  3180. /* BPF_FUNC_sysctl_get_name flags. */
  3181. enum {
  3182. BPF_F_SYSCTL_BASE_NAME = (1ULL << 0),
  3183. };
  3184. /* BPF_FUNC_sk_storage_get flags */
  3185. enum {
  3186. BPF_SK_STORAGE_GET_F_CREATE = (1ULL << 0),
  3187. };
  3188. /* BPF_FUNC_read_branch_records flags. */
  3189. enum {
  3190. BPF_F_GET_BRANCH_RECORDS_SIZE = (1ULL << 0),
  3191. };
  3192. /* Mode for BPF_FUNC_skb_adjust_room helper. */
  3193. enum bpf_adj_room_mode {
  3194. BPF_ADJ_ROOM_NET,
  3195. BPF_ADJ_ROOM_MAC,
  3196. };
  3197. /* Mode for BPF_FUNC_skb_load_bytes_relative helper. */
  3198. enum bpf_hdr_start_off {
  3199. BPF_HDR_START_MAC,
  3200. BPF_HDR_START_NET,
  3201. };
  3202. /* Encapsulation type for BPF_FUNC_lwt_push_encap helper. */
  3203. enum bpf_lwt_encap_mode {
  3204. BPF_LWT_ENCAP_SEG6,
  3205. BPF_LWT_ENCAP_SEG6_INLINE,
  3206. BPF_LWT_ENCAP_IP,
  3207. };
  3208. #define __bpf_md_ptr(type, name) \
  3209. union { \
  3210. type name; \
  3211. __u64 :64; \
  3212. } __attribute__((aligned(8)))
  3213. /* user accessible mirror of in-kernel sk_buff.
  3214. * new fields can only be added to the end of this structure
  3215. */
  3216. struct __sk_buff {
  3217. __u32 len;
  3218. __u32 pkt_type;
  3219. __u32 mark;
  3220. __u32 queue_mapping;
  3221. __u32 protocol;
  3222. __u32 vlan_present;
  3223. __u32 vlan_tci;
  3224. __u32 vlan_proto;
  3225. __u32 priority;
  3226. __u32 ingress_ifindex;
  3227. __u32 ifindex;
  3228. __u32 tc_index;
  3229. __u32 cb[5];
  3230. __u32 hash;
  3231. __u32 tc_classid;
  3232. __u32 data;
  3233. __u32 data_end;
  3234. __u32 napi_id;
  3235. /* Accessed by BPF_PROG_TYPE_sk_skb types from here to ... */
  3236. __u32 family;
  3237. __u32 remote_ip4; /* Stored in network byte order */
  3238. __u32 local_ip4; /* Stored in network byte order */
  3239. __u32 remote_ip6[4]; /* Stored in network byte order */
  3240. __u32 local_ip6[4]; /* Stored in network byte order */
  3241. __u32 remote_port; /* Stored in network byte order */
  3242. __u32 local_port; /* stored in host byte order */
  3243. /* ... here. */
  3244. __u32 data_meta;
  3245. __bpf_md_ptr(struct bpf_flow_keys *, flow_keys);
  3246. __u64 tstamp;
  3247. __u32 wire_len;
  3248. __u32 gso_segs;
  3249. __bpf_md_ptr(struct bpf_sock *, sk);
  3250. __u32 gso_size;
  3251. };
  3252. struct bpf_tunnel_key {
  3253. __u32 tunnel_id;
  3254. union {
  3255. __u32 remote_ipv4;
  3256. __u32 remote_ipv6[4];
  3257. };
  3258. __u8 tunnel_tos;
  3259. __u8 tunnel_ttl;
  3260. __u16 tunnel_ext; /* Padding, future use. */
  3261. __u32 tunnel_label;
  3262. };
  3263. /* user accessible mirror of in-kernel xfrm_state.
  3264. * new fields can only be added to the end of this structure
  3265. */
  3266. struct bpf_xfrm_state {
  3267. __u32 reqid;
  3268. __u32 spi; /* Stored in network byte order */
  3269. __u16 family;
  3270. __u16 ext; /* Padding, future use. */
  3271. union {
  3272. __u32 remote_ipv4; /* Stored in network byte order */
  3273. __u32 remote_ipv6[4]; /* Stored in network byte order */
  3274. };
  3275. };
  3276. /* Generic BPF return codes which all BPF program types may support.
