You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

fuzzy_storage.c 82KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522252325242525252625272528252925302531253225332534253525362537253825392540254125422543254425452546254725482549255025512552255325542555255625572558255925602561256225632564256525662567256825692570257125722573257425752576257725782579258025812582258325842585258625872588258925902591259225932594259525962597259825992600260126022603260426052606260726082609261026112612261326142615261626172618261926202621262226232624262526262627262826292630263126322633263426352636263726382639264026412642264326442645264626472648264926502651265226532654265526562657265826592660266126622663266426652666266726682669267026712672267326742675267626772678267926802681268226832684268526862687268826892690269126922693269426952696269726982699270027012702270327042705270627072708270927102711271227132714271527162717271827192720272127222723272427252726272727282729273027312732273327342735273627372738273927402741274227432744274527462747274827492750275127522753275427552756275727582759276027612762276327642765276627672768276927702771277227732774277527762777277827792780278127822783278427852786278727882789279027912792279327942795279627972798279928002801280228032804280528062807280828092810281128122813281428152816281728182819282028212822282328242825282628272828282928302831283228332834283528362837283828392840284128422843284428452846284728482849285028512852285328542855285628572858285928602861286228632864286528662867286828692870287128722873287428752876287728782879288028812882288328842885288628872888288928902891289228932894289528962897289828992900290129022903290429052906290729082909291029112912291329142915291629172918291929202921292229232924292529262927292829292930293129322933293429352936293729382939294029412942294329442945294629472948294929502951295229532954295529562957295829592960296129622963296429652966296729682969297029712972297329742975297629772978297929802981298229832984298529862987298829892990299129922993299429952996299729982999300030013002300330043005300630073008300930103011301230133014301530163017301830193020302130223023302430253026302730283029303030313032303330343035303630373038303930403041304230433044304530463047304830493050305130523053305430553056305730583059306030613062306330643065306630673068306930703071307230733074307530763077307830793080308130823083308430853086308730883089309030913092309330943095309630973098309931003101310231033104310531063107310831093110311131123113311431153116311731183119312031213122312331243125312631273128312931303131313231333134313531363137313831393140314131423143314431453146314731483149315031513152315331543155315631573158315931603161316231633164316531663167316831693170317131723173317431753176317731783179318031813182318331843185318631873188318931903191319231933194319531963197319831993200320132023203320432053206320732083209321032113212321332143215321632173218321932203221322232233224322532263227322832293230323132323233323432353236323732383239324032413242324332443245324632473248
  1. /*-
  2. * Copyright 2016 Vsevolod Stakhov
  3. *
  4. * Licensed under the Apache License, Version 2.0 (the "License");
  5. * you may not use this file except in compliance with the License.
  6. * You may obtain a copy of the License at
  7. *
  8. * http://www.apache.org/licenses/LICENSE-2.0
  9. *
  10. * Unless required by applicable law or agreed to in writing, software
  11. * distributed under the License is distributed on an "AS IS" BASIS,
  12. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. * See the License for the specific language governing permissions and
  14. * limitations under the License.
  15. */
  16. /*
  17. * Rspamd fuzzy storage server
  18. */
  19. #include "config.h"
  20. #include "libserver/fuzzy_wire.h"
  21. #include "util.h"
  22. #include "rspamd.h"
  23. #include "map.h"
  24. #include "map_helpers.h"
  25. #include "fuzzy_wire.h"
  26. #include "fuzzy_backend.h"
  27. #include "ottery.h"
  28. #include "ref.h"
  29. #include "xxhash.h"
  30. #include "libserver/worker_util.h"
  31. #include "libserver/rspamd_control.h"
  32. #include "libcryptobox/cryptobox.h"
  33. #include "libcryptobox/keypairs_cache.h"
  34. #include "libcryptobox/keypair.h"
  35. #include "libserver/rspamd_control.h"
  36. #include "libutil/hash.h"
  37. #include "libutil/map_private.h"
  38. #include "libutil/http_private.h"
  39. #include "libutil/http_router.h"
  40. #include "unix-std.h"
  41. #include <math.h>
  42. /* Resync value in seconds */
  43. #define DEFAULT_SYNC_TIMEOUT 60.0
  44. #define DEFAULT_KEYPAIR_CACHE_SIZE 512
  45. #define DEFAULT_MASTER_TIMEOUT 10.0
  46. #define DEFAULT_UPDATES_MAXFAIL 3
  47. #define COOKIE_SIZE 128
  48. #define DEFAULT_MAX_BUCKETS 2000
  49. #define DEFAULT_BUCKET_TTL 3600
  50. #define DEFAULT_BUCKET_MASK 24
  51. static const gchar *local_db_name = "local";
  52. #define msg_err_fuzzy_update(...) rspamd_default_log_function (G_LOG_LEVEL_CRITICAL, \
  53. session->name, session->uid, \
  54. G_STRFUNC, \
  55. __VA_ARGS__)
  56. #define msg_warn_fuzzy_update(...) rspamd_default_log_function (G_LOG_LEVEL_WARNING, \
  57. session->name, session->uid, \
  58. G_STRFUNC, \
  59. __VA_ARGS__)
  60. #define msg_info_fuzzy_update(...) rspamd_default_log_function (G_LOG_LEVEL_INFO, \
  61. session->name, session->uid, \
  62. G_STRFUNC, \
  63. __VA_ARGS__)
  64. #define msg_err_fuzzy_collection(...) rspamd_default_log_function (G_LOG_LEVEL_CRITICAL, \
  65. "fuzzy_collection", session->uid, \
  66. G_STRFUNC, \
  67. __VA_ARGS__)
  68. #define msg_warn_fuzzy_collection(...) rspamd_default_log_function (G_LOG_LEVEL_WARNING, \
  69. "fuzzy_collection", session->uid, \
  70. G_STRFUNC, \
  71. __VA_ARGS__)
  72. #define msg_info_fuzzy_collection(...) rspamd_default_log_function (G_LOG_LEVEL_INFO, \
  73. "fuzzy_collection", session->uid, \
  74. G_STRFUNC, \
  75. __VA_ARGS__)
  76. /* Init functions */
  77. gpointer init_fuzzy (struct rspamd_config *cfg);
  78. void start_fuzzy (struct rspamd_worker *worker);
  79. worker_t fuzzy_worker = {
  80. "fuzzy", /* Name */
  81. init_fuzzy, /* Init function */
  82. start_fuzzy, /* Start function */
  83. RSPAMD_WORKER_HAS_SOCKET,
  84. RSPAMD_WORKER_SOCKET_UDP|RSPAMD_WORKER_SOCKET_TCP, /* Both socket */
  85. RSPAMD_WORKER_VER /* Version info */
  86. };
  87. struct fuzzy_global_stat {
  88. guint64 fuzzy_hashes;
  89. /**< number of fuzzy hashes stored */
  90. guint64 fuzzy_hashes_expired;
  91. /**< number of fuzzy hashes expired */
  92. guint64 fuzzy_hashes_checked[RSPAMD_FUZZY_EPOCH_MAX];
  93. /**< amount of check requests for each epoch */
  94. guint64 fuzzy_shingles_checked[RSPAMD_FUZZY_EPOCH_MAX];
  95. /**< amount of shingle check requests for each epoch */
  96. guint64 fuzzy_hashes_found[RSPAMD_FUZZY_EPOCH_MAX];
  97. /**< amount of hashes found by epoch */
  98. guint64 invalid_requests;
  99. };
  100. struct fuzzy_key_stat {
  101. guint64 checked;
  102. guint64 matched;
  103. guint64 added;
  104. guint64 deleted;
  105. guint64 errors;
  106. rspamd_lru_hash_t *last_ips;
  107. };
  108. struct rspamd_fuzzy_mirror {
  109. gchar *name;
  110. struct upstream_list *u;
  111. struct rspamd_cryptobox_pubkey *key;
  112. };
  113. struct rspamd_leaky_bucket_elt {
  114. rspamd_inet_addr_t *addr;
  115. gdouble last;
  116. gdouble cur;
  117. };
  118. static const guint64 rspamd_fuzzy_storage_magic = 0x291a3253eb1b3ea5ULL;
  119. struct rspamd_fuzzy_storage_ctx {
  120. guint64 magic;
  121. /* Events base */
  122. struct event_base *ev_base;
  123. /* DNS resolver */
  124. struct rspamd_dns_resolver *resolver;
  125. /* Config */
  126. struct rspamd_config *cfg;
  127. /* END OF COMMON PART */
  128. struct fuzzy_global_stat stat;
  129. gdouble expire;
  130. gdouble sync_timeout;
  131. struct rspamd_radix_map_helper *update_ips;
  132. struct rspamd_radix_map_helper *master_ips;
  133. struct rspamd_radix_map_helper *blocked_ips;
  134. struct rspamd_radix_map_helper *ratelimit_whitelist;
  135. struct rspamd_cryptobox_keypair *sync_keypair;
  136. struct rspamd_cryptobox_pubkey *master_key;
  137. struct timeval master_io_tv;
  138. gdouble master_timeout;
  139. GPtrArray *mirrors;
  140. const ucl_object_t *update_map;
  141. const ucl_object_t *masters_map;
  142. const ucl_object_t *blocked_map;
  143. const ucl_object_t *ratelimit_whitelist_map;
  144. GHashTable *master_flags;
  145. guint keypair_cache_size;
  146. gint peer_fd;
  147. struct event peer_ev;
  148. struct event stat_ev;
  149. struct timeval stat_tv;
  150. /* Local keypair */
  151. struct rspamd_cryptobox_keypair *default_keypair; /* Bad clash, need for parse keypair */
  152. struct fuzzy_key *default_key;
  153. GHashTable *keys;
  154. gboolean encrypted_only;
  155. gboolean collection_mode;
  156. gboolean read_only;
  157. struct rspamd_cryptobox_keypair *collection_keypair;
