skbuff.h 79 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427242824292430243124322433243424352436243724382439244024412442244324442445244624472448244924502451245224532454245524562457245824592460246124622463246424652466246724682469247024712472247324742475247624772478247924802481248224832484248524862487248824892490249124922493249424952496249724982499250025012502250325042505250625072508250925102511251225132514251525162517251825192520252125222523252425252526252725282529253025312532253325342535253625372538253925402541254225432544254525462547254825492550255125522553255425552556255725582559256025612562256325642565256625672568256925702571257225732574257525762577257825792580258125822583258425852586258725882589259025912592259325942595259625972598259926002601260226032604260526062607260826092610261126122613261426152616261726182619262026212622262326242625262626272628262926302631263226332634263526362637263826392640264126422643264426452646264726482649265026512652265326542655265626572658265926602661266226632664266526662667266826692670267126722673267426752676267726782679268026812682268326842685268626872688268926902691269226932694269526962697269826992700270127022703270427052706270727082709271027112712271327142715271627172718271927202721272227232724272527262727272827292730273127322733273427352736273727382739274027412742274327442745274627472748274927502751275227532754275527562757275827592760276127622763276427652766276727682769277027712772277327742775277627772778277927802781278227832784278527862787278827892790279127922793279427952796279727982799280028012802280328042805280628072808280928102811281228132814281528162817281828192820282128222823282428252826282728282829283028312832283328342835283628372838283928402841284228432844284528462847
  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_UDP_TUNNEL = 1 << 7,
  279. SKB_GSO_MPLS = 1 << 8,
  280. };
  281. #if BITS_PER_LONG > 32
  282. #define NET_SKBUFF_DATA_USES_OFFSET 1
  283. #endif
  284. #ifdef NET_SKBUFF_DATA_USES_OFFSET
  285. typedef unsigned int sk_buff_data_t;
  286. #else
  287. typedef unsigned char *sk_buff_data_t;
  288. #endif
  289. #if defined(CONFIG_NF_DEFRAG_IPV4) || defined(CONFIG_NF_DEFRAG_IPV4_MODULE) || \
  290. defined(CONFIG_NF_DEFRAG_IPV6) || defined(CONFIG_NF_DEFRAG_IPV6_MODULE)
  291. #define NET_SKBUFF_NF_DEFRAG_NEEDED 1
  292. #endif
  293. /**
  294. * struct sk_buff - socket buffer
  295. * @next: Next buffer in list
  296. * @prev: Previous buffer in list
  297. * @tstamp: Time we arrived
  298. * @sk: Socket we are owned by
  299. * @dev: Device we arrived on/are leaving by
  300. * @cb: Control buffer. Free for use by every layer. Put private vars here
  301. * @_skb_refdst: destination entry (with norefcount bit)
  302. * @sp: the security path, used for xfrm
  303. * @len: Length of actual data
  304. * @data_len: Data length
  305. * @mac_len: Length of link layer header
  306. * @hdr_len: writable header length of cloned skb
  307. * @csum: Checksum (must include start/offset pair)
  308. * @csum_start: Offset from skb->head where checksumming should start
  309. * @csum_offset: Offset from csum_start where checksum should be stored
  310. * @priority: Packet queueing priority
  311. * @local_df: allow local fragmentation
  312. * @cloned: Head may be cloned (check refcnt to be sure)
  313. * @ip_summed: Driver fed us an IP checksum
  314. * @nohdr: Payload reference only, must not modify header
  315. * @nfctinfo: Relationship of this skb to the connection
  316. * @pkt_type: Packet class
  317. * @fclone: skbuff clone status
  318. * @ipvs_property: skbuff is owned by ipvs
  319. * @peeked: this packet has been seen already, so stats have been
  320. * done for it, don't do them again
  321. * @nf_trace: netfilter packet trace flag
  322. * @protocol: Packet protocol from driver
  323. * @destructor: Destruct function
  324. * @nfct: Associated connection, if any
  325. * @nfct_reasm: netfilter conntrack re-assembly pointer
  326. * @nf_bridge: Saved data about a bridged frame - see br_netfilter.c
  327. * @skb_iif: ifindex of device we arrived on
  328. * @tc_index: Traffic control index
  329. * @tc_verd: traffic control verdict
  330. * @rxhash: the packet hash computed on receive
  331. * @queue_mapping: Queue mapping for multiqueue devices
  332. * @ndisc_nodetype: router type (from link layer)
  333. * @ooo_okay: allow the mapping of a socket to a queue to be changed
  334. * @l4_rxhash: indicate rxhash is a canonical 4-tuple hash over transport
  335. * ports.
  336. * @wifi_acked_valid: wifi_acked was set
  337. * @wifi_acked: whether frame was acked on wifi or not
  338. * @no_fcs: Request NIC to treat last 4 bytes as Ethernet FCS
  339. * @dma_cookie: a cookie to one of several possible DMA operations
  340. * done by skb DMA functions
  341. * @napi_id: id of the NAPI struct this skb came from
  342. * @secmark: security marking
  343. * @mark: Generic packet mark
  344. * @dropcount: total number of sk_receive_queue overflows
  345. * @vlan_proto: vlan encapsulation protocol
  346. * @vlan_tci: vlan tag control information
  347. * @inner_protocol: Protocol (encapsulation)
  348. * @inner_transport_header: Inner transport layer header (encapsulation)
  349. * @inner_network_header: Network layer header (encapsulation)
  350. * @inner_mac_header: Link layer header (encapsulation)
  351. * @transport_header: Transport layer header
  352. * @network_header: Network layer header
  353. * @mac_header: Link layer header
  354. * @tail: Tail pointer
  355. * @end: End pointer
  356. * @head: Head of buffer
  357. * @data: Data head pointer
  358. * @truesize: Buffer size
  359. * @users: User count - see {datagram,tcp}.c
  360. */
  361. struct sk_buff {
  362. /* These two members must be first. */
  363. struct sk_buff *next;
  364. struct sk_buff *prev;
  365. ktime_t tstamp;
  366. struct sock *sk;
  367. struct net_device *dev;
  368. /*
  369. * This is the control buffer. It is free to use for every
  370. * layer. Please put your private variables there. If you
  371. * want to keep them across layers you have to do a skb_clone()
  372. * first. This is owned by whoever has the skb queued ATM.
  373. */
  374. char cb[48] __aligned(8);
  375. unsigned long _skb_refdst;
  376. #ifdef CONFIG_XFRM
  377. struct sec_path *sp;
  378. #endif
  379. unsigned int len,
  380. data_len;
  381. __u16 mac_len,
  382. hdr_len;
  383. union {
  384. __wsum csum;
  385. struct {
  386. __u16 csum_start;
  387. __u16 csum_offset;
  388. };
  389. };
  390. __u32 priority;
  391. kmemcheck_bitfield_begin(flags1);
  392. __u8 local_df:1,
  393. cloned:1,
  394. ip_summed:2,
  395. nohdr:1,
  396. nfctinfo:3;
  397. __u8 pkt_type:3,
  398. fclone:2,
  399. ipvs_property:1,
  400. peeked:1,
  401. nf_trace:1;
  402. kmemcheck_bitfield_end(flags1);
  403. __be16 protocol;
  404. void (*destructor)(struct sk_buff *skb);
  405. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  406. struct nf_conntrack *nfct;
  407. #endif
  408. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  409. struct sk_buff *nfct_reasm;
  410. #endif
  411. #ifdef CONFIG_BRIDGE_NETFILTER
  412. struct nf_bridge_info *nf_bridge;
  413. #endif
  414. int skb_iif;
  415. __u32 rxhash;
  416. __be16 vlan_proto;
  417. __u16 vlan_tci;
  418. #ifdef CONFIG_NET_SCHED
  419. __u16 tc_index; /* traffic control index */
  420. #ifdef CONFIG_NET_CLS_ACT
  421. __u16 tc_verd; /* traffic control verdict */
  422. #endif
  423. #endif
  424. __u16 queue_mapping;
  425. kmemcheck_bitfield_begin(flags2);
  426. #ifdef CONFIG_IPV6_NDISC_NODETYPE
  427. __u8 ndisc_nodetype:2;
  428. #endif
  429. __u8 pfmemalloc:1;
  430. __u8 ooo_okay:1;
  431. __u8 l4_rxhash:1;
  432. __u8 wifi_acked_valid:1;
  433. __u8 wifi_acked:1;
  434. __u8 no_fcs:1;
  435. __u8 head_frag:1;
  436. /* Encapsulation protocol and NIC drivers should use
  437. * this flag to indicate to each other if the skb contains
  438. * encapsulated packet or not and maybe use the inner packet
  439. * headers if needed
  440. */
  441. __u8 encapsulation:1;
  442. /* 7/9 bit hole (depending on ndisc_nodetype presence) */
  443. kmemcheck_bitfield_end(flags2);
  444. #if defined CONFIG_NET_DMA || defined CONFIG_NET_RX_BUSY_POLL
  445. union {
  446. unsigned int napi_id;
  447. dma_cookie_t dma_cookie;
  448. };
  449. #endif
  450. #ifdef CONFIG_NETWORK_SECMARK
  451. __u32 secmark;
  452. #endif
  453. union {
  454. __u32 mark;
  455. __u32 dropcount;
  456. __u32 reserved_tailroom;
  457. };
  458. __be16 inner_protocol;
  459. __u16 inner_transport_header;
  460. __u16 inner_network_header;
  461. __u16 inner_mac_header;
  462. __u16 transport_header;
  463. __u16 network_header;
  464. __u16 mac_header;
  465. /* These elements must be at the end, see alloc_skb() for details. */
  466. sk_buff_data_t tail;
  467. sk_buff_data_t end;
  468. unsigned char *head,
  469. *data;
  470. unsigned int truesize;
  471. atomic_t users;
  472. };
  473. #ifdef __KERNEL__
  474. /*
  475. * Handling routines are only of interest to the kernel
  476. */
  477. #include <linux/slab.h>
  478. #define SKB_ALLOC_FCLONE 0x01
  479. #define SKB_ALLOC_RX 0x02
  480. /* Returns true if the skb was allocated from PFMEMALLOC reserves */
  481. static inline bool skb_pfmemalloc(const struct sk_buff *skb)
  482. {
  483. return unlikely(skb->pfmemalloc);
  484. }
  485. /*
  486. * skb might have a dst pointer attached, refcounted or not.
  487. * _skb_refdst low order bit is set if refcount was _not_ taken
  488. */
  489. #define SKB_DST_NOREF 1UL
  490. #define SKB_DST_PTRMASK ~(SKB_DST_NOREF)
  491. /**
  492. * skb_dst - returns skb dst_entry
  493. * @skb: buffer
  494. *
  495. * Returns skb dst_entry, regardless of reference taken or not.
  496. */
  497. static inline struct dst_entry *skb_dst(const struct sk_buff *skb)
  498. {
  499. /* If refdst was not refcounted, check we still are in a
  500. * rcu_read_lock section
  501. */
  502. WARN_ON((skb->_skb_refdst & SKB_DST_NOREF) &&
  503. !rcu_read_lock_held() &&
  504. !rcu_read_lock_bh_held());
  505. return (struct dst_entry *)(skb->_skb_refdst & SKB_DST_PTRMASK);
  506. }
  507. /**
  508. * skb_dst_set - sets skb dst
  509. * @skb: buffer
  510. * @dst: dst entry
  511. *
  512. * Sets skb dst, assuming a reference was taken on dst and should
  513. * be released by skb_dst_drop()
  514. */
  515. static inline void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst)
  516. {
  517. skb->_skb_refdst = (unsigned long)dst;
  518. }
  519. extern void __skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst,
  520. bool force);
  521. /**
  522. * skb_dst_set_noref - sets skb dst, hopefully, without taking reference
  523. * @skb: buffer
  524. * @dst: dst entry
  525. *
  526. * Sets skb dst, assuming a reference was not taken on dst.
  527. * If dst entry is cached, we do not take reference and dst_release
  528. * will be avoided by refdst_drop. If dst entry is not cached, we take
  529. * reference, so that last dst_release can destroy the dst immediately.
  530. */
  531. static inline void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst)
  532. {
  533. __skb_dst_set_noref(skb, dst, false);
  534. }
  535. /**
  536. * skb_dst_set_noref_force - sets skb dst, without taking reference
  537. * @skb: buffer
  538. * @dst: dst entry
  539. *
  540. * Sets skb dst, assuming a reference was not taken on dst.