  3277. * The values are binary compatible with their TC_ACT_* counter-part to
  3278. * provide backwards compatibility with existing SCHED_CLS and SCHED_ACT
  3279. * programs.
  3280. *
  3281. * XDP is handled seprately, see XDP_*.
  3282. */
  3283. enum bpf_ret_code {
  3284. BPF_OK = 0,
  3285. /* 1 reserved */
  3286. BPF_DROP = 2,
  3287. /* 3-6 reserved */
  3288. BPF_REDIRECT = 7,
  3289. /* >127 are reserved for prog type specific return codes.
  3290. *
  3291. * BPF_LWT_REROUTE: used by BPF_PROG_TYPE_LWT_IN and
  3292. * BPF_PROG_TYPE_LWT_XMIT to indicate that skb had been
  3293. * changed and should be routed based on its new L3 header.
  3294. * (This is an L3 redirect, as opposed to L2 redirect
  3295. * represented by BPF_REDIRECT above).
  3296. */
  3297. BPF_LWT_REROUTE = 128,
  3298. };
  3299. struct bpf_sock {
  3300. __u32 bound_dev_if;
  3301. __u32 family;
  3302. __u32 type;
  3303. __u32 protocol;
  3304. __u32 mark;
  3305. __u32 priority;
  3306. /* IP address also allows 1 and 2 bytes access */
  3307. __u32 src_ip4;
  3308. __u32 src_ip6[4];
  3309. __u32 src_port; /* host byte order */
  3310. __u32 dst_port; /* network byte order */
  3311. __u32 dst_ip4;
  3312. __u32 dst_ip6[4];
  3313. __u32 state;
  3314. };
  3315. struct bpf_tcp_sock {
  3316. __u32 snd_cwnd; /* Sending congestion window */
  3317. __u32 srtt_us; /* smoothed round trip time << 3 in usecs */
  3318. __u32 rtt_min;
  3319. __u32 snd_ssthresh; /* Slow start size threshold */
  3320. __u32 rcv_nxt; /* What we want to receive next */
  3321. __u32 snd_nxt; /* Next sequence we send */
  3322. __u32 snd_una; /* First byte we want an ack for */
  3323. __u32 mss_cache; /* Cached effective mss, not including SACKS */
  3324. __u32 ecn_flags; /* ECN status bits. */
  3325. __u32 rate_delivered; /* saved rate sample: packets delivered */
  3326. __u32 rate_interval_us; /* saved rate sample: time elapsed */
  3327. __u32 packets_out; /* Packets which are "in flight" */
  3328. __u32 retrans_out; /* Retransmitted packets out */
  3329. __u32 total_retrans; /* Total retransmits for entire connection */
  3330. __u32 segs_in; /* RFC4898 tcpEStatsPerfSegsIn
  3331. * total number of segments in.
  3332. */
  3333. __u32 data_segs_in; /* RFC4898 tcpEStatsPerfDataSegsIn
  3334. * total number of data segments in.
  3335. */
  3336. __u32 segs_out; /* RFC4898 tcpEStatsPerfSegsOut
  3337. * The total number of segments sent.
  3338. */
  3339. __u32 data_segs_out; /* RFC4898 tcpEStatsPerfDataSegsOut
  3340. * total number of data segments sent.
  3341. */
  3342. __u32 lost_out; /* Lost packets */
  3343. __u32 sacked_out; /* SACK'd packets */
  3344. __u64 bytes_received; /* RFC4898 tcpEStatsAppHCThruOctetsReceived
  3345. * sum(delta(rcv_nxt)), or how many bytes
  3346. * were acked.
  3347. */
  3348. __u64 bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked
  3349. * sum(delta(snd_una)), or how many bytes
  3350. * were acked.
  3351. */
  3352. __u32 dsack_dups; /* RFC4898 tcpEStatsStackDSACKDups
  3353. * total number of DSACK blocks received
  3354. */
  3355. __u32 delivered; /* Total data packets delivered incl. rexmits */
  3356. __u32 delivered_ce; /* Like the above but only ECE marked packets */
  3357. __u32 icsk_retransmits; /* Number of unrecovered [RTO] timeouts */
  3358. };
  3359. struct bpf_sock_tuple {
  3360. union {
  3361. struct {
  3362. __be32 saddr;
  3363. __be32 daddr;
  3364. __be16 sport;
  3365. __be16 dport;
  3366. } ipv4;
  3367. struct {
  3368. __be32 saddr[4];
  3369. __be32 daddr[4];
  3370. __be16 sport;
  3371. __be16 dport;
  3372. } ipv6;
  3373. };
  3374. };
  3375. struct bpf_xdp_sock {
  3376. __u32 queue_id;
  3377. };
  3378. #define XDP_PACKET_HEADROOM 256
  3379. /* User return codes for XDP prog type.
