skbuff.h 79 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522252325242525252625272528252925302531253225332534253525362537253825392540254125422543254425452546254725482549255025512552255325542555255625572558255925602561256225632564256525662567256825692570257125722573257425752576257725782579258025812582258325842585258625872588258925902591259225932594259525962597259825992600260126022603260426052606260726082609261026112612261326142615261626172618261926202621262226232624262526262627262826292630263126322633263426352636263726382639264026412642264326442645264626472648264926502651265226532654265526562657265826592660266126622663266426652666266726682669267026712672267326742675267626772678267926802681268226832684268526862687268826892690269126922693269426952696269726982699270027012702270327042705270627072708270927102711271227132714271527162717271827192720272127222723272427252726272727282729273027312732273327342735273627372738273927402741274227432744274527462747274827492750275127522753275427552756275727582759276027612762276327642765276627672768276927702771277227732774277527762777277827792780278127822783278427852786278727882789279027912792279327942795279627972798279928002801280228032804280528062807280828092810281128122813281428152816281728182819282028212822282328242825282628272828282928302831283228332834283528362837
  1. /*
  2. * Definitions for the 'struct sk_buff' memory handlers.
  3. *
  4. * Authors:
  5. * Alan Cox, <gw4pts@gw4pts.ampr.org>
  6. * Florian La Roche, <rzsfl@rz.uni-sb.de>
  7. *
  8. * This program is free software; you can redistribute it and/or
  9. * modify it under the terms of the GNU General Public License
  10. * as published by the Free Software Foundation; either version
  11. * 2 of the License, or (at your option) any later version.
  12. */
  13. #ifndef _LINUX_SKBUFF_H
  14. #define _LINUX_SKBUFF_H
  15. #include <linux/kernel.h>
  16. #include <linux/kmemcheck.h>
  17. #include <linux/compiler.h>
  18. #include <linux/time.h>
  19. #include <linux/bug.h>
  20. #include <linux/cache.h>
  21. #include <linux/atomic.h>
  22. #include <asm/types.h>
  23. #include <linux/spinlock.h>
  24. #include <linux/net.h>
  25. #include <linux/textsearch.h>
  26. #include <net/checksum.h>
  27. #include <linux/rcupdate.h>
  28. #include <linux/dmaengine.h>
  29. #include <linux/hrtimer.h>
  30. #include <linux/dma-mapping.h>
  31. #include <linux/netdev_features.h>
  32. #include <net/flow_keys.h>
  33. /* Don't change this without changing skb_csum_unnecessary! */
  34. #define CHECKSUM_NONE 0
  35. #define CHECKSUM_UNNECESSARY 1
  36. #define CHECKSUM_COMPLETE 2
  37. #define CHECKSUM_PARTIAL 3
  38. #define SKB_DATA_ALIGN(X) (((X) + (SMP_CACHE_BYTES - 1)) & \
  39. ~(SMP_CACHE_BYTES - 1))
  40. #define SKB_WITH_OVERHEAD(X) \
  41. ((X) - SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))
  42. #define SKB_MAX_ORDER(X, ORDER) \
  43. SKB_WITH_OVERHEAD((PAGE_SIZE << (ORDER)) - (X))
  44. #define SKB_MAX_HEAD(X) (SKB_MAX_ORDER((X), 0))
  45. #define SKB_MAX_ALLOC (SKB_MAX_ORDER(0, 2))
  46. /* return minimum truesize of one skb containing X bytes of data */
  47. #define SKB_TRUESIZE(X) ((X) + \
  48. SKB_DATA_ALIGN(sizeof(struct sk_buff)) + \
  49. SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))
  50. /* A. Checksumming of received packets by device.
  51. *
  52. * NONE: device failed to checksum this packet.
  53. * skb->csum is undefined.
  54. *
  55. * UNNECESSARY: device parsed packet and wouldbe verified checksum.
  56. * skb->csum is undefined.
  57. * It is bad option, but, unfortunately, many of vendors do this.
  58. * Apparently with secret goal to sell you new device, when you
  59. * will add new protocol to your host. F.e. IPv6. 8)
  60. *
  61. * COMPLETE: the most generic way. Device supplied checksum of _all_
  62. * the packet as seen by netif_rx in skb->csum.
  63. * NOTE: Even if device supports only some protocols, but
  64. * is able to produce some skb->csum, it MUST use COMPLETE,
  65. * not UNNECESSARY.
  66. *
  67. * PARTIAL: identical to the case for output below. This may occur
  68. * on a packet received directly from another Linux OS, e.g.,
  69. * a virtualised Linux kernel on the same host. The packet can
  70. * be treated in the same way as UNNECESSARY except that on
  71. * output (i.e., forwarding) the checksum must be filled in
  72. * by the OS or the hardware.
  73. *
  74. * B. Checksumming on output.
  75. *
  76. * NONE: skb is checksummed by protocol or csum is not required.
  77. *
  78. * PARTIAL: device is required to csum packet as seen by hard_start_xmit
  79. * from skb->csum_start to the end and to record the checksum
  80. * at skb->csum_start + skb->csum_offset.
  81. *
  82. * Device must show its capabilities in dev->features, set
  83. * at device setup time.
  84. * NETIF_F_HW_CSUM - it is clever device, it is able to checksum
  85. * everything.
  86. * NETIF_F_IP_CSUM - device is dumb. It is able to csum only
  87. * TCP/UDP over IPv4. Sigh. Vendors like this
  88. * way by an unknown reason. Though, see comment above
  89. * about CHECKSUM_UNNECESSARY. 8)
  90. * NETIF_F_IPV6_CSUM about as dumb as the last one but does IPv6 instead.
  91. *
  92. * UNNECESSARY: device will do per protocol specific csum. Protocol drivers
  93. * that do not want net to perform the checksum calculation should use
  94. * this flag in their outgoing skbs.
  95. * NETIF_F_FCOE_CRC this indicates the device can do FCoE FC CRC
  96. * offload. Correspondingly, the FCoE protocol driver
  97. * stack should use CHECKSUM_UNNECESSARY.
  98. *
  99. * Any questions? No questions, good. --ANK
  100. */
  101. struct net_device;
  102. struct scatterlist;
  103. struct pipe_inode_info;
  104. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  105. struct nf_conntrack {
  106. atomic_t use;
  107. };
  108. #endif
  109. #ifdef CONFIG_BRIDGE_NETFILTER
  110. struct nf_bridge_info {
  111. atomic_t use;
  112. unsigned int mask;
  113. struct net_device *physindev;
  114. struct net_device *physoutdev;
  115. unsigned long data[32 / sizeof(unsigned long)];
  116. };
  117. #endif
  118. struct sk_buff_head {
  119. /* These two members must be first. */
  120. struct sk_buff *next;
  121. struct sk_buff *prev;
  122. __u32 qlen;
  123. spinlock_t lock;
  124. };
  125. struct sk_buff;
  126. /* To allow 64K frame to be packed as single skb without frag_list we
  127. * require 64K/PAGE_SIZE pages plus 1 additional page to allow for
  128. * buffers which do not start on a page boundary.
  129. *
  130. * Since GRO uses frags we allocate at least 16 regardless of page
  131. * size.
  132. */
  133. #if (65536/PAGE_SIZE + 1) < 16
  134. #define MAX_SKB_FRAGS 16UL
  135. #else
  136. #define MAX_SKB_FRAGS (65536/PAGE_SIZE + 1)
  137. #endif
  138. typedef struct skb_frag_struct skb_frag_t;
  139. struct skb_frag_struct {
  140. struct {
  141. struct page *p;
  142. } page;
  143. #if (BITS_PER_LONG > 32) || (PAGE_SIZE >= 65536)
  144. __u32 page_offset;
  145. __u32 size;
  146. #else
  147. __u16 page_offset;
  148. __u16 size;
  149. #endif
  150. };
  151. static inline unsigned int skb_frag_size(const skb_frag_t *frag)
  152. {
  153. return frag->size;
  154. }
  155. static inline void skb_frag_size_set(skb_frag_t *frag, unsigned int size)
  156. {
  157. frag->size = size;
  158. }
  159. static inline void skb_frag_size_add(skb_frag_t *frag, int delta)
  160. {
  161. frag->size += delta;
  162. }
  163. static inline void skb_frag_size_sub(skb_frag_t *frag, int delta)
  164. {
  165. frag->size -= delta;
  166. }
  167. #define HAVE_HW_TIME_STAMP
  168. /**
  169. * struct skb_shared_hwtstamps - hardware time stamps
  170. * @hwtstamp: hardware time stamp transformed into duration
  171. * since arbitrary point in time
  172. * @syststamp: hwtstamp transformed to system time base
  173. *
  174. * Software time stamps generated by ktime_get_real() are stored in
  175. * skb->tstamp. The relation between the different kinds of time
  176. * stamps is as follows:
  177. *
  178. * syststamp and tstamp can be compared against each other in
  179. * arbitrary combinations. The accuracy of a
  180. * syststamp/tstamp/"syststamp from other device" comparison is
  181. * limited by the accuracy of the transformation into system time
  182. * base. This depends on the device driver and its underlying
  183. * hardware.
  184. *
  185. * hwtstamps can only be compared against other hwtstamps from
  186. * the same device.
  187. *
  188. * This structure is attached to packets as part of the
  189. * &skb_shared_info. Use skb_hwtstamps() to get a pointer.
  190. */
  191. struct skb_shared_hwtstamps {
  192. ktime_t hwtstamp;
  193. ktime_t syststamp;
  194. };
  195. /* Definitions for tx_flags in struct skb_shared_info */
  196. enum {
  197. /* generate hardware time stamp */
  198. SKBTX_HW_TSTAMP = 1 << 0,
  199. /* generate software time stamp */
  200. SKBTX_SW_TSTAMP = 1 << 1,
  201. /* device driver is going to provide hardware time stamp */
  202. SKBTX_IN_PROGRESS = 1 << 2,
  203. /* device driver supports TX zero-copy buffers */
  204. SKBTX_DEV_ZEROCOPY = 1 << 3,
  205. /* generate wifi status information (where possible) */
  206. SKBTX_WIFI_STATUS = 1 << 4,
  207. /* This indicates at least one fragment might be overwritten
  208. * (as in vmsplice(), sendfile() ...)
  209. * If we need to compute a TX checksum, we'll need to copy
  210. * all frags to avoid possible bad checksum
  211. */
  212. SKBTX_SHARED_FRAG = 1 << 5,
  213. };
  214. /*
  215. * The callback notifies userspace to release buffers when skb DMA is done in
  216. * lower device, the skb last reference should be 0 when calling this.
  217. * The zerocopy_success argument is true if zero copy transmit occurred,
  218. * false on data copy or out of memory error caused by data copy attempt.
  219. * The ctx field is used to track device context.
  220. * The desc field is used to track userspace buffer index.
  221. */
  222. struct ubuf_info {
  223. void (*callback)(struct ubuf_info *, bool zerocopy_success);
  224. void *ctx;
  225. unsigned long desc;
  226. };
  227. /* This data is invariant across clones and lives at
  228. * the end of the header data, ie. at skb->end.
  229. */
  230. struct skb_shared_info {
  231. unsigned char nr_frags;
  232. __u8 tx_flags;
  233. unsigned short gso_size;
  234. /* Warning: this field is not always filled in (UFO)! */
  235. unsigned short gso_segs;
  236. unsigned short gso_type;
  237. struct sk_buff *frag_list;
  238. struct skb_shared_hwtstamps hwtstamps;
  239. __be32 ip6_frag_id;
  240. /*
  241. * Warning : all fields before dataref are cleared in __alloc_skb()
  242. */
  243. atomic_t dataref;
  244. /* Intermediate layers must ensure that destructor_arg
  245. * remains valid until skb destructor */
  246. void * destructor_arg;
  247. /* must be last field, see pskb_expand_head() */
  248. skb_frag_t frags[MAX_SKB_FRAGS];
  249. };
  250. /* We divide dataref into two halves. The higher 16 bits hold references
  251. * to the payload part of skb->data. The lower 16 bits hold references to
  252. * the entire skb->data. A clone of a headerless skb holds the length of
  253. * the header in skb->hdr_len.
  254. *
  255. * All users must obey the rule that the skb->data reference count must be
  256. * greater than or equal to the payload reference count.
  257. *
  258. * Holding a reference to the payload part means that the user does not
  259. * care about modifications to the header part of skb->data.