  541. * No reference is taken and no dst_release will be called. While for
  542. * cached dsts deferred reclaim is a basic feature, for entries that are
  543. * not cached it is caller's job to guarantee that last dst_release for
  544. * provided dst happens when nobody uses it, eg. after a RCU grace period.
  545. */
  546. static inline void skb_dst_set_noref_force(struct sk_buff *skb,
  547. struct dst_entry *dst)
  548. {
  549. __skb_dst_set_noref(skb, dst, true);
  550. }
  551. /**
  552. * skb_dst_is_noref - Test if skb dst isn't refcounted
  553. * @skb: buffer
  554. */
  555. static inline bool skb_dst_is_noref(const struct sk_buff *skb)
  556. {
  557. return (skb->_skb_refdst & SKB_DST_NOREF) && skb_dst(skb);
  558. }
  559. static inline struct rtable *skb_rtable(const struct sk_buff *skb)
  560. {
  561. return (struct rtable *)skb_dst(skb);
  562. }
  563. extern void kfree_skb(struct sk_buff *skb);
  564. extern void kfree_skb_list(struct sk_buff *segs);
  565. extern void skb_tx_error(struct sk_buff *skb);
  566. extern void consume_skb(struct sk_buff *skb);
  567. extern void __kfree_skb(struct sk_buff *skb);
  568. extern struct kmem_cache *skbuff_head_cache;
  569. extern void kfree_skb_partial(struct sk_buff *skb, bool head_stolen);
  570. extern bool skb_try_coalesce(struct sk_buff *to, struct sk_buff *from,
  571. bool *fragstolen, int *delta_truesize);
  572. extern struct sk_buff *__alloc_skb(unsigned int size,
  573. gfp_t priority, int flags, int node);
  574. extern struct sk_buff *build_skb(void *data, unsigned int frag_size);
  575. static inline struct sk_buff *alloc_skb(unsigned int size,
  576. gfp_t priority)
  577. {
  578. return __alloc_skb(size, priority, 0, NUMA_NO_NODE);
  579. }
  580. static inline struct sk_buff *alloc_skb_fclone(unsigned int size,
  581. gfp_t priority)
  582. {
  583. return __alloc_skb(size, priority, SKB_ALLOC_FCLONE, NUMA_NO_NODE);
  584. }
  585. extern struct sk_buff *__alloc_skb_head(gfp_t priority, int node);
  586. static inline struct sk_buff *alloc_skb_head(gfp_t priority)
  587. {
  588. return __alloc_skb_head(priority, -1);
  589. }
  590. extern struct sk_buff *skb_morph(struct sk_buff *dst, struct sk_buff *src);
  591. extern int skb_copy_ubufs(struct sk_buff *skb, gfp_t gfp_mask);
  592. extern struct sk_buff *skb_clone(struct sk_buff *skb,
  593. gfp_t priority);
  594. extern struct sk_buff *skb_copy(const struct sk_buff *skb,
  595. gfp_t priority);
  596. extern struct sk_buff *__pskb_copy(struct sk_buff *skb,
  597. int headroom, gfp_t gfp_mask);
  598. extern int pskb_expand_head(struct sk_buff *skb,
  599. int nhead, int ntail,
  600. gfp_t gfp_mask);
  601. extern struct sk_buff *skb_realloc_headroom(struct sk_buff *skb,
  602. unsigned int headroom);
  603. extern struct sk_buff *skb_copy_expand(const struct sk_buff *skb,
  604. int newheadroom, int newtailroom,
  605. gfp_t priority);
  606. extern int skb_to_sgvec(struct sk_buff *skb,
  607. struct scatterlist *sg, int offset,
  608. int len);
  609. extern int skb_cow_data(struct sk_buff *skb, int tailbits,
  610. struct sk_buff **trailer);
  611. extern int skb_pad(struct sk_buff *skb, int pad);
  612. #define dev_kfree_skb(a) consume_skb(a)
  613. extern int skb_append_datato_frags(struct sock *sk, struct sk_buff *skb,
  614. int getfrag(void *from, char *to, int offset,
  615. int len,int odd, struct sk_buff *skb),
  616. void *from, int length);
  617. struct skb_seq_state {
  618. __u32 lower_offset;
  619. __u32 upper_offset;
  620. __u32 frag_idx;
  621. __u32 stepped_offset;
  622. struct sk_buff *root_skb;
  623. struct sk_buff *cur_skb;
  624. __u8 *frag_data;
  625. };
  626. extern void skb_prepare_seq_read(struct sk_buff *skb,
  627. unsigned int from, unsigned int to,
  628. struct skb_seq_state *st);
  629. extern unsigned int skb_seq_read(unsigned int consumed, const u8 **data,
  630. struct skb_seq_state *st);
  631. extern void skb_abort_seq_read(struct skb_seq_state *st);
  632. extern unsigned int skb_find_text(struct sk_buff *skb, unsigned int from,
  633. unsigned int to, struct ts_config *config,
  634. struct ts_state *state);
  635. extern void __skb_get_rxhash(struct sk_buff *skb);
  636. static inline __u32 skb_get_rxhash(struct sk_buff *skb)
  637. {
  638. if (!skb->l4_rxhash)
  639. __skb_get_rxhash(skb);
  640. return skb->rxhash;
  641. }
  642. #ifdef NET_SKBUFF_DATA_USES_OFFSET
  643. static inline unsigned char *skb_end_pointer(const struct sk_buff *skb)
  644. {
  645. return skb->head + skb->end;
  646. }
  647. static inline unsigned int skb_end_offset(const struct sk_buff *skb)
  648. {
  649. return skb->end;
  650. }
  651. #else
  652. static inline unsigned char *skb_end_pointer(const struct sk_buff *skb)
  653. {
  654. return skb->end;
  655. }
  656. static inline unsigned int skb_end_offset(const struct sk_buff *skb)
  657. {
  658. return skb->end - skb->head;
  659. }
  660. #endif
  661. /* Internal */
  662. #define skb_shinfo(SKB) ((struct skb_shared_info *)(skb_end_pointer(SKB)))
  663. static inline struct skb_shared_hwtstamps *skb_hwtstamps(struct sk_buff *skb)
  664. {
  665. return &skb_shinfo(skb)->hwtstamps;
  666. }
  667. /**
  668. * skb_queue_empty - check if a queue is empty
  669. * @list: queue head
  670. *
  671. * Returns true if the queue is empty, false otherwise.
  672. */
  673. static inline int skb_queue_empty(const struct sk_buff_head *list)
  674. {
  675. return list->next == (struct sk_buff *)list;
  676. }
  677. /**
  678. * skb_queue_is_last - check if skb is the last entry in the queue
  679. * @list: queue head
  680. * @skb: buffer
  681. *
  682. * Returns true if @skb is the last buffer on the list.
  683. */
  684. static inline bool skb_queue_is_last(const struct sk_buff_head *list,
  685. const struct sk_buff *skb)
  686. {
  687. return skb->next == (struct sk_buff *)list;
  688. }
  689. /**
  690. * skb_queue_is_first - check if skb is the first entry in the queue
  691. * @list: queue head
  692. * @skb: buffer
  693. *
  694. * Returns true if @skb is the first buffer on the list.
  695. */
  696. static inline bool skb_queue_is_first(const struct sk_buff_head *list,
  697. const struct sk_buff *skb)
  698. {
  699. return skb->prev == (struct sk_buff *)list;
  700. }
  701. /**
  702. * skb_queue_next - return the next packet in the queue
  703. * @list: queue head
  704. * @skb: current buffer
  705. *
  706. * Return the next packet in @list after @skb. It is only valid to
  707. * call this if skb_queue_is_last() evaluates to false.
  708. */
  709. static inline struct sk_buff *skb_queue_next(const struct sk_buff_head *list,
  710. const struct sk_buff *skb)
  711. {
  712. /* This BUG_ON may seem severe, but if we just return then we
  713. * are going to dereference garbage.
  714. */
  715. BUG_ON(skb_queue_is_last(list, skb));
  716. return skb->next;
  717. }
  718. /**
  719. * skb_queue_prev - return the prev packet in the queue
  720. * @list: queue head
  721. * @skb: current buffer
  722. *
  723. * Return the prev packet in @list before @skb. It is only valid to
  724. * call this if skb_queue_is_first() evaluates to false.
  725. */
  726. static inline struct sk_buff *skb_queue_prev(const struct sk_buff_head *list,
  727. const struct sk_buff *skb)
  728. {
  729. /* This BUG_ON may seem severe, but if we just return then we
  730. * are going to dereference garbage.
  731. */
  732. BUG_ON(skb_queue_is_first(list, skb));
  733. return skb->prev;
  734. }
  735. /**
  736. * skb_get - reference buffer
  737. * @skb: buffer to reference
  738. *
  739. * Makes another reference to a socket buffer and returns a pointer
  740. * to the buffer.
  741. */
  742. static inline struct sk_buff *skb_get(struct sk_buff *skb)
  743. {
  744. atomic_inc(&skb->users);
  745. return skb;
  746. }
  747. /*
  748. * If users == 1, we are the only owner and are can avoid redundant
  749. * atomic change.
  750. */
  751. /**
  752. * skb_cloned - is the buffer a clone
  753. * @skb: buffer to check
  754. *
  755. * Returns true if the buffer was generated with skb_clone() and is
  756. * one of multiple shared copies of the buffer. Cloned buffers are
  757. * shared data so must not be written to under normal circumstances.
  758. */
  759. static inline int skb_cloned(const struct sk_buff *skb)
  760. {
  761. return skb->cloned &&
  762. (atomic_read(&skb_shinfo(skb)->dataref) & SKB_DATAREF_MASK) != 1;
  763. }
  764. static inline int skb_unclone(struct sk_buff *skb, gfp_t pri)
  765. {
  766. might_sleep_if(pri & __GFP_WAIT);
  767. if (skb_cloned(skb))
  768. return pskb_expand_head(skb, 0, 0, pri);
  769. return 0;
  770. }
  771. /**
  772. * skb_header_cloned - is the header a clone
  773. * @skb: buffer to check
  774. *
  775. * Returns true if modifying the header part of the buffer requires
  776. * the data to be copied.
  777. */
  778. static inline int skb_header_cloned(const struct sk_buff *skb)
  779. {
  780. int dataref;
  781. if (!skb->cloned)
  782. return 0;
  783. dataref = atomic_read(&skb_shinfo(skb)->dataref);
  784. dataref = (dataref & SKB_DATAREF_MASK) - (dataref >> SKB_DATAREF_SHIFT);
  785. return dataref != 1;
  786. }
  787. /**
  788. * skb_header_release - release reference to header
  789. * @skb: buffer to operate on
  790. *
  791. * Drop a reference to the header part of the buffer. This is done
  792. * by acquiring a payload reference. You must not read from the header
  793. * part of skb->data after this.
  794. */
  795. static inline void skb_header_release(struct sk_buff *skb)
  796. {
  797. BUG_ON(skb->nohdr);
  798. skb->nohdr = 1;
  799. atomic_add(1 << SKB_DATAREF_SHIFT, &skb_shinfo(skb)->dataref);
  800. }
  801. /**
  802. * skb_shared - is the buffer shared
  803. * @skb: buffer to check
  804. *
  805. * Returns true if more than one person has a reference to this
  806. * buffer.
  807. */
  808. static inline int skb_shared(const struct sk_buff *skb)
  809. {
  810. return atomic_read(&skb->users) != 1;
  811. }
  812. /**
  813. * skb_share_check - check if buffer is shared and if so clone it
  814. * @skb: buffer to check
  815. * @pri: priority for memory allocation
  816. *
  817. * If the buffer is shared the buffer is cloned and the old copy
  818. * drops a reference. A new clone with a single reference is returned.
  819. * If the buffer is not shared the original buffer is returned. When
  820. * being called from interrupt status or with spinlocks held pri must
  821. * be GFP_ATOMIC.
  822. *
  823. * NULL is returned on a memory allocation failure.