  3380. * A valid XDP program must return one of these defined values. All other
  3381. * return codes are reserved for future use. Unknown return codes will
  3382. * result in packet drops and a warning via bpf_warn_invalid_xdp_action().
  3383. */
  3384. enum xdp_action {
  3385. XDP_ABORTED = 0,
  3386. XDP_DROP,
  3387. XDP_PASS,
  3388. XDP_TX,
  3389. XDP_REDIRECT,
  3390. };
  3391. /* user accessible metadata for XDP packet hook
  3392. * new fields must be added to the end of this structure
  3393. */
  3394. struct xdp_md {
  3395. __u32 data;
  3396. __u32 data_end;
  3397. __u32 data_meta;
  3398. /* Below access go through struct xdp_rxq_info */
  3399. __u32 ingress_ifindex; /* rxq->dev->ifindex */
  3400. __u32 rx_queue_index; /* rxq->queue_index */
  3401. };
  3402. enum sk_action {
  3403. SK_DROP = 0,
  3404. SK_PASS,
  3405. };
  3406. /* user accessible metadata for SK_MSG packet hook, new fields must
  3407. * be added to the end of this structure
  3408. */
  3409. struct sk_msg_md {
  3410. __bpf_md_ptr(void *, data);
  3411. __bpf_md_ptr(void *, data_end);
  3412. __u32 family;
  3413. __u32 remote_ip4; /* Stored in network byte order */
  3414. __u32 local_ip4; /* Stored in network byte order */
  3415. __u32 remote_ip6[4]; /* Stored in network byte order */
  3416. __u32 local_ip6[4]; /* Stored in network byte order */
  3417. __u32 remote_port; /* Stored in network byte order */
  3418. __u32 local_port; /* stored in host byte order */
  3419. __u32 size; /* Total size of sk_msg */
  3420. };
  3421. struct sk_reuseport_md {
  3422. /*
  3423. * Start of directly accessible data. It begins from
  3424. * the tcp/udp header.
  3425. */
  3426. __bpf_md_ptr(void *, data);
  3427. /* End of directly accessible data */
  3428. __bpf_md_ptr(void *, data_end);
  3429. /*
  3430. * Total length of packet (starting from the tcp/udp header).
  3431. * Note that the directly accessible bytes (data_end - data)
  3432. * could be less than this "len". Those bytes could be
  3433. * indirectly read by a helper "bpf_skb_load_bytes()".
  3434. */
  3435. __u32 len;
  3436. /*
  3437. * Eth protocol in the mac header (network byte order). e.g.