  260. */
  261. #define SKB_DATAREF_SHIFT 16
  262. #define SKB_DATAREF_MASK ((1 << SKB_DATAREF_SHIFT) - 1)
  263. enum {
  264. SKB_FCLONE_UNAVAILABLE,
  265. SKB_FCLONE_ORIG,
  266. SKB_FCLONE_CLONE,
  267. };
  268. enum {
  269. SKB_GSO_TCPV4 = 1 << 0,
  270. SKB_GSO_UDP = 1 << 1,
  271. /* This indicates the skb is from an untrusted source. */
  272. SKB_GSO_DODGY = 1 << 2,
  273. /* This indicates the tcp segment has CWR set. */
  274. SKB_GSO_TCP_ECN = 1 << 3,
  275. SKB_GSO_TCPV6 = 1 << 4,
  276. SKB_GSO_FCOE = 1 << 5,
  277. SKB_GSO_GRE = 1 << 6,
  278. SKB_GSO_IPIP = 1 << 7,
  279. SKB_GSO_SIT = 1 << 8,
  280. SKB_GSO_UDP_TUNNEL = 1 << 9,
  281. SKB_GSO_MPLS = 1 << 10,
  282. };
  283. #if BITS_PER_LONG > 32
  284. #define NET_SKBUFF_DATA_USES_OFFSET 1
  285. #endif
  286. #ifdef NET_SKBUFF_DATA_USES_OFFSET
  287. typedef unsigned int sk_buff_data_t;
  288. #else
  289. typedef unsigned char *sk_buff_data_t;
  290. #endif
  291. #if defined(CONFIG_NF_DEFRAG_IPV4) || defined(CONFIG_NF_DEFRAG_IPV4_MODULE) || \
  292. defined(CONFIG_NF_DEFRAG_IPV6) || defined(CONFIG_NF_DEFRAG_IPV6_MODULE)
  293. #define NET_SKBUFF_NF_DEFRAG_NEEDED 1
  294. #endif
  295. /**
  296. * struct sk_buff - socket buffer
  297. * @next: Next buffer in list
  298. * @prev: Previous buffer in list
  299. * @tstamp: Time we arrived
  300. * @sk: Socket we are owned by
  301. * @dev: Device we arrived on/are leaving by
  302. * @cb: Control buffer. Free for use by every layer. Put private vars here
  303. * @_skb_refdst: destination entry (with norefcount bit)
  304. * @sp: the security path, used for xfrm
  305. * @len: Length of actual data
  306. * @data_len: Data length
  307. * @mac_len: Length of link layer header
  308. * @hdr_len: writable header length of cloned skb
  309. * @csum: Checksum (must include start/offset pair)
  310. * @csum_start: Offset from skb->head where checksumming should start
  311. * @csum_offset: Offset from csum_start where checksum should be stored
  312. * @priority: Packet queueing priority
  313. * @local_df: allow local fragmentation
  314. * @cloned: Head may be cloned (check refcnt to be sure)
  315. * @ip_summed: Driver fed us an IP checksum
  316. * @nohdr: Payload reference only, must not modify header
  317. * @nfctinfo: Relationship of this skb to the connection
  318. * @pkt_type: Packet class
  319. * @fclone: skbuff clone status
  320. * @ipvs_property: skbuff is owned by ipvs
  321. * @peeked: this packet has been seen already, so stats have been
  322. * done for it, don't do them again
  323. * @nf_trace: netfilter packet trace flag
  324. * @protocol: Packet protocol from driver
  325. * @destructor: Destruct function
  326. * @nfct: Associated connection, if any
  327. * @nfct_reasm: netfilter conntrack re-assembly pointer
  328. * @nf_bridge: Saved data about a bridged frame - see br_netfilter.c
  329. * @skb_iif: ifindex of device we arrived on
  330. * @tc_index: Traffic control index
  331. * @tc_verd: traffic control verdict
  332. * @rxhash: the packet hash computed on receive
  333. * @queue_mapping: Queue mapping for multiqueue devices
  334. * @ndisc_nodetype: router type (from link layer)
  335. * @ooo_okay: allow the mapping of a socket to a queue to be changed
  336. * @l4_rxhash: indicate rxhash is a canonical 4-tuple hash over transport
  337. * ports.
  338. * @wifi_acked_valid: wifi_acked was set
  339. * @wifi_acked: whether frame was acked on wifi or not
  340. * @no_fcs: Request NIC to treat last 4 bytes as Ethernet FCS
  341. * @dma_cookie: a cookie to one of several possible DMA operations
  342. * done by skb DMA functions
  343. * @napi_id: id of the NAPI struct this skb came from
  344. * @secmark: security marking
  345. * @mark: Generic packet mark
  346. * @dropcount: total number of sk_receive_queue overflows
  347. * @vlan_proto: vlan encapsulation protocol
  348. * @vlan_tci: vlan tag control information
  349. * @inner_protocol: Protocol (encapsulation)
  350. * @inner_transport_header: Inner transport layer header (encapsulation)
  351. * @inner_network_header: Network layer header (encapsulation)
  352. * @inner_mac_header: Link layer header (encapsulation)
  353. * @transport_header: Transport layer header
  354. * @network_header: Network layer header
  355. * @mac_header: Link layer header
  356. * @tail: Tail pointer
  357. * @end: End pointer
  358. * @head: Head of buffer
  359. * @data: Data head pointer
  360. * @truesize: Buffer size
  361. * @users: User count - see {datagram,tcp}.c
  362. */
  363. struct sk_buff {
  364. /* These two members must be first. */
  365. struct sk_buff *next;
  366. struct sk_buff *prev;
  367. ktime_t tstamp;
  368. struct sock *sk;
  369. struct net_device *dev;
  370. /*
  371. * This is the control buffer. It is free to use for every
  372. * layer. Please put your private variables there. If you
  373. * want to keep them across layers you have to do a skb_clone()
  374. * first. This is owned by whoever has the skb queued ATM.
  375. */
  376. char cb[48] __aligned(8);
  377. unsigned long _skb_refdst;
  378. #ifdef CONFIG_XFRM
  379. struct sec_path *sp;
  380. #endif
  381. unsigned int len,
  382. data_len;
  383. __u16 mac_len,
  384. hdr_len;
  385. union {
  386. __wsum csum;
  387. struct {
  388. __u16 csum_start;
  389. __u16 csum_offset;
  390. };
  391. };
  392. __u32 priority;
  393. kmemcheck_bitfield_begin(flags1);
  394. __u8 local_df:1,
  395. cloned:1,
  396. ip_summed:2,
  397. nohdr:1,
  398. nfctinfo:3;
  399. __u8 pkt_type:3,
  400. fclone:2,
  401. ipvs_property:1,
  402. peeked:1,
  403. nf_trace:1;
  404. kmemcheck_bitfield_end(flags1);
  405. __be16 protocol;
  406. void (*destructor)(struct sk_buff *skb);
  407. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  408. struct nf_conntrack *nfct;
  409. #endif
  410. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  411. struct sk_buff *nfct_reasm;
  412. #endif
  413. #ifdef CONFIG_BRIDGE_NETFILTER
  414. struct nf_bridge_info *nf_bridge;
  415. #endif
  416. int skb_iif;
  417. __u32 rxhash;
  418. __be16 vlan_proto;
  419. __u16 vlan_tci;
  420. #ifdef CONFIG_NET_SCHED
  421. __u16 tc_index; /* traffic control index */
  422. #ifdef CONFIG_NET_CLS_ACT
  423. __u16 tc_verd; /* traffic control verdict */
  424. #endif
  425. #endif
  426. __u16 queue_mapping;
  427. kmemcheck_bitfield_begin(flags2);
  428. #ifdef CONFIG_IPV6_NDISC_NODETYPE
  429. __u8 ndisc_nodetype:2;
  430. #endif
  431. __u8 pfmemalloc:1;
  432. __u8 ooo_okay:1;
  433. __u8 l4_rxhash:1;
  434. __u8 wifi_acked_valid:1;
  435. __u8 wifi_acked:1;
  436. __u8 no_fcs:1;
  437. __u8 head_frag:1;
  438. /* Encapsulation protocol and NIC drivers should use
  439. * this flag to indicate to each other if the skb contains
  440. * encapsulated packet or not and maybe use the inner packet
  441. * headers if needed
  442. */
  443. __u8 encapsulation:1;
  444. /* 6/8 bit hole (depending on ndisc_nodetype presence) */
  445. kmemcheck_bitfield_end(flags2);
  446. #if defined CONFIG_NET_DMA || defined CONFIG_NET_RX_BUSY_POLL
  447. union {
  448. unsigned int napi_id;
  449. dma_cookie_t dma_cookie;
  450. };
  451. #endif
  452. #ifdef CONFIG_NETWORK_SECMARK
  453. __u32 secmark;
  454. #endif
  455. union {
  456. __u32 mark;
  457. __u32 dropcount;
  458. __u32 reserved_tailroom;
  459. };
  460. __be16 inner_protocol;
  461. __u16 inner_transport_header;
  462. __u16 inner_network_header;
  463. __u16 inner_mac_header;
  464. __u16 transport_header;
  465. __u16 network_header;
  466. __u16 mac_header;
  467. /* These elements must be at the end, see alloc_skb() for details. */
  468. sk_buff_data_t tail;
  469. sk_buff_data_t end;
  470. unsigned char *head,
  471. *data;
  472. unsigned int truesize;
  473. atomic_t users;
  474. };
  475. #ifdef __KERNEL__
  476. /*
  477. * Handling routines are only of interest to the kernel
  478. */
  479. #include <linux/slab.h>
  480. #define SKB_ALLOC_FCLONE 0x01
  481. #define SKB_ALLOC_RX 0x02
  482. /* Returns true if the skb was allocated from PFMEMALLOC reserves */
  483. static inline bool skb_pfmemalloc(const struct sk_buff *skb)
  484. {
  485. return unlikely(skb->pfmemalloc);
  486. }
  487. /*
  488. * skb might have a dst pointer attached, refcounted or not.
  489. * _skb_refdst low order bit is set if refcount was _not_ taken
  490. */
  491. #define SKB_DST_NOREF 1UL
  492. #define SKB_DST_PTRMASK ~(SKB_DST_NOREF)
  493. /**
  494. * skb_dst - returns skb dst_entry
  495. * @skb: buffer
  496. *
  497. * Returns skb dst_entry, regardless of reference taken or not.
  498. */
  499. static inline struct dst_entry *skb_dst(const struct sk_buff *skb)
  500. {
  501. /* If refdst was not refcounted, check we still are in a
  502. * rcu_read_lock section
  503. */
  504. WARN_ON((skb->_skb_refdst & SKB_DST_NOREF) &&
  505. !rcu_read_lock_held() &&
  506. !rcu_read_lock_bh_held());
  507. return (struct dst_entry *)(skb->_skb_refdst & SKB_DST_PTRMASK);
  508. }
  509. /**
  510. * skb_dst_set - sets skb dst
  511. * @skb: buffer
  512. * @dst: dst entry
  513. *
  514. * Sets skb dst, assuming a reference was taken on dst and should
  515. * be released by skb_dst_drop()
  516. */
  517. static inline void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst)
  518. {
  519. skb->_skb_refdst = (unsigned long)dst;
  520. }
  521. void __skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst,
  522. bool force);
  523. /**
  524. * skb_dst_set_noref - sets skb dst, hopefully, without taking reference
  525. * @skb: buffer
  526. * @dst: dst entry
  527. *
  528. * Sets skb dst, assuming a reference was not taken on dst.
  529. * If dst entry is cached, we do not take reference and dst_release
  530. * will be avoided by refdst_drop. If dst entry is not cached, we take
  531. * reference, so that last dst_release can destroy the dst immediately.
  532. */
  533. static inline void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst)
  534. {
  535. __skb_dst_set_noref(skb, dst, false);
  536. }
  537. /**
  538. * skb_dst_set_noref_force - sets skb dst, without taking reference
  539. * @skb: buffer
  540. * @dst: dst entry
  541. *
  542. * Sets skb dst, assuming a reference was not taken on dst.
  543. * No reference is taken and no dst_release will be called. While for
  544. * cached dsts deferred reclaim is a basic feature, for entries that are
  545. * not cached it is caller's job to guarantee that last dst_release for
  546. * provided dst happens when nobody uses it, eg. after a RCU grace period.