  824. */
  825. static inline struct sk_buff *skb_share_check(struct sk_buff *skb, gfp_t pri)
  826. {
  827. might_sleep_if(pri & __GFP_WAIT);
  828. if (skb_shared(skb)) {
  829. struct sk_buff *nskb = skb_clone(skb, pri);
  830. if (likely(nskb))
  831. consume_skb(skb);
  832. else
  833. kfree_skb(skb);
  834. skb = nskb;
  835. }
  836. return skb;
  837. }
  838. /*
  839. * Copy shared buffers into a new sk_buff. We effectively do COW on
  840. * packets to handle cases where we have a local reader and forward
  841. * and a couple of other messy ones. The normal one is tcpdumping
  842. * a packet thats being forwarded.
  843. */
  844. /**
  845. * skb_unshare - make a copy of a shared buffer
  846. * @skb: buffer to check
  847. * @pri: priority for memory allocation
  848. *
  849. * If the socket buffer is a clone then this function creates a new
  850. * copy of the data, drops a reference count on the old copy and returns
  851. * the new copy with the reference count at 1. If the buffer is not a clone
  852. * the original buffer is returned. When called with a spinlock held or
  853. * from interrupt state @pri must be %GFP_ATOMIC
  854. *
  855. * %NULL is returned on a memory allocation failure.
  856. */
  857. static inline struct sk_buff *skb_unshare(struct sk_buff *skb,
  858. gfp_t pri)
  859. {
  860. might_sleep_if(pri & __GFP_WAIT);
  861. if (skb_cloned(skb)) {
  862. struct sk_buff *nskb = skb_copy(skb, pri);
  863. kfree_skb(skb); /* Free our shared copy */
  864. skb = nskb;
  865. }
  866. return skb;
  867. }
  868. /**
  869. * skb_peek - peek at the head of an &sk_buff_head
  870. * @list_: list to peek at
  871. *
  872. * Peek an &sk_buff. Unlike most other operations you _MUST_
  873. * be careful with this one. A peek leaves the buffer on the
  874. * list and someone else may run off with it. You must hold
  875. * the appropriate locks or have a private queue to do this.
  876. *
  877. * Returns %NULL for an empty list or a pointer to the head element.
  878. * The reference count is not incremented and the reference is therefore
  879. * volatile. Use with caution.
  880. */
  881. static inline struct sk_buff *skb_peek(const struct sk_buff_head *list_)
  882. {
  883. struct sk_buff *skb = list_->next;
  884. if (skb == (struct sk_buff *)list_)
  885. skb = NULL;
  886. return skb;
  887. }
  888. /**
  889. * skb_peek_next - peek skb following the given one from a queue
  890. * @skb: skb to start from
  891. * @list_: list to peek at
  892. *
  893. * Returns %NULL when the end of the list is met or a pointer to the
  894. * next element. The reference count is not incremented and the
  895. * reference is therefore volatile. Use with caution.
  896. */
  897. static inline struct sk_buff *skb_peek_next(struct sk_buff *skb,
  898. const struct sk_buff_head *list_)
  899. {
  900. struct sk_buff *next = skb->next;
  901. if (next == (struct sk_buff *)list_)
  902. next = NULL;
  903. return next;
  904. }
  905. /**
  906. * skb_peek_tail - peek at the tail of an &sk_buff_head
  907. * @list_: list to peek at
  908. *
  909. * Peek an &sk_buff. Unlike most other operations you _MUST_
  910. * be careful with this one. A peek leaves the buffer on the
  911. * list and someone else may run off with it. You must hold
  912. * the appropriate locks or have a private queue to do this.
  913. *
  914. * Returns %NULL for an empty list or a pointer to the tail element.
  915. * The reference count is not incremented and the reference is therefore
  916. * volatile. Use with caution.
  917. */
  918. static inline struct sk_buff *skb_peek_tail(const struct sk_buff_head *list_)
  919. {
  920. struct sk_buff *skb = list_->prev;
  921. if (skb == (struct sk_buff *)list_)
  922. skb = NULL;
  923. return skb;
  924. }
  925. /**
  926. * skb_queue_len - get queue length
  927. * @list_: list to measure
  928. *
  929. * Return the length of an &sk_buff queue.
  930. */
  931. static inline __u32 skb_queue_len(const struct sk_buff_head *list_)
  932. {
  933. return list_->qlen;
  934. }
  935. /**
  936. * __skb_queue_head_init - initialize non-spinlock portions of sk_buff_head
  937. * @list: queue to initialize
  938. *
  939. * This initializes only the list and queue length aspects of
  940. * an sk_buff_head object. This allows to initialize the list
  941. * aspects of an sk_buff_head without reinitializing things like
  942. * the spinlock. It can also be used for on-stack sk_buff_head
  943. * objects where the spinlock is known to not be used.
  944. */
  945. static inline void __skb_queue_head_init(struct sk_buff_head *list)
  946. {
  947. list->prev = list->next = (struct sk_buff *)list;
  948. list->qlen = 0;
  949. }
  950. /*
  951. * This function creates a split out lock class for each invocation;
  952. * this is needed for now since a whole lot of users of the skb-queue
  953. * infrastructure in drivers have different locking usage (in hardirq)
  954. * than the networking core (in softirq only). In the long run either the
  955. * network layer or drivers should need annotation to consolidate the
  956. * main types of usage into 3 classes.
  957. */
  958. static inline void skb_queue_head_init(struct sk_buff_head *list)
  959. {
  960. spin_lock_init(&list->lock);
  961. __skb_queue_head_init(list);
  962. }
  963. static inline void skb_queue_head_init_class(struct sk_buff_head *list,
  964. struct lock_class_key *class)
  965. {
  966. skb_queue_head_init(list);
  967. lockdep_set_class(&list->lock, class);
  968. }
  969. /*
  970. * Insert an sk_buff on a list.
  971. *
  972. * The "__skb_xxxx()" functions are the non-atomic ones that
  973. * can only be called with interrupts disabled.
  974. */
  975. extern void skb_insert(struct sk_buff *old, struct sk_buff *newsk, struct sk_buff_head *list);
  976. static inline void __skb_insert(struct sk_buff *newsk,
  977. struct sk_buff *prev, struct sk_buff *next,
  978. struct sk_buff_head *list)
  979. {
  980. newsk->next = next;
  981. newsk->prev = prev;
  982. next->prev = prev->next = newsk;
  983. list->qlen++;
  984. }
  985. static inline void __skb_queue_splice(const struct sk_buff_head *list,
  986. struct sk_buff *prev,
  987. struct sk_buff *next)
  988. {
  989. struct sk_buff *first = list->next;
  990. struct sk_buff *last = list->prev;
  991. first->prev = prev;
  992. prev->next = first;
  993. last->next = next;
  994. next->prev = last;
  995. }
  996. /**
  997. * skb_queue_splice - join two skb lists, this is designed for stacks
  998. * @list: the new list to add
  999. * @head: the place to add it in the first list
  1000. */
  1001. static inline void skb_queue_splice(const struct sk_buff_head *list,
  1002. struct sk_buff_head *head)
  1003. {
  1004. if (!skb_queue_empty(list)) {
  1005. __skb_queue_splice(list, (struct sk_buff *) head, head->next);
  1006. head->qlen += list->qlen;
  1007. }
  1008. }
  1009. /**
  1010. * skb_queue_splice_init - join two skb lists and reinitialise the emptied list
  1011. * @list: the new list to add
  1012. * @head: the place to add it in the first list
  1013. *
  1014. * The list at @list is reinitialised
  1015. */
  1016. static inline void skb_queue_splice_init(struct sk_buff_head *list,
  1017. struct sk_buff_head *head)
  1018. {
  1019. if (!skb_queue_empty(list)) {
  1020. __skb_queue_splice(list, (struct sk_buff *) head, head->next);
  1021. head->qlen += list->qlen;
  1022. __skb_queue_head_init(list);
  1023. }
  1024. }
  1025. /**
  1026. * skb_queue_splice_tail - join two skb lists, each list being a queue
  1027. * @list: the new list to add
  1028. * @head: the place to add it in the first list
  1029. */
  1030. static inline void skb_queue_splice_tail(const struct sk_buff_head *list,
  1031. struct sk_buff_head *head)
  1032. {
  1033. if (!skb_queue_empty(list)) {
  1034. __skb_queue_splice(list, head->prev, (struct sk_buff *) head);
  1035. head->qlen += list->qlen;
  1036. }
  1037. }
  1038. /**
  1039. * skb_queue_splice_tail_init - join two skb lists and reinitialise the emptied list
  1040. * @list: the new list to add
  1041. * @head: the place to add it in the first list
  1042. *
  1043. * Each of the lists is a queue.
  1044. * The list at @list is reinitialised
  1045. */
  1046. static inline void skb_queue_splice_tail_init(struct sk_buff_head *list,
  1047. struct sk_buff_head *head)
  1048. {
  1049. if (!skb_queue_empty(list)) {
  1050. __skb_queue_splice(list, head->prev, (struct sk_buff *) head);
  1051. head->qlen += list->qlen;
  1052. __skb_queue_head_init(list);
  1053. }
  1054. }
  1055. /**
  1056. * __skb_queue_after - queue a buffer at the list head
  1057. * @list: list to use
  1058. * @prev: place after this buffer
  1059. * @newsk: buffer to queue
  1060. *
  1061. * Queue a buffer int the middle of a list. This function takes no locks
  1062. * and you must therefore hold required locks before calling it.
  1063. *
  1064. * A buffer cannot be placed on two lists at the same time.
  1065. */
  1066. static inline void __skb_queue_after(struct sk_buff_head *list,
  1067. struct sk_buff *prev,
  1068. struct sk_buff *newsk)
  1069. {
  1070. __skb_insert(newsk, prev, prev->next, list);
  1071. }
  1072. extern void skb_append(struct sk_buff *old, struct sk_buff *newsk,
  1073. struct sk_buff_head *list);
  1074. static inline void __skb_queue_before(struct sk_buff_head *list,
  1075. struct sk_buff *next,
  1076. struct sk_buff *newsk)
  1077. {
  1078. __skb_insert(newsk, next->prev, next, list);
  1079. }
  1080. /**
  1081. * __skb_queue_head - queue a buffer at the list head
  1082. * @list: list to use
  1083. * @newsk: buffer to queue
  1084. *
  1085. * Queue a buffer at the start of a list. This function takes no locks
  1086. * and you must therefore hold required locks before calling it.
  1087. *
  1088. * A buffer cannot be placed on two lists at the same time.
  1089. */
  1090. extern void skb_queue_head(struct sk_buff_head *list, struct sk_buff *newsk);
  1091. static inline void __skb_queue_head(struct sk_buff_head *list,
  1092. struct sk_buff *newsk)
  1093. {
  1094. __skb_queue_after(list, (struct sk_buff *)list, newsk);
  1095. }
  1096. /**
  1097. * __skb_queue_tail - queue a buffer at the list tail
  1098. * @list: list to use
  1099. * @newsk: buffer to queue
  1100. *
  1101. * Queue a buffer at the end of a list. This function takes no locks
  1102. * and you must therefore hold required locks before calling it.
  1103. *
  1104. * A buffer cannot be placed on two lists at the same time.
  1105. */
  1106. extern void skb_queue_tail(struct sk_buff_head *list, struct sk_buff *newsk);
  1107. static inline void __skb_queue_tail(struct sk_buff_head *list,
  1108. struct sk_buff *newsk)
  1109. {
  1110. __skb_queue_before(list, (struct sk_buff *)list, newsk);
  1111. }
  1112. /*
  1113. * remove sk_buff from list. _Must_ be called atomically, and with
  1114. * the list known..
  1115. */
  1116. extern void skb_unlink(struct sk_buff *skb, struct sk_buff_head *list);
  1117. static inline void __skb_unlink(struct sk_buff *skb, struct sk_buff_head *list)
  1118. {
  1119. struct sk_buff *next, *prev;
  1120. list->qlen--;
  1121. next = skb->next;
  1122. prev = skb->prev;
  1123. skb->next = skb->prev = NULL;
  1124. next->prev = prev;
  1125. prev->next = next;
  1126. }
  1127. /**
  1128. * __skb_dequeue - remove from the head of the queue
  1129. * @list: list to dequeue from
  1130. *
  1131. * Remove the head of the list. This function does not take any locks
  1132. * so must be used with appropriate locks held only. The head item is
  1133. * returned or %NULL if the list is empty.