  3438. * ETH_P_IP(0x0800) and ETH_P_IPV6(0x86DD)
  3439. */
  3440. __u32 eth_protocol;
  3441. __u32 ip_protocol; /* IP protocol. e.g. IPPROTO_TCP, IPPROTO_UDP */
  3442. __u32 bind_inany; /* Is sock bound to an INANY address? */
  3443. __u32 hash; /* A hash of the packet 4 tuples */
  3444. };
  3445. #define BPF_TAG_SIZE 8
  3446. struct bpf_prog_info {
  3447. __u32 type;
  3448. __u32 id;
  3449. __u8 tag[BPF_TAG_SIZE];
  3450. __u32 jited_prog_len;
  3451. __u32 xlated_prog_len;
  3452. __aligned_u64 jited_prog_insns;
  3453. __aligned_u64 xlated_prog_insns;
  3454. __u64 load_time; /* ns since boottime */
  3455. __u32 created_by_uid;
  3456. __u32 nr_map_ids;
  3457. __aligned_u64 map_ids;
  3458. char name[BPF_OBJ_NAME_LEN];
  3459. __u32 ifindex;
  3460. __u32 gpl_compatible:1;
  3461. __u32 :31; /* alignment pad */
  3462. __u64 netns_dev;
  3463. __u64 netns_ino;
  3464. __u32 nr_jited_ksyms;
  3465. __u32 nr_jited_func_lens;
  3466. __aligned_u64 jited_ksyms;
  3467. __aligned_u64 jited_func_lens;
  3468. __u32 btf_id;
  3469. __u32 func_info_rec_size;
  3470. __aligned_u64 func_info;
  3471. __u32 nr_func_info;
  3472. __u32 nr_line_info;
  3473. __aligned_u64 line_info;
  3474. __aligned_u64 jited_line_info;
  3475. __u32 nr_jited_line_info;
  3476. __u32 line_info_rec_size;
  3477. __u32 jited_line_info_rec_size;
  3478. __u32 nr_prog_tags;
  3479. __aligned_u64 prog_tags;
  3480. __u64 run_time_ns;
  3481. __u64 run_cnt;
  3482. } __attribute__((aligned(8)));
  3483. struct bpf_map_info {
  3484. __u32 type;
  3485. __u32 id;
  3486. __u32 key_size;
  3487. __u32 value_size;
  3488. __u32 max_entries;
  3489. __u32 map_flags;
  3490. char name[BPF_OBJ_NAME_LEN];
  3491. __u32 ifindex;
  3492. __u32 btf_vmlinux_value_type_id;
  3493. __u64 netns_dev;
  3494. __u64 netns_ino;
  3495. __u32 btf_id;
  3496. __u32 btf_key_type_id;
  3497. __u32 btf_value_type_id;
  3498. } __attribute__((aligned(8)));
  3499. struct bpf_btf_info {
  3500. __aligned_u64 btf;
  3501. __u32 btf_size;
  3502. __u32 id;
  3503. } __attribute__((aligned(8)));
  3504. /* User bpf_sock_addr struct to access socket fields and sockaddr struct passed
  3505. * by user and intended to be used by socket (e.g. to bind to, depends on
  3506. * attach attach type).
  3507. */
  3508. struct bpf_sock_addr {
  3509. __u32 user_family; /* Allows 4-byte read, but no write. */
  3510. __u32 user_ip4; /* Allows 1,2,4-byte read and 4-byte write.
  3511. * Stored in network byte order.
  3512. */
  3513. __u32 user_ip6[4]; /* Allows 1,2,4,8-byte read and 4,8-byte write.
  3514. * Stored in network byte order.
  3515. */
  3516. __u32 user_port; /* Allows 4-byte read and write.
  3517. * Stored in network byte order
  3518. */
  3519. __u32 family; /* Allows 4-byte read, but no write */
  3520. __u32 type; /* Allows 4-byte read, but no write */
  3521. __u32 protocol; /* Allows 4-byte read, but no write */
  3522. __u32 msg_src_ip4; /* Allows 1,2,4-byte read and 4-byte write.
  3523. * Stored in network byte order.
  3524. */
  3525. __u32 msg_src_ip6[4]; /* Allows 1,2,4,8-byte read and 4,8-byte write.
  3526. * Stored in network byte order.
  3527. */
  3528. __bpf_md_ptr(struct bpf_sock *, sk);
  3529. };
  3530. /* User bpf_sock_ops struct to access socket values and specify request ops
  3531. * and their replies.
  3532. * Some of this fields are in network (bigendian) byte order and may need
  3533. * to be converted before use (bpf_ntohl() defined in samples/bpf/bpf_endian.h).
  3534. * New fields can only be added at the end of this structure
  3535. */
  3536. struct bpf_sock_ops {
  3537. __u32 op;
  3538. union {
  3539. __u32 args[4]; /* Optionally passed to bpf program */
  3540. __u32 reply; /* Returned by bpf program */
  3541. __u32 replylong[4]; /* Optionally returned by bpf prog */
  3542. };
  3543. __u32 family;
  3544. __u32 remote_ip4; /* Stored in network byte order */
  3545. __u32 local_ip4; /* Stored in network byte order */
  3546. __u32 remote_ip6[4]; /* Stored in network byte order */
  3547. __u32 local_ip6[4]; /* Stored in network byte order */
  3548. __u32 remote_port; /* Stored in network byte order */
  3549. __u32 local_port; /* stored in host byte order */
  3550. __u32 is_fullsock; /* Some TCP fields are only valid if
  3551. * there is a full socket. If not, the
  3552. * fields read as zero.