  547. */
  548. static inline void skb_dst_set_noref_force(struct sk_buff *skb,
  549. struct dst_entry *dst)
  550. {
  551. __skb_dst_set_noref(skb, dst, true);
  552. }
  553. /**
  554. * skb_dst_is_noref - Test if skb dst isn't refcounted
  555. * @skb: buffer
  556. */
  557. static inline bool skb_dst_is_noref(const struct sk_buff *skb)
  558. {
  559. return (skb->_skb_refdst & SKB_DST_NOREF) && skb_dst(skb);
  560. }
  561. static inline struct rtable *skb_rtable(const struct sk_buff *skb)
  562. {
  563. return (struct rtable *)skb_dst(skb);
  564. }
  565. void kfree_skb(struct sk_buff *skb);
  566. void kfree_skb_list(struct sk_buff *segs);
  567. void skb_tx_error(struct sk_buff *skb);
  568. void consume_skb(struct sk_buff *skb);
  569. void __kfree_skb(struct sk_buff *skb);
  570. extern struct kmem_cache *skbuff_head_cache;
  571. void kfree_skb_partial(struct sk_buff *skb, bool head_stolen);
  572. bool skb_try_coalesce(struct sk_buff *to, struct sk_buff *from,
  573. bool *fragstolen, int *delta_truesize);
  574. struct sk_buff *__alloc_skb(unsigned int size, gfp_t priority, int flags,
  575. int node);
  576. struct sk_buff *build_skb(void *data, unsigned int frag_size);
  577. static inline struct sk_buff *alloc_skb(unsigned int size,
  578. gfp_t priority)
  579. {
  580. return __alloc_skb(size, priority, 0, NUMA_NO_NODE);
  581. }
  582. static inline struct sk_buff *alloc_skb_fclone(unsigned int size,
  583. gfp_t priority)
  584. {
  585. return __alloc_skb(size, priority, SKB_ALLOC_FCLONE, NUMA_NO_NODE);
  586. }
  587. struct sk_buff *__alloc_skb_head(gfp_t priority, int node);
  588. static inline struct sk_buff *alloc_skb_head(gfp_t priority)
  589. {
  590. return __alloc_skb_head(priority, -1);
  591. }
  592. struct sk_buff *skb_morph(struct sk_buff *dst, struct sk_buff *src);
  593. int skb_copy_ubufs(struct sk_buff *skb, gfp_t gfp_mask);
  594. struct sk_buff *skb_clone(struct sk_buff *skb, gfp_t priority);
  595. struct sk_buff *skb_copy(const struct sk_buff *skb, gfp_t priority);
  596. struct sk_buff *__pskb_copy(struct sk_buff *skb, int headroom, gfp_t gfp_mask);
  597. int pskb_expand_head(struct sk_buff *skb, int nhead, int ntail, gfp_t gfp_mask);
  598. struct sk_buff *skb_realloc_headroom(struct sk_buff *skb,
  599. unsigned int headroom);
  600. struct sk_buff *skb_copy_expand(const struct sk_buff *skb, int newheadroom,
  601. int newtailroom, gfp_t priority);
  602. int skb_to_sgvec(struct sk_buff *skb, struct scatterlist *sg, int offset,
  603. int len);
  604. int skb_cow_data(struct sk_buff *skb, int tailbits, struct sk_buff **trailer);
  605. int skb_pad(struct sk_buff *skb, int pad);
  606. #define dev_kfree_skb(a) consume_skb(a)
  607. int skb_append_datato_frags(struct sock *sk, struct sk_buff *skb,
  608. int getfrag(void *from, char *to, int offset,
  609. int len, int odd, struct sk_buff *skb),
  610. void *from, int length);
  611. struct skb_seq_state {
  612. __u32 lower_offset;
  613. __u32 upper_offset;
  614. __u32 frag_idx;
  615. __u32 stepped_offset;
  616. struct sk_buff *root_skb;
  617. struct sk_buff *cur_skb;
  618. __u8 *frag_data;
  619. };
  620. void skb_prepare_seq_read(struct sk_buff *skb, unsigned int from,
  621. unsigned int to, struct skb_seq_state *st);
  622. unsigned int skb_seq_read(unsigned int consumed, const u8 **data,
  623. struct skb_seq_state *st);
  624. void skb_abort_seq_read(struct skb_seq_state *st);
  625. unsigned int skb_find_text(struct sk_buff *skb, unsigned int from,
  626. unsigned int to, struct ts_config *config,
  627. struct ts_state *state);
  628. void __skb_get_rxhash(struct sk_buff *skb);
  629. static inline __u32 skb_get_rxhash(struct sk_buff *skb)
  630. {
  631. if (!skb->l4_rxhash)
  632. __skb_get_rxhash(skb);
  633. return skb->rxhash;
  634. }
  635. #ifdef NET_SKBUFF_DATA_USES_OFFSET
  636. static inline unsigned char *skb_end_pointer(const struct sk_buff *skb)
  637. {
  638. return skb->head + skb->end;
  639. }
  640. static inline unsigned int skb_end_offset(const struct sk_buff *skb)
  641. {
  642. return skb->end;
  643. }
  644. #else
  645. static inline unsigned char *skb_end_pointer(const struct sk_buff *skb)
  646. {
  647. return skb->end;
  648. }
  649. static inline unsigned int skb_end_offset(const struct sk_buff *skb)
  650. {
  651. return skb->end - skb->head;
  652. }
  653. #endif
  654. /* Internal */
  655. #define skb_shinfo(SKB) ((struct skb_shared_info *)(skb_end_pointer(SKB)))
  656. static inline struct skb_shared_hwtstamps *skb_hwtstamps(struct sk_buff *skb)
  657. {
  658. return &skb_shinfo(skb)->hwtstamps;
  659. }
  660. /**
  661. * skb_queue_empty - check if a queue is empty
  662. * @list: queue head
  663. *
  664. * Returns true if the queue is empty, false otherwise.
  665. */
  666. static inline int skb_queue_empty(const struct sk_buff_head *list)
  667. {
  668. return list->next == (struct sk_buff *)list;
  669. }
  670. /**
  671. * skb_queue_is_last - check if skb is the last entry in the queue
  672. * @list: queue head
  673. * @skb: buffer
  674. *
  675. * Returns true if @skb is the last buffer on the list.
  676. */
  677. static inline bool skb_queue_is_last(const struct sk_buff_head *list,
  678. const struct sk_buff *skb)
  679. {
  680. return skb->next == (struct sk_buff *)list;
  681. }
  682. /**
  683. * skb_queue_is_first - check if skb is the first entry in the queue
  684. * @list: queue head
  685. * @skb: buffer
  686. *
  687. * Returns true if @skb is the first buffer on the list.
  688. */
  689. static inline bool skb_queue_is_first(const struct sk_buff_head *list,
  690. const struct sk_buff *skb)
  691. {
  692. return skb->prev == (struct sk_buff *)list;
  693. }
  694. /**
  695. * skb_queue_next - return the next packet in the queue
  696. * @list: queue head
  697. * @skb: current buffer
  698. *
  699. * Return the next packet in @list after @skb. It is only valid to
  700. * call this if skb_queue_is_last() evaluates to false.
  701. */
  702. static inline struct sk_buff *skb_queue_next(const struct sk_buff_head *list,
  703. const struct sk_buff *skb)
  704. {
  705. /* This BUG_ON may seem severe, but if we just return then we
  706. * are going to dereference garbage.
  707. */
  708. BUG_ON(skb_queue_is_last(list, skb));
  709. return skb->next;
  710. }
  711. /**
  712. * skb_queue_prev - return the prev packet in the queue
  713. * @list: queue head
  714. * @skb: current buffer
  715. *
  716. * Return the prev packet in @list before @skb. It is only valid to
  717. * call this if skb_queue_is_first() evaluates to false.
  718. */
  719. static inline struct sk_buff *skb_queue_prev(const struct sk_buff_head *list,
  720. const struct sk_buff *skb)
  721. {
  722. /* This BUG_ON may seem severe, but if we just return then we
  723. * are going to dereference garbage.
  724. */
  725. BUG_ON(skb_queue_is_first(list, skb));
  726. return skb->prev;
  727. }
  728. /**
  729. * skb_get - reference buffer
  730. * @skb: buffer to reference
  731. *
  732. * Makes another reference to a socket buffer and returns a pointer
  733. * to the buffer.
  734. */
  735. static inline struct sk_buff *skb_get(struct sk_buff *skb)
  736. {
  737. atomic_inc(&skb->users);
  738. return skb;
  739. }
  740. /*
  741. * If users == 1, we are the only owner and are can avoid redundant
  742. * atomic change.
  743. */
  744. /**
  745. * skb_cloned - is the buffer a clone
  746. * @skb: buffer to check
  747. *
  748. * Returns true if the buffer was generated with skb_clone() and is
  749. * one of multiple shared copies of the buffer. Cloned buffers are
  750. * shared data so must not be written to under normal circumstances.
  751. */
  752. static inline int skb_cloned(const struct sk_buff *skb)
  753. {
  754. return skb->cloned &&
  755. (atomic_read(&skb_shinfo(skb)->dataref) & SKB_DATAREF_MASK) != 1;
  756. }
  757. static inline int skb_unclone(struct sk_buff *skb, gfp_t pri)
  758. {
  759. might_sleep_if(pri & __GFP_WAIT);
  760. if (skb_cloned(skb))
  761. return pskb_expand_head(skb, 0, 0, pri);
  762. return 0;
  763. }
  764. /**
  765. * skb_header_cloned - is the header a clone
  766. * @skb: buffer to check
  767. *
  768. * Returns true if modifying the header part of the buffer requires
  769. * the data to be copied.
  770. */
  771. static inline int skb_header_cloned(const struct sk_buff *skb)
  772. {
  773. int dataref;
  774. if (!skb->cloned)
  775. return 0;
  776. dataref = atomic_read(&skb_shinfo(skb)->dataref);
  777. dataref = (dataref & SKB_DATAREF_MASK) - (dataref >> SKB_DATAREF_SHIFT);
  778. return dataref != 1;
  779. }
  780. /**
  781. * skb_header_release - release reference to header
  782. * @skb: buffer to operate on
  783. *
  784. * Drop a reference to the header part of the buffer. This is done
  785. * by acquiring a payload reference. You must not read from the header
  786. * part of skb->data after this.
  787. */
  788. static inline void skb_header_release(struct sk_buff *skb)
  789. {
  790. BUG_ON(skb->nohdr);
  791. skb->nohdr = 1;
  792. atomic_add(1 << SKB_DATAREF_SHIFT, &skb_shinfo(skb)->dataref);
  793. }
  794. /**
  795. * skb_shared - is the buffer shared
  796. * @skb: buffer to check
  797. *
  798. * Returns true if more than one person has a reference to this
  799. * buffer.
  800. */
  801. static inline int skb_shared(const struct sk_buff *skb)
  802. {
  803. return atomic_read(&skb->users) != 1;
  804. }
  805. /**
  806. * skb_share_check - check if buffer is shared and if so clone it
  807. * @skb: buffer to check
  808. * @pri: priority for memory allocation
  809. *
  810. * If the buffer is shared the buffer is cloned and the old copy
  811. * drops a reference. A new clone with a single reference is returned.
  812. * If the buffer is not shared the original buffer is returned. When
  813. * being called from interrupt status or with spinlocks held pri must
  814. * be GFP_ATOMIC.
  815. *
  816. * NULL is returned on a memory allocation failure.
  817. */
  818. static inline struct sk_buff *skb_share_check(struct sk_buff *skb, gfp_t pri)
  819. {
  820. might_sleep_if(pri & __GFP_WAIT);
  821. if (skb_shared(skb)) {
  822. struct sk_buff *nskb = skb_clone(skb, pri);
  823. if (likely(nskb))
  824. consume_skb(skb);
  825. else
  826. kfree_skb(skb);
  827. skb = nskb;
  828. }
  829. return skb;
  830. }
  831. /*
  832. * Copy shared buffers into a new sk_buff. We effectively do COW on
  833. * packets to handle cases where we have a local reader and forward
  834. * and a couple of other messy ones. The normal one is tcpdumping
  835. * a packet thats being forwarded.
  836. */
  837. /**
  838. * skb_unshare - make a copy of a shared buffer
  839. * @skb: buffer to check
  840. * @pri: priority for memory allocation
  841. *
  842. * If the socket buffer is a clone then this function creates a new
  843. * copy of the data, drops a reference count on the old copy and returns
  844. * the new copy with the reference count at 1. If the buffer is not a clone
  845. * the original buffer is returned. When called with a spinlock held or
  846. * from interrupt state @pri must be %GFP_ATOMIC
  847. *
  848. * %NULL is returned on a memory allocation failure.
  849. */
  850. static inline struct sk_buff *skb_unshare(struct sk_buff *skb,
  851. gfp_t pri)
  852. {
  853. might_sleep_if(pri & __GFP_WAIT);
  854. if (skb_cloned(skb)) {
  855. struct sk_buff *nskb = skb_copy(skb, pri);
  856. kfree_skb(skb); /* Free our shared copy */
  857. skb = nskb;
  858. }
  859. return skb;
  860. }
  861. /**
  862. * skb_peek - peek at the head of an &sk_buff_head
  863. * @list_: list to peek at
  864. *
  865. * Peek an &sk_buff. Unlike most other operations you _MUST_
  866. * be careful with this one. A peek leaves the buffer on the
  867. * list and someone else may run off with it. You must hold
  868. * the appropriate locks or have a private queue to do this.
  869. *
  870. * Returns %NULL for an empty list or a pointer to the head element.
  871. * The reference count is not incremented and the reference is therefore
  872. * volatile. Use with caution.
  873. */
  874. static inline struct sk_buff *skb_peek(const struct sk_buff_head *list_)
  875. {
  876. struct sk_buff *skb = list_->next;
  877. if (skb == (struct sk_buff *)list_)
  878. skb = NULL;
  879. return skb;
  880. }
  881. /**
  882. * skb_peek_next - peek skb following the given one from a queue
  883. * @skb: skb to start from
  884. * @list_: list to peek at
  885. *
  886. * Returns %NULL when the end of the list is met or a pointer to the
  887. * next element. The reference count is not incremented and the
  888. * reference is therefore volatile. Use with caution.
  889. */
  890. static inline struct sk_buff *skb_peek_next(struct sk_buff *skb,
  891. const struct sk_buff_head *list_)
  892. {
  893. struct sk_buff *next = skb->next;
  894. if (next == (struct sk_buff *)list_)
  895. next = NULL;
  896. return next;
  897. }
  898. /**
  899. * skb_peek_tail - peek at the tail of an &sk_buff_head
  900. * @list_: list to peek at
  901. *
  902. * Peek an &sk_buff. Unlike most other operations you _MUST_
  903. * be careful with this one. A peek leaves the buffer on the
  904. * list and someone else may run off with it. You must hold
  905. * the appropriate locks or have a private queue to do this.
  906. *
  907. * Returns %NULL for an empty list or a pointer to the tail element.
  908. * The reference count is not incremented and the reference is therefore
  909. * volatile. Use with caution.
  910. */
  911. static inline struct sk_buff *skb_peek_tail(const struct sk_buff_head *list_)
  912. {
  913. struct sk_buff *skb = list_->prev;
  914. if (skb == (struct sk_buff *)list_)
  915. skb = NULL;
  916. return skb;
  917. }
  918. /**
  919. * skb_queue_len - get queue length
  920. * @list_: list to measure
  921. *
  922. * Return the length of an &sk_buff queue.
  923. */
  924. static inline __u32 skb_queue_len(const struct sk_buff_head *list_)
  925. {
  926. return list_->qlen;
  927. }
  928. /**
  929. * __skb_queue_head_init - initialize non-spinlock portions of sk_buff_head
  930. * @list: queue to initialize
  931. *
  932. * This initializes only the list and queue length aspects of
  933. * an sk_buff_head object. This allows to initialize the list
  934. * aspects of an sk_buff_head without reinitializing things like
  935. * the spinlock. It can also be used for on-stack sk_buff_head
  936. * objects where the spinlock is known to not be used.
  937. */
  938. static inline void __skb_queue_head_init(struct sk_buff_head *list)
  939. {
  940. list->prev = list->next = (struct sk_buff *)list;
  941. list->qlen = 0;
  942. }
  943. /*
  944. * This function creates a split out lock class for each invocation;
  945. * this is needed for now since a whole lot of users of the skb-queue
  946. * infrastructure in drivers have different locking usage (in hardirq)
  947. * than the networking core (in softirq only). In the long run either the
  948. * network layer or drivers should need annotation to consolidate the
  949. * main types of usage into 3 classes.