  158. struct rspamd_cryptobox_pubkey *collection_sign_key;
  159. gchar *collection_id_file;
  160. struct rspamd_keypair_cache *keypair_cache;
  161. rspamd_lru_hash_t *errors_ips;
  162. rspamd_lru_hash_t *ratelimit_buckets;
  163. struct rspamd_fuzzy_backend *backend;
  164. GArray *updates_pending;
  165. guint updates_failed;
  166. guint updates_maxfail;
  167. guint32 collection_id;
  168. /* Ratelimits */
  169. guint leaky_bucket_ttl;
  170. guint leaky_bucket_mask;
  171. guint max_buckets;
  172. gboolean ratelimit_log_only;
  173. gdouble leaky_bucket_burst;
  174. gdouble leaky_bucket_rate;
  175. struct rspamd_worker *worker;
  176. struct rspamd_http_connection_router *collection_rt;
  177. const ucl_object_t *skip_map;
  178. struct rspamd_hash_map_helper *skip_hashes;
  179. guchar cookie[COOKIE_SIZE];
  180. };
  181. enum fuzzy_cmd_type {
  182. CMD_NORMAL,
  183. CMD_SHINGLE,
  184. CMD_ENCRYPTED_NORMAL,
  185. CMD_ENCRYPTED_SHINGLE
  186. };
  187. struct fuzzy_session {
  188. struct rspamd_worker *worker;
  189. rspamd_inet_addr_t *addr;
  190. struct rspamd_fuzzy_storage_ctx *ctx;
  191. union {
  192. struct rspamd_fuzzy_encrypted_shingle_cmd enc_shingle;
  193. struct rspamd_fuzzy_encrypted_cmd enc_normal;
  194. struct rspamd_fuzzy_cmd normal;
  195. struct rspamd_fuzzy_shingle_cmd shingle;
  196. } cmd;
  197. struct rspamd_fuzzy_encrypted_reply reply;
  198. struct fuzzy_key_stat *ip_stat;
  199. enum rspamd_fuzzy_epoch epoch;
  200. enum fuzzy_cmd_type cmd_type;
  201. gint fd;
  202. guint64 time;
  203. struct event io;
  204. ref_entry_t ref;
  205. struct fuzzy_key_stat *key_stat;
  206. guchar nm[rspamd_cryptobox_MAX_NMBYTES];
  207. };
  208. struct fuzzy_peer_request {
  209. struct event io_ev;
  210. struct fuzzy_peer_cmd cmd;
  211. };
  212. struct fuzzy_key {
  213. struct rspamd_cryptobox_keypair *key;
  214. struct rspamd_cryptobox_pubkey *pk;
  215. struct fuzzy_key_stat *stat;
  216. };
  217. struct fuzzy_master_update_session {
  218. const gchar *name;
  219. gchar uid[16];
  220. struct rspamd_http_connection *conn;
  221. struct rspamd_http_message *msg;
  222. struct rspamd_fuzzy_storage_ctx *ctx;
  223. const gchar *src;
  224. gchar *psrc;
  225. rspamd_inet_addr_t *addr;
  226. gboolean replied;
  227. gint sock;
  228. };
  229. static void rspamd_fuzzy_write_reply (struct fuzzy_session *session);
  230. static gboolean
  231. rspamd_fuzzy_check_ratelimit (struct fuzzy_session *session)
  232. {
  233. rspamd_inet_addr_t *masked;
  234. struct rspamd_leaky_bucket_elt *elt;
  235. struct timeval tv;
  236. gdouble now;
  237. if (session->ctx->ratelimit_whitelist != NULL) {
  238. if (rspamd_match_radix_map_addr (session->ctx->ratelimit_whitelist,
  239. session->addr) != NULL) {
  240. return TRUE;
  241. }
  242. }
  243. /*
  244. if (rspamd_inet_address_is_local (session->addr, TRUE)) {
  245. return TRUE;
  246. }
  247. */
  248. masked = rspamd_inet_address_copy (session->addr);
  249. if (rspamd_inet_address_get_af (masked) == AF_INET) {
  250. rspamd_inet_address_apply_mask (masked,
  251. MIN (session->ctx->leaky_bucket_mask, 32));
  252. }
  253. else {
  254. /* Must be at least /64 */
  255. rspamd_inet_address_apply_mask (masked,
  256. MIN (MAX (session->ctx->leaky_bucket_mask * 4, 64), 128));
  257. }
  258. #ifdef HAVE_EVENT_NO_CACHE_TIME_FUNC
  259. event_base_gettimeofday_cached (session->ctx->ev_base, &tv);
  260. #else
  261. gettimeofday (&tv, NULL);
  262. #endif
  263. now = tv_to_double (&tv);
  264. elt = rspamd_lru_hash_lookup (session->ctx->ratelimit_buckets, masked,
  265. tv.tv_sec);
  266. if (elt) {
  267. gboolean ratelimited = FALSE;
  268. if (isnan (elt->cur)) {
  269. /* Ratelimit exceeded, preserve it for the whole ttl */
  270. ratelimited = TRUE;
  271. }
  272. else {
  273. /* Update bucket */
  274. if (elt->last < now) {
  275. elt->cur -= session->ctx->leaky_bucket_rate * (now - elt->last);
  276. elt->last = now;
  277. if (elt->cur < 0) {
  278. elt->cur = 0;
  279. }
  280. }
  281. else {
  282. elt->last = now;
  283. }
  284. /* Check bucket */
  285. if (elt->cur >= session->ctx->leaky_bucket_burst) {
  286. msg_info ("ratelimiting %s (%s), %.1f max elts",
  287. rspamd_inet_address_to_string (session->addr),
  288. rspamd_inet_address_to_string (masked),
  289. session->ctx->leaky_bucket_burst);
  290. elt->cur = NAN;
  291. }
  292. else {
  293. elt->cur ++; /* Allow one more request */
  294. }
  295. }
  296. rspamd_inet_address_free (masked);
  297. return !ratelimited;
  298. }
  299. else {
  300. /* New bucket */
  301. elt = g_malloc (sizeof (*elt));
  302. elt->addr = masked; /* transfer ownership */
  303. elt->cur = 1;
  304. elt->last = now;
  305. rspamd_lru_hash_insert (session->ctx->ratelimit_buckets,
  306. masked,
  307. elt,
  308. tv.tv_sec,
  309. session->ctx->leaky_bucket_ttl);
  310. }
  311. return TRUE;
  312. }
  313. static gboolean
  314. rspamd_fuzzy_check_client (struct fuzzy_session *session, gboolean is_write)
  315. {
  316. if (session->ctx->blocked_ips != NULL) {
  317. if (rspamd_match_radix_map_addr (session->ctx->blocked_ips,
  318. session->addr) != NULL) {
  319. return FALSE;
  320. }
  321. }
  322. if (is_write) {
  323. if (session->ctx->read_only) {
  324. return FALSE;
  325. }
  326. if (session->ctx->update_ips != NULL) {
  327. if (rspamd_match_radix_map_addr (session->ctx->update_ips,
  328. session->addr) == NULL) {
  329. return FALSE;
  330. }
  331. else {
  332. return TRUE;
  333. }
  334. }
  335. return FALSE;
  336. }
  337. /* Non write */
  338. if (session->ctx->ratelimit_buckets) {
  339. if (session->ctx->ratelimit_log_only) {
  340. (void)rspamd_fuzzy_check_ratelimit (session); /* Check but ignore */
  341. }
  342. else {
  343. return rspamd_fuzzy_check_ratelimit (session);
  344. }
  345. }
  346. return TRUE;
  347. }
  348. static void
  349. fuzzy_key_stat_dtor (gpointer p)
  350. {
  351. struct fuzzy_key_stat *st = p;
  352. if (st->last_ips) {
  353. rspamd_lru_hash_destroy (st->last_ips);
  354. }
  355. }
  356. static void
  357. fuzzy_key_dtor (gpointer p)
  358. {
  359. struct fuzzy_key *key = p;
  360. if (key->stat) {
  361. fuzzy_key_stat_dtor (key->stat);
  362. }
  363. }
  364. static void
  365. fuzzy_count_callback (guint64 count, void *ud)
  366. {
  367. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  368. ctx->stat.fuzzy_hashes = count;
  369. }
  370. struct fuzzy_slave_connection {
  371. struct rspamd_cryptobox_keypair *local_key;
  372. struct rspamd_cryptobox_pubkey *remote_key;
  373. struct upstream *up;
  374. struct rspamd_http_connection *http_conn;
  375. struct rspamd_fuzzy_mirror *mirror;
  376. gint sock;
  377. };
  378. static void
  379. fuzzy_rl_bucket_free (gpointer p)
  380. {
  381. struct rspamd_leaky_bucket_elt *elt = (struct rspamd_leaky_bucket_elt *)p;
  382. rspamd_inet_address_free (elt->addr);
  383. g_free (elt);
  384. }
  385. static void
  386. fuzzy_mirror_close_connection (struct fuzzy_slave_connection *conn)
  387. {
  388. if (conn) {
  389. if (conn->http_conn) {
  390. rspamd_http_connection_reset (conn->http_conn);
  391. rspamd_http_connection_unref (conn->http_conn);
  392. }
  393. close (conn->sock);
  394. g_free (conn);
  395. }
  396. }
  397. struct rspamd_fuzzy_updates_cbdata {
  398. struct rspamd_fuzzy_storage_ctx *ctx;
  399. struct rspamd_http_message *msg;
  400. struct fuzzy_slave_connection *conn;
  401. struct rspamd_fuzzy_mirror *m;
  402. GArray *updates_pending;
  403. };
  404. static void
  405. fuzzy_mirror_updates_version_cb (guint64 rev64, void *ud)
  406. {
  407. struct rspamd_fuzzy_updates_cbdata *cbdata = ud;
  408. struct fuzzy_peer_cmd *io_cmd;
  409. guint32 rev32 = rev64, len;
  410. const gchar *p;
  411. rspamd_fstring_t *reply;
  412. struct fuzzy_slave_connection *conn;
  413. struct rspamd_fuzzy_storage_ctx *ctx;
  414. struct rspamd_http_message *msg;
  415. struct rspamd_fuzzy_mirror *m;
  416. struct timeval tv;
  417. guint i;
  418. conn = cbdata->conn;
  419. ctx = cbdata->ctx;
  420. msg = cbdata->msg;
  421. m = cbdata->m;
  422. rev32 = GUINT32_TO_LE (rev32);
  423. len = sizeof (guint32) * 2; /* revision + last chunk */
  424. for (i = 0; i < cbdata->updates_pending->len; i ++) {
  425. io_cmd = &g_array_index (cbdata->updates_pending,
  426. struct fuzzy_peer_cmd, i);
  427. if (io_cmd->is_shingle) {
  428. len += sizeof (guint32) + sizeof (guint32) +
  429. sizeof (struct rspamd_fuzzy_shingle_cmd);
  430. }
  431. else {
  432. len += sizeof (guint32) + sizeof (guint32) +
  433. sizeof (struct rspamd_fuzzy_cmd);
  434. }
  435. }
  436. reply = rspamd_fstring_sized_new (len);
  437. reply = rspamd_fstring_append (reply, (const char *)&rev32,
  438. sizeof (rev32));
  439. for (i = 0; i < cbdata->updates_pending->len; i ++) {