  1134. */
  1135. extern struct sk_buff *skb_dequeue(struct sk_buff_head *list);
  1136. static inline struct sk_buff *__skb_dequeue(struct sk_buff_head *list)
  1137. {
  1138. struct sk_buff *skb = skb_peek(list);
  1139. if (skb)
  1140. __skb_unlink(skb, list);
  1141. return skb;
  1142. }
  1143. /**
  1144. * __skb_dequeue_tail - remove from the tail of the queue
  1145. * @list: list to dequeue from
  1146. *
  1147. * Remove the tail of the list. This function does not take any locks
  1148. * so must be used with appropriate locks held only. The tail item is
  1149. * returned or %NULL if the list is empty.
  1150. */
  1151. extern struct sk_buff *skb_dequeue_tail(struct sk_buff_head *list);
  1152. static inline struct sk_buff *__skb_dequeue_tail(struct sk_buff_head *list)
  1153. {
  1154. struct sk_buff *skb = skb_peek_tail(list);
  1155. if (skb)
  1156. __skb_unlink(skb, list);
  1157. return skb;
  1158. }
  1159. static inline bool skb_is_nonlinear(const struct sk_buff *skb)
  1160. {
  1161. return skb->data_len;
  1162. }
  1163. static inline unsigned int skb_headlen(const struct sk_buff *skb)
  1164. {
  1165. return skb->len - skb->data_len;
  1166. }
  1167. static inline int skb_pagelen(const struct sk_buff *skb)
  1168. {
  1169. int i, len = 0;
  1170. for (i = (int)skb_shinfo(skb)->nr_frags - 1; i >= 0; i--)
  1171. len += skb_frag_size(&skb_shinfo(skb)->frags[i]);
  1172. return len + skb_headlen(skb);
  1173. }
  1174. /**
  1175. * __skb_fill_page_desc - initialise a paged fragment in an skb
  1176. * @skb: buffer containing fragment to be initialised
  1177. * @i: paged fragment index to initialise
  1178. * @page: the page to use for this fragment
  1179. * @off: the offset to the data with @page
  1180. * @size: the length of the data
  1181. *
  1182. * Initialises the @i'th fragment of @skb to point to &size bytes at
  1183. * offset @off within @page.
  1184. *
  1185. * Does not take any additional reference on the fragment.
  1186. */
  1187. static inline void __skb_fill_page_desc(struct sk_buff *skb, int i,
  1188. struct page *page, int off, int size)
  1189. {
  1190. skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
  1191. /*
  1192. * Propagate page->pfmemalloc to the skb if we can. The problem is
  1193. * that not all callers have unique ownership of the page. If
  1194. * pfmemalloc is set, we check the mapping as a mapping implies
  1195. * page->index is set (index and pfmemalloc share space).
  1196. * If it's a valid mapping, we cannot use page->pfmemalloc but we
  1197. * do not lose pfmemalloc information as the pages would not be
  1198. * allocated using __GFP_MEMALLOC.
  1199. */
  1200. frag->page.p = page;
  1201. frag->page_offset = off;
  1202. skb_frag_size_set(frag, size);
  1203. page = compound_head(page);
  1204. if (page->pfmemalloc && !page->mapping)
  1205. skb->pfmemalloc = true;
  1206. }
  1207. /**
  1208. * skb_fill_page_desc - initialise a paged fragment in an skb
  1209. * @skb: buffer containing fragment to be initialised
  1210. * @i: paged fragment index to initialise
  1211. * @page: the page to use for this fragment
  1212. * @off: the offset to the data with @page
  1213. * @size: the length of the data
  1214. *
  1215. * As per __skb_fill_page_desc() -- initialises the @i'th fragment of
  1216. * @skb to point to &size bytes at offset @off within @page. In
  1217. * addition updates @skb such that @i is the last fragment.
  1218. *
  1219. * Does not take any additional reference on the fragment.
  1220. */
  1221. static inline void skb_fill_page_desc(struct sk_buff *skb, int i,
  1222. struct page *page, int off, int size)
  1223. {
  1224. __skb_fill_page_desc(skb, i, page, off, size);
  1225. skb_shinfo(skb)->nr_frags = i + 1;
  1226. }
  1227. extern void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page,
  1228. int off, int size, unsigned int truesize);
  1229. #define SKB_PAGE_ASSERT(skb) BUG_ON(skb_shinfo(skb)->nr_frags)
  1230. #define SKB_FRAG_ASSERT(skb) BUG_ON(skb_has_frag_list(skb))
  1231. #define SKB_LINEAR_ASSERT(skb) BUG_ON(skb_is_nonlinear(skb))
  1232. #ifdef NET_SKBUFF_DATA_USES_OFFSET
  1233. static inline unsigned char *skb_tail_pointer(const struct sk_buff *skb)
  1234. {
  1235. return skb->head + skb->tail;
  1236. }
  1237. static inline void skb_reset_tail_pointer(struct sk_buff *skb)
  1238. {
  1239. skb->tail = skb->data - skb->head;
  1240. }
  1241. static inline void skb_set_tail_pointer(struct sk_buff *skb, const int offset)
  1242. {
  1243. skb_reset_tail_pointer(skb);
  1244. skb->tail += offset;
  1245. }
  1246. #else /* NET_SKBUFF_DATA_USES_OFFSET */
  1247. static inline unsigned char *skb_tail_pointer(const struct sk_buff *skb)
  1248. {
  1249. return skb->tail;
  1250. }
  1251. static inline void skb_reset_tail_pointer(struct sk_buff *skb)
  1252. {
  1253. skb->tail = skb->data;
  1254. }
  1255. static inline void skb_set_tail_pointer(struct sk_buff *skb, const int offset)
  1256. {
  1257. skb->tail = skb->data + offset;
  1258. }
  1259. #endif /* NET_SKBUFF_DATA_USES_OFFSET */
  1260. /*
  1261. * Add data to an sk_buff
  1262. */
  1263. extern unsigned char *skb_put(struct sk_buff *skb, unsigned int len);
  1264. static inline unsigned char *__skb_put(struct sk_buff *skb, unsigned int len)
  1265. {
  1266. unsigned char *tmp = skb_tail_pointer(skb);
  1267. SKB_LINEAR_ASSERT(skb);
  1268. skb->tail += len;
  1269. skb->len += len;
  1270. return tmp;
  1271. }
  1272. extern unsigned char *skb_push(struct sk_buff *skb, unsigned int len);
  1273. static inline unsigned char *__skb_push(struct sk_buff *skb, unsigned int len)
  1274. {
  1275. skb->data -= len;
  1276. skb->len += len;
  1277. return skb->data;
  1278. }
  1279. extern unsigned char *skb_pull(struct sk_buff *skb, unsigned int len);
  1280. static inline unsigned char *__skb_pull(struct sk_buff *skb, unsigned int len)
  1281. {
  1282. skb->len -= len;
  1283. BUG_ON(skb->len < skb->data_len);
  1284. return skb->data += len;
  1285. }
  1286. static inline unsigned char *skb_pull_inline(struct sk_buff *skb, unsigned int len)
  1287. {
  1288. return unlikely(len > skb->len) ? NULL : __skb_pull(skb, len);
  1289. }
  1290. extern unsigned char *__pskb_pull_tail(struct sk_buff *skb, int delta);
  1291. static inline unsigned char *__pskb_pull(struct sk_buff *skb, unsigned int len)
  1292. {
  1293. if (len > skb_headlen(skb) &&
  1294. !__pskb_pull_tail(skb, len - skb_headlen(skb)))
  1295. return NULL;
  1296. skb->len -= len;
  1297. return skb->data += len;
  1298. }
  1299. static inline unsigned char *pskb_pull(struct sk_buff *skb, unsigned int len)
  1300. {
  1301. return unlikely(len > skb->len) ? NULL : __pskb_pull(skb, len);
  1302. }
  1303. static inline int pskb_may_pull(struct sk_buff *skb, unsigned int len)
  1304. {
  1305. if (likely(len <= skb_headlen(skb)))
  1306. return 1;
  1307. if (unlikely(len > skb->len))
  1308. return 0;
  1309. return __pskb_pull_tail(skb, len - skb_headlen(skb)) != NULL;
  1310. }
  1311. /**
  1312. * skb_headroom - bytes at buffer head
  1313. * @skb: buffer to check
  1314. *
  1315. * Return the number of bytes of free space at the head of an &sk_buff.
  1316. */
  1317. static inline unsigned int skb_headroom(const struct sk_buff *skb)
  1318. {
  1319. return skb->data - skb->head;
  1320. }
  1321. /**
  1322. * skb_tailroom - bytes at buffer end
  1323. * @skb: buffer to check
  1324. *
  1325. * Return the number of bytes of free space at the tail of an sk_buff
  1326. */
  1327. static inline int skb_tailroom(const struct sk_buff *skb)
  1328. {
  1329. return skb_is_nonlinear(skb) ? 0 : skb->end - skb->tail;
  1330. }
  1331. /**
  1332. * skb_availroom - bytes at buffer end
  1333. * @skb: buffer to check
  1334. *
  1335. * Return the number of bytes of free space at the tail of an sk_buff
  1336. * allocated by sk_stream_alloc()
  1337. */
  1338. static inline int skb_availroom(const struct sk_buff *skb)
  1339. {
  1340. if (skb_is_nonlinear(skb))
  1341. return 0;
  1342. return skb->end - skb->tail - skb->reserved_tailroom;
  1343. }
  1344. /**
  1345. * skb_reserve - adjust headroom
  1346. * @skb: buffer to alter
  1347. * @len: bytes to move
  1348. *
  1349. * Increase the headroom of an empty &sk_buff by reducing the tail
  1350. * room. This is only allowed for an empty buffer.