  950. */
  951. static inline void skb_queue_head_init(struct sk_buff_head *list)
  952. {
  953. spin_lock_init(&list->lock);
  954. __skb_queue_head_init(list);
  955. }
  956. static inline void skb_queue_head_init_class(struct sk_buff_head *list,
  957. struct lock_class_key *class)
  958. {
  959. skb_queue_head_init(list);
  960. lockdep_set_class(&list->lock, class);
  961. }
  962. /*
  963. * Insert an sk_buff on a list.
  964. *
  965. * The "__skb_xxxx()" functions are the non-atomic ones that
  966. * can only be called with interrupts disabled.
  967. */
  968. void skb_insert(struct sk_buff *old, struct sk_buff *newsk,
  969. struct sk_buff_head *list);
  970. static inline void __skb_insert(struct sk_buff *newsk,
  971. struct sk_buff *prev, struct sk_buff *next,
  972. struct sk_buff_head *list)
  973. {
  974. newsk->next = next;
  975. newsk->prev = prev;
  976. next->prev = prev->next = newsk;
  977. list->qlen++;
  978. }
  979. static inline void __skb_queue_splice(const struct sk_buff_head *list,
  980. struct sk_buff *prev,
  981. struct sk_buff *next)
  982. {
  983. struct sk_buff *first = list->next;
  984. struct sk_buff *last = list->prev;
  985. first->prev = prev;
  986. prev->next = first;
  987. last->next = next;
  988. next->prev = last;
  989. }
  990. /**
  991. * skb_queue_splice - join two skb lists, this is designed for stacks
  992. * @list: the new list to add
  993. * @head: the place to add it in the first list
  994. */
  995. static inline void skb_queue_splice(const struct sk_buff_head *list,
  996. struct sk_buff_head *head)
  997. {
  998. if (!skb_queue_empty(list)) {
  999. __skb_queue_splice(list, (struct sk_buff *) head, head->next);
  1000. head->qlen += list->qlen;
  1001. }
  1002. }
  1003. /**
  1004. * skb_queue_splice_init - join two skb lists and reinitialise the emptied list
  1005. * @list: the new list to add
  1006. * @head: the place to add it in the first list
  1007. *
  1008. * The list at @list is reinitialised
  1009. */
  1010. static inline void skb_queue_splice_init(struct sk_buff_head *list,
  1011. struct sk_buff_head *head)
  1012. {
  1013. if (!skb_queue_empty(list)) {
  1014. __skb_queue_splice(list, (struct sk_buff *) head, head->next);
  1015. head->qlen += list->qlen;
  1016. __skb_queue_head_init(list);
  1017. }
  1018. }
  1019. /**
  1020. * skb_queue_splice_tail - join two skb lists, each list being a queue
  1021. * @list: the new list to add
  1022. * @head: the place to add it in the first list
  1023. */
  1024. static inline void skb_queue_splice_tail(const struct sk_buff_head *list,
  1025. struct sk_buff_head *head)
  1026. {
  1027. if (!skb_queue_empty(list)) {
  1028. __skb_queue_splice(list, head->prev, (struct sk_buff *) head);
  1029. head->qlen += list->qlen;
  1030. }
  1031. }
  1032. /**
  1033. * skb_queue_splice_tail_init - join two skb lists and reinitialise the emptied list
  1034. * @list: the new list to add
  1035. * @head: the place to add it in the first list
  1036. *
  1037. * Each of the lists is a queue.
  1038. * The list at @list is reinitialised
  1039. */
  1040. static inline void skb_queue_splice_tail_init(struct sk_buff_head *list,
  1041. struct sk_buff_head *head)
  1042. {
  1043. if (!skb_queue_empty(list)) {
  1044. __skb_queue_splice(list, head->prev, (struct sk_buff *) head);
  1045. head->qlen += list->qlen;
  1046. __skb_queue_head_init(list);
  1047. }
  1048. }
  1049. /**
  1050. * __skb_queue_after - queue a buffer at the list head
  1051. * @list: list to use
  1052. * @prev: place after this buffer
  1053. * @newsk: buffer to queue
  1054. *
  1055. * Queue a buffer int the middle of a list. This function takes no locks
  1056. * and you must therefore hold required locks before calling it.
  1057. *
  1058. * A buffer cannot be placed on two lists at the same time.
  1059. */
  1060. static inline void __skb_queue_after(struct sk_buff_head *list,
  1061. struct sk_buff *prev,
  1062. struct sk_buff *newsk)
  1063. {
  1064. __skb_insert(newsk, prev, prev->next, list);
  1065. }
  1066. void skb_append(struct sk_buff *old, struct sk_buff *newsk,
  1067. struct sk_buff_head *list);
  1068. static inline void __skb_queue_before(struct sk_buff_head *list,
  1069. struct sk_buff *next,
  1070. struct sk_buff *newsk)
  1071. {
  1072. __skb_insert(newsk, next->prev, next, list);
  1073. }
  1074. /**
  1075. * __skb_queue_head - queue a buffer at the list head
  1076. * @list: list to use
  1077. * @newsk: buffer to queue
  1078. *
  1079. * Queue a buffer at the start of a list. This function takes no locks
  1080. * and you must therefore hold required locks before calling it.
  1081. *
  1082. * A buffer cannot be placed on two lists at the same time.
  1083. */
  1084. void skb_queue_head(struct sk_buff_head *list, struct sk_buff *newsk);
  1085. static inline void __skb_queue_head(struct sk_buff_head *list,
  1086. struct sk_buff *newsk)
  1087. {
  1088. __skb_queue_after(list, (struct sk_buff *)list, newsk);
  1089. }
  1090. /**
  1091. * __skb_queue_tail - queue a buffer at the list tail
  1092. * @list: list to use
  1093. * @newsk: buffer to queue
  1094. *
  1095. * Queue a buffer at the end of a list. This function takes no locks
  1096. * and you must therefore hold required locks before calling it.
  1097. *
  1098. * A buffer cannot be placed on two lists at the same time.
  1099. */
  1100. void skb_queue_tail(struct sk_buff_head *list, struct sk_buff *newsk);
  1101. static inline void __skb_queue_tail(struct sk_buff_head *list,
  1102. struct sk_buff *newsk)
  1103. {
  1104. __skb_queue_before(list, (struct sk_buff *)list, newsk);
  1105. }
  1106. /*
  1107. * remove sk_buff from list. _Must_ be called atomically, and with
  1108. * the list known..
  1109. */
  1110. void skb_unlink(struct sk_buff *skb, struct sk_buff_head *list);
  1111. static inline void __skb_unlink(struct sk_buff *skb, struct sk_buff_head *list)
  1112. {
  1113. struct sk_buff *next, *prev;
  1114. list->qlen--;
  1115. next = skb->next;
  1116. prev = skb->prev;
  1117. skb->next = skb->prev = NULL;
  1118. next->prev = prev;
  1119. prev->next = next;
  1120. }
  1121. /**
  1122. * __skb_dequeue - remove from the head of the queue
  1123. * @list: list to dequeue from
  1124. *
  1125. * Remove the head of the list. This function does not take any locks
  1126. * so must be used with appropriate locks held only. The head item is
  1127. * returned or %NULL if the list is empty.
  1128. */
  1129. struct sk_buff *skb_dequeue(struct sk_buff_head *list);
  1130. static inline struct sk_buff *__skb_dequeue(struct sk_buff_head *list)
  1131. {
  1132. struct sk_buff *skb = skb_peek(list);
  1133. if (skb)
  1134. __skb_unlink(skb, list);
  1135. return skb;
  1136. }
  1137. /**
  1138. * __skb_dequeue_tail - remove from the tail of the queue
  1139. * @list: list to dequeue from
  1140. *
  1141. * Remove the tail of the list. This function does not take any locks
  1142. * so must be used with appropriate locks held only. The tail item is
  1143. * returned or %NULL if the list is empty.
  1144. */
  1145. struct sk_buff *skb_dequeue_tail(struct sk_buff_head *list);
  1146. static inline struct sk_buff *__skb_dequeue_tail(struct sk_buff_head *list)
  1147. {
  1148. struct sk_buff *skb = skb_peek_tail(list);
  1149. if (skb)
  1150. __skb_unlink(skb, list);
  1151. return skb;
  1152. }
  1153. static inline bool skb_is_nonlinear(const struct sk_buff *skb)
  1154. {
  1155. return skb->data_len;
  1156. }
  1157. static inline unsigned int skb_headlen(const struct sk_buff *skb)
  1158. {
  1159. return skb->len - skb->data_len;
  1160. }
  1161. static inline int skb_pagelen(const struct sk_buff *skb)
  1162. {
  1163. int i, len = 0;
  1164. for (i = (int)skb_shinfo(skb)->nr_frags - 1; i >= 0; i--)
  1165. len += skb_frag_size(&skb_shinfo(skb)->frags[i]);
  1166. return len + skb_headlen(skb);
  1167. }
  1168. /**
  1169. * __skb_fill_page_desc - initialise a paged fragment in an skb
  1170. * @skb: buffer containing fragment to be initialised
  1171. * @i: paged fragment index to initialise
  1172. * @page: the page to use for this fragment
  1173. * @off: the offset to the data with @page
  1174. * @size: the length of the data
  1175. *
  1176. * Initialises the @i'th fragment of @skb to point to &size bytes at
  1177. * offset @off within @page.
  1178. *
  1179. * Does not take any additional reference on the fragment.
  1180. */
  1181. static inline void __skb_fill_page_desc(struct sk_buff *skb, int i,
  1182. struct page *page, int off, int size)
  1183. {
  1184. skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
  1185. /*
  1186. * Propagate page->pfmemalloc to the skb if we can. The problem is
  1187. * that not all callers have unique ownership of the page. If
  1188. * pfmemalloc is set, we check the mapping as a mapping implies
  1189. * page->index is set (index and pfmemalloc share space).
  1190. * If it's a valid mapping, we cannot use page->pfmemalloc but we
  1191. * do not lose pfmemalloc information as the pages would not be
  1192. * allocated using __GFP_MEMALLOC.
  1193. */
  1194. frag->page.p = page;
  1195. frag->page_offset = off;
  1196. skb_frag_size_set(frag, size);
  1197. page = compound_head(page);
  1198. if (page->pfmemalloc && !page->mapping)
  1199. skb->pfmemalloc = true;
  1200. }
  1201. /**
  1202. * skb_fill_page_desc - initialise a paged fragment in an skb
  1203. * @skb: buffer containing fragment to be initialised
  1204. * @i: paged fragment index to initialise
  1205. * @page: the page to use for this fragment
  1206. * @off: the offset to the data with @page
  1207. * @size: the length of the data
  1208. *
  1209. * As per __skb_fill_page_desc() -- initialises the @i'th fragment of
  1210. * @skb to point to @size bytes at offset @off within @page. In
  1211. * addition updates @skb such that @i is the last fragment.
  1212. *
  1213. * Does not take any additional reference on the fragment.
  1214. */
  1215. static inline void skb_fill_page_desc(struct sk_buff *skb, int i,
  1216. struct page *page, int off, int size)
  1217. {
  1218. __skb_fill_page_desc(skb, i, page, off, size);
  1219. skb_shinfo(skb)->nr_frags = i + 1;
  1220. }
  1221. void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
  1222. int size, unsigned int truesize);
  1223. void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int size,
  1224. unsigned int truesize);
  1225. #define SKB_PAGE_ASSERT(skb) BUG_ON(skb_shinfo(skb)->nr_frags)
  1226. #define SKB_FRAG_ASSERT(skb) BUG_ON(skb_has_frag_list(skb))
  1227. #define SKB_LINEAR_ASSERT(skb) BUG_ON(skb_is_nonlinear(skb))
  1228. #ifdef NET_SKBUFF_DATA_USES_OFFSET
  1229. static inline unsigned char *skb_tail_pointer(const struct sk_buff *skb)
  1230. {
  1231. return skb->head + skb->tail;
  1232. }
  1233. static inline void skb_reset_tail_pointer(struct sk_buff *skb)
  1234. {
  1235. skb->tail = skb->data - skb->head;
  1236. }
  1237. static inline void skb_set_tail_pointer(struct sk_buff *skb, const int offset)
  1238. {
  1239. skb_reset_tail_pointer(skb);
  1240. skb->tail += offset;
  1241. }
  1242. #else /* NET_SKBUFF_DATA_USES_OFFSET */
  1243. static inline unsigned char *skb_tail_pointer(const struct sk_buff *skb)
  1244. {
  1245. return skb->tail;
  1246. }
  1247. static inline void skb_reset_tail_pointer(struct sk_buff *skb)
  1248. {
  1249. skb->tail = skb->data;
  1250. }
  1251. static inline void skb_set_tail_pointer(struct sk_buff *skb, const int offset)
  1252. {
  1253. skb->tail = skb->data + offset;
  1254. }
  1255. #endif /* NET_SKBUFF_DATA_USES_OFFSET */
  1256. /*
  1257. * Add data to an sk_buff
  1258. */
  1259. unsigned char *pskb_put(struct sk_buff *skb, struct sk_buff *tail, int len);
  1260. unsigned char *skb_put(struct sk_buff *skb, unsigned int len);
  1261. static inline unsigned char *__skb_put(struct sk_buff *skb, unsigned int len)
  1262. {
  1263. unsigned char *tmp = skb_tail_pointer(skb);
  1264. SKB_LINEAR_ASSERT(skb);
  1265. skb->tail += len;
  1266. skb->len += len;
  1267. return tmp;
  1268. }
  1269. unsigned char *skb_push(struct sk_buff *skb, unsigned int len);
  1270. static inline unsigned char *__skb_push(struct sk_buff *skb, unsigned int len)
  1271. {
  1272. skb->data -= len;
  1273. skb->len += len;
  1274. return skb->data;
  1275. }
  1276. unsigned char *skb_pull(struct sk_buff *skb, unsigned int len);
  1277. static inline unsigned char *__skb_pull(struct sk_buff *skb, unsigned int len)
  1278. {
  1279. skb->len -= len;
  1280. BUG_ON(skb->len < skb->data_len);
  1281. return skb->data += len;
  1282. }
  1283. static inline unsigned char *skb_pull_inline(struct sk_buff *skb, unsigned int len)
  1284. {
  1285. return unlikely(len > skb->len) ? NULL : __skb_pull(skb, len);
  1286. }
  1287. unsigned char *__pskb_pull_tail(struct sk_buff *skb, int delta);
  1288. static inline unsigned char *__pskb_pull(struct sk_buff *skb, unsigned int len)
  1289. {
  1290. if (len > skb_headlen(skb) &&
  1291. !__pskb_pull_tail(skb, len - skb_headlen(skb)))
  1292. return NULL;
  1293. skb->len -= len;
  1294. return skb->data += len;
  1295. }
  1296. static inline unsigned char *pskb_pull(struct sk_buff *skb, unsigned int len)
  1297. {
  1298. return unlikely(len > skb->len) ? NULL : __pskb_pull(skb, len);
  1299. }
  1300. static inline int pskb_may_pull(struct sk_buff *skb, unsigned int len)
  1301. {
  1302. if (likely(len <= skb_headlen(skb)))
  1303. return 1;
  1304. if (unlikely(len > skb->len))
  1305. return 0;
  1306. return __pskb_pull_tail(skb, len - skb_headlen(skb)) != NULL;
  1307. }
  1308. /**
  1309. * skb_headroom - bytes at buffer head
  1310. * @skb: buffer to check
  1311. *
  1312. * Return the number of bytes of free space at the head of an &sk_buff.