  3553. */
  3554. __u32 snd_cwnd;
  3555. __u32 srtt_us; /* Averaged RTT << 3 in usecs */
  3556. __u32 bpf_sock_ops_cb_flags; /* flags defined in uapi/linux/tcp.h */
  3557. __u32 state;
  3558. __u32 rtt_min;
  3559. __u32 snd_ssthresh;
  3560. __u32 rcv_nxt;
  3561. __u32 snd_nxt;
  3562. __u32 snd_una;
  3563. __u32 mss_cache;
  3564. __u32 ecn_flags;
  3565. __u32 rate_delivered;
  3566. __u32 rate_interval_us;
  3567. __u32 packets_out;
  3568. __u32 retrans_out;
  3569. __u32 total_retrans;
  3570. __u32 segs_in;
  3571. __u32 data_segs_in;
  3572. __u32 segs_out;
  3573. __u32 data_segs_out;
  3574. __u32 lost_out;
  3575. __u32 sacked_out;
  3576. __u32 sk_txhash;
  3577. __u64 bytes_received;
  3578. __u64 bytes_acked;
  3579. __bpf_md_ptr(struct bpf_sock *, sk);
  3580. };
  3581. /* Definitions for bpf_sock_ops_cb_flags */
  3582. enum {
  3583. BPF_SOCK_OPS_RTO_CB_FLAG = (1<<0),
  3584. BPF_SOCK_OPS_RETRANS_CB_FLAG = (1<<1),
  3585. BPF_SOCK_OPS_STATE_CB_FLAG = (1<<2),
  3586. BPF_SOCK_OPS_RTT_CB_FLAG = (1<<3),
  3587. /* Mask of all currently supported cb flags */
  3588. BPF_SOCK_OPS_ALL_CB_FLAGS = 0xF,
  3589. };
  3590. /* List of known BPF sock_ops operators.
  3591. * New entries can only be added at the end
  3592. */
  3593. enum {
  3594. BPF_SOCK_OPS_VOID,
  3595. BPF_SOCK_OPS_TIMEOUT_INIT, /* Should return SYN-RTO value to use or
  3596. * -1 if default value should be used
  3597. */
  3598. BPF_SOCK_OPS_RWND_INIT, /* Should return initial advertized
  3599. * window (in packets) or -1 if default
  3600. * value should be used
  3601. */
  3602. BPF_SOCK_OPS_TCP_CONNECT_CB, /* Calls BPF program right before an
  3603. * active connection is initialized
  3604. */
  3605. BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB, /* Calls BPF program when an
  3606. * active connection is
  3607. * established
  3608. */
  3609. BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB, /* Calls BPF program when a
  3610. * passive connection is
  3611. * established
  3612. */
  3613. BPF_SOCK_OPS_NEEDS_ECN, /* If connection's congestion control
  3614. * needs ECN
  3615. */
  3616. BPF_SOCK_OPS_BASE_RTT, /* Get base RTT. The correct value is
  3617. * based on the path and may be
  3618. * dependent on the congestion control
  3619. * algorithm. In general it indicates
  3620. * a congestion threshold. RTTs above
  3621. * this indicate congestion
  3622. */
  3623. BPF_SOCK_OPS_RTO_CB, /* Called when an RTO has triggered.
  3624. * Arg1: value of icsk_retransmits
  3625. * Arg2: value of icsk_rto
  3626. * Arg3: whether RTO has expired
  3627. */
  3628. BPF_SOCK_OPS_RETRANS_CB, /* Called when skb is retransmitted.
  3629. * Arg1: sequence number of 1st byte
  3630. * Arg2: # segments
  3631. * Arg3: return value of
  3632. * tcp_transmit_skb (0 => success)
  3633. */
  3634. BPF_SOCK_OPS_STATE_CB, /* Called when TCP changes state.
  3635. * Arg1: old_state
  3636. * Arg2: new_state
  3637. */
  3638. BPF_SOCK_OPS_TCP_LISTEN_CB, /* Called on listen(2), right after
  3639. * socket transition to LISTEN state.
  3640. */
  3641. BPF_SOCK_OPS_RTT_CB, /* Called on every RTT.
  3642. */
  3643. };
  3644. /* List of TCP states. There is a build check in net/ipv4/tcp.c to detect
  3645. * changes between the TCP and BPF versions. Ideally this should never happen.