  1351. */
  1352. static inline void skb_reserve(struct sk_buff *skb, int len)
  1353. {
  1354. skb->data += len;
  1355. skb->tail += len;
  1356. }
  1357. static inline void skb_reset_inner_headers(struct sk_buff *skb)
  1358. {
  1359. skb->inner_mac_header = skb->mac_header;
  1360. skb->inner_network_header = skb->network_header;
  1361. skb->inner_transport_header = skb->transport_header;
  1362. }
  1363. static inline void skb_reset_mac_len(struct sk_buff *skb)
  1364. {
  1365. skb->mac_len = skb->network_header - skb->mac_header;
  1366. }
  1367. static inline unsigned char *skb_inner_transport_header(const struct sk_buff
  1368. *skb)
  1369. {
  1370. return skb->head + skb->inner_transport_header;
  1371. }
  1372. static inline void skb_reset_inner_transport_header(struct sk_buff *skb)
  1373. {
  1374. skb->inner_transport_header = skb->data - skb->head;
  1375. }
  1376. static inline void skb_set_inner_transport_header(struct sk_buff *skb,
  1377. const int offset)
  1378. {
  1379. skb_reset_inner_transport_header(skb);
  1380. skb->inner_transport_header += offset;
  1381. }
  1382. static inline unsigned char *skb_inner_network_header(const struct sk_buff *skb)
  1383. {
  1384. return skb->head + skb->inner_network_header;
  1385. }
  1386. static inline void skb_reset_inner_network_header(struct sk_buff *skb)
  1387. {
  1388. skb->inner_network_header = skb->data - skb->head;
  1389. }
  1390. static inline void skb_set_inner_network_header(struct sk_buff *skb,
  1391. const int offset)
  1392. {
  1393. skb_reset_inner_network_header(skb);
  1394. skb->inner_network_header += offset;
  1395. }
  1396. static inline unsigned char *skb_inner_mac_header(const struct sk_buff *skb)
  1397. {
  1398. return skb->head + skb->inner_mac_header;
  1399. }
  1400. static inline void skb_reset_inner_mac_header(struct sk_buff *skb)
  1401. {
  1402. skb->inner_mac_header = skb->data - skb->head;
  1403. }
  1404. static inline void skb_set_inner_mac_header(struct sk_buff *skb,
  1405. const int offset)
  1406. {
  1407. skb_reset_inner_mac_header(skb);
  1408. skb->inner_mac_header += offset;
  1409. }
  1410. static inline bool skb_transport_header_was_set(const struct sk_buff *skb)
  1411. {
  1412. return skb->transport_header != (typeof(skb->transport_header))~0U;
  1413. }
  1414. static inline unsigned char *skb_transport_header(const struct sk_buff *skb)
  1415. {
  1416. return skb->head + skb->transport_header;
  1417. }
  1418. static inline void skb_reset_transport_header(struct sk_buff *skb)
  1419. {
  1420. skb->transport_header = skb->data - skb->head;
  1421. }
  1422. static inline void skb_set_transport_header(struct sk_buff *skb,
  1423. const int offset)
  1424. {
  1425. skb_reset_transport_header(skb);
  1426. skb->transport_header += offset;
  1427. }
  1428. static inline unsigned char *skb_network_header(const struct sk_buff *skb)
  1429. {
  1430. return skb->head + skb->network_header;
  1431. }
  1432. static inline void skb_reset_network_header(struct sk_buff *skb)
  1433. {
  1434. skb->network_header = skb->data - skb->head;
  1435. }
  1436. static inline void skb_set_network_header(struct sk_buff *skb, const int offset)
  1437. {
  1438. skb_reset_network_header(skb);
  1439. skb->network_header += offset;
  1440. }
  1441. static inline unsigned char *skb_mac_header(const struct sk_buff *skb)
  1442. {
  1443. return skb->head + skb->mac_header;
  1444. }
  1445. static inline int skb_mac_header_was_set(const struct sk_buff *skb)
  1446. {
  1447. return skb->mac_header != (typeof(skb->mac_header))~0U;
  1448. }
  1449. static inline void skb_reset_mac_header(struct sk_buff *skb)
  1450. {
  1451. skb->mac_header = skb->data - skb->head;
  1452. }
  1453. static inline void skb_set_mac_header(struct sk_buff *skb, const int offset)
  1454. {
  1455. skb_reset_mac_header(skb);
  1456. skb->mac_header += offset;
  1457. }
  1458. static inline void skb_probe_transport_header(struct sk_buff *skb,
  1459. const int offset_hint)
  1460. {
  1461. struct flow_keys keys;
  1462. if (skb_transport_header_was_set(skb))
  1463. return;
  1464. else if (skb_flow_dissect(skb, &keys))
  1465. skb_set_transport_header(skb, keys.thoff);
  1466. else
  1467. skb_set_transport_header(skb, offset_hint);
  1468. }
  1469. static inline void skb_mac_header_rebuild(struct sk_buff *skb)
  1470. {
  1471. if (skb_mac_header_was_set(skb)) {
  1472. const unsigned char *old_mac = skb_mac_header(skb);
  1473. skb_set_mac_header(skb, -skb->mac_len);
  1474. memmove(skb_mac_header(skb), old_mac, skb->mac_len);
  1475. }
  1476. }
  1477. static inline int skb_checksum_start_offset(const struct sk_buff *skb)
  1478. {
  1479. return skb->csum_start - skb_headroom(skb);
  1480. }
  1481. static inline int skb_transport_offset(const struct sk_buff *skb)
  1482. {
  1483. return skb_transport_header(skb) - skb->data;
  1484. }
  1485. static inline u32 skb_network_header_len(const struct sk_buff *skb)
  1486. {
  1487. return skb->transport_header - skb->network_header;
  1488. }
  1489. static inline u32 skb_inner_network_header_len(const struct sk_buff *skb)
  1490. {
  1491. return skb->inner_transport_header - skb->inner_network_header;
  1492. }
  1493. static inline int skb_network_offset(const struct sk_buff *skb)
  1494. {
  1495. return skb_network_header(skb) - skb->data;
  1496. }
  1497. static inline int skb_inner_network_offset(const struct sk_buff *skb)
  1498. {
  1499. return skb_inner_network_header(skb) - skb->data;
  1500. }
  1501. static inline int pskb_network_may_pull(struct sk_buff *skb, unsigned int len)
  1502. {
  1503. return pskb_may_pull(skb, skb_network_offset(skb) + len);
  1504. }
  1505. /*
  1506. * CPUs often take a performance hit when accessing unaligned memory
  1507. * locations. The actual performance hit varies, it can be small if the
  1508. * hardware handles it or large if we have to take an exception and fix it
  1509. * in software.
  1510. *
  1511. * Since an ethernet header is 14 bytes network drivers often end up with
  1512. * the IP header at an unaligned offset. The IP header can be aligned by
  1513. * shifting the start of the packet by 2 bytes. Drivers should do this
  1514. * with:
  1515. *
  1516. * skb_reserve(skb, NET_IP_ALIGN);
  1517. *
  1518. * The downside to this alignment of the IP header is that the DMA is now
  1519. * unaligned. On some architectures the cost of an unaligned DMA is high
  1520. * and this cost outweighs the gains made by aligning the IP header.
  1521. *
  1522. * Since this trade off varies between architectures, we allow NET_IP_ALIGN
  1523. * to be overridden.
  1524. */
  1525. #ifndef NET_IP_ALIGN
  1526. #define NET_IP_ALIGN 2
  1527. #endif
  1528. /*
  1529. * The networking layer reserves some headroom in skb data (via
  1530. * dev_alloc_skb). This is used to avoid having to reallocate skb data when
  1531. * the header has to grow. In the default case, if the header has to grow
  1532. * 32 bytes or less we avoid the reallocation.
  1533. *
  1534. * Unfortunately this headroom changes the DMA alignment of the resulting
  1535. * network packet. As for NET_IP_ALIGN, this unaligned DMA is expensive
  1536. * on some architectures. An architecture can override this value,
  1537. * perhaps setting it to a cacheline in size (since that will maintain
  1538. * cacheline alignment of the DMA). It must be a power of 2.
  1539. *
  1540. * Various parts of the networking layer expect at least 32 bytes of
  1541. * headroom, you should not reduce this.
  1542. *
  1543. * Using max(32, L1_CACHE_BYTES) makes sense (especially with RPS)
  1544. * to reduce average number of cache lines per packet.
  1545. * get_rps_cpus() for example only access one 64 bytes aligned block :
  1546. * NET_IP_ALIGN(2) + ethernet_header(14) + IP_header(20/40) + ports(8)
  1547. */
  1548. #ifndef NET_SKB_PAD
  1549. #define NET_SKB_PAD max(32, L1_CACHE_BYTES)
  1550. #endif
  1551. extern int ___pskb_trim(struct sk_buff *skb, unsigned int len);
  1552. static inline void __skb_trim(struct sk_buff *skb, unsigned int len)
  1553. {
  1554. if (unlikely(skb_is_nonlinear(skb))) {
  1555. WARN_ON(1);
  1556. return;
  1557. }
  1558. skb->len = len;
  1559. skb_set_tail_pointer(skb, len);
  1560. }
  1561. extern void skb_trim(struct sk_buff *skb, unsigned int len);
  1562. static inline int __pskb_trim(struct sk_buff *skb, unsigned int len)
  1563. {
  1564. if (skb->data_len)
  1565. return ___pskb_trim(skb, len);
  1566. __skb_trim(skb, len);
  1567. return 0;
  1568. }
  1569. static inline int pskb_trim(struct sk_buff *skb, unsigned int len)
  1570. {
  1571. return (len < skb->len) ? __pskb_trim(skb, len) : 0;
  1572. }
  1573. /**
  1574. * pskb_trim_unique - remove end from a paged unique (not cloned) buffer
  1575. * @skb: buffer to alter
  1576. * @len: new length
  1577. *
  1578. * This is identical to pskb_trim except that the caller knows that
  1579. * the skb is not cloned so we should never get an error due to out-
  1580. * of-memory.
  1581. */
  1582. static inline void pskb_trim_unique(struct sk_buff *skb, unsigned int len)
  1583. {
  1584. int err = pskb_trim(skb, len);
  1585. BUG_ON(err);
  1586. }
  1587. /**
  1588. * skb_orphan - orphan a buffer
  1589. * @skb: buffer to orphan
  1590. *
  1591. * If a buffer currently has an owner then we call the owner's
  1592. * destructor function and make the @skb unowned. The buffer continues
  1593. * to exist but is no longer charged to its former owner.
  1594. */
  1595. static inline void skb_orphan(struct sk_buff *skb)
  1596. {
  1597. if (skb->destructor) {
  1598. skb->destructor(skb);
  1599. skb->destructor = NULL;
  1600. skb->sk = NULL;
  1601. } else {
  1602. BUG_ON(skb->sk);
  1603. }
  1604. }
  1605. /**
  1606. * skb_orphan_frags - orphan the frags contained in a buffer
  1607. * @skb: buffer to orphan frags from
  1608. * @gfp_mask: allocation mask for replacement pages
  1609. *
  1610. * For each frag in the SKB which needs a destructor (i.e. has an
  1611. * owner) create a copy of that frag and release the original
  1612. * page by calling the destructor.
  1613. */
  1614. static inline int skb_orphan_frags(struct sk_buff *skb, gfp_t gfp_mask)
  1615. {
  1616. if (likely(!(skb_shinfo(skb)->tx_flags & SKBTX_DEV_ZEROCOPY)))
  1617. return 0;
  1618. return skb_copy_ubufs(skb, gfp_mask);
  1619. }
  1620. /**
  1621. * __skb_queue_purge - empty a list
  1622. * @list: list to empty
  1623. *
  1624. * Delete all buffers on an &sk_buff list. Each buffer is removed from
  1625. * the list and one reference dropped. This function does not take the
  1626. * list lock and the caller must hold the relevant locks to use it.
  1627. */
  1628. extern void skb_queue_purge(struct sk_buff_head *list);
  1629. static inline void __skb_queue_purge(struct sk_buff_head *list)
  1630. {
  1631. struct sk_buff *skb;
  1632. while ((skb = __skb_dequeue(list)) != NULL)
  1633. kfree_skb(skb);
  1634. }
  1635. #define NETDEV_FRAG_PAGE_MAX_ORDER get_order(32768)
  1636. #define NETDEV_FRAG_PAGE_MAX_SIZE (PAGE_SIZE << NETDEV_FRAG_PAGE_MAX_ORDER)
  1637. #define NETDEV_PAGECNT_MAX_BIAS NETDEV_FRAG_PAGE_MAX_SIZE
  1638. extern void *netdev_alloc_frag(unsigned int fragsz);
  1639. extern struct sk_buff *__netdev_alloc_skb(struct net_device *dev,
  1640. unsigned int length,
  1641. gfp_t gfp_mask);
  1642. /**
  1643. * netdev_alloc_skb - allocate an skbuff for rx on a specific device
  1644. * @dev: network device to receive on
  1645. * @length: length to allocate
  1646. *
  1647. * Allocate a new &sk_buff and assign it a usage count of one. The
  1648. * buffer has unspecified headroom built in. Users should allocate
  1649. * the headroom they think they need without accounting for the
  1650. * built in space. The built in space is used for optimisations.
  1651. *
  1652. * %NULL is returned if there is no free memory. Although this function
  1653. * allocates memory it can be called from an interrupt.
  1654. */
  1655. static inline struct sk_buff *netdev_alloc_skb(struct net_device *dev,
  1656. unsigned int length)
  1657. {
  1658. return __netdev_alloc_skb(dev, length, GFP_ATOMIC);
  1659. }
  1660. /* legacy helper around __netdev_alloc_skb() */
  1661. static inline struct sk_buff *__dev_alloc_skb(unsigned int length,
  1662. gfp_t gfp_mask)
  1663. {
  1664. return __netdev_alloc_skb(NULL, length, gfp_mask);
  1665. }
  1666. /* legacy helper around netdev_alloc_skb() */
  1667. static inline struct sk_buff *dev_alloc_skb(unsigned int length)
  1668. {
  1669. return netdev_alloc_skb(NULL, length);
  1670. }
  1671. static inline struct sk_buff *__netdev_alloc_skb_ip_align(struct net_device *dev,
  1672. unsigned int length, gfp_t gfp)
  1673. {
  1674. struct sk_buff *skb = __netdev_alloc_skb(dev, length + NET_IP_ALIGN, gfp);
  1675. if (NET_IP_ALIGN && skb)
  1676. skb_reserve(skb, NET_IP_ALIGN);
  1677. return skb;
  1678. }
  1679. static inline struct sk_buff *netdev_alloc_skb_ip_align(struct net_device *dev,
  1680. unsigned int length)
  1681. {
  1682. return __netdev_alloc_skb_ip_align(dev, length, GFP_ATOMIC);
  1683. }
  1684. /**
  1685. * __skb_alloc_pages - allocate pages for ps-rx on a skb and preserve pfmemalloc data
  1686. * @gfp_mask: alloc_pages_node mask. Set __GFP_NOMEMALLOC if not for network packet RX
  1687. * @skb: skb to set pfmemalloc on if __GFP_MEMALLOC is used
  1688. * @order: size of the allocation
  1689. *
  1690. * Allocate a new page.