  1313. */
  1314. static inline unsigned int skb_headroom(const struct sk_buff *skb)
  1315. {
  1316. return skb->data - skb->head;
  1317. }
  1318. /**
  1319. * skb_tailroom - bytes at buffer end
  1320. * @skb: buffer to check
  1321. *
  1322. * Return the number of bytes of free space at the tail of an sk_buff
  1323. */
  1324. static inline int skb_tailroom(const struct sk_buff *skb)
  1325. {
  1326. return skb_is_nonlinear(skb) ? 0 : skb->end - skb->tail;
  1327. }
  1328. /**
  1329. * skb_availroom - bytes at buffer end
  1330. * @skb: buffer to check
  1331. *
  1332. * Return the number of bytes of free space at the tail of an sk_buff
  1333. * allocated by sk_stream_alloc()
  1334. */
  1335. static inline int skb_availroom(const struct sk_buff *skb)
  1336. {
  1337. if (skb_is_nonlinear(skb))
  1338. return 0;
  1339. return skb->end - skb->tail - skb->reserved_tailroom;
  1340. }
  1341. /**
  1342. * skb_reserve - adjust headroom
  1343. * @skb: buffer to alter
  1344. * @len: bytes to move
  1345. *
  1346. * Increase the headroom of an empty &sk_buff by reducing the tail
  1347. * room. This is only allowed for an empty buffer.
  1348. */
  1349. static inline void skb_reserve(struct sk_buff *skb, int len)
  1350. {
  1351. skb->data += len;
  1352. skb->tail += len;
  1353. }
  1354. static inline void skb_reset_inner_headers(struct sk_buff *skb)
  1355. {
  1356. skb->inner_mac_header = skb->mac_header;
  1357. skb->inner_network_header = skb->network_header;
  1358. skb->inner_transport_header = skb->transport_header;
  1359. }
  1360. static inline void skb_reset_mac_len(struct sk_buff *skb)
  1361. {
  1362. skb->mac_len = skb->network_header - skb->mac_header;
  1363. }
  1364. static inline unsigned char *skb_inner_transport_header(const struct sk_buff
  1365. *skb)
  1366. {
  1367. return skb->head + skb->inner_transport_header;
  1368. }
  1369. static inline void skb_reset_inner_transport_header(struct sk_buff *skb)
  1370. {
  1371. skb->inner_transport_header = skb->data - skb->head;
  1372. }
  1373. static inline void skb_set_inner_transport_header(struct sk_buff *skb,
  1374. const int offset)
  1375. {
  1376. skb_reset_inner_transport_header(skb);
  1377. skb->inner_transport_header += offset;
  1378. }
  1379. static inline unsigned char *skb_inner_network_header(const struct sk_buff *skb)
  1380. {
  1381. return skb->head + skb->inner_network_header;
  1382. }
  1383. static inline void skb_reset_inner_network_header(struct sk_buff *skb)
  1384. {
  1385. skb->inner_network_header = skb->data - skb->head;
  1386. }
  1387. static inline void skb_set_inner_network_header(struct sk_buff *skb,
  1388. const int offset)
  1389. {
  1390. skb_reset_inner_network_header(skb);
  1391. skb->inner_network_header += offset;
  1392. }
  1393. static inline unsigned char *skb_inner_mac_header(const struct sk_buff *skb)
  1394. {
  1395. return skb->head + skb->inner_mac_header;
  1396. }
  1397. static inline void skb_reset_inner_mac_header(struct sk_buff *skb)
  1398. {
  1399. skb->inner_mac_header = skb->data - skb->head;
  1400. }
  1401. static inline void skb_set_inner_mac_header(struct sk_buff *skb,
  1402. const int offset)
  1403. {
  1404. skb_reset_inner_mac_header(skb);
  1405. skb->inner_mac_header += offset;
  1406. }
  1407. static inline bool skb_transport_header_was_set(const struct sk_buff *skb)
  1408. {
  1409. return skb->transport_header != (typeof(skb->transport_header))~0U;
  1410. }
  1411. static inline unsigned char *skb_transport_header(const struct sk_buff *skb)
  1412. {
  1413. return skb->head + skb->transport_header;
  1414. }
  1415. static inline void skb_reset_transport_header(struct sk_buff *skb)
  1416. {
  1417. skb->transport_header = skb->data - skb->head;
  1418. }
  1419. static inline void skb_set_transport_header(struct sk_buff *skb,
  1420. const int offset)
  1421. {
  1422. skb_reset_transport_header(skb);
  1423. skb->transport_header += offset;
  1424. }
  1425. static inline unsigned char *skb_network_header(const struct sk_buff *skb)
  1426. {
  1427. return skb->head + skb->network_header;
  1428. }
  1429. static inline void skb_reset_network_header(struct sk_buff *skb)
  1430. {
  1431. skb->network_header = skb->data - skb->head;
  1432. }
  1433. static inline void skb_set_network_header(struct sk_buff *skb, const int offset)
  1434. {
  1435. skb_reset_network_header(skb);
  1436. skb->network_header += offset;
  1437. }
  1438. static inline unsigned char *skb_mac_header(const struct sk_buff *skb)
  1439. {
  1440. return skb->head + skb->mac_header;
  1441. }
  1442. static inline int skb_mac_header_was_set(const struct sk_buff *skb)
  1443. {
  1444. return skb->mac_header != (typeof(skb->mac_header))~0U;
  1445. }
  1446. static inline void skb_reset_mac_header(struct sk_buff *skb)
  1447. {
  1448. skb->mac_header = skb->data - skb->head;
  1449. }
  1450. static inline void skb_set_mac_header(struct sk_buff *skb, const int offset)
  1451. {
  1452. skb_reset_mac_header(skb);
  1453. skb->mac_header += offset;
  1454. }
  1455. static inline void skb_probe_transport_header(struct sk_buff *skb,
  1456. const int offset_hint)
  1457. {
  1458. struct flow_keys keys;
  1459. if (skb_transport_header_was_set(skb))
  1460. return;
  1461. else if (skb_flow_dissect(skb, &keys))
  1462. skb_set_transport_header(skb, keys.thoff);
  1463. else
  1464. skb_set_transport_header(skb, offset_hint);
  1465. }
  1466. static inline void skb_mac_header_rebuild(struct sk_buff *skb)
  1467. {
  1468. if (skb_mac_header_was_set(skb)) {
  1469. const unsigned char *old_mac = skb_mac_header(skb);
  1470. skb_set_mac_header(skb, -skb->mac_len);
  1471. memmove(skb_mac_header(skb), old_mac, skb->mac_len);
  1472. }
  1473. }
  1474. static inline int skb_checksum_start_offset(const struct sk_buff *skb)
  1475. {
  1476. return skb->csum_start - skb_headroom(skb);
  1477. }
  1478. static inline int skb_transport_offset(const struct sk_buff *skb)
  1479. {
  1480. return skb_transport_header(skb) - skb->data;
  1481. }
  1482. static inline u32 skb_network_header_len(const struct sk_buff *skb)
  1483. {
  1484. return skb->transport_header - skb->network_header;
  1485. }
  1486. static inline u32 skb_inner_network_header_len(const struct sk_buff *skb)
  1487. {
  1488. return skb->inner_transport_header - skb->inner_network_header;
  1489. }
  1490. static inline int skb_network_offset(const struct sk_buff *skb)
  1491. {
  1492. return skb_network_header(skb) - skb->data;
  1493. }
  1494. static inline int skb_inner_network_offset(const struct sk_buff *skb)
  1495. {
  1496. return skb_inner_network_header(skb) - skb->data;
  1497. }
  1498. static inline int pskb_network_may_pull(struct sk_buff *skb, unsigned int len)
  1499. {
  1500. return pskb_may_pull(skb, skb_network_offset(skb) + len);
  1501. }
  1502. /*
  1503. * CPUs often take a performance hit when accessing unaligned memory
  1504. * locations. The actual performance hit varies, it can be small if the
  1505. * hardware handles it or large if we have to take an exception and fix it
  1506. * in software.
  1507. *
  1508. * Since an ethernet header is 14 bytes network drivers often end up with
  1509. * the IP header at an unaligned offset. The IP header can be aligned by
  1510. * shifting the start of the packet by 2 bytes. Drivers should do this
  1511. * with:
  1512. *
  1513. * skb_reserve(skb, NET_IP_ALIGN);
  1514. *
  1515. * The downside to this alignment of the IP header is that the DMA is now
  1516. * unaligned. On some architectures the cost of an unaligned DMA is high
  1517. * and this cost outweighs the gains made by aligning the IP header.
  1518. *
  1519. * Since this trade off varies between architectures, we allow NET_IP_ALIGN
  1520. * to be overridden.
  1521. */
  1522. #ifndef NET_IP_ALIGN
  1523. #define NET_IP_ALIGN 2
  1524. #endif
  1525. /*
  1526. * The networking layer reserves some headroom in skb data (via
  1527. * dev_alloc_skb). This is used to avoid having to reallocate skb data when
  1528. * the header has to grow. In the default case, if the header has to grow
  1529. * 32 bytes or less we avoid the reallocation.
  1530. *
  1531. * Unfortunately this headroom changes the DMA alignment of the resulting
  1532. * network packet. As for NET_IP_ALIGN, this unaligned DMA is expensive
  1533. * on some architectures. An architecture can override this value,
  1534. * perhaps setting it to a cacheline in size (since that will maintain
  1535. * cacheline alignment of the DMA). It must be a power of 2.
  1536. *
  1537. * Various parts of the networking layer expect at least 32 bytes of
  1538. * headroom, you should not reduce this.
  1539. *
  1540. * Using max(32, L1_CACHE_BYTES) makes sense (especially with RPS)
  1541. * to reduce average number of cache lines per packet.
  1542. * get_rps_cpus() for example only access one 64 bytes aligned block :
  1543. * NET_IP_ALIGN(2) + ethernet_header(14) + IP_header(20/40) + ports(8)
  1544. */
  1545. #ifndef NET_SKB_PAD
  1546. #define NET_SKB_PAD max(32, L1_CACHE_BYTES)
  1547. #endif
  1548. int ___pskb_trim(struct sk_buff *skb, unsigned int len);
  1549. static inline void __skb_trim(struct sk_buff *skb, unsigned int len)
  1550. {
  1551. if (unlikely(skb_is_nonlinear(skb))) {
  1552. WARN_ON(1);
  1553. return;
  1554. }
  1555. skb->len = len;
  1556. skb_set_tail_pointer(skb, len);
  1557. }
  1558. void skb_trim(struct sk_buff *skb, unsigned int len);
  1559. static inline int __pskb_trim(struct sk_buff *skb, unsigned int len)
  1560. {
  1561. if (skb->data_len)
  1562. return ___pskb_trim(skb, len);
  1563. __skb_trim(skb, len);
  1564. return 0;
  1565. }
  1566. static inline int pskb_trim(struct sk_buff *skb, unsigned int len)
  1567. {
  1568. return (len < skb->len) ? __pskb_trim(skb, len) : 0;
  1569. }
  1570. /**
  1571. * pskb_trim_unique - remove end from a paged unique (not cloned) buffer
  1572. * @skb: buffer to alter
  1573. * @len: new length
  1574. *
  1575. * This is identical to pskb_trim except that the caller knows that
  1576. * the skb is not cloned so we should never get an error due to out-
  1577. * of-memory.
  1578. */
  1579. static inline void pskb_trim_unique(struct sk_buff *skb, unsigned int len)
  1580. {
  1581. int err = pskb_trim(skb, len);
  1582. BUG_ON(err);
  1583. }
  1584. /**
  1585. * skb_orphan - orphan a buffer
  1586. * @skb: buffer to orphan
  1587. *
  1588. * If a buffer currently has an owner then we call the owner's
  1589. * destructor function and make the @skb unowned. The buffer continues
  1590. * to exist but is no longer charged to its former owner.