  440. io_cmd = &g_array_index (cbdata->updates_pending, struct fuzzy_peer_cmd, i);
  441. if (io_cmd->is_shingle) {
  442. len = sizeof (guint32) +
  443. sizeof (struct rspamd_fuzzy_shingle_cmd);
  444. }
  445. else {
  446. len = sizeof (guint32) +
  447. sizeof (struct rspamd_fuzzy_cmd);
  448. }
  449. p = (const char *)io_cmd;
  450. len = GUINT32_TO_LE (len);
  451. reply = rspamd_fstring_append (reply, (const char *)&len, sizeof (len));
  452. reply = rspamd_fstring_append (reply, p, len);
  453. }
  454. /* Last chunk */
  455. len = 0;
  456. reply = rspamd_fstring_append (reply, (const char *)&len, sizeof (len));
  457. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  458. double_to_tv (ctx->sync_timeout, &tv);
  459. rspamd_http_connection_write_message (conn->http_conn,
  460. msg, NULL, NULL, conn,
  461. conn->sock,
  462. &tv, ctx->ev_base);
  463. msg_info ("send update request to %s", m->name);
  464. g_array_free (cbdata->updates_pending, TRUE);
  465. g_free (cbdata);
  466. }
  467. static void
  468. fuzzy_mirror_updates_to_http (struct rspamd_fuzzy_mirror *m,
  469. struct fuzzy_slave_connection *conn,
  470. struct rspamd_fuzzy_storage_ctx *ctx,
  471. struct rspamd_http_message *msg,
  472. GArray *updates)
  473. {
  474. struct rspamd_fuzzy_updates_cbdata *cbdata;
  475. cbdata = g_malloc (sizeof (*cbdata));
  476. cbdata->ctx = ctx;
  477. cbdata->msg = msg;
  478. cbdata->conn = conn;
  479. cbdata->m = m;
  480. /* Copy queue */
  481. cbdata->updates_pending = g_array_sized_new (FALSE, FALSE,
  482. sizeof (struct fuzzy_peer_cmd), updates->len);
  483. g_array_append_vals (cbdata->updates_pending, updates->data, updates->len);
  484. rspamd_fuzzy_backend_version (ctx->backend, local_db_name,
  485. fuzzy_mirror_updates_version_cb, cbdata);
  486. }
  487. static void
  488. fuzzy_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  489. {
  490. struct fuzzy_slave_connection *bk_conn = conn->ud;
  491. msg_info ("abnormally closing connection from backend: %s:%s, "
  492. "error: %e",
  493. bk_conn->mirror->name,
  494. rspamd_inet_address_to_string (rspamd_upstream_addr (bk_conn->up)),
  495. err);
  496. fuzzy_mirror_close_connection (bk_conn);
  497. }
  498. static gint
  499. fuzzy_mirror_finish_handler (struct rspamd_http_connection *conn,
  500. struct rspamd_http_message *msg)
  501. {
  502. struct fuzzy_slave_connection *bk_conn = conn->ud;
  503. msg_info ("finished mirror connection to %s", bk_conn->mirror->name);
  504. fuzzy_mirror_close_connection (bk_conn);
  505. return 0;
  506. }
  507. static void
  508. rspamd_fuzzy_send_update_mirror (struct rspamd_fuzzy_storage_ctx *ctx,
  509. struct rspamd_fuzzy_mirror *m, GArray *updates)
  510. {
  511. struct fuzzy_slave_connection *conn;
  512. struct rspamd_http_message *msg;
  513. conn = g_malloc0 (sizeof (*conn));
  514. conn->up = rspamd_upstream_get (m->u,
  515. RSPAMD_UPSTREAM_MASTER_SLAVE, NULL, 0);
  516. conn->mirror = m;
  517. if (conn->up == NULL) {
  518. msg_err ("cannot select upstream for %s", m->name);
  519. return;
  520. }
  521. conn->sock = rspamd_inet_address_connect (
  522. rspamd_upstream_addr (conn->up),
  523. SOCK_STREAM, TRUE);
  524. if (conn->sock == -1) {
  525. msg_err ("cannot connect upstream for %s", m->name);
  526. rspamd_upstream_fail (conn->up, TRUE);
  527. return;
  528. }
  529. msg = rspamd_http_new_message (HTTP_REQUEST);
  530. rspamd_printf_fstring (&msg->url, "/update_v1/%s", m->name);
  531. conn->http_conn = rspamd_http_connection_new (NULL,
  532. fuzzy_mirror_error_handler,
  533. fuzzy_mirror_finish_handler,
  534. RSPAMD_HTTP_CLIENT_SIMPLE,
  535. RSPAMD_HTTP_CLIENT,
  536. ctx->keypair_cache,
  537. NULL);
  538. rspamd_http_connection_set_key (conn->http_conn,
  539. ctx->sync_keypair);
  540. msg->peer_key = rspamd_pubkey_ref (m->key);
  541. fuzzy_mirror_updates_to_http (m, conn, ctx, msg, updates);
  542. }
  543. struct rspamd_updates_cbdata {
  544. GArray *updates_pending;
  545. struct rspamd_fuzzy_storage_ctx *ctx;
  546. gchar *source;
  547. };
  548. static void
  549. fuzzy_update_version_callback (guint64 ver, void *ud)
  550. {
  551. msg_info ("updated fuzzy storage from %s: version: %d",
  552. (const char *)ud, (gint)ver);
  553. g_free (ud);
  554. }
  555. static void
  556. fuzzy_stat_count_callback (guint64 count, void *ud)
  557. {
  558. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  559. event_add (&ctx->stat_ev, &ctx->stat_tv);
  560. ctx->stat.fuzzy_hashes = count;
  561. }
  562. static void
  563. rspamd_fuzzy_stat_callback (gint fd, gshort what, gpointer ud)
  564. {
  565. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  566. event_del (&ctx->stat_ev);
  567. rspamd_fuzzy_backend_count (ctx->backend, fuzzy_stat_count_callback, ctx);
  568. }
  569. static void
  570. rspamd_fuzzy_updates_cb (gboolean success,
  571. guint nadded,
  572. guint ndeleted,
  573. guint nextended,
  574. guint nignored,
  575. void *ud)
  576. {
  577. struct rspamd_updates_cbdata *cbdata = ud;
  578. struct rspamd_fuzzy_mirror *m;
  579. guint i;
  580. struct rspamd_fuzzy_storage_ctx *ctx;
  581. const gchar *source;
  582. ctx = cbdata->ctx;
  583. source = cbdata->source;
  584. if (success) {
  585. rspamd_fuzzy_backend_count (ctx->backend, fuzzy_count_callback, ctx);
  586. if (ctx->updates_pending->len > 0) {
  587. for (i = 0; i < ctx->mirrors->len; i ++) {
  588. m = g_ptr_array_index (ctx->mirrors, i);
  589. rspamd_fuzzy_send_update_mirror (ctx, m,
  590. cbdata->updates_pending);
  591. }
  592. }
  593. msg_info ("successfully updated fuzzy storage: %d updates in queue; "
  594. "%d pending currently; "
  595. "%d added, %d deleted, %d extended, %d duplicates",
  596. cbdata->updates_pending->len,
  597. ctx->updates_pending->len,
  598. nadded, ndeleted, nextended, nignored);
  599. rspamd_fuzzy_backend_version (ctx->backend, source,
  600. fuzzy_update_version_callback, g_strdup (source));
  601. ctx->updates_failed = 0;
  602. }
  603. else {
  604. if (++ctx->updates_failed > ctx->updates_maxfail) {
  605. msg_err ("cannot commit update transaction to fuzzy backend, discard "
  606. "%ud updates after %d retries",
  607. cbdata->updates_pending->len,
  608. ctx->updates_maxfail);
  609. ctx->updates_failed = 0;
  610. }
  611. else {
  612. msg_err ("cannot commit update transaction to fuzzy backend, "
  613. "%ud updates are still left; %ud currently pending;"
  614. " %d updates left",
  615. cbdata->updates_pending->len,
  616. ctx->updates_pending->len,
  617. ctx->updates_maxfail - ctx->updates_failed);
  618. /* Move the remaining updates to ctx queue */
  619. g_array_append_vals (ctx->updates_pending,
  620. cbdata->updates_pending->data,
  621. cbdata->updates_pending->len);
  622. }
  623. }
  624. if (ctx->worker->wanna_die) {
  625. /* Plan exit */
  626. struct timeval tv;
  627. tv.tv_sec = 0;
  628. tv.tv_usec = 0;
  629. event_base_loopexit (ctx->ev_base, &tv);
  630. }
  631. g_array_free (cbdata->updates_pending, TRUE);
  632. g_free (cbdata->source);
  633. g_free (cbdata);
  634. }
  635. static void
  636. rspamd_fuzzy_process_updates_queue (struct rspamd_fuzzy_storage_ctx *ctx,
  637. const gchar *source, gboolean forced)
  638. {
  639. struct rspamd_updates_cbdata *cbdata;
  640. if ((forced ||ctx->updates_pending->len > 0)) {
  641. cbdata = g_malloc (sizeof (*cbdata));
  642. cbdata->ctx = ctx;
  643. cbdata->updates_pending = ctx->updates_pending;
  644. ctx->updates_pending = g_array_sized_new (FALSE, FALSE,
  645. sizeof (struct fuzzy_peer_cmd),
  646. MAX (cbdata->updates_pending->len, 1024));
  647. cbdata->source = g_strdup (source);
  648. rspamd_fuzzy_backend_process_updates (ctx->backend,
  649. cbdata->updates_pending,
  650. source, rspamd_fuzzy_updates_cb, cbdata);
  651. }
  652. }
  653. static void
  654. rspamd_fuzzy_reply_io (gint fd, gshort what, gpointer d)
  655. {
  656. struct fuzzy_session *session = d;
  657. rspamd_fuzzy_write_reply (session);
  658. REF_RELEASE (session);
  659. }
  660. static void
  661. rspamd_fuzzy_write_reply (struct fuzzy_session *session)
  662. {
  663. gssize r;
  664. gsize len;
  665. gconstpointer data;
  666. if (session->cmd_type == CMD_ENCRYPTED_NORMAL ||
  667. session->cmd_type == CMD_ENCRYPTED_SHINGLE) {
  668. /* Encrypted reply */
  669. data = &session->reply;
  670. if (session->epoch > RSPAMD_FUZZY_EPOCH10) {
  671. len = sizeof (session->reply);
  672. }
  673. else {
  674. len = sizeof (session->reply.hdr) + sizeof (session->reply.rep.v1);
  675. }
  676. }
  677. else {
  678. data = &session->reply.rep;
  679. if (session->epoch > RSPAMD_FUZZY_EPOCH10) {
  680. len = sizeof (session->reply.rep);
  681. }
  682. else {
  683. len = sizeof (session->reply.rep.v1);
  684. }
  685. }
  686. r = rspamd_inet_address_sendto (session->fd, data, len, 0,
  687. session->addr);
  688. if (r == -1) {