  1691. *
  1692. * %NULL is returned if there is no free memory.
  1693. */
  1694. static inline struct page *__skb_alloc_pages(gfp_t gfp_mask,
  1695. struct sk_buff *skb,
  1696. unsigned int order)
  1697. {
  1698. struct page *page;
  1699. gfp_mask |= __GFP_COLD;
  1700. if (!(gfp_mask & __GFP_NOMEMALLOC))
  1701. gfp_mask |= __GFP_MEMALLOC;
  1702. page = alloc_pages_node(NUMA_NO_NODE, gfp_mask, order);
  1703. if (skb && page && page->pfmemalloc)
  1704. skb->pfmemalloc = true;
  1705. return page;
  1706. }
  1707. /**
  1708. * __skb_alloc_page - allocate a page for ps-rx for a given skb and preserve pfmemalloc data
  1709. * @gfp_mask: alloc_pages_node mask. Set __GFP_NOMEMALLOC if not for network packet RX
  1710. * @skb: skb to set pfmemalloc on if __GFP_MEMALLOC is used
  1711. *
  1712. * Allocate a new page.
  1713. *
  1714. * %NULL is returned if there is no free memory.
  1715. */
  1716. static inline struct page *__skb_alloc_page(gfp_t gfp_mask,
  1717. struct sk_buff *skb)
  1718. {
  1719. return __skb_alloc_pages(gfp_mask, skb, 0);
  1720. }
  1721. /**
  1722. * skb_propagate_pfmemalloc - Propagate pfmemalloc if skb is allocated after RX page
  1723. * @page: The page that was allocated from skb_alloc_page
  1724. * @skb: The skb that may need pfmemalloc set
  1725. */
  1726. static inline void skb_propagate_pfmemalloc(struct page *page,
  1727. struct sk_buff *skb)
  1728. {
  1729. if (page && page->pfmemalloc)
  1730. skb->pfmemalloc = true;
  1731. }
  1732. /**
  1733. * skb_frag_page - retrieve the page refered to by a paged fragment
  1734. * @frag: the paged fragment
  1735. *
  1736. * Returns the &struct page associated with @frag.
  1737. */
  1738. static inline struct page *skb_frag_page(const skb_frag_t *frag)
  1739. {
  1740. return frag->page.p;
  1741. }
  1742. /**
  1743. * __skb_frag_ref - take an addition reference on a paged fragment.
  1744. * @frag: the paged fragment
  1745. *
  1746. * Takes an additional reference on the paged fragment @frag.
  1747. */
  1748. static inline void __skb_frag_ref(skb_frag_t *frag)
  1749. {
  1750. get_page(skb_frag_page(frag));
  1751. }
  1752. /**
  1753. * skb_frag_ref - take an addition reference on a paged fragment of an skb.
  1754. * @skb: the buffer
  1755. * @f: the fragment offset.
  1756. *
  1757. * Takes an additional reference on the @f'th paged fragment of @skb.
  1758. */
  1759. static inline void skb_frag_ref(struct sk_buff *skb, int f)
  1760. {
  1761. __skb_frag_ref(&skb_shinfo(skb)->frags[f]);
  1762. }
  1763. /**
  1764. * __skb_frag_unref - release a reference on a paged fragment.
  1765. * @frag: the paged fragment
  1766. *
  1767. * Releases a reference on the paged fragment @frag.
  1768. */
  1769. static inline void __skb_frag_unref(skb_frag_t *frag)
  1770. {
  1771. put_page(skb_frag_page(frag));
  1772. }
  1773. /**
  1774. * skb_frag_unref - release a reference on a paged fragment of an skb.
  1775. * @skb: the buffer
  1776. * @f: the fragment offset
  1777. *
  1778. * Releases a reference on the @f'th paged fragment of @skb.
  1779. */
  1780. static inline void skb_frag_unref(struct sk_buff *skb, int f)
  1781. {
  1782. __skb_frag_unref(&skb_shinfo(skb)->frags[f]);
  1783. }
  1784. /**
  1785. * skb_frag_address - gets the address of the data contained in a paged fragment
  1786. * @frag: the paged fragment buffer
  1787. *
  1788. * Returns the address of the data within @frag. The page must already
  1789. * be mapped.
  1790. */
  1791. static inline void *skb_frag_address(const skb_frag_t *frag)
  1792. {
  1793. return page_address(skb_frag_page(frag)) + frag->page_offset;
  1794. }
  1795. /**
  1796. * skb_frag_address_safe - gets the address of the data contained in a paged fragment
  1797. * @frag: the paged fragment buffer
  1798. *
  1799. * Returns the address of the data within @frag. Checks that the page
  1800. * is mapped and returns %NULL otherwise.
  1801. */
  1802. static inline void *skb_frag_address_safe(const skb_frag_t *frag)
  1803. {
  1804. void *ptr = page_address(skb_frag_page(frag));
  1805. if (unlikely(!ptr))
  1806. return NULL;
  1807. return ptr + frag->page_offset;
  1808. }
  1809. /**
  1810. * __skb_frag_set_page - sets the page contained in a paged fragment
  1811. * @frag: the paged fragment
  1812. * @page: the page to set
  1813. *
  1814. * Sets the fragment @frag to contain @page.
  1815. */
  1816. static inline void __skb_frag_set_page(skb_frag_t *frag, struct page *page)
  1817. {
  1818. frag->page.p = page;
  1819. }
  1820. /**
  1821. * skb_frag_set_page - sets the page contained in a paged fragment of an skb
  1822. * @skb: the buffer
  1823. * @f: the fragment offset
  1824. * @page: the page to set
  1825. *
  1826. * Sets the @f'th fragment of @skb to contain @page.
  1827. */
  1828. static inline void skb_frag_set_page(struct sk_buff *skb, int f,
  1829. struct page *page)
  1830. {
  1831. __skb_frag_set_page(&skb_shinfo(skb)->frags[f], page);
  1832. }
  1833. /**
  1834. * skb_frag_dma_map - maps a paged fragment via the DMA API
  1835. * @dev: the device to map the fragment to
  1836. * @frag: the paged fragment to map
  1837. * @offset: the offset within the fragment (starting at the
  1838. * fragment's own offset)
  1839. * @size: the number of bytes to map
  1840. * @dir: the direction of the mapping (%PCI_DMA_*)
  1841. *
  1842. * Maps the page associated with @frag to @device.
  1843. */
  1844. static inline dma_addr_t skb_frag_dma_map(struct device *dev,
  1845. const skb_frag_t *frag,
  1846. size_t offset, size_t size,
  1847. enum dma_data_direction dir)
  1848. {
  1849. return dma_map_page(dev, skb_frag_page(frag),
  1850. frag->page_offset + offset, size, dir);
  1851. }
  1852. static inline struct sk_buff *pskb_copy(struct sk_buff *skb,
  1853. gfp_t gfp_mask)
  1854. {
  1855. return __pskb_copy(skb, skb_headroom(skb), gfp_mask);
  1856. }
  1857. /**
  1858. * skb_clone_writable - is the header of a clone writable
  1859. * @skb: buffer to check
  1860. * @len: length up to which to write
  1861. *
  1862. * Returns true if modifying the header part of the cloned buffer
  1863. * does not requires the data to be copied.
  1864. */
  1865. static inline int skb_clone_writable(const struct sk_buff *skb, unsigned int len)
  1866. {
  1867. return !skb_header_cloned(skb) &&
  1868. skb_headroom(skb) + len <= skb->hdr_len;
  1869. }
  1870. static inline int __skb_cow(struct sk_buff *skb, unsigned int headroom,
  1871. int cloned)
  1872. {
  1873. int delta = 0;
  1874. if (headroom > skb_headroom(skb))
  1875. delta = headroom - skb_headroom(skb);
  1876. if (delta || cloned)
  1877. return pskb_expand_head(skb, ALIGN(delta, NET_SKB_PAD), 0,
  1878. GFP_ATOMIC);
  1879. return 0;
  1880. }
  1881. /**
  1882. * skb_cow - copy header of skb when it is required
  1883. * @skb: buffer to cow
  1884. * @headroom: needed headroom
  1885. *
  1886. * If the skb passed lacks sufficient headroom or its data part
  1887. * is shared, data is reallocated. If reallocation fails, an error
  1888. * is returned and original skb is not changed.
  1889. *
  1890. * The result is skb with writable area skb->head...skb->tail
  1891. * and at least @headroom of space at head.
  1892. */
  1893. static inline int skb_cow(struct sk_buff *skb, unsigned int headroom)
  1894. {
  1895. return __skb_cow(skb, headroom, skb_cloned(skb));
  1896. }
  1897. /**
  1898. * skb_cow_head - skb_cow but only making the head writable
  1899. * @skb: buffer to cow
  1900. * @headroom: needed headroom
  1901. *
  1902. * This function is identical to skb_cow except that we replace the
  1903. * skb_cloned check by skb_header_cloned. It should be used when
  1904. * you only need to push on some header and do not need to modify
  1905. * the data.
  1906. */
  1907. static inline int skb_cow_head(struct sk_buff *skb, unsigned int headroom)
  1908. {
  1909. return __skb_cow(skb, headroom, skb_header_cloned(skb));
  1910. }
  1911. /**
  1912. * skb_padto - pad an skbuff up to a minimal size
  1913. * @skb: buffer to pad
  1914. * @len: minimal length
  1915. *
  1916. * Pads up a buffer to ensure the trailing bytes exist and are
  1917. * blanked. If the buffer already contains sufficient data it
  1918. * is untouched. Otherwise it is extended. Returns zero on
  1919. * success. The skb is freed on error.
  1920. */
  1921. static inline int skb_padto(struct sk_buff *skb, unsigned int len)
  1922. {
  1923. unsigned int size = skb->len;
  1924. if (likely(size >= len))
  1925. return 0;
  1926. return skb_pad(skb, len - size);
  1927. }
  1928. static inline int skb_add_data(struct sk_buff *skb,
  1929. char __user *from, int copy)
  1930. {
  1931. const int off = skb->len;
  1932. if (skb->ip_summed == CHECKSUM_NONE) {
  1933. int err = 0;
  1934. __wsum csum = csum_and_copy_from_user(from, skb_put(skb, copy),
  1935. copy, 0, &err);
  1936. if (!err) {
  1937. skb->csum = csum_block_add(skb->csum, csum, off);
  1938. return 0;
  1939. }
  1940. } else if (!copy_from_user(skb_put(skb, copy), from, copy))
  1941. return 0;
  1942. __skb_trim(skb, off);
  1943. return -EFAULT;
  1944. }
  1945. static inline bool skb_can_coalesce(struct sk_buff *skb, int i,
  1946. const struct page *page, int off)
  1947. {
  1948. if (i) {
  1949. const struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[i - 1];
  1950. return page == skb_frag_page(frag) &&
  1951. off == frag->page_offset + skb_frag_size(frag);
  1952. }
  1953. return false;
  1954. }
  1955. static inline int __skb_linearize(struct sk_buff *skb)
  1956. {
  1957. return __pskb_pull_tail(skb, skb->data_len) ? 0 : -ENOMEM;
  1958. }
  1959. /**
  1960. * skb_linearize - convert paged skb to linear one
  1961. * @skb: buffer to linarize
  1962. *
  1963. * If there is no free memory -ENOMEM is returned, otherwise zero
  1964. * is returned and the old skb data released.
  1965. */
  1966. static inline int skb_linearize(struct sk_buff *skb)
  1967. {
  1968. return skb_is_nonlinear(skb) ? __skb_linearize(skb) : 0;
  1969. }
  1970. /**
  1971. * skb_has_shared_frag - can any frag be overwritten
  1972. * @skb: buffer to test
  1973. *
  1974. * Return true if the skb has at least one frag that might be modified
  1975. * by an external entity (as in vmsplice()/sendfile())
  1976. */
  1977. static inline bool skb_has_shared_frag(const struct sk_buff *skb)
  1978. {
  1979. return skb_is_nonlinear(skb) &&
  1980. skb_shinfo(skb)->tx_flags & SKBTX_SHARED_FRAG;
  1981. }
  1982. /**
  1983. * skb_linearize_cow - make sure skb is linear and writable
  1984. * @skb: buffer to process
  1985. *
  1986. * If there is no free memory -ENOMEM is returned, otherwise zero
  1987. * is returned and the old skb data released.