  1591. */
  1592. static inline void skb_orphan(struct sk_buff *skb)
  1593. {
  1594. if (skb->destructor) {
  1595. skb->destructor(skb);
  1596. skb->destructor = NULL;
  1597. skb->sk = NULL;
  1598. } else {
  1599. BUG_ON(skb->sk);
  1600. }
  1601. }
  1602. /**
  1603. * skb_orphan_frags - orphan the frags contained in a buffer
  1604. * @skb: buffer to orphan frags from
  1605. * @gfp_mask: allocation mask for replacement pages
  1606. *
  1607. * For each frag in the SKB which needs a destructor (i.e. has an
  1608. * owner) create a copy of that frag and release the original
  1609. * page by calling the destructor.
  1610. */
  1611. static inline int skb_orphan_frags(struct sk_buff *skb, gfp_t gfp_mask)
  1612. {
  1613. if (likely(!(skb_shinfo(skb)->tx_flags & SKBTX_DEV_ZEROCOPY)))
  1614. return 0;
  1615. return skb_copy_ubufs(skb, gfp_mask);
  1616. }
  1617. /**
  1618. * __skb_queue_purge - empty a list
  1619. * @list: list to empty
  1620. *
  1621. * Delete all buffers on an &sk_buff list. Each buffer is removed from
  1622. * the list and one reference dropped. This function does not take the
  1623. * list lock and the caller must hold the relevant locks to use it.
  1624. */
  1625. void skb_queue_purge(struct sk_buff_head *list);
  1626. static inline void __skb_queue_purge(struct sk_buff_head *list)
  1627. {
  1628. struct sk_buff *skb;
  1629. while ((skb = __skb_dequeue(list)) != NULL)
  1630. kfree_skb(skb);
  1631. }
  1632. #define NETDEV_FRAG_PAGE_MAX_ORDER get_order(32768)
  1633. #define NETDEV_FRAG_PAGE_MAX_SIZE (PAGE_SIZE << NETDEV_FRAG_PAGE_MAX_ORDER)
  1634. #define NETDEV_PAGECNT_MAX_BIAS NETDEV_FRAG_PAGE_MAX_SIZE
  1635. void *netdev_alloc_frag(unsigned int fragsz);
  1636. struct sk_buff *__netdev_alloc_skb(struct net_device *dev, unsigned int length,
  1637. gfp_t gfp_mask);
  1638. /**
  1639. * netdev_alloc_skb - allocate an skbuff for rx on a specific device
  1640. * @dev: network device to receive on
  1641. * @length: length to allocate
  1642. *
  1643. * Allocate a new &sk_buff and assign it a usage count of one. The
  1644. * buffer has unspecified headroom built in. Users should allocate
  1645. * the headroom they think they need without accounting for the
  1646. * built in space. The built in space is used for optimisations.
  1647. *
  1648. * %NULL is returned if there is no free memory. Although this function
  1649. * allocates memory it can be called from an interrupt.
  1650. */
  1651. static inline struct sk_buff *netdev_alloc_skb(struct net_device *dev,
  1652. unsigned int length)
  1653. {
  1654. return __netdev_alloc_skb(dev, length, GFP_ATOMIC);
  1655. }
  1656. /* legacy helper around __netdev_alloc_skb() */
  1657. static inline struct sk_buff *__dev_alloc_skb(unsigned int length,
  1658. gfp_t gfp_mask)
  1659. {
  1660. return __netdev_alloc_skb(NULL, length, gfp_mask);
  1661. }
  1662. /* legacy helper around netdev_alloc_skb() */
  1663. static inline struct sk_buff *dev_alloc_skb(unsigned int length)
  1664. {
  1665. return netdev_alloc_skb(NULL, length);
  1666. }
  1667. static inline struct sk_buff *__netdev_alloc_skb_ip_align(struct net_device *dev,
  1668. unsigned int length, gfp_t gfp)
  1669. {
  1670. struct sk_buff *skb = __netdev_alloc_skb(dev, length + NET_IP_ALIGN, gfp);
  1671. if (NET_IP_ALIGN && skb)
  1672. skb_reserve(skb, NET_IP_ALIGN);
  1673. return skb;
  1674. }
  1675. static inline struct sk_buff *netdev_alloc_skb_ip_align(struct net_device *dev,
  1676. unsigned int length)
  1677. {
  1678. return __netdev_alloc_skb_ip_align(dev, length, GFP_ATOMIC);
  1679. }
  1680. /**
  1681. * __skb_alloc_pages - allocate pages for ps-rx on a skb and preserve pfmemalloc data
  1682. * @gfp_mask: alloc_pages_node mask. Set __GFP_NOMEMALLOC if not for network packet RX
  1683. * @skb: skb to set pfmemalloc on if __GFP_MEMALLOC is used
  1684. * @order: size of the allocation
  1685. *
  1686. * Allocate a new page.
  1687. *
  1688. * %NULL is returned if there is no free memory.
  1689. */
  1690. static inline struct page *__skb_alloc_pages(gfp_t gfp_mask,
  1691. struct sk_buff *skb,
  1692. unsigned int order)
  1693. {
  1694. struct page *page;
  1695. gfp_mask |= __GFP_COLD;
  1696. if (!(gfp_mask & __GFP_NOMEMALLOC))
  1697. gfp_mask |= __GFP_MEMALLOC;
  1698. page = alloc_pages_node(NUMA_NO_NODE, gfp_mask, order);
  1699. if (skb && page && page->pfmemalloc)
  1700. skb->pfmemalloc = true;
  1701. return page;
  1702. }
  1703. /**
  1704. * __skb_alloc_page - allocate a page for ps-rx for a given skb and preserve pfmemalloc data
  1705. * @gfp_mask: alloc_pages_node mask. Set __GFP_NOMEMALLOC if not for network packet RX
  1706. * @skb: skb to set pfmemalloc on if __GFP_MEMALLOC is used
  1707. *
  1708. * Allocate a new page.
  1709. *
  1710. * %NULL is returned if there is no free memory.
  1711. */
  1712. static inline struct page *__skb_alloc_page(gfp_t gfp_mask,
  1713. struct sk_buff *skb)
  1714. {
  1715. return __skb_alloc_pages(gfp_mask, skb, 0);
  1716. }
  1717. /**
  1718. * skb_propagate_pfmemalloc - Propagate pfmemalloc if skb is allocated after RX page
  1719. * @page: The page that was allocated from skb_alloc_page
  1720. * @skb: The skb that may need pfmemalloc set
  1721. */
  1722. static inline void skb_propagate_pfmemalloc(struct page *page,
  1723. struct sk_buff *skb)
  1724. {
  1725. if (page && page->pfmemalloc)
  1726. skb->pfmemalloc = true;
  1727. }
  1728. /**
  1729. * skb_frag_page - retrieve the page refered to by a paged fragment
  1730. * @frag: the paged fragment
  1731. *
  1732. * Returns the &struct page associated with @frag.
  1733. */
  1734. static inline struct page *skb_frag_page(const skb_frag_t *frag)
  1735. {
  1736. return frag->page.p;
  1737. }
  1738. /**
  1739. * __skb_frag_ref - take an addition reference on a paged fragment.
  1740. * @frag: the paged fragment
  1741. *
  1742. * Takes an additional reference on the paged fragment @frag.
  1743. */
  1744. static inline void __skb_frag_ref(skb_frag_t *frag)
  1745. {
  1746. get_page(skb_frag_page(frag));
  1747. }
  1748. /**
  1749. * skb_frag_ref - take an addition reference on a paged fragment of an skb.
  1750. * @skb: the buffer
  1751. * @f: the fragment offset.
  1752. *
  1753. * Takes an additional reference on the @f'th paged fragment of @skb.
  1754. */
  1755. static inline void skb_frag_ref(struct sk_buff *skb, int f)
  1756. {
  1757. __skb_frag_ref(&skb_shinfo(skb)->frags[f]);
  1758. }
  1759. /**
  1760. * __skb_frag_unref - release a reference on a paged fragment.
  1761. * @frag: the paged fragment
  1762. *
  1763. * Releases a reference on the paged fragment @frag.
  1764. */
  1765. static inline void __skb_frag_unref(skb_frag_t *frag)
  1766. {
  1767. put_page(skb_frag_page(frag));
  1768. }
  1769. /**
  1770. * skb_frag_unref - release a reference on a paged fragment of an skb.
  1771. * @skb: the buffer
  1772. * @f: the fragment offset
  1773. *
  1774. * Releases a reference on the @f'th paged fragment of @skb.
  1775. */
  1776. static inline void skb_frag_unref(struct sk_buff *skb, int f)
  1777. {
  1778. __skb_frag_unref(&skb_shinfo(skb)->frags[f]);
  1779. }
  1780. /**
  1781. * skb_frag_address - gets the address of the data contained in a paged fragment
  1782. * @frag: the paged fragment buffer
  1783. *
  1784. * Returns the address of the data within @frag. The page must already
  1785. * be mapped.
  1786. */
  1787. static inline void *skb_frag_address(const skb_frag_t *frag)
  1788. {
  1789. return page_address(skb_frag_page(frag)) + frag->page_offset;
  1790. }
  1791. /**
  1792. * skb_frag_address_safe - gets the address of the data contained in a paged fragment
  1793. * @frag: the paged fragment buffer
  1794. *
  1795. * Returns the address of the data within @frag. Checks that the page
  1796. * is mapped and returns %NULL otherwise.
  1797. */
  1798. static inline void *skb_frag_address_safe(const skb_frag_t *frag)
  1799. {
  1800. void *ptr = page_address(skb_frag_page(frag));
  1801. if (unlikely(!ptr))
  1802. return NULL;
  1803. return ptr + frag->page_offset;
  1804. }
  1805. /**
  1806. * __skb_frag_set_page - sets the page contained in a paged fragment
  1807. * @frag: the paged fragment
  1808. * @page: the page to set
  1809. *
  1810. * Sets the fragment @frag to contain @page.
  1811. */
  1812. static inline void __skb_frag_set_page(skb_frag_t *frag, struct page *page)
  1813. {
  1814. frag->page.p = page;
  1815. }
  1816. /**
  1817. * skb_frag_set_page - sets the page contained in a paged fragment of an skb
  1818. * @skb: the buffer
  1819. * @f: the fragment offset
  1820. * @page: the page to set
  1821. *
  1822. * Sets the @f'th fragment of @skb to contain @page.
  1823. */
  1824. static inline void skb_frag_set_page(struct sk_buff *skb, int f,
  1825. struct page *page)
  1826. {
  1827. __skb_frag_set_page(&skb_shinfo(skb)->frags[f], page);
  1828. }
  1829. bool skb_page_frag_refill(unsigned int sz, struct page_frag *pfrag, gfp_t prio);
  1830. /**
  1831. * skb_frag_dma_map - maps a paged fragment via the DMA API
  1832. * @dev: the device to map the fragment to
  1833. * @frag: the paged fragment to map
  1834. * @offset: the offset within the fragment (starting at the
  1835. * fragment's own offset)
  1836. * @size: the number of bytes to map
  1837. * @dir: the direction of the mapping (%PCI_DMA_*)
  1838. *
  1839. * Maps the page associated with @frag to @device.
  1840. */
  1841. static inline dma_addr_t skb_frag_dma_map(struct device *dev,
  1842. const skb_frag_t *frag,
  1843. size_t offset, size_t size,
  1844. enum dma_data_direction dir)
  1845. {
  1846. return dma_map_page(dev, skb_frag_page(frag),
  1847. frag->page_offset + offset, size, dir);
  1848. }
  1849. static inline struct sk_buff *pskb_copy(struct sk_buff *skb,
  1850. gfp_t gfp_mask)
  1851. {
  1852. return __pskb_copy(skb, skb_headroom(skb), gfp_mask);
  1853. }
  1854. /**
  1855. * skb_clone_writable - is the header of a clone writable
  1856. * @skb: buffer to check
  1857. * @len: length up to which to write
  1858. *
  1859. * Returns true if modifying the header part of the cloned buffer
  1860. * does not requires the data to be copied.
  1861. */
  1862. static inline int skb_clone_writable(const struct sk_buff *skb, unsigned int len)
  1863. {
  1864. return !skb_header_cloned(skb) &&
  1865. skb_headroom(skb) + len <= skb->hdr_len;
  1866. }
  1867. static inline int __skb_cow(struct sk_buff *skb, unsigned int headroom,
  1868. int cloned)
  1869. {
  1870. int delta = 0;
  1871. if (headroom > skb_headroom(skb))
  1872. delta = headroom - skb_headroom(skb);
  1873. if (delta || cloned)
  1874. return pskb_expand_head(skb, ALIGN(delta, NET_SKB_PAD), 0,
  1875. GFP_ATOMIC);
  1876. return 0;
  1877. }
  1878. /**
  1879. * skb_cow - copy header of skb when it is required
  1880. * @skb: buffer to cow
  1881. * @headroom: needed headroom
  1882. *
  1883. * If the skb passed lacks sufficient headroom or its data part
  1884. * is shared, data is reallocated. If reallocation fails, an error
  1885. * is returned and original skb is not changed.
  1886. *
  1887. * The result is skb with writable area skb->head...skb->tail
  1888. * and at least @headroom of space at head.
  1889. */
  1890. static inline int skb_cow(struct sk_buff *skb, unsigned int headroom)
  1891. {
  1892. return __skb_cow(skb, headroom, skb_cloned(skb));
  1893. }
  1894. /**
  1895. * skb_cow_head - skb_cow but only making the head writable
  1896. * @skb: buffer to cow
  1897. * @headroom: needed headroom
  1898. *
  1899. * This function is identical to skb_cow except that we replace the
  1900. * skb_cloned check by skb_header_cloned. It should be used when
  1901. * you only need to push on some header and do not need to modify
  1902. * the data.