  3646. * If it does, we need to add code to convert them before calling
  3647. * the BPF sock_ops function.
  3648. */
  3649. enum {
  3650. BPF_TCP_ESTABLISHED = 1,
  3651. BPF_TCP_SYN_SENT,
  3652. BPF_TCP_SYN_RECV,
  3653. BPF_TCP_FIN_WAIT1,
  3654. BPF_TCP_FIN_WAIT2,
  3655. BPF_TCP_TIME_WAIT,
  3656. BPF_TCP_CLOSE,
  3657. BPF_TCP_CLOSE_WAIT,
  3658. BPF_TCP_LAST_ACK,
  3659. BPF_TCP_LISTEN,
  3660. BPF_TCP_CLOSING, /* Now a valid state */
  3661. BPF_TCP_NEW_SYN_RECV,
  3662. BPF_TCP_MAX_STATES /* Leave at the end! */
  3663. };
  3664. enum {
  3665. TCP_BPF_IW = 1001, /* Set TCP initial congestion window */
  3666. TCP_BPF_SNDCWND_CLAMP = 1002, /* Set sndcwnd_clamp */
  3667. };
  3668. struct bpf_perf_event_value {
  3669. __u64 counter;
  3670. __u64 enabled;
  3671. __u64 running;
  3672. };
  3673. enum {
  3674. BPF_DEVCG_ACC_MKNOD = (1ULL << 0),
  3675. BPF_DEVCG_ACC_READ = (1ULL << 1),
  3676. BPF_DEVCG_ACC_WRITE = (1ULL << 2),
  3677. };
  3678. enum {
  3679. BPF_DEVCG_DEV_BLOCK = (1ULL << 0),
  3680. BPF_DEVCG_DEV_CHAR = (1ULL << 1),
  3681. };
  3682. struct bpf_cgroup_dev_ctx {
  3683. /* access_type encoded as (BPF_DEVCG_ACC_* << 16) | BPF_DEVCG_DEV_* */
  3684. __u32 access_type;
  3685. __u32 major;
  3686. __u32 minor;
  3687. };
  3688. struct bpf_raw_tracepoint_args {
  3689. __u64 args[0];
  3690. };
  3691. /* DIRECT: Skip the FIB rules and go to FIB table associated with device
  3692. * OUTPUT: Do lookup from egress perspective; default is ingress
  3693. */
  3694. enum {
  3695. BPF_FIB_LOOKUP_DIRECT = (1U << 0),
  3696. BPF_FIB_LOOKUP_OUTPUT = (1U << 1),
  3697. };
  3698. enum {
  3699. BPF_FIB_LKUP_RET_SUCCESS, /* lookup successful */
  3700. BPF_FIB_LKUP_RET_BLACKHOLE, /* dest is blackholed; can be dropped */
  3701. BPF_FIB_LKUP_RET_UNREACHABLE, /* dest is unreachable; can be dropped */
  3702. BPF_FIB_LKUP_RET_PROHIBIT, /* dest not allowed; can be dropped */
  3703. BPF_FIB_LKUP_RET_NOT_FWDED, /* packet is not forwarded */
  3704. BPF_FIB_LKUP_RET_FWD_DISABLED, /* fwding is not enabled on ingress */
  3705. BPF_FIB_LKUP_RET_UNSUPP_LWT, /* fwd requires encapsulation */
  3706. BPF_FIB_LKUP_RET_NO_NEIGH, /* no neighbor entry for nh */
  3707. BPF_FIB_LKUP_RET_FRAG_NEEDED, /* fragmentation required to fwd */
  3708. };
  3709. struct bpf_fib_lookup {
  3710. /* input: network family for lookup (AF_INET, AF_INET6)
  3711. * output: network family of egress nexthop
  3712. */
  3713. __u8 family;
  3714. /* set if lookup is to consider L4 data - e.g., FIB rules */
  3715. __u8 l4_protocol;
  3716. __be16 sport;
  3717. __be16 dport;
  3718. /* total length of packet from network header - used for MTU check */
  3719. __u16 tot_len;
  3720. /* input: L3 device index for lookup
  3721. * output: device index from FIB lookup
  3722. */
  3723. __u32 ifindex;
  3724. union {
  3725. /* inputs to lookup */
  3726. __u8 tos; /* AF_INET */
  3727. __be32 flowinfo; /* AF_INET6, flow_label + priority */