  689. if (errno == EINTR || errno == EWOULDBLOCK || errno == EAGAIN) {
  690. /* Grab reference to avoid early destruction */
  691. REF_RETAIN (session);
  692. event_set (&session->io, session->fd, EV_WRITE,
  693. rspamd_fuzzy_reply_io, session);
  694. event_base_set (session->ctx->ev_base, &session->io);
  695. event_add (&session->io, NULL);
  696. }
  697. else {
  698. msg_err ("error while writing reply: %s", strerror (errno));
  699. }
  700. }
  701. }
  702. static void
  703. fuzzy_peer_send_io (gint fd, gshort what, gpointer d)
  704. {
  705. struct fuzzy_peer_request *up_req = d;
  706. gssize r;
  707. r = write (fd, &up_req->cmd, sizeof (up_req->cmd));
  708. if (r != sizeof (up_req->cmd)) {
  709. msg_err ("cannot send update request to the peer: %s", strerror (errno));
  710. }
  711. event_del (&up_req->io_ev);
  712. g_free (up_req);
  713. }
  714. static void
  715. rspamd_fuzzy_update_stats (struct rspamd_fuzzy_storage_ctx *ctx,
  716. enum rspamd_fuzzy_epoch epoch,
  717. gboolean matched,
  718. gboolean is_shingle,
  719. struct fuzzy_key_stat *key_stat,
  720. struct fuzzy_key_stat *ip_stat,
  721. guint cmd, guint reply)
  722. {
  723. ctx->stat.fuzzy_hashes_checked[epoch] ++;
  724. if (matched) {
  725. ctx->stat.fuzzy_hashes_found[epoch]++;
  726. }
  727. if (is_shingle) {
  728. ctx->stat.fuzzy_shingles_checked[epoch]++;
  729. }
  730. if (key_stat) {
  731. if (!matched && reply != 0) {
  732. key_stat->errors ++;
  733. }
  734. else {
  735. if (cmd == FUZZY_CHECK) {
  736. key_stat->checked++;
  737. if (matched) {
  738. key_stat->matched ++;
  739. }
  740. }
  741. else if (cmd == FUZZY_WRITE) {
  742. key_stat->added++;
  743. }
  744. else if (cmd == FUZZY_DEL) {
  745. key_stat->deleted++;
  746. }
  747. }
  748. }
  749. if (ip_stat) {
  750. if (!matched && reply != 0) {
  751. ip_stat->errors++;
  752. }
  753. else {
  754. if (cmd == FUZZY_CHECK) {
  755. ip_stat->checked++;
  756. if (matched) {
  757. ip_stat->matched++;
  758. }
  759. }
  760. else if (cmd == FUZZY_WRITE) {
  761. ip_stat->added++;
  762. }
  763. else if (cmd == FUZZY_DEL) {
  764. ip_stat->deleted++;
  765. }
  766. }
  767. }
  768. }
  769. static void
  770. rspamd_fuzzy_make_reply (struct rspamd_fuzzy_cmd *cmd,
  771. struct rspamd_fuzzy_reply *result,
  772. struct fuzzy_session *session,
  773. gboolean encrypted, gboolean is_shingle)
  774. {
  775. gsize len;
  776. if (cmd) {
  777. result->v1.tag = cmd->tag;
  778. memcpy (&session->reply.rep, result, sizeof (*result));
  779. rspamd_fuzzy_update_stats (session->ctx,
  780. session->epoch,
  781. result->v1.prob > 0.5,
  782. is_shingle,
  783. session->key_stat,
  784. session->ip_stat,
  785. cmd->cmd,
  786. result->v1.value);
  787. if (encrypted) {
  788. /* We need also to encrypt reply */
  789. ottery_rand_bytes (session->reply.hdr.nonce,
  790. sizeof (session->reply.hdr.nonce));
  791. /*
  792. * For old replies we need to encrypt just old part, otherwise
  793. * decryption would fail due to mac verification mistake
  794. */
  795. if (session->epoch > RSPAMD_FUZZY_EPOCH10) {
  796. len = sizeof (session->reply.rep);
  797. }
  798. else {
  799. len = sizeof (session->reply.rep.v1);
  800. }
  801. rspamd_cryptobox_encrypt_nm_inplace ((guchar *)&session->reply.rep,
  802. len,
  803. session->reply.hdr.nonce,
  804. session->nm,
  805. session->reply.hdr.mac,
  806. RSPAMD_CRYPTOBOX_MODE_25519);
  807. }
  808. }
  809. rspamd_fuzzy_write_reply (session);
  810. }
  811. static void
  812. rspamd_fuzzy_check_callback (struct rspamd_fuzzy_reply *result, void *ud)
  813. {
  814. struct fuzzy_session *session = ud;
  815. gboolean encrypted = FALSE, is_shingle = FALSE;
  816. struct rspamd_fuzzy_cmd *cmd = NULL;
  817. const struct rspamd_shingle *shingle = NULL;
  818. struct rspamd_shingle sgl_cpy;
  819. switch (session->cmd_type) {
  820. case CMD_NORMAL:
  821. cmd = &session->cmd.normal;
  822. break;
  823. case CMD_SHINGLE:
  824. cmd = &session->cmd.shingle.basic;
  825. memcpy (&sgl_cpy, &session->cmd.shingle.sgl, sizeof (sgl_cpy));
  826. shingle = &sgl_cpy;
  827. is_shingle = TRUE;
  828. break;
  829. case CMD_ENCRYPTED_NORMAL:
  830. cmd = &session->cmd.enc_normal.cmd;
  831. encrypted = TRUE;
  832. break;
  833. case CMD_ENCRYPTED_SHINGLE:
  834. cmd = &session->cmd.enc_shingle.cmd.basic;
  835. memcpy (&sgl_cpy, &session->cmd.enc_shingle.cmd.sgl, sizeof (sgl_cpy));
  836. shingle = &sgl_cpy;
  837. encrypted = TRUE;
  838. is_shingle = TRUE;
  839. break;
  840. }
  841. rspamd_fuzzy_make_reply (cmd, result, session, encrypted, is_shingle);
  842. /* Refresh hash if found with strong confidence */
  843. if (result->v1.prob > 0.9 && !session->ctx->read_only) {
  844. struct fuzzy_peer_cmd up_cmd;
  845. struct fuzzy_peer_request *up_req;
  846. if (session->worker->index == 0 || session->ctx->peer_fd == -1) {
  847. /* Just add to the queue */
  848. memset (&up_cmd, 0, sizeof (up_cmd));
  849. up_cmd.is_shingle = is_shingle;
  850. memcpy (up_cmd.cmd.normal.digest, result->digest,
  851. sizeof (up_cmd.cmd.normal.digest));
  852. up_cmd.cmd.normal.flag = result->v1.flag;
  853. up_cmd.cmd.normal.cmd = FUZZY_REFRESH;
  854. up_cmd.cmd.normal.shingles_count = cmd->shingles_count;
  855. if (is_shingle && shingle) {
  856. memcpy (&up_cmd.cmd.shingle.sgl, shingle,
  857. sizeof (up_cmd.cmd.shingle.sgl));
  858. }
  859. g_array_append_val (session->ctx->updates_pending, up_cmd);
  860. }
  861. else {
  862. /* We need to send request to the peer */
  863. up_req = g_malloc0 (sizeof (*up_req));
  864. up_req->cmd.is_shingle = is_shingle;
  865. memcpy (up_req->cmd.cmd.normal.digest, result->digest,
  866. sizeof (up_req->cmd.cmd.normal.digest));
  867. up_req->cmd.cmd.normal.flag = result->v1.flag;
  868. up_req->cmd.cmd.normal.cmd = FUZZY_REFRESH;
  869. up_req->cmd.cmd.normal.shingles_count = cmd->shingles_count;
  870. if (is_shingle && shingle) {
  871. memcpy (&up_req->cmd.cmd.shingle.sgl, shingle,
  872. sizeof (up_req->cmd.cmd.shingle.sgl));
  873. }
  874. event_set (&up_req->io_ev, session->ctx->peer_fd, EV_WRITE,
  875. fuzzy_peer_send_io, up_req);
  876. event_base_set (session->ctx->ev_base, &up_req->io_ev);
  877. event_add (&up_req->io_ev, NULL);
  878. }
  879. }
  880. REF_RELEASE (session);
  881. }
  882. static void
  883. rspamd_fuzzy_process_command (struct fuzzy_session *session)
  884. {
  885. gboolean encrypted = FALSE, is_shingle = FALSE;
  886. struct rspamd_fuzzy_cmd *cmd = NULL;
  887. struct rspamd_fuzzy_reply result;
  888. struct fuzzy_peer_cmd up_cmd;
  889. struct fuzzy_peer_request *up_req;
  890. struct fuzzy_key_stat *ip_stat = NULL;
  891. gchar hexbuf[rspamd_cryptobox_HASHBYTES * 2 + 1];
  892. rspamd_inet_addr_t *naddr;
  893. gpointer ptr;
  894. gsize up_len = 0;
  895. switch (session->cmd_type) {
  896. case CMD_NORMAL:
  897. cmd = &session->cmd.normal;
  898. up_len = sizeof (session->cmd.normal);
  899. break;
  900. case CMD_SHINGLE:
  901. cmd = &session->cmd.shingle.basic;
  902. up_len = sizeof (session->cmd.shingle);
  903. is_shingle = TRUE;
  904. break;
  905. case CMD_ENCRYPTED_NORMAL:
  906. cmd = &session->cmd.enc_normal.cmd;
  907. up_len = sizeof (session->cmd.normal);
  908. encrypted = TRUE;
  909. break;
  910. case CMD_ENCRYPTED_SHINGLE:
  911. cmd = &session->cmd.enc_shingle.cmd.basic;
  912. up_len = sizeof (session->cmd.shingle);
  913. encrypted = TRUE;
  914. is_shingle = TRUE;
  915. break;
  916. default:
  917. msg_err ("invalid command type: %d", session->cmd_type);
  918. return;
  919. }
  920. memset (&result, 0, sizeof (result));
  921. memcpy (result.digest, cmd->digest, sizeof (result.digest));
  922. result.v1.flag = cmd->flag;
  923. result.v1.tag = cmd->tag;
  924. if (G_UNLIKELY (cmd == NULL || up_len == 0)) {
  925. result.v1.value = 500;
  926. result.v1.prob = 0.0;
  927. rspamd_fuzzy_make_reply (cmd, &result, session, encrypted, is_shingle);
  928. return;
  929. }
  930. if (session->ctx->encrypted_only && !encrypted) {
  931. /* Do not accept unencrypted commands */
  932. result.v1.value = 403;
  933. result.v1.prob = 0.0;
  934. rspamd_fuzzy_make_reply (cmd, &result, session, encrypted, is_shingle);
  935. return;
  936. }
  937. if (session->key_stat) {
  938. ip_stat = rspamd_lru_hash_lookup (session->key_stat->last_ips,
  939. session->addr, -1);
  940. if (ip_stat == NULL) {
  941. naddr = rspamd_inet_address_copy (session->addr);
  942. ip_stat = g_malloc0 (sizeof (*ip_stat));
  943. rspamd_lru_hash_insert (session->key_stat->last_ips,
  944. naddr, ip_stat, -1, 0);
  945. }
  946. session->ip_stat = ip_stat;
  947. }
  948. if (cmd->cmd == FUZZY_CHECK) {
  949. if (rspamd_fuzzy_check_client (session, FALSE)) {
  950. if (G_UNLIKELY (session->ctx->collection_mode)) {
  951. result.v1.prob = 0;
  952. result.v1.value = 500;
  953. result.v1.flag = 0;