  1903. */
  1904. static inline int skb_cow_head(struct sk_buff *skb, unsigned int headroom)
  1905. {
  1906. return __skb_cow(skb, headroom, skb_header_cloned(skb));
  1907. }
  1908. /**
  1909. * skb_padto - pad an skbuff up to a minimal size
  1910. * @skb: buffer to pad
  1911. * @len: minimal length
  1912. *
  1913. * Pads up a buffer to ensure the trailing bytes exist and are
  1914. * blanked. If the buffer already contains sufficient data it
  1915. * is untouched. Otherwise it is extended. Returns zero on
  1916. * success. The skb is freed on error.
  1917. */
  1918. static inline int skb_padto(struct sk_buff *skb, unsigned int len)
  1919. {
  1920. unsigned int size = skb->len;
  1921. if (likely(size >= len))
  1922. return 0;
  1923. return skb_pad(skb, len - size);
  1924. }
  1925. static inline int skb_add_data(struct sk_buff *skb,
  1926. char __user *from, int copy)
  1927. {
  1928. const int off = skb->len;
  1929. if (skb->ip_summed == CHECKSUM_NONE) {
  1930. int err = 0;
  1931. __wsum csum = csum_and_copy_from_user(from, skb_put(skb, copy),
  1932. copy, 0, &err);
  1933. if (!err) {
  1934. skb->csum = csum_block_add(skb->csum, csum, off);
  1935. return 0;
  1936. }
  1937. } else if (!copy_from_user(skb_put(skb, copy), from, copy))
  1938. return 0;
  1939. __skb_trim(skb, off);
  1940. return -EFAULT;
  1941. }
  1942. static inline bool skb_can_coalesce(struct sk_buff *skb, int i,
  1943. const struct page *page, int off)
  1944. {
  1945. if (i) {
  1946. const struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[i - 1];
  1947. return page == skb_frag_page(frag) &&
  1948. off == frag->page_offset + skb_frag_size(frag);
  1949. }
  1950. return false;
  1951. }
  1952. static inline int __skb_linearize(struct sk_buff *skb)
  1953. {
  1954. return __pskb_pull_tail(skb, skb->data_len) ? 0 : -ENOMEM;
  1955. }
  1956. /**
  1957. * skb_linearize - convert paged skb to linear one
  1958. * @skb: buffer to linarize
  1959. *
  1960. * If there is no free memory -ENOMEM is returned, otherwise zero
  1961. * is returned and the old skb data released.
  1962. */
  1963. static inline int skb_linearize(struct sk_buff *skb)
  1964. {
  1965. return skb_is_nonlinear(skb) ? __skb_linearize(skb) : 0;
  1966. }
  1967. /**
  1968. * skb_has_shared_frag - can any frag be overwritten
  1969. * @skb: buffer to test
  1970. *
  1971. * Return true if the skb has at least one frag that might be modified
  1972. * by an external entity (as in vmsplice()/sendfile())
  1973. */
  1974. static inline bool skb_has_shared_frag(const struct sk_buff *skb)
  1975. {
  1976. return skb_is_nonlinear(skb) &&
  1977. skb_shinfo(skb)->tx_flags & SKBTX_SHARED_FRAG;
  1978. }
  1979. /**
  1980. * skb_linearize_cow - make sure skb is linear and writable
  1981. * @skb: buffer to process
  1982. *
  1983. * If there is no free memory -ENOMEM is returned, otherwise zero
  1984. * is returned and the old skb data released.
  1985. */
  1986. static inline int skb_linearize_cow(struct sk_buff *skb)
  1987. {
  1988. return skb_is_nonlinear(skb) || skb_cloned(skb) ?
  1989. __skb_linearize(skb) : 0;
  1990. }
  1991. /**
  1992. * skb_postpull_rcsum - update checksum for received skb after pull
  1993. * @skb: buffer to update
  1994. * @start: start of data before pull
  1995. * @len: length of data pulled
  1996. *
  1997. * After doing a pull on a received packet, you need to call this to
  1998. * update the CHECKSUM_COMPLETE checksum, or set ip_summed to
  1999. * CHECKSUM_NONE so that it can be recomputed from scratch.
  2000. */
  2001. static inline void skb_postpull_rcsum(struct sk_buff *skb,
  2002. const void *start, unsigned int len)
  2003. {
  2004. if (skb->ip_summed == CHECKSUM_COMPLETE)
  2005. skb->csum = csum_sub(skb->csum, csum_partial(start, len, 0));
  2006. }
  2007. unsigned char *skb_pull_rcsum(struct sk_buff *skb, unsigned int len);
  2008. /**
  2009. * pskb_trim_rcsum - trim received skb and update checksum
  2010. * @skb: buffer to trim
  2011. * @len: new length
  2012. *
  2013. * This is exactly the same as pskb_trim except that it ensures the
  2014. * checksum of received packets are still valid after the operation.
  2015. */
  2016. static inline int pskb_trim_rcsum(struct sk_buff *skb, unsigned int len)
  2017. {
  2018. if (likely(len >= skb->len))
  2019. return 0;
  2020. if (skb->ip_summed == CHECKSUM_COMPLETE)
  2021. skb->ip_summed = CHECKSUM_NONE;
  2022. return __pskb_trim(skb, len);
  2023. }
  2024. #define skb_queue_walk(queue, skb) \
  2025. for (skb = (queue)->next; \
  2026. skb != (struct sk_buff *)(queue); \
  2027. skb = skb->next)
  2028. #define skb_queue_walk_safe(queue, skb, tmp) \
  2029. for (skb = (queue)->next, tmp = skb->next; \
  2030. skb != (struct sk_buff *)(queue); \
  2031. skb = tmp, tmp = skb->next)
  2032. #define skb_queue_walk_from(queue, skb) \
  2033. for (; skb != (struct sk_buff *)(queue); \
  2034. skb = skb->next)
  2035. #define skb_queue_walk_from_safe(queue, skb, tmp) \
  2036. for (tmp = skb->next; \
  2037. skb != (struct sk_buff *)(queue); \
  2038. skb = tmp, tmp = skb->next)
  2039. #define skb_queue_reverse_walk(queue, skb) \
  2040. for (skb = (queue)->prev; \
  2041. skb != (struct sk_buff *)(queue); \
  2042. skb = skb->prev)
  2043. #define skb_queue_reverse_walk_safe(queue, skb, tmp) \
  2044. for (skb = (queue)->prev, tmp = skb->prev; \
  2045. skb != (struct sk_buff *)(queue); \
  2046. skb = tmp, tmp = skb->prev)
  2047. #define skb_queue_reverse_walk_from_safe(queue, skb, tmp) \
  2048. for (tmp = skb->prev; \
  2049. skb != (struct sk_buff *)(queue); \
  2050. skb = tmp, tmp = skb->prev)
  2051. static inline bool skb_has_frag_list(const struct sk_buff *skb)
  2052. {
  2053. return skb_shinfo(skb)->frag_list != NULL;
  2054. }
  2055. static inline void skb_frag_list_init(struct sk_buff *skb)
  2056. {
  2057. skb_shinfo(skb)->frag_list = NULL;
  2058. }
  2059. static inline void skb_frag_add_head(struct sk_buff *skb, struct sk_buff *frag)
  2060. {
  2061. frag->next = skb_shinfo(skb)->frag_list;
  2062. skb_shinfo(skb)->frag_list = frag;
  2063. }
  2064. #define skb_walk_frags(skb, iter) \
  2065. for (iter = skb_shinfo(skb)->frag_list; iter; iter = iter->next)
  2066. struct sk_buff *__skb_recv_datagram(struct sock *sk, unsigned flags,
  2067. int *peeked, int *off, int *err);
  2068. struct sk_buff *skb_recv_datagram(struct sock *sk, unsigned flags, int noblock,
  2069. int *err);
  2070. unsigned int datagram_poll(struct file *file, struct socket *sock,
  2071. struct poll_table_struct *wait);
  2072. int skb_copy_datagram_iovec(const struct sk_buff *from, int offset,
  2073. struct iovec *to, int size);
  2074. int skb_copy_and_csum_datagram_iovec(struct sk_buff *skb, int hlen,
  2075. struct iovec *iov);
  2076. int skb_copy_datagram_from_iovec(struct sk_buff *skb, int offset,
  2077. const struct iovec *from, int from_offset,
  2078. int len);
  2079. int zerocopy_sg_from_iovec(struct sk_buff *skb, const struct iovec *frm,
  2080. int offset, size_t count);
  2081. int skb_copy_datagram_const_iovec(const struct sk_buff *from, int offset,
  2082. const struct iovec *to, int to_offset,
  2083. int size);
  2084. void skb_free_datagram(struct sock *sk, struct sk_buff *skb);
  2085. void skb_free_datagram_locked(struct sock *sk, struct sk_buff *skb);
  2086. int skb_kill_datagram(struct sock *sk, struct sk_buff *skb, unsigned int flags);
  2087. int skb_copy_bits(const struct sk_buff *skb, int offset, void *to, int len);
  2088. int skb_store_bits(struct sk_buff *skb, int offset, const void *from, int len);
  2089. __wsum skb_copy_and_csum_bits(const struct sk_buff *skb, int offset, u8 *to,
  2090. int len, __wsum csum);
  2091. int skb_splice_bits(struct sk_buff *skb, unsigned int offset,
  2092. struct pipe_inode_info *pipe, unsigned int len,
  2093. unsigned int flags);
  2094. void skb_copy_and_csum_dev(const struct sk_buff *skb, u8 *to);
  2095. void skb_split(struct sk_buff *skb, struct sk_buff *skb1, const u32 len);
  2096. int skb_shift(struct sk_buff *tgt, struct sk_buff *skb, int shiftlen);
  2097. void skb_scrub_packet(struct sk_buff *skb, bool xnet);
  2098. struct sk_buff *skb_segment(struct sk_buff *skb, netdev_features_t features);
  2099. struct skb_checksum_ops {
  2100. __wsum (*update)(const void *mem, int len, __wsum wsum);
  2101. __wsum (*combine)(__wsum csum, __wsum csum2, int offset, int len);
  2102. };
  2103. __wsum __skb_checksum(const struct sk_buff *skb, int offset, int len,
  2104. __wsum csum, const struct skb_checksum_ops *ops);
  2105. __wsum skb_checksum(const struct sk_buff *skb, int offset, int len,
  2106. __wsum csum);
  2107. static inline void *skb_header_pointer(const struct sk_buff *skb, int offset,
  2108. int len, void *buffer)
  2109. {
  2110. int hlen = skb_headlen(skb);
  2111. if (hlen - offset >= len)
  2112. return skb->data + offset;
  2113. if (skb_copy_bits(skb, offset, buffer, len) < 0)
  2114. return NULL;
  2115. return buffer;
  2116. }
  2117. static inline void skb_copy_from_linear_data(const struct sk_buff *skb,
  2118. void *to,
  2119. const unsigned int len)
  2120. {
  2121. memcpy(to, skb->data, len);
  2122. }
  2123. static inline void skb_copy_from_linear_data_offset(const struct sk_buff *skb,
  2124. const int offset, void *to,
  2125. const unsigned int len)
  2126. {
  2127. memcpy(to, skb->data + offset, len);
  2128. }
  2129. static inline void skb_copy_to_linear_data(struct sk_buff *skb,
  2130. const void *from,
  2131. const unsigned int len)
  2132. {
  2133. memcpy(skb->data, from, len);
  2134. }
  2135. static inline void skb_copy_to_linear_data_offset(struct sk_buff *skb,
  2136. const int offset,
  2137. const void *from,
  2138. const unsigned int len)
  2139. {
  2140. memcpy(skb->data + offset, from, len);
  2141. }
  2142. void skb_init(void);
  2143. static inline ktime_t skb_get_ktime(const struct sk_buff *skb)
  2144. {
  2145. return skb->tstamp;
  2146. }
  2147. /**
  2148. * skb_get_timestamp - get timestamp from a skb
  2149. * @skb: skb to get stamp from
  2150. * @stamp: pointer to struct timeval to store stamp in
  2151. *
  2152. * Timestamps are stored in the skb as offsets to a base timestamp.
  2153. * This function converts the offset back to a struct timeval and stores
  2154. * it in stamp.
  2155. */
  2156. static inline void skb_get_timestamp(const struct sk_buff *skb,
  2157. struct timeval *stamp)
  2158. {
  2159. *stamp = ktime_to_timeval(skb->tstamp);
  2160. }
  2161. static inline void skb_get_timestampns(const struct sk_buff *skb,
  2162. struct timespec *stamp)
  2163. {
  2164. *stamp = ktime_to_timespec(skb->tstamp);
  2165. }
  2166. static inline void __net_timestamp(struct sk_buff *skb)
  2167. {
  2168. skb->tstamp = ktime_get_real();
  2169. }
  2170. static inline ktime_t net_timedelta(ktime_t t)
  2171. {
  2172. return ktime_sub(ktime_get_real(), t);
  2173. }
  2174. static inline ktime_t net_invalid_timestamp(void)
  2175. {
  2176. return ktime_set(0, 0);
  2177. }
  2178. void skb_timestamping_init(void);
  2179. #ifdef CONFIG_NETWORK_PHY_TIMESTAMPING
  2180. void skb_clone_tx_timestamp(struct sk_buff *skb);
  2181. bool skb_defer_rx_timestamp(struct sk_buff *skb);
  2182. #else /* CONFIG_NETWORK_PHY_TIMESTAMPING */
  2183. static inline void skb_clone_tx_timestamp(struct sk_buff *skb)
  2184. {
  2185. }
  2186. static inline bool skb_defer_rx_timestamp(struct sk_buff *skb)
  2187. {
  2188. return false;
  2189. }
  2190. #endif /* !CONFIG_NETWORK_PHY_TIMESTAMPING */
  2191. /**
  2192. * skb_complete_tx_timestamp() - deliver cloned skb with tx timestamps
  2193. *
  2194. * PHY drivers may accept clones of transmitted packets for
  2195. * timestamping via their phy_driver.txtstamp method. These drivers
  2196. * must call this function to return the skb back to the stack, with
  2197. * or without a timestamp.