  1988. */
  1989. static inline int skb_linearize_cow(struct sk_buff *skb)
  1990. {
  1991. return skb_is_nonlinear(skb) || skb_cloned(skb) ?
  1992. __skb_linearize(skb) : 0;
  1993. }
  1994. /**
  1995. * skb_postpull_rcsum - update checksum for received skb after pull
  1996. * @skb: buffer to update
  1997. * @start: start of data before pull
  1998. * @len: length of data pulled
  1999. *
  2000. * After doing a pull on a received packet, you need to call this to
  2001. * update the CHECKSUM_COMPLETE checksum, or set ip_summed to
  2002. * CHECKSUM_NONE so that it can be recomputed from scratch.
  2003. */
  2004. static inline void skb_postpull_rcsum(struct sk_buff *skb,
  2005. const void *start, unsigned int len)
  2006. {
  2007. if (skb->ip_summed == CHECKSUM_COMPLETE)
  2008. skb->csum = csum_sub(skb->csum, csum_partial(start, len, 0));
  2009. }
  2010. unsigned char *skb_pull_rcsum(struct sk_buff *skb, unsigned int len);
  2011. /**
  2012. * pskb_trim_rcsum - trim received skb and update checksum
  2013. * @skb: buffer to trim
  2014. * @len: new length
  2015. *
  2016. * This is exactly the same as pskb_trim except that it ensures the
  2017. * checksum of received packets are still valid after the operation.
  2018. */
  2019. static inline int pskb_trim_rcsum(struct sk_buff *skb, unsigned int len)
  2020. {
  2021. if (likely(len >= skb->len))
  2022. return 0;
  2023. if (skb->ip_summed == CHECKSUM_COMPLETE)
  2024. skb->ip_summed = CHECKSUM_NONE;
  2025. return __pskb_trim(skb, len);
  2026. }
  2027. #define skb_queue_walk(queue, skb) \
  2028. for (skb = (queue)->next; \
  2029. skb != (struct sk_buff *)(queue); \
  2030. skb = skb->next)
  2031. #define skb_queue_walk_safe(queue, skb, tmp) \
  2032. for (skb = (queue)->next, tmp = skb->next; \
  2033. skb != (struct sk_buff *)(queue); \
  2034. skb = tmp, tmp = skb->next)
  2035. #define skb_queue_walk_from(queue, skb) \
  2036. for (; skb != (struct sk_buff *)(queue); \
  2037. skb = skb->next)
  2038. #define skb_queue_walk_from_safe(queue, skb, tmp) \
  2039. for (tmp = skb->next; \
  2040. skb != (struct sk_buff *)(queue); \
  2041. skb = tmp, tmp = skb->next)
  2042. #define skb_queue_reverse_walk(queue, skb) \
  2043. for (skb = (queue)->prev; \
  2044. skb != (struct sk_buff *)(queue); \
  2045. skb = skb->prev)
  2046. #define skb_queue_reverse_walk_safe(queue, skb, tmp) \
  2047. for (skb = (queue)->prev, tmp = skb->prev; \
  2048. skb != (struct sk_buff *)(queue); \
  2049. skb = tmp, tmp = skb->prev)
  2050. #define skb_queue_reverse_walk_from_safe(queue, skb, tmp) \
  2051. for (tmp = skb->prev; \
  2052. skb != (struct sk_buff *)(queue); \
  2053. skb = tmp, tmp = skb->prev)
  2054. static inline bool skb_has_frag_list(const struct sk_buff *skb)
  2055. {
  2056. return skb_shinfo(skb)->frag_list != NULL;
  2057. }
  2058. static inline void skb_frag_list_init(struct sk_buff *skb)
  2059. {
  2060. skb_shinfo(skb)->frag_list = NULL;
  2061. }
  2062. static inline void skb_frag_add_head(struct sk_buff *skb, struct sk_buff *frag)
  2063. {
  2064. frag->next = skb_shinfo(skb)->frag_list;
  2065. skb_shinfo(skb)->frag_list = frag;
  2066. }
  2067. #define skb_walk_frags(skb, iter) \
  2068. for (iter = skb_shinfo(skb)->frag_list; iter; iter = iter->next)
  2069. extern struct sk_buff *__skb_recv_datagram(struct sock *sk, unsigned flags,
  2070. int *peeked, int *off, int *err);
  2071. extern struct sk_buff *skb_recv_datagram(struct sock *sk, unsigned flags,
  2072. int noblock, int *err);
  2073. extern unsigned int datagram_poll(struct file *file, struct socket *sock,
  2074. struct poll_table_struct *wait);
  2075. extern int skb_copy_datagram_iovec(const struct sk_buff *from,
  2076. int offset, struct iovec *to,
  2077. int size);
  2078. extern int skb_copy_and_csum_datagram_iovec(struct sk_buff *skb,
  2079. int hlen,
  2080. struct iovec *iov);
  2081. extern int skb_copy_datagram_from_iovec(struct sk_buff *skb,
  2082. int offset,
  2083. const struct iovec *from,
  2084. int from_offset,
  2085. int len);
  2086. extern int zerocopy_sg_from_iovec(struct sk_buff *skb,
  2087. const struct iovec *frm,
  2088. int offset,
  2089. size_t count);
  2090. extern int skb_copy_datagram_const_iovec(const struct sk_buff *from,
  2091. int offset,
  2092. const struct iovec *to,
  2093. int to_offset,
  2094. int size);
  2095. extern void skb_free_datagram(struct sock *sk, struct sk_buff *skb);
  2096. extern void skb_free_datagram_locked(struct sock *sk,
  2097. struct sk_buff *skb);
  2098. extern int skb_kill_datagram(struct sock *sk, struct sk_buff *skb,
  2099. unsigned int flags);
  2100. extern __wsum skb_checksum(const struct sk_buff *skb, int offset,
  2101. int len, __wsum csum);
  2102. extern int skb_copy_bits(const struct sk_buff *skb, int offset,
  2103. void *to, int len);
  2104. extern int skb_store_bits(struct sk_buff *skb, int offset,
  2105. const void *from, int len);
  2106. extern __wsum skb_copy_and_csum_bits(const struct sk_buff *skb,
  2107. int offset, u8 *to, int len,
  2108. __wsum csum);
  2109. extern int skb_splice_bits(struct sk_buff *skb,
  2110. unsigned int offset,
  2111. struct pipe_inode_info *pipe,
  2112. unsigned int len,
  2113. unsigned int flags);
  2114. extern void skb_copy_and_csum_dev(const struct sk_buff *skb, u8 *to);
  2115. extern void skb_split(struct sk_buff *skb,
  2116. struct sk_buff *skb1, const u32 len);
  2117. extern int skb_shift(struct sk_buff *tgt, struct sk_buff *skb,
  2118. int shiftlen);
  2119. extern void skb_scrub_packet(struct sk_buff *skb);
  2120. extern struct sk_buff *skb_segment(struct sk_buff *skb,
  2121. netdev_features_t features);
  2122. static inline void *skb_header_pointer(const struct sk_buff *skb, int offset,
  2123. int len, void *buffer)
  2124. {
  2125. int hlen = skb_headlen(skb);
  2126. if (hlen - offset >= len)
  2127. return skb->data + offset;
  2128. if (skb_copy_bits(skb, offset, buffer, len) < 0)
  2129. return NULL;
  2130. return buffer;
  2131. }
  2132. static inline void skb_copy_from_linear_data(const struct sk_buff *skb,
  2133. void *to,
  2134. const unsigned int len)
  2135. {
  2136. memcpy(to, skb->data, len);
  2137. }
  2138. static inline void skb_copy_from_linear_data_offset(const struct sk_buff *skb,
  2139. const int offset, void *to,
  2140. const unsigned int len)
  2141. {
  2142. memcpy(to, skb->data + offset, len);
  2143. }
  2144. static inline void skb_copy_to_linear_data(struct sk_buff *skb,
  2145. const void *from,
  2146. const unsigned int len)
  2147. {
  2148. memcpy(skb->data, from, len);
  2149. }
  2150. static inline void skb_copy_to_linear_data_offset(struct sk_buff *skb,
  2151. const int offset,
  2152. const void *from,
  2153. const unsigned int len)
  2154. {
  2155. memcpy(skb->data + offset, from, len);
  2156. }
  2157. extern void skb_init(void);
  2158. static inline ktime_t skb_get_ktime(const struct sk_buff *skb)
  2159. {
  2160. return skb->tstamp;
  2161. }
  2162. /**
  2163. * skb_get_timestamp - get timestamp from a skb
  2164. * @skb: skb to get stamp from
  2165. * @stamp: pointer to struct timeval to store stamp in
  2166. *
  2167. * Timestamps are stored in the skb as offsets to a base timestamp.
  2168. * This function converts the offset back to a struct timeval and stores
  2169. * it in stamp.
  2170. */
  2171. static inline void skb_get_timestamp(const struct sk_buff *skb,
  2172. struct timeval *stamp)
  2173. {
  2174. *stamp = ktime_to_timeval(skb->tstamp);
  2175. }
  2176. static inline void skb_get_timestampns(const struct sk_buff *skb,
  2177. struct timespec *stamp)
  2178. {
  2179. *stamp = ktime_to_timespec(skb->tstamp);
  2180. }
  2181. static inline void __net_timestamp(struct sk_buff *skb)
  2182. {
  2183. skb->tstamp = ktime_get_real();
  2184. }
  2185. static inline ktime_t net_timedelta(ktime_t t)
  2186. {
  2187. return ktime_sub(ktime_get_real(), t);
  2188. }
  2189. static inline ktime_t net_invalid_timestamp(void)
  2190. {
  2191. return ktime_set(0, 0);
  2192. }
  2193. extern void skb_timestamping_init(void);
  2194. #ifdef CONFIG_NETWORK_PHY_TIMESTAMPING
  2195. extern void skb_clone_tx_timestamp(struct sk_buff *skb);
  2196. extern bool skb_defer_rx_timestamp(struct sk_buff *skb);
  2197. #else /* CONFIG_NETWORK_PHY_TIMESTAMPING */
  2198. static inline void skb_clone_tx_timestamp(struct sk_buff *skb)
  2199. {
  2200. }
  2201. static inline bool skb_defer_rx_timestamp(struct sk_buff *skb)
  2202. {
  2203. return false;
  2204. }
  2205. #endif /* !CONFIG_NETWORK_PHY_TIMESTAMPING */
  2206. /**
  2207. * skb_complete_tx_timestamp() - deliver cloned skb with tx timestamps
  2208. *
  2209. * PHY drivers may accept clones of transmitted packets for
  2210. * timestamping via their phy_driver.txtstamp method. These drivers
  2211. * must call this function to return the skb back to the stack, with
  2212. * or without a timestamp.
  2213. *
  2214. * @skb: clone of the the original outgoing packet
  2215. * @hwtstamps: hardware time stamps, may be NULL if not available
  2216. *
  2217. */
  2218. void skb_complete_tx_timestamp(struct sk_buff *skb,
  2219. struct skb_shared_hwtstamps *hwtstamps);
  2220. /**
  2221. * skb_tstamp_tx - queue clone of skb with send time stamps
  2222. * @orig_skb: the original outgoing packet
  2223. * @hwtstamps: hardware time stamps, may be NULL if not available
  2224. *
  2225. * If the skb has a socket associated, then this function clones the
  2226. * skb (thus sharing the actual data and optional structures), stores
  2227. * the optional hardware time stamping information (if non NULL) or
  2228. * generates a software time stamp (otherwise), then queues the clone
  2229. * to the error queue of the socket. Errors are silently ignored.