  954. rspamd_fuzzy_make_reply (cmd, &result, session, encrypted,
  955. is_shingle);
  956. } else {
  957. REF_RETAIN (session);
  958. rspamd_fuzzy_backend_check (session->ctx->backend, cmd,
  959. rspamd_fuzzy_check_callback, session);
  960. }
  961. }
  962. else {
  963. result.v1.value = 403;
  964. result.v1.prob = 0.0;
  965. result.v1.flag = 0;
  966. rspamd_fuzzy_make_reply (cmd, &result, session, encrypted, is_shingle);
  967. }
  968. }
  969. else if (cmd->cmd == FUZZY_STAT) {
  970. if (G_UNLIKELY (session->ctx->collection_mode)) {
  971. result.v1.prob = 0;
  972. result.v1.value = 500;
  973. result.v1.flag = 0;
  974. rspamd_fuzzy_make_reply (cmd, &result, session, encrypted, is_shingle);
  975. }
  976. else {
  977. result.v1.prob = 1.0;
  978. result.v1.value = 0;
  979. result.v1.flag = session->ctx->stat.fuzzy_hashes;
  980. rspamd_fuzzy_make_reply (cmd, &result, session, encrypted, is_shingle);
  981. }
  982. }
  983. else {
  984. if (rspamd_fuzzy_check_client (session, TRUE)) {
  985. /* Check whitelist */
  986. if (session->ctx->skip_hashes && cmd->cmd == FUZZY_WRITE) {
  987. rspamd_encode_hex_buf (cmd->digest, sizeof (cmd->digest),
  988. hexbuf, sizeof (hexbuf) - 1);
  989. hexbuf[sizeof (hexbuf) - 1] = '\0';
  990. if (rspamd_match_hash_map (session->ctx->skip_hashes, hexbuf)) {
  991. result.v1.value = 401;
  992. result.v1.prob = 0.0;
  993. goto reply;
  994. }
  995. }
  996. if (session->worker->index == 0 || session->ctx->peer_fd == -1) {
  997. /* Just add to the queue */
  998. up_cmd.is_shingle = is_shingle;
  999. ptr = is_shingle ?
  1000. (gpointer)&up_cmd.cmd.shingle :
  1001. (gpointer)&up_cmd.cmd.normal;
  1002. memcpy (ptr, cmd, up_len);
  1003. g_array_append_val (session->ctx->updates_pending, up_cmd);
  1004. }
  1005. else {
  1006. /* We need to send request to the peer */
  1007. up_req = g_malloc0 (sizeof (*up_req));
  1008. up_req->cmd.is_shingle = is_shingle;
  1009. ptr = is_shingle ?
  1010. (gpointer)&up_req->cmd.cmd.shingle :
  1011. (gpointer)&up_req->cmd.cmd.normal;
  1012. memcpy (ptr, cmd, up_len);
  1013. event_set (&up_req->io_ev, session->ctx->peer_fd, EV_WRITE,
  1014. fuzzy_peer_send_io, up_req);
  1015. event_base_set (session->ctx->ev_base, &up_req->io_ev);
  1016. event_add (&up_req->io_ev, NULL);
  1017. }
  1018. result.v1.value = 0;
  1019. result.v1.prob = 1.0;
  1020. }
  1021. else {
  1022. result.v1.value = 403;
  1023. result.v1.prob = 0.0;
  1024. }
  1025. reply:
  1026. rspamd_fuzzy_make_reply (cmd, &result, session, encrypted, is_shingle);
  1027. }
  1028. }
  1029. static enum rspamd_fuzzy_epoch
  1030. rspamd_fuzzy_command_valid (struct rspamd_fuzzy_cmd *cmd, gint r)
  1031. {
  1032. enum rspamd_fuzzy_epoch ret = RSPAMD_FUZZY_EPOCH_MAX;
  1033. switch (cmd->version) {
  1034. case 4:
  1035. if (cmd->shingles_count > 0) {
  1036. if (r == sizeof (struct rspamd_fuzzy_shingle_cmd)) {
  1037. ret = RSPAMD_FUZZY_EPOCH11;
  1038. }
  1039. }
  1040. else {
  1041. if (r == sizeof (*cmd)) {
  1042. ret = RSPAMD_FUZZY_EPOCH11;
  1043. }
  1044. }
  1045. break;
  1046. case 3:
  1047. if (cmd->shingles_count > 0) {
  1048. if (r == sizeof (struct rspamd_fuzzy_shingle_cmd)) {
  1049. ret = RSPAMD_FUZZY_EPOCH10;
  1050. }
  1051. }
  1052. else {
  1053. if (r == sizeof (*cmd)) {
  1054. ret = RSPAMD_FUZZY_EPOCH10;
  1055. }
  1056. }
  1057. break;
  1058. case 2:
  1059. /*
  1060. * rspamd 0.8 has slightly different tokenizer then it might be not
  1061. * 100% compatible
  1062. */
  1063. if (cmd->shingles_count > 0) {
  1064. if (r == sizeof (struct rspamd_fuzzy_shingle_cmd)) {
  1065. ret = RSPAMD_FUZZY_EPOCH8;
  1066. }
  1067. }
  1068. else {
  1069. ret = RSPAMD_FUZZY_EPOCH8;
  1070. }
  1071. break;
  1072. default:
  1073. break;
  1074. }
  1075. return ret;
  1076. }
  1077. static gboolean
  1078. rspamd_fuzzy_decrypt_command (struct fuzzy_session *s)
  1079. {
  1080. struct rspamd_fuzzy_encrypted_req_hdr *hdr;
  1081. guchar *payload;
  1082. gsize payload_len;
  1083. struct rspamd_cryptobox_pubkey *rk;
  1084. struct fuzzy_key *key;
  1085. if (s->ctx->default_key == NULL) {
  1086. msg_warn ("received encrypted request when encryption is not enabled");
  1087. return FALSE;
  1088. }
  1089. if (s->cmd_type == CMD_ENCRYPTED_NORMAL) {
  1090. hdr = &s->cmd.enc_normal.hdr;
  1091. payload = (guchar *)&s->cmd.enc_normal.cmd;
  1092. payload_len = sizeof (s->cmd.enc_normal.cmd);
  1093. }
  1094. else {
  1095. hdr = &s->cmd.enc_shingle.hdr;
  1096. payload = (guchar *) &s->cmd.enc_shingle.cmd;
  1097. payload_len = sizeof (s->cmd.enc_shingle.cmd);
  1098. }
  1099. /* Compare magic */
  1100. if (memcmp (hdr->magic, fuzzy_encrypted_magic, sizeof (hdr->magic)) != 0) {
  1101. msg_debug ("invalid magic for the encrypted packet");
  1102. return FALSE;
  1103. }
  1104. /* Try to find the desired key */
  1105. key = g_hash_table_lookup (s->ctx->keys, hdr->key_id);
  1106. if (key == NULL) {
  1107. /* Unknown key, assume default one */
  1108. key = s->ctx->default_key;
  1109. }
  1110. s->key_stat = key->stat;
  1111. /* Now process keypair */
  1112. rk = rspamd_pubkey_from_bin (hdr->pubkey, sizeof (hdr->pubkey),
  1113. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  1114. if (rk == NULL) {
  1115. msg_err ("bad key");
  1116. return FALSE;
  1117. }
  1118. rspamd_keypair_cache_process (s->ctx->keypair_cache, key->key, rk);
  1119. /* Now decrypt request */
  1120. if (!rspamd_cryptobox_decrypt_nm_inplace (payload, payload_len, hdr->nonce,
  1121. rspamd_pubkey_get_nm (rk, key->key),
  1122. hdr->mac, RSPAMD_CRYPTOBOX_MODE_25519)) {
  1123. msg_err ("decryption failed");
  1124. rspamd_pubkey_unref (rk);
  1125. return FALSE;
  1126. }
  1127. memcpy (s->nm, rspamd_pubkey_get_nm (rk, key->key), sizeof (s->nm));
  1128. rspamd_pubkey_unref (rk);
  1129. return TRUE;
  1130. }
  1131. static gboolean
  1132. rspamd_fuzzy_cmd_from_wire (guchar *buf, guint buflen, struct fuzzy_session *s)
  1133. {
  1134. enum rspamd_fuzzy_epoch epoch;
  1135. /* For now, we assume that recvfrom returns a complete datagramm */
  1136. switch (buflen) {
  1137. case sizeof (struct rspamd_fuzzy_cmd):
  1138. s->cmd_type = CMD_NORMAL;
  1139. memcpy (&s->cmd.normal, buf, sizeof (s->cmd.normal));
  1140. epoch = rspamd_fuzzy_command_valid (&s->cmd.normal, buflen);
  1141. if (epoch == RSPAMD_FUZZY_EPOCH_MAX) {
  1142. msg_debug ("invalid fuzzy command of size %d received", buflen);
  1143. return FALSE;
  1144. }
  1145. s->epoch = epoch;
  1146. break;
  1147. case sizeof (struct rspamd_fuzzy_shingle_cmd):
  1148. s->cmd_type = CMD_SHINGLE;
  1149. memcpy (&s->cmd.shingle, buf, sizeof (s->cmd.shingle));
  1150. epoch = rspamd_fuzzy_command_valid (&s->cmd.shingle.basic, buflen);
  1151. if (epoch == RSPAMD_FUZZY_EPOCH_MAX) {
  1152. msg_debug ("invalid fuzzy command of size %d received", buflen);
  1153. return FALSE;
  1154. }
  1155. s->epoch = epoch;
  1156. break;
  1157. case sizeof (struct rspamd_fuzzy_encrypted_cmd):
  1158. s->cmd_type = CMD_ENCRYPTED_NORMAL;
  1159. memcpy (&s->cmd.enc_normal, buf, sizeof (s->cmd.enc_normal));
  1160. if (!rspamd_fuzzy_decrypt_command (s)) {
  1161. return FALSE;
  1162. }
  1163. epoch = rspamd_fuzzy_command_valid (&s->cmd.enc_normal.cmd,
  1164. sizeof (s->cmd.enc_normal.cmd));
  1165. if (epoch == RSPAMD_FUZZY_EPOCH_MAX) {
  1166. msg_debug ("invalid fuzzy command of size %d received", buflen);
  1167. return FALSE;
  1168. }
  1169. /* Encrypted is epoch 10 at least */
  1170. s->epoch = epoch;
  1171. break;
  1172. case sizeof (struct rspamd_fuzzy_encrypted_shingle_cmd):
  1173. s->cmd_type = CMD_ENCRYPTED_SHINGLE;
  1174. memcpy (&s->cmd.enc_shingle, buf, sizeof (s->cmd.enc_shingle));
  1175. if (!rspamd_fuzzy_decrypt_command (s)) {
  1176. return FALSE;
  1177. }
  1178. epoch = rspamd_fuzzy_command_valid (&s->cmd.enc_shingle.cmd.basic,
  1179. sizeof (s->cmd.enc_shingle.cmd));
  1180. if (epoch == RSPAMD_FUZZY_EPOCH_MAX) {
  1181. msg_debug ("invalid fuzzy command of size %d received", buflen);
  1182. return FALSE;
  1183. }
  1184. s->epoch = epoch;
  1185. break;
  1186. default:
  1187. msg_debug ("invalid fuzzy command of size %d received", buflen);
  1188. return FALSE;
  1189. }
  1190. return TRUE;
  1191. }
  1192. static void
  1193. rspamd_fuzzy_mirror_process_update (struct fuzzy_master_update_session *session,
  1194. struct rspamd_http_message *msg, guint our_rev)
  1195. {
  1196. const guchar *p;
  1197. gsize remain;
  1198. gint32 revision;
  1199. guint32 len = 0, cnt = 0;
  1200. struct fuzzy_peer_cmd cmd;
  1201. enum {
  1202. read_len = 0,
  1203. read_data,
  1204. finish_processing
  1205. } state = read_len;
  1206. gpointer flag_ptr;
  1207. /*
  1208. * Message format:
  1209. * <uint32_le> - revision
  1210. * <uint32_le> - size of the next element
  1211. * <data> - command data
  1212. * ...