  2198. *
  2199. * @skb: clone of the the original outgoing packet
  2200. * @hwtstamps: hardware time stamps, may be NULL if not available
  2201. *
  2202. */
  2203. void skb_complete_tx_timestamp(struct sk_buff *skb,
  2204. struct skb_shared_hwtstamps *hwtstamps);
  2205. /**
  2206. * skb_tstamp_tx - queue clone of skb with send time stamps
  2207. * @orig_skb: the original outgoing packet
  2208. * @hwtstamps: hardware time stamps, may be NULL if not available
  2209. *
  2210. * If the skb has a socket associated, then this function clones the
  2211. * skb (thus sharing the actual data and optional structures), stores
  2212. * the optional hardware time stamping information (if non NULL) or
  2213. * generates a software time stamp (otherwise), then queues the clone
  2214. * to the error queue of the socket. Errors are silently ignored.
  2215. */
  2216. void skb_tstamp_tx(struct sk_buff *orig_skb,
  2217. struct skb_shared_hwtstamps *hwtstamps);
  2218. static inline void sw_tx_timestamp(struct sk_buff *skb)
  2219. {
  2220. if (skb_shinfo(skb)->tx_flags & SKBTX_SW_TSTAMP &&
  2221. !(skb_shinfo(skb)->tx_flags & SKBTX_IN_PROGRESS))
  2222. skb_tstamp_tx(skb, NULL);
  2223. }
  2224. /**
  2225. * skb_tx_timestamp() - Driver hook for transmit timestamping
  2226. *
  2227. * Ethernet MAC Drivers should call this function in their hard_xmit()
  2228. * function immediately before giving the sk_buff to the MAC hardware.
  2229. *
  2230. * @skb: A socket buffer.
  2231. */
  2232. static inline void skb_tx_timestamp(struct sk_buff *skb)
  2233. {
  2234. skb_clone_tx_timestamp(skb);
  2235. sw_tx_timestamp(skb);
  2236. }
  2237. /**
  2238. * skb_complete_wifi_ack - deliver skb with wifi status
  2239. *
  2240. * @skb: the original outgoing packet
  2241. * @acked: ack status
  2242. *
  2243. */
  2244. void skb_complete_wifi_ack(struct sk_buff *skb, bool acked);
  2245. __sum16 __skb_checksum_complete_head(struct sk_buff *skb, int len);
  2246. __sum16 __skb_checksum_complete(struct sk_buff *skb);
  2247. static inline int skb_csum_unnecessary(const struct sk_buff *skb)
  2248. {
  2249. return skb->ip_summed & CHECKSUM_UNNECESSARY;
  2250. }
  2251. /**
  2252. * skb_checksum_complete - Calculate checksum of an entire packet
  2253. * @skb: packet to process
  2254. *
  2255. * This function calculates the checksum over the entire packet plus
  2256. * the value of skb->csum. The latter can be used to supply the
  2257. * checksum of a pseudo header as used by TCP/UDP. It returns the
  2258. * checksum.
  2259. *
  2260. * For protocols that contain complete checksums such as ICMP/TCP/UDP,
  2261. * this function can be used to verify that checksum on received
  2262. * packets. In that case the function should return zero if the
  2263. * checksum is correct. In particular, this function will return zero
  2264. * if skb->ip_summed is CHECKSUM_UNNECESSARY which indicates that the
  2265. * hardware has already verified the correctness of the checksum.
  2266. */
  2267. static inline __sum16 skb_checksum_complete(struct sk_buff *skb)
  2268. {
  2269. return skb_csum_unnecessary(skb) ?
  2270. 0 : __skb_checksum_complete(skb);
  2271. }
  2272. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2273. void nf_conntrack_destroy(struct nf_conntrack *nfct);
  2274. static inline void nf_conntrack_put(struct nf_conntrack *nfct)
  2275. {
  2276. if (nfct && atomic_dec_and_test(&nfct->use))
  2277. nf_conntrack_destroy(nfct);
  2278. }
  2279. static inline void nf_conntrack_get(struct nf_conntrack *nfct)
  2280. {
  2281. if (nfct)
  2282. atomic_inc(&nfct->use);
  2283. }
  2284. #endif
  2285. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2286. static inline void nf_conntrack_get_reasm(struct sk_buff *skb)
  2287. {
  2288. if (skb)
  2289. atomic_inc(&skb->users);
  2290. }
  2291. static inline void nf_conntrack_put_reasm(struct sk_buff *skb)
  2292. {
  2293. if (skb)
  2294. kfree_skb(skb);
  2295. }
  2296. #endif
  2297. #ifdef CONFIG_BRIDGE_NETFILTER
  2298. static inline void nf_bridge_put(struct nf_bridge_info *nf_bridge)
  2299. {
  2300. if (nf_bridge && atomic_dec_and_test(&nf_bridge->use))
  2301. kfree(nf_bridge);
  2302. }
  2303. static inline void nf_bridge_get(struct nf_bridge_info *nf_bridge)
  2304. {
  2305. if (nf_bridge)
  2306. atomic_inc(&nf_bridge->use);
  2307. }
  2308. #endif /* CONFIG_BRIDGE_NETFILTER */
  2309. static inline void nf_reset(struct sk_buff *skb)
  2310. {
  2311. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2312. nf_conntrack_put(skb->nfct);
  2313. skb->nfct = NULL;
  2314. #endif
  2315. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2316. nf_conntrack_put_reasm(skb->nfct_reasm);
  2317. skb->nfct_reasm = NULL;
  2318. #endif
  2319. #ifdef CONFIG_BRIDGE_NETFILTER
  2320. nf_bridge_put(skb->nf_bridge);
  2321. skb->nf_bridge = NULL;
  2322. #endif
  2323. }
  2324. static inline void nf_reset_trace(struct sk_buff *skb)
  2325. {
  2326. #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
  2327. skb->nf_trace = 0;
  2328. #endif
  2329. }
  2330. /* Note: This doesn't put any conntrack and bridge info in dst. */
  2331. static inline void __nf_copy(struct sk_buff *dst, const struct sk_buff *src)
  2332. {
  2333. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2334. dst->nfct = src->nfct;
  2335. nf_conntrack_get(src->nfct);
  2336. dst->nfctinfo = src->nfctinfo;
  2337. #endif
  2338. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2339. dst->nfct_reasm = src->nfct_reasm;
  2340. nf_conntrack_get_reasm(src->nfct_reasm);
  2341. #endif
  2342. #ifdef CONFIG_BRIDGE_NETFILTER
  2343. dst->nf_bridge = src->nf_bridge;
  2344. nf_bridge_get(src->nf_bridge);
  2345. #endif
  2346. }
  2347. static inline void nf_copy(struct sk_buff *dst, const struct sk_buff *src)
  2348. {
  2349. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2350. nf_conntrack_put(dst->nfct);
  2351. #endif
  2352. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2353. nf_conntrack_put_reasm(dst->nfct_reasm);
  2354. #endif
  2355. #ifdef CONFIG_BRIDGE_NETFILTER
  2356. nf_bridge_put(dst->nf_bridge);
  2357. #endif
  2358. __nf_copy(dst, src);
  2359. }
  2360. #ifdef CONFIG_NETWORK_SECMARK
  2361. static inline void skb_copy_secmark(struct sk_buff *to, const struct sk_buff *from)
  2362. {
  2363. to->secmark = from->secmark;
  2364. }
  2365. static inline void skb_init_secmark(struct sk_buff *skb)
  2366. {
  2367. skb->secmark = 0;
  2368. }
  2369. #else
  2370. static inline void skb_copy_secmark(struct sk_buff *to, const struct sk_buff *from)
  2371. { }
  2372. static inline void skb_init_secmark(struct sk_buff *skb)
  2373. { }
  2374. #endif
  2375. static inline void skb_set_queue_mapping(struct sk_buff *skb, u16 queue_mapping)
  2376. {
  2377. skb->queue_mapping = queue_mapping;
  2378. }
  2379. static inline u16 skb_get_queue_mapping(const struct sk_buff *skb)
  2380. {
  2381. return skb->queue_mapping;
  2382. }
  2383. static inline void skb_copy_queue_mapping(struct sk_buff *to, const struct sk_buff *from)
  2384. {
  2385. to->queue_mapping = from->queue_mapping;
  2386. }
  2387. static inline void skb_record_rx_queue(struct sk_buff *skb, u16 rx_queue)
  2388. {
  2389. skb->queue_mapping = rx_queue + 1;
  2390. }
  2391. static inline u16 skb_get_rx_queue(const struct sk_buff *skb)
  2392. {
  2393. return skb->queue_mapping - 1;
  2394. }
  2395. static inline bool skb_rx_queue_recorded(const struct sk_buff *skb)
  2396. {
  2397. return skb->queue_mapping != 0;
  2398. }
  2399. u16 __skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb,
  2400. unsigned int num_tx_queues);
  2401. static inline struct sec_path *skb_sec_path(struct sk_buff *skb)
  2402. {
  2403. #ifdef CONFIG_XFRM
  2404. return skb->sp;
  2405. #else
  2406. return NULL;
  2407. #endif
  2408. }
  2409. /* Keeps track of mac header offset relative to skb->head.
  2410. * It is useful for TSO of Tunneling protocol. e.g. GRE.
  2411. * For non-tunnel skb it points to skb_mac_header() and for
  2412. * tunnel skb it points to outer mac header.
  2413. * Keeps track of level of encapsulation of network headers.
  2414. */
  2415. struct skb_gso_cb {
  2416. int mac_offset;
  2417. int encap_level;
  2418. };
  2419. #define SKB_GSO_CB(skb) ((struct skb_gso_cb *)(skb)->cb)
  2420. static inline int skb_tnl_header_len(const struct sk_buff *inner_skb)
  2421. {
  2422. return (skb_mac_header(inner_skb) - inner_skb->head) -
  2423. SKB_GSO_CB(inner_skb)->mac_offset;
  2424. }
  2425. static inline int gso_pskb_expand_head(struct sk_buff *skb, int extra)
  2426. {
  2427. int new_headroom, headroom;
  2428. int ret;
  2429. headroom = skb_headroom(skb);
  2430. ret = pskb_expand_head(skb, extra, 0, GFP_ATOMIC);
  2431. if (ret)
  2432. return ret;
  2433. new_headroom = skb_headroom(skb);
  2434. SKB_GSO_CB(skb)->mac_offset += (new_headroom - headroom);
  2435. return 0;
  2436. }
  2437. static inline bool skb_is_gso(const struct sk_buff *skb)
  2438. {
  2439. return skb_shinfo(skb)->gso_size;
  2440. }
  2441. /* Note: Should be called only if skb_is_gso(skb) is true */
  2442. static inline bool skb_is_gso_v6(const struct sk_buff *skb)
  2443. {
  2444. return skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6;
  2445. }
  2446. void __skb_warn_lro_forwarding(const struct sk_buff *skb);
  2447. static inline bool skb_warn_if_lro(const struct sk_buff *skb)
  2448. {
  2449. /* LRO sets gso_size but not gso_type, whereas if GSO is really
  2450. * wanted then gso_type will be set. */
  2451. const struct skb_shared_info *shinfo = skb_shinfo(skb);
  2452. if (skb_is_nonlinear(skb) && shinfo->gso_size != 0 &&
  2453. unlikely(shinfo->gso_type == 0)) {
  2454. __skb_warn_lro_forwarding(skb);
  2455. return true;
  2456. }
  2457. return false;
  2458. }
  2459. static inline void skb_forward_csum(struct sk_buff *skb)
  2460. {
  2461. /* Unfortunately we don't support this one. Any brave souls? */
  2462. if (skb->ip_summed == CHECKSUM_COMPLETE)
  2463. skb->ip_summed = CHECKSUM_NONE;
  2464. }
  2465. /**
  2466. * skb_checksum_none_assert - make sure skb ip_summed is CHECKSUM_NONE
  2467. * @skb: skb to check
  2468. *
  2469. * fresh skbs have their ip_summed set to CHECKSUM_NONE.
  2470. * Instead of forcing ip_summed to CHECKSUM_NONE, we can
  2471. * use this helper, to document places where we make this assertion.
  2472. */
  2473. static inline void skb_checksum_none_assert(const struct sk_buff *skb)
  2474. {
  2475. #ifdef DEBUG
  2476. BUG_ON(skb->ip_summed != CHECKSUM_NONE);
  2477. #endif
  2478. }
  2479. bool skb_partial_csum_set(struct sk_buff *skb, u16 start, u16 off);
  2480. u32 __skb_get_poff(const struct sk_buff *skb);
  2481. /**
  2482. * skb_head_is_locked - Determine if the skb->head is locked down
  2483. * @skb: skb to check
  2484. *
  2485. * The head on skbs build around a head frag can be removed if they are
  2486. * not cloned. This function returns true if the skb head is locked down
  2487. * due to either being allocated via kmalloc, or by being a clone with
  2488. * multiple references to the head.
  2489. */
  2490. static inline bool skb_head_is_locked(const struct sk_buff *skb)
  2491. {
  2492. return !skb->head_frag || skb_cloned(skb);
  2493. }
  2494. #endif /* __KERNEL__ */
  2495. #endif /* _LINUX_SKBUFF_H */