  3728. /* output: metric of fib result (IPv4/IPv6 only) */
  3729. __u32 rt_metric;
  3730. };
  3731. union {
  3732. __be32 ipv4_src;
  3733. __u32 ipv6_src[4]; /* in6_addr; network order */
  3734. };
  3735. /* input to bpf_fib_lookup, ipv{4,6}_dst is destination address in
  3736. * network header. output: bpf_fib_lookup sets to gateway address
  3737. * if FIB lookup returns gateway route
  3738. */
  3739. union {
  3740. __be32 ipv4_dst;
  3741. __u32 ipv6_dst[4]; /* in6_addr; network order */
  3742. };
  3743. /* output */
  3744. __be16 h_vlan_proto;
  3745. __be16 h_vlan_TCI;
  3746. __u8 smac[6]; /* ETH_ALEN */
  3747. __u8 dmac[6]; /* ETH_ALEN */
  3748. };
  3749. enum bpf_task_fd_type {
  3750. BPF_FD_TYPE_RAW_TRACEPOINT, /* tp name */
  3751. BPF_FD_TYPE_TRACEPOINT, /* tp name */
  3752. BPF_FD_TYPE_KPROBE, /* (symbol + offset) or addr */
  3753. BPF_FD_TYPE_KRETPROBE, /* (symbol + offset) or addr */
  3754. BPF_FD_TYPE_UPROBE, /* filename + offset */
  3755. BPF_FD_TYPE_URETPROBE, /* filename + offset */
  3756. };
  3757. enum {
  3758. BPF_FLOW_DISSECTOR_F_PARSE_1ST_FRAG = (1U << 0),
  3759. BPF_FLOW_DISSECTOR_F_STOP_AT_FLOW_LABEL = (1U << 1),
  3760. BPF_FLOW_DISSECTOR_F_STOP_AT_ENCAP = (1U << 2),
  3761. };
  3762. struct bpf_flow_keys {
  3763. __u16 nhoff;
  3764. __u16 thoff;
  3765. __u16 addr_proto; /* ETH_P_* of valid addrs */
  3766. __u8 is_frag;
  3767. __u8 is_first_frag;
  3768. __u8 is_encap;
  3769. __u8 ip_proto;
  3770. __be16 n_proto;
  3771. __be16 sport;
  3772. __be16 dport;
  3773. union {
  3774. struct {
  3775. __be32 ipv4_src;
  3776. __be32 ipv4_dst;
  3777. };
  3778. struct {
  3779. __u32 ipv6_src[4]; /* in6_addr; network order */
  3780. __u32 ipv6_dst[4]; /* in6_addr; network order */
  3781. };
  3782. };
  3783. __u32 flags;
  3784. __be32 flow_label;
  3785. };
  3786. struct bpf_func_info {
  3787. __u32 insn_off;
  3788. __u32 type_id;
  3789. };
  3790. #define BPF_LINE_INFO_LINE_NUM(line_col) ((line_col) >> 10)
  3791. #define BPF_LINE_INFO_LINE_COL(line_col) ((line_col) & 0x3ff)
  3792. struct bpf_line_info {
  3793. __u32 insn_off;
  3794. __u32 file_name_off;
  3795. __u32 line_off;
  3796. __u32 line_col;
  3797. };
  3798. struct bpf_spin_lock {
  3799. __u32 val;
  3800. };
  3801. struct bpf_sysctl {
  3802. __u32 write; /* Sysctl is being read (= 0) or written (= 1).
  3803. * Allows 1,2,4-byte read, but no write.
  3804. */
  3805. __u32 file_pos; /* Sysctl file position to read from, write to.
  3806. * Allows 1,2,4-byte read an 4-byte write.
  3807. */
  3808. };
  3809. struct bpf_sockopt {
  3810. __bpf_md_ptr(struct bpf_sock *, sk);
  3811. __bpf_md_ptr(void *, optval);
  3812. __bpf_md_ptr(void *, optval_end);
  3813. __s32 level;
  3814. __s32 optname;
  3815. __s32 optlen;
  3816. __s32 retval;
  3817. };
  3818. struct bpf_pidns_info {
  3819. __u32 pid;
  3820. __u32 tgid;
  3821. };
  3822. #endif /* __LINUX_BPF_H__ */