  1213. * <0> - end of data
  1214. * ... - ignored
  1215. */
  1216. p = rspamd_http_message_get_body (msg, &remain);
  1217. if (p && remain >= sizeof (gint32) * 2) {
  1218. memcpy (&revision, p, sizeof (gint32));
  1219. revision = GINT32_TO_LE (revision);
  1220. if (revision <= our_rev) {
  1221. msg_err_fuzzy_update ("remote revision: %d is older than ours: %d, "
  1222. "refusing update",
  1223. revision, our_rev);
  1224. return;
  1225. }
  1226. else if (revision - our_rev > 1) {
  1227. msg_warn_fuzzy_update ("remote revision: %d is newer more than one revision "
  1228. "than ours: %d, cold sync is recommended",
  1229. revision, our_rev);
  1230. }
  1231. remain -= sizeof (gint32);
  1232. p += sizeof (gint32);
  1233. }
  1234. else {
  1235. msg_err_fuzzy_update ("short update message, not processing");
  1236. goto err;
  1237. }
  1238. while (remain > 0) {
  1239. switch (state) {
  1240. case read_len:
  1241. if (remain < sizeof (guint32)) {
  1242. msg_err_fuzzy_update ("short update message while reading "
  1243. "length, not processing");
  1244. goto err;
  1245. }
  1246. memcpy (&len, p, sizeof (guint32));
  1247. len = GUINT32_TO_LE (len);
  1248. remain -= sizeof (guint32);
  1249. p += sizeof (guint32);
  1250. if (len == 0) {
  1251. remain = 0;
  1252. state = finish_processing;
  1253. }
  1254. else {
  1255. state = read_data;
  1256. }
  1257. break;
  1258. case read_data:
  1259. if (remain < len) {
  1260. msg_err_fuzzy_update ("short update message while reading data, "
  1261. "not processing"
  1262. " (%zd is available, %d is required)", remain, len);
  1263. return;
  1264. }
  1265. if (len < sizeof (struct rspamd_fuzzy_cmd) + sizeof (guint32) ||
  1266. len > sizeof (cmd)) {
  1267. /* Bad size command */
  1268. msg_err_fuzzy_update ("incorrect element size: %d, at least "
  1269. "%d expected", len,
  1270. (gint)(sizeof (struct rspamd_fuzzy_cmd) + sizeof (guint32)));
  1271. goto err;
  1272. }
  1273. memcpy (&cmd, p, len);
  1274. if (cmd.is_shingle && len != sizeof (cmd)) {
  1275. /* Short command */
  1276. msg_err_fuzzy_update ("incorrect element size: %d, at least "
  1277. "%d expected", len,
  1278. (gint)(sizeof (cmd)));
  1279. goto err;
  1280. }
  1281. if (cmd.is_shingle) {
  1282. if ((flag_ptr = g_hash_table_lookup (session->ctx->master_flags,
  1283. GUINT_TO_POINTER (cmd.cmd.shingle.basic.flag))) != NULL) {
  1284. cmd.cmd.shingle.basic.flag = GPOINTER_TO_UINT (flag_ptr);
  1285. }
  1286. }
  1287. else {
  1288. if ((flag_ptr = g_hash_table_lookup (session->ctx->master_flags,
  1289. GUINT_TO_POINTER (cmd.cmd.normal.flag))) != NULL) {
  1290. cmd.cmd.normal.flag = GPOINTER_TO_UINT (flag_ptr);
  1291. }
  1292. }
  1293. g_array_append_val (session->ctx->updates_pending, cmd);
  1294. p += len;
  1295. remain -= len;
  1296. len = 0;
  1297. state = read_len;
  1298. cnt ++;
  1299. break;
  1300. case finish_processing:
  1301. /* Do nothing */
  1302. remain = 0;
  1303. break;
  1304. }
  1305. }
  1306. rspamd_fuzzy_process_updates_queue (session->ctx, session->src, TRUE);
  1307. msg_info_fuzzy_update ("processed updates from the master %s, "
  1308. "%ud operations processed,"
  1309. " revision: %d (local revision: %d)",
  1310. rspamd_inet_address_to_string (session->addr),
  1311. cnt, revision, our_rev);
  1312. err:
  1313. return;
  1314. }
  1315. static void
  1316. fuzzy_session_destroy (gpointer d)
  1317. {
  1318. struct fuzzy_session *session = d;
  1319. rspamd_inet_address_free (session->addr);
  1320. rspamd_explicit_memzero (session->nm, sizeof (session->nm));
  1321. session->worker->nconns--;
  1322. g_free (session);
  1323. }
  1324. static void
  1325. rspamd_fuzzy_mirror_session_destroy (struct fuzzy_master_update_session *session)
  1326. {
  1327. if (session) {
  1328. rspamd_http_connection_reset (session->conn);
  1329. rspamd_http_connection_unref (session->conn);
  1330. rspamd_inet_address_free (session->addr);
  1331. close (session->sock);
  1332. if (session->psrc) {
  1333. g_free (session->psrc);
  1334. }
  1335. g_free (session);
  1336. }
  1337. }
  1338. static void
  1339. rspamd_fuzzy_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  1340. {
  1341. struct fuzzy_master_update_session *session = conn->ud;
  1342. msg_err_fuzzy_update ("abnormally closing connection from: %s, error: %e",
  1343. rspamd_inet_address_to_string (session->addr), err);
  1344. /* Terminate session immediately */
  1345. rspamd_fuzzy_mirror_session_destroy (session);
  1346. }
  1347. static void
  1348. rspamd_fuzzy_mirror_send_reply (struct fuzzy_master_update_session *session,
  1349. guint code, const gchar *str)
  1350. {
  1351. struct rspamd_http_message *msg;
  1352. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1353. msg->url = rspamd_fstring_new_init (str, strlen (str));
  1354. msg->code = code;
  1355. session->replied = TRUE;
  1356. rspamd_http_connection_reset (session->conn);
  1357. rspamd_http_connection_write_message (session->conn, msg, NULL, "text/plain",
  1358. session, session->sock, &session->ctx->master_io_tv,
  1359. session->ctx->ev_base);
  1360. }
  1361. static void
  1362. rspamd_fuzzy_update_version_callback (guint64 version, void *ud)
  1363. {
  1364. struct fuzzy_master_update_session *session = ud;
  1365. rspamd_fuzzy_mirror_process_update (session, session->msg, version);
  1366. rspamd_fuzzy_mirror_send_reply (session, 200, "OK");
  1367. }
  1368. static gint
  1369. rspamd_fuzzy_mirror_finish_handler (struct rspamd_http_connection *conn,
  1370. struct rspamd_http_message *msg)
  1371. {
  1372. struct fuzzy_master_update_session *session = conn->ud;
  1373. const struct rspamd_cryptobox_pubkey *rk;
  1374. const gchar *err_str = NULL;
  1375. gchar *psrc;
  1376. const gchar *src = NULL;
  1377. gsize remain;
  1378. if (session->replied) {
  1379. rspamd_fuzzy_mirror_session_destroy (session);
  1380. return 0;
  1381. }
  1382. /* Check key */
  1383. if (!rspamd_http_connection_is_encrypted (conn)) {
  1384. msg_err_fuzzy_update ("refuse unencrypted update from: %s",
  1385. rspamd_inet_address_to_string (session->addr));
  1386. err_str = "Unencrypted update is not allowed";
  1387. goto end;
  1388. }
  1389. else {
  1390. if (session->ctx->master_key) {
  1391. rk = rspamd_http_connection_get_peer_key (conn);
  1392. g_assert (rk != NULL);
  1393. if (!rspamd_pubkey_equal (rk, session->ctx->master_key)) {
  1394. msg_err_fuzzy_update ("refuse unknown pubkey update from: %s",
  1395. rspamd_inet_address_to_string (session->addr));
  1396. err_str = "Unknown pubkey";
  1397. goto end;
  1398. }
  1399. }
  1400. else {
  1401. msg_warn_fuzzy_update ("no trusted key specified, accept any update from %s",
  1402. rspamd_inet_address_to_string (session->addr));
  1403. }
  1404. if (!rspamd_http_message_get_body (msg, NULL) || !msg->url
  1405. || msg->url->len == 0) {
  1406. msg_err_fuzzy_update ("empty update message, not processing");
  1407. err_str = "Empty update";
  1408. goto end;
  1409. }
  1410. /* Detect source from url: /update_v1/<source>, so we look for the last '/' */
  1411. remain = msg->url->len;
  1412. psrc = rspamd_fstringdup (msg->url);
  1413. src = psrc;
  1414. while (remain--) {
  1415. if (src[remain] == '/') {
  1416. src = &src[remain + 1];
  1417. break;
  1418. }
  1419. }
  1420. session->src = src;
  1421. session->psrc = psrc;
  1422. session->msg = msg;
  1423. rspamd_fuzzy_backend_version (session->ctx->backend, src,
  1424. rspamd_fuzzy_update_version_callback, session);
  1425. return 0;
  1426. }
  1427. end:
  1428. rspamd_fuzzy_mirror_send_reply (session, 403, err_str);
  1429. return 0;
  1430. }
  1431. struct rspamd_fuzzy_collection_session {
  1432. struct rspamd_fuzzy_storage_ctx *ctx;
  1433. struct rspamd_worker *worker;
  1434. rspamd_inet_addr_t *from_addr;
  1435. guchar uid[16];
  1436. };
  1437. static void
  1438. rspamd_fuzzy_collection_error_handler (struct rspamd_http_connection_entry *conn_ent,
  1439. GError *err)
  1440. {
  1441. struct rspamd_fuzzy_collection_session *session = conn_ent->ud;
  1442. msg_err_fuzzy_collection ("http error occurred: %s", err->message);
  1443. }
  1444. static void
  1445. rspamd_fuzzy_collection_finish_handler (struct rspamd_http_connection_entry *conn_ent)
  1446. {
  1447. struct rspamd_fuzzy_collection_session *session = conn_ent->ud;
  1448. rspamd_inet_address_free (session->from_addr);
  1449. g_free (session);
  1450. }
  1451. void
  1452. rspamd_fuzzy_collection_send_error (struct rspamd_http_connection_entry *entry,
  1453. gint code, const gchar *error_msg, ...)
  1454. {
  1455. struct rspamd_http_message *msg;
  1456. va_list args;
  1457. rspamd_fstring_t *reply;
  1458. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1459. va_start (args, error_msg);
  1460. msg->status = rspamd_fstring_new ();
  1461. rspamd_vprintf_fstring (&msg->status, error_msg, args);
  1462. va_end (args);
  1463. msg->date = time (NULL);
  1464. msg->code = code;
  1465. reply = rspamd_fstring_sized_new (msg->status->len + 16);
  1466. rspamd_printf_fstring (&reply, "%V", msg->status);
  1467. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1468. rspamd_http_connection_reset (entry->conn);
  1469. rspamd_http_router_insert_headers (entry->rt, msg);
  1470. rspamd_http_connection_write_message (entry->conn,
  1471. msg,
  1472. NULL,
  1473. "text/plain",
  1474. entry,
  1475. entry->conn->fd,
  1476. entry->rt->ptv,
  1477. entry->rt->ev_base);
  1478. entry->is_reply = TRUE;
  1479. }
  1480. /*
  1481. * Note: this function steals fstring
  1482. */
  1483. void
  1484. rspamd_fuzzy_collection_send_fstring (struct rspamd_http_connection_entry *entry,
  1485. rspamd_fstring_t *fstr)
  1486. {
  1487. struct rspamd_http_message *msg;
  1488. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1489. msg->status = rspamd_fstring_new_init ("OK", 2);
  1490. msg->date = time (NULL);
  1491. msg->code = 200;
  1492. rspamd_http_message_set_body_from_fstring_steal (msg, fstr);
  1493. rspamd_http_connection_reset (entry->conn);
  1494. rspamd_http_router_insert_headers (entry->rt, msg);
  1495. rspamd_http_connection_write_message (entry->conn,
  1496. msg,
  1497. NULL,
  1498. "application/octet-stream",
  1499. entry,
  1500. entry->conn->fd,
  1501. entry->rt->ptv,
  1502. entry->rt->ev_base);
  1503. entry->is_reply = TRUE;
  1504. }
  1505. static int
  1506. rspamd_fuzzy_collection_cookie (struct rspamd_http_connection_entry *conn_ent,
  1507. struct rspamd_http_message *msg)
  1508. {
  1509. struct rspamd_fuzzy_collection_session *session = conn_ent->ud;
  1510. rspamd_fstring_t *cookie;
  1511. cookie = rspamd_fstring_new_init (session->ctx->cookie,
  1512. sizeof (session->ctx->cookie));
  1513. rspamd_fuzzy_collection_send_fstring (conn_ent, cookie);
  1514. return 0;
  1515. }
  1516. static int
  1517. rspamd_fuzzy_collection_data (struct rspamd_http_connection_entry *conn_ent,
  1518. struct rspamd_http_message *msg)
  1519. {
  1520. struct rspamd_fuzzy_collection_session *session = conn_ent->ud;
  1521. const rspamd_ftok_t *sign_header;
  1522. struct rspamd_fuzzy_storage_ctx *ctx;
  1523. guint i;
  1524. struct fuzzy_peer_cmd *io_cmd;
  1525. rspamd_fstring_t *reply;
  1526. GError *err = NULL;
  1527. guchar *decoded_signature;
  1528. gsize dec_len;
  1529. guint32 cmdlen, nupdates = 0;
  1530. sign_header = rspamd_http_message_find_header (msg, "Signature");
  1531. if (sign_header == NULL) {
  1532. rspamd_fuzzy_collection_send_error (conn_ent, 403, "Missing signature");
  1533. return 0;
  1534. }
  1535. ctx = session->ctx;
  1536. if (ctx->collection_sign_key == NULL) {
  1537. rspamd_fuzzy_collection_send_error (conn_ent, 500, "Misconfigured signature key");
  1538. return 0;
  1539. }
  1540. decoded_signature = g_malloc (sign_header->len * 2 + 1);
  1541. dec_len = rspamd_decode_hex_buf (sign_header->begin, sign_header->len,
  1542. decoded_signature, sign_header->len * 2 + 1);
  1543. if (dec_len == -1 || !rspamd_keypair_verify (ctx->collection_sign_key,
  1544. ctx->cookie, sizeof (ctx->cookie),
  1545. decoded_signature, dec_len, &err)) {
  1546. if (err) {
  1547. rspamd_fuzzy_collection_send_error (conn_ent, 403, "Signature verification error: %e",
  1548. err);
  1549. g_error_free (err);
  1550. }
  1551. else {
  1552. rspamd_fuzzy_collection_send_error (conn_ent, 403, "Signature verification error");
  1553. }
  1554. g_free (decoded_signature);
  1555. return 0;
  1556. }
  1557. g_free (decoded_signature);
  1558. /* Generate new cookie */
  1559. ottery_rand_bytes (ctx->cookie, sizeof (ctx->cookie));
  1560. /* Send&Clear updates */
  1561. reply = rspamd_fstring_sized_new (8192);
  1562. /*
  1563. * Message format:
  1564. * <uint32_le> - revision
  1565. * <uint32_le> - size of the next element
  1566. * <data> - command data
  1567. * ...
  1568. * <0> - end of data
  1569. * ... - ignored
  1570. */
  1571. reply = rspamd_fstring_append (reply, (const gchar *)&ctx->collection_id,
  1572. sizeof (ctx->collection_id));
  1573. for (i = 0; i < ctx->updates_pending->len; i ++) {
  1574. io_cmd = &g_array_index (ctx->updates_pending, struct fuzzy_peer_cmd, i);
  1575. if (io_cmd->is_shingle) {
  1576. cmdlen = sizeof (io_cmd->cmd.shingle) + sizeof (guint32);
  1577. }
  1578. else {
  1579. cmdlen = sizeof (io_cmd->cmd.normal) + sizeof (guint32);
  1580. }
  1581. cmdlen = GUINT32_TO_LE (cmdlen);
  1582. reply = rspamd_fstring_append (reply, (const gchar *)&cmdlen,
  1583. sizeof (cmdlen));
  1584. reply = rspamd_fstring_append (reply, (const gchar *)io_cmd,
  1585. cmdlen);
  1586. nupdates ++;
  1587. }
  1588. msg_info_fuzzy_collection ("collection %d done, send %d updates",
  1589. ctx->collection_id, nupdates);
  1590. /* Last command */
  1591. cmdlen = 0;
  1592. reply = rspamd_fstring_append (reply, (const gchar *)&cmdlen,
  1593. sizeof (cmdlen));
  1594. ctx->updates_pending->len = 0;
  1595. /* Clear failed attempts counter */
  1596. ctx->updates_failed = 0;
  1597. ctx->collection_id ++;
  1598. rspamd_fuzzy_collection_send_fstring (conn_ent, reply);
  1599. return 0;
  1600. }
  1601. static void
  1602. accept_fuzzy_collection_socket (gint fd, short what, void *arg)
  1603. {
  1604. struct rspamd_worker *worker = (struct rspamd_worker *)arg;
  1605. rspamd_inet_addr_t *addr;
  1606. gint nfd;
  1607. struct rspamd_fuzzy_storage_ctx *ctx;
  1608. struct rspamd_fuzzy_collection_session *session;
  1609. if ((nfd =
  1610. rspamd_accept_from_socket (fd, &addr, worker->accept_events)) == -1) {
  1611. msg_warn ("accept failed: %s", strerror (errno));
  1612. return;
  1613. }
  1614. /* Check for EAGAIN */
  1615. if (nfd == 0) {
  1616. return;
  1617. }
  1618. ctx = worker->ctx;
  1619. if (!ctx->collection_keypair) {
  1620. msg_err ("deny request from %s, as no local keypair is specified",
  1621. rspamd_inet_address_to_string (addr));
  1622. rspamd_inet_address_free (addr);
  1623. close (nfd);
  1624. return;
  1625. }
  1626. session = g_malloc0 (sizeof (*session));
  1627. session->ctx = ctx;
  1628. session->worker = worker;
  1629. rspamd_random_hex (session->uid, sizeof (session->uid) - 1);
  1630. session->uid[sizeof (session->uid) - 1] = '\0';
  1631. session->from_addr = addr;
  1632. rspamd_http_router_handle_socket (ctx->collection_rt, nfd, session);
  1633. msg_info_fuzzy_collection ("accepted connection from %s port %d, session ptr: %p",
  1634. rspamd_inet_address_to_string (addr),
  1635. rspamd_inet_address_get_port (addr),
  1636. session);
  1637. }
  1638. static void
  1639. rspamd_fuzzy_collection_periodic (gint fd, gshort what, gpointer ud)
  1640. {
  1641. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1642. if (++ctx->updates_failed > ctx->updates_maxfail) {
  1643. msg_err ("cannot store more data in workqueue, discard "
  1644. "%ud updates after %d missed collection points",
  1645. ctx->updates_pending->len,
  1646. ctx->updates_maxfail);
  1647. ctx->updates_failed = 0;
  1648. ctx->updates_pending->len = 0;
  1649. /* Regenerate cookie */
  1650. ottery_rand_bytes (ctx->cookie, sizeof (ctx->cookie));
  1651. }
  1652. else {
  1653. msg_err ("fuzzy data has not been collected in time, "
  1654. "%ud updates are still pending, %d updates left",
  1655. ctx->updates_pending->len,
  1656. ctx->updates_maxfail - ctx->updates_failed);
  1657. }
  1658. if (ctx->worker->wanna_die) {
  1659. /* Plan exit */
  1660. struct timeval tv;
  1661. tv.tv_sec = 0;
  1662. tv.tv_usec = 0;
  1663. event_base_loopexit (ctx->ev_base, &tv);
  1664. }
  1665. }
  1666. static void
  1667. accept_fuzzy_mirror_socket (gint fd, short what, void *arg)
  1668. {
  1669. struct rspamd_worker *worker = (struct rspamd_worker *)arg;
  1670. rspamd_inet_addr_t *addr;
  1671. gint nfd;
  1672. struct rspamd_http_connection *http_conn;
  1673. struct rspamd_fuzzy_storage_ctx *ctx;
  1674. struct fuzzy_master_update_session *session;
  1675. if ((nfd =
  1676. rspamd_accept_from_socket (fd, &addr, worker->accept_events)) == -1) {
  1677. msg_warn ("accept failed: %s", strerror (errno));
  1678. return;
  1679. }
  1680. /* Check for EAGAIN */
  1681. if (nfd == 0) {
  1682. return;
  1683. }
  1684. ctx = worker->ctx;
  1685. if (!ctx->master_ips) {
  1686. msg_err ("deny update request from %s as no masters defined",
  1687. rspamd_inet_address_to_string (addr));
  1688. rspamd_inet_address_free (addr);
  1689. close (nfd);
  1690. return;
  1691. }
  1692. else if (rspamd_match_radix_map_addr (ctx->master_ips, addr) == NULL) {
  1693. msg_err ("deny update request from %s",
  1694. rspamd_inet_address_to_string (addr));
  1695. rspamd_inet_address_free (addr);
  1696. close (nfd);
  1697. return;
  1698. }
  1699. if (!ctx->sync_keypair) {
  1700. msg_err ("deny update request from %s, as no local keypair is specified",
  1701. rspamd_inet_address_to_string (addr));
  1702. rspamd_inet_address_free (addr);
  1703. close (nfd);
  1704. return;
  1705. }
  1706. session = g_malloc0 (sizeof (*session));
  1707. session->name = rspamd_inet_address_to_string (addr);
  1708. rspamd_random_hex (session->uid, sizeof (session->uid) - 1);
  1709. session->uid[sizeof (session->uid) - 1] = '\0';
  1710. http_conn = rspamd_http_connection_new (NULL,
  1711. rspamd_fuzzy_mirror_error_handler,
  1712. rspamd_fuzzy_mirror_finish_handler,
  1713. 0,
  1714. RSPAMD_HTTP_SERVER,
  1715. ctx->keypair_cache,
  1716. NULL);
  1717. rspamd_http_connection_set_key (http_conn, ctx->sync_keypair);
  1718. session->ctx = ctx;
  1719. session->conn = http_conn;
  1720. session->addr = addr;
  1721. session->sock = nfd;
  1722. rspamd_http_connection_read_message (http_conn,
  1723. session,
  1724. nfd,
  1725. &ctx->master_io_tv,
  1726. ctx->ev_base);
  1727. }
  1728. /*
  1729. * Accept new connection and construct task
  1730. */
  1731. static void
  1732. accept_fuzzy_socket (gint fd, short what, void *arg)
  1733. {
  1734. struct rspamd_worker *worker = (struct rspamd_worker *)arg;
  1735. struct fuzzy_session *session;
  1736. rspamd_inet_addr_t *addr;
  1737. gssize r;
  1738. guint8 buf[512];
  1739. guint64 *nerrors;
  1740. /* Got some data */
  1741. if (what == EV_READ) {
  1742. for (;;) {
  1743. worker->nconns++;
  1744. r = rspamd_inet_address_recvfrom (fd,
  1745. buf,
  1746. sizeof (buf),
  1747. 0,
  1748. &addr);
  1749. if (r == -1) {
  1750. if (errno == EINTR) {
  1751. continue;
  1752. }
  1753. else if (errno == EAGAIN || errno == EWOULDBLOCK) {
  1754. return;
  1755. }
  1756. msg_err ("got error while reading from socket: %d, %s",
  1757. errno,
  1758. strerror (errno));
  1759. return;
  1760. }
  1761. session = g_malloc0 (sizeof (*session));
  1762. REF_INIT_RETAIN (session, fuzzy_session_destroy);
  1763. session->worker = worker;
  1764. session->fd = fd;
  1765. session->ctx = worker->ctx;
  1766. session->time = (guint64) time (NULL);
  1767. session->addr = addr;
  1768. if (rspamd_fuzzy_cmd_from_wire (buf, r, session)) {
  1769. /* Check shingles count sanity */
  1770. rspamd_fuzzy_process_command (session);
  1771. }
  1772. else {
  1773. /* Discard input */
  1774. session->ctx->stat.invalid_requests ++;
  1775. msg_debug ("invalid fuzzy command of size %z received", r);
  1776. nerrors = rspamd_lru_hash_lookup (session->ctx->errors_ips,
  1777. addr, -1);
  1778. if (nerrors == NULL) {
  1779. nerrors = g_malloc (sizeof (*nerrors));
  1780. *nerrors = 1;
  1781. rspamd_lru_hash_insert (session->ctx->errors_ips,
  1782. rspamd_inet_address_copy (addr),
  1783. nerrors, -1, -1);
  1784. }
  1785. else {
  1786. *nerrors = *nerrors + 1;
  1787. }
  1788. }
  1789. REF_RELEASE (session);
  1790. }
  1791. }
  1792. }
  1793. static gboolean
  1794. rspamd_fuzzy_storage_periodic_callback (void *ud)
  1795. {
  1796. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1797. if (ctx->updates_pending->len > 0) {
  1798. rspamd_fuzzy_process_updates_queue (ctx, local_db_name, FALSE);
  1799. return TRUE;
  1800. }
  1801. return FALSE;
  1802. }
  1803. static gboolean
  1804. rspamd_fuzzy_storage_sync (struct rspamd_main *rspamd_main,
  1805. struct rspamd_worker *worker, gint fd,
  1806. gint attached_fd,
  1807. struct rspamd_control_command *cmd,
  1808. gpointer ud)
  1809. {
  1810. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1811. struct rspamd_control_reply rep;
  1812. rep.reply.fuzzy_sync.status = 0;
  1813. if (ctx->backend && worker->index == 0) {
  1814. rspamd_fuzzy_process_updates_queue (ctx, local_db_name, FALSE);
  1815. rspamd_fuzzy_backend_start_update (ctx->backend, ctx->sync_timeout,
  1816. rspamd_fuzzy_storage_periodic_callback, ctx);
  1817. }
  1818. if (write (fd, &rep, sizeof (rep)) != sizeof (rep)) {
  1819. msg_err ("cannot write reply to the control socket: %s",
  1820. strerror (errno));
  1821. }
  1822. return TRUE;
  1823. }
  1824. static gboolean
  1825. rspamd_fuzzy_storage_reload (struct rspamd_main *rspamd_main,
  1826. struct rspamd_worker *worker, gint fd,
  1827. gint attached_fd,
  1828. struct rspamd_control_command *cmd,
  1829. gpointer ud)
  1830. {
  1831. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1832. GError *err = NULL;
  1833. struct rspamd_control_reply rep;
  1834. msg_info ("reloading fuzzy storage after receiving reload command");
  1835. if (ctx->backend) {
  1836. /* Close backend and reopen it one more time */
  1837. rspamd_fuzzy_backend_close (ctx->backend);
  1838. }
  1839. memset (&rep, 0, sizeof (rep));
  1840. rep.type = RSPAMD_CONTROL_RELOAD;
  1841. if ((ctx->backend = rspamd_fuzzy_backend_create (ctx->ev_base,
  1842. worker->cf->options, rspamd_main->cfg,
  1843. &err)) == NULL) {
  1844. msg_err ("cannot open backend after reload: %e", err);
  1845. g_error_free (err);
  1846. rep.reply.reload.status = err->code;
  1847. }
  1848. else {
  1849. rep.reply.reload.status = 0;
  1850. }
  1851. if (ctx->backend && worker->index == 0) {
  1852. rspamd_fuzzy_backend_start_update (ctx->backend, ctx->sync_timeout,
  1853. rspamd_fuzzy_storage_periodic_callback, ctx);
  1854. }
  1855. if (write (fd, &rep, sizeof (rep)) != sizeof (rep)) {
  1856. msg_err ("cannot write reply to the control socket: %s",
  1857. strerror (errno));
  1858. }
  1859. return TRUE;
  1860. }
  1861. static ucl_object_t *
  1862. rspamd_fuzzy_storage_stat_key (struct fuzzy_key_stat *key_stat)
  1863. {
  1864. ucl_object_t *res;
  1865. res = ucl_object_typed_new (UCL_OBJECT);
  1866. ucl_object_insert_key (res, ucl_object_fromint (key_stat->checked),
  1867. "checked", 0, false);
  1868. ucl_object_insert_key (res, ucl_object_fromint (key_stat->matched),
  1869. "matched", 0, false);
  1870. ucl_object_insert_key (res, ucl_object_fromint (key_stat->added),
  1871. "added", 0, false);
  1872. ucl_object_insert_key (res, ucl_object_fromint (key_stat->deleted),
  1873. "deleted", 0, false);
  1874. ucl_object_insert_key (res, ucl_object_fromint (key_stat->errors),
  1875. "errors", 0, false);
  1876. return res;
  1877. }
  1878. static ucl_object_t *
  1879. rspamd_fuzzy_stat_to_ucl (struct rspamd_