  2230. */
  2231. extern void skb_tstamp_tx(struct sk_buff *orig_skb,
  2232. struct skb_shared_hwtstamps *hwtstamps);
  2233. static inline void sw_tx_timestamp(struct sk_buff *skb)
  2234. {
  2235. if (skb_shinfo(skb)->tx_flags & SKBTX_SW_TSTAMP &&
  2236. !(skb_shinfo(skb)->tx_flags & SKBTX_IN_PROGRESS))
  2237. skb_tstamp_tx(skb, NULL);
  2238. }
  2239. /**
  2240. * skb_tx_timestamp() - Driver hook for transmit timestamping
  2241. *
  2242. * Ethernet MAC Drivers should call this function in their hard_xmit()
  2243. * function immediately before giving the sk_buff to the MAC hardware.
  2244. *
  2245. * @skb: A socket buffer.
  2246. */
  2247. static inline void skb_tx_timestamp(struct sk_buff *skb)
  2248. {
  2249. skb_clone_tx_timestamp(skb);
  2250. sw_tx_timestamp(skb);
  2251. }
  2252. /**
  2253. * skb_complete_wifi_ack - deliver skb with wifi status
  2254. *
  2255. * @skb: the original outgoing packet
  2256. * @acked: ack status
  2257. *
  2258. */
  2259. void skb_complete_wifi_ack(struct sk_buff *skb, bool acked);
  2260. extern __sum16 __skb_checksum_complete_head(struct sk_buff *skb, int len);
  2261. extern __sum16 __skb_checksum_complete(struct sk_buff *skb);
  2262. static inline int skb_csum_unnecessary(const struct sk_buff *skb)
  2263. {
  2264. return skb->ip_summed & CHECKSUM_UNNECESSARY;
  2265. }
  2266. /**
  2267. * skb_checksum_complete - Calculate checksum of an entire packet
  2268. * @skb: packet to process
  2269. *
  2270. * This function calculates the checksum over the entire packet plus
  2271. * the value of skb->csum. The latter can be used to supply the
  2272. * checksum of a pseudo header as used by TCP/UDP. It returns the
  2273. * checksum.
  2274. *
  2275. * For protocols that contain complete checksums such as ICMP/TCP/UDP,
  2276. * this function can be used to verify that checksum on received
  2277. * packets. In that case the function should return zero if the
  2278. * checksum is correct. In particular, this function will return zero
  2279. * if skb->ip_summed is CHECKSUM_UNNECESSARY which indicates that the
  2280. * hardware has already verified the correctness of the checksum.
  2281. */
  2282. static inline __sum16 skb_checksum_complete(struct sk_buff *skb)
  2283. {
  2284. return skb_csum_unnecessary(skb) ?
  2285. 0 : __skb_checksum_complete(skb);
  2286. }
  2287. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2288. extern void nf_conntrack_destroy(struct nf_conntrack *nfct);
  2289. static inline void nf_conntrack_put(struct nf_conntrack *nfct)
  2290. {
  2291. if (nfct && atomic_dec_and_test(&nfct->use))
  2292. nf_conntrack_destroy(nfct);
  2293. }
  2294. static inline void nf_conntrack_get(struct nf_conntrack *nfct)
  2295. {
  2296. if (nfct)
  2297. atomic_inc(&nfct->use);
  2298. }
  2299. #endif
  2300. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2301. static inline void nf_conntrack_get_reasm(struct sk_buff *skb)
  2302. {
  2303. if (skb)
  2304. atomic_inc(&skb->users);
  2305. }
  2306. static inline void nf_conntrack_put_reasm(struct sk_buff *skb)
  2307. {
  2308. if (skb)
  2309. kfree_skb(skb);
  2310. }
  2311. #endif
  2312. #ifdef CONFIG_BRIDGE_NETFILTER
  2313. static inline void nf_bridge_put(struct nf_bridge_info *nf_bridge)
  2314. {
  2315. if (nf_bridge && atomic_dec_and_test(&nf_bridge->use))
  2316. kfree(nf_bridge);
  2317. }
  2318. static inline void nf_bridge_get(struct nf_bridge_info *nf_bridge)
  2319. {
  2320. if (nf_bridge)
  2321. atomic_inc(&nf_bridge->use);
  2322. }
  2323. #endif /* CONFIG_BRIDGE_NETFILTER */
  2324. static inline void nf_reset(struct sk_buff *skb)
  2325. {
  2326. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2327. nf_conntrack_put(skb->nfct);
  2328. skb->nfct = NULL;
  2329. #endif
  2330. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2331. nf_conntrack_put_reasm(skb->nfct_reasm);
  2332. skb->nfct_reasm = NULL;
  2333. #endif
  2334. #ifdef CONFIG_BRIDGE_NETFILTER
  2335. nf_bridge_put(skb->nf_bridge);
  2336. skb->nf_bridge = NULL;
  2337. #endif
  2338. }
  2339. static inline void nf_reset_trace(struct sk_buff *skb)
  2340. {
  2341. #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
  2342. skb->nf_trace = 0;
  2343. #endif
  2344. }
  2345. /* Note: This doesn't put any conntrack and bridge info in dst. */
  2346. static inline void __nf_copy(struct sk_buff *dst, const struct sk_buff *src)
  2347. {
  2348. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2349. dst->nfct = src->nfct;
  2350. nf_conntrack_get(src->nfct);
  2351. dst->nfctinfo = src->nfctinfo;
  2352. #endif
  2353. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2354. dst->nfct_reasm = src->nfct_reasm;
  2355. nf_conntrack_get_reasm(src->nfct_reasm);
  2356. #endif
  2357. #ifdef CONFIG_BRIDGE_NETFILTER
  2358. dst->nf_bridge = src->nf_bridge;
  2359. nf_bridge_get(src->nf_bridge);
  2360. #endif
  2361. }
  2362. static inline void nf_copy(struct sk_buff *dst, const struct sk_buff *src)
  2363. {
  2364. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  2365. nf_conntrack_put(dst->nfct);
  2366. #endif
  2367. #ifdef NET_SKBUFF_NF_DEFRAG_NEEDED
  2368. nf_conntrack_put_reasm(dst->nfct_reasm);
  2369. #endif
  2370. #ifdef CONFIG_BRIDGE_NETFILTER
  2371. nf_bridge_put(dst->nf_bridge);
  2372. #endif
  2373. __nf_copy(dst, src);
  2374. }
  2375. #ifdef CONFIG_NETWORK_SECMARK
  2376. static inline void skb_copy_secmark(struct sk_buff *to, const struct sk_buff *from)
  2377. {
  2378. to->secmark = from->secmark;
  2379. }
  2380. static inline void skb_init_secmark(struct sk_buff *skb)
  2381. {
  2382. skb->secmark = 0;
  2383. }
  2384. #else
  2385. static inline void skb_copy_secmark(struct sk_buff *to, const struct sk_buff *from)
  2386. { }
  2387. static inline void skb_init_secmark(struct sk_buff *skb)
  2388. { }
  2389. #endif
  2390. static inline void skb_set_queue_mapping(struct sk_buff *skb, u16 queue_mapping)
  2391. {
  2392. skb->queue_mapping = queue_mapping;
  2393. }
  2394. static inline u16 skb_get_queue_mapping(const struct sk_buff *skb)
  2395. {
  2396. return skb->queue_mapping;
  2397. }
  2398. static inline void skb_copy_queue_mapping(struct sk_buff *to, const struct sk_buff *from)
  2399. {
  2400. to->queue_mapping = from->queue_mapping;
  2401. }
  2402. static inline void skb_record_rx_queue(struct sk_buff *skb, u16 rx_queue)
  2403. {
  2404. skb->queue_mapping = rx_queue + 1;
  2405. }
  2406. static inline u16 skb_get_rx_queue(const struct sk_buff *skb)
  2407. {
  2408. return skb->queue_mapping - 1;
  2409. }
  2410. static inline bool skb_rx_queue_recorded(const struct sk_buff *skb)
  2411. {
  2412. return skb->queue_mapping != 0;
  2413. }
  2414. extern u16 __skb_tx_hash(const struct net_device *dev,
  2415. const struct sk_buff *skb,
  2416. unsigned int num_tx_queues);
  2417. #ifdef CONFIG_XFRM
  2418. static inline struct sec_path *skb_sec_path(struct sk_buff *skb)
  2419. {
  2420. return skb->sp;
  2421. }
  2422. #else
  2423. static inline struct sec_path *skb_sec_path(struct sk_buff *skb)
  2424. {
  2425. return NULL;
  2426. }
  2427. #endif
  2428. /* Keeps track of mac header offset relative to skb->head.
  2429. * It is useful for TSO of Tunneling protocol. e.g. GRE.
  2430. * For non-tunnel skb it points to skb_mac_header() and for
  2431. * tunnel skb it points to outer mac header. */
  2432. struct skb_gso_cb {
  2433. int mac_offset;
  2434. };
  2435. #define SKB_GSO_CB(skb) ((struct skb_gso_cb *)(skb)->cb)
  2436. static inline int skb_tnl_header_len(const struct sk_buff *inner_skb)
  2437. {
  2438. return (skb_mac_header(inner_skb) - inner_skb->head) -
  2439. SKB_GSO_CB(inner_skb)->mac_offset;
  2440. }
  2441. static inline int gso_pskb_expand_head(struct sk_buff *skb, int extra)
  2442. {
  2443. int new_headroom, headroom;
  2444. int ret;
  2445. headroom = skb_headroom(skb);
  2446. ret = pskb_expand_head(skb, extra, 0, GFP_ATOMIC);
  2447. if (ret)
  2448. return ret;
  2449. new_headroom = skb_headroom(skb);
  2450. SKB_GSO_CB(skb)->mac_offset += (new_headroom - headroom);
  2451. return 0;
  2452. }
  2453. static inline bool skb_is_gso(const struct sk_buff *skb)
  2454. {
  2455. return skb_shinfo(skb)->gso_size;
  2456. }
  2457. static inline bool skb_is_gso_v6(const struct sk_buff *skb)
  2458. {
  2459. return skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6;
  2460. }
  2461. extern void __skb_warn_lro_forwarding(const struct sk_buff *skb);
  2462. static inline bool skb_warn_if_lro(const struct sk_buff *skb)
  2463. {
  2464. /* LRO sets gso_size but not gso_type, whereas if GSO is really
  2465. * wanted then gso_type will be set. */
  2466. const struct skb_shared_info *shinfo = skb_shinfo(skb);
  2467. if (skb_is_nonlinear(skb) && shinfo->gso_size != 0 &&
  2468. unlikely(shinfo->gso_type == 0)) {
  2469. __skb_warn_lro_forwarding(skb);
  2470. return true;
  2471. }
  2472. return false;
  2473. }
  2474. static inline void skb_forward_csum(struct sk_buff *skb)
  2475. {
  2476. /* Unfortunately we don't support this one. Any brave souls? */
  2477. if (skb->ip_summed == CHECKSUM_COMPLETE)
  2478. skb->ip_summed = CHECKSUM_NONE;
  2479. }
  2480. /**
  2481. * skb_checksum_none_assert - make sure skb ip_summed is CHECKSUM_NONE
  2482. * @skb: skb to check
  2483. *
  2484. * fresh skbs have their ip_summed set to CHECKSUM_NONE.
  2485. * Instead of forcing ip_summed to CHECKSUM_NONE, we can
  2486. * use this helper, to document places where we make this assertion.
  2487. */
  2488. static inline void skb_checksum_none_assert(const struct sk_buff *skb)
  2489. {
  2490. #ifdef DEBUG
  2491. BUG_ON(skb->ip_summed != CHECKSUM_NONE);
  2492. #endif
  2493. }
  2494. bool skb_partial_csum_set(struct sk_buff *skb, u16 start, u16 off);
  2495. u32 __skb_get_poff(const struct sk_buff *skb);
  2496. /**
  2497. * skb_head_is_locked - Determine if the skb->head is locked down
  2498. * @skb: skb to check
  2499. *
  2500. * The head on skbs build around a head frag can be removed if they are
  2501. * not cloned. This function returns true if the skb head is locked down
  2502. * due to either being allocated via kmalloc, or by being a clone with
  2503. * multiple references to the head.
  2504. */
  2505. static inline bool skb_head_is_locked(const struct sk_buff *skb)
  2506. {
  2507. return !skb->head_frag || skb_cloned(skb);
  2508. }
  2509. #endif /* __KERNEL__ */
  2510. #endif /* _LINUX_SKBUFF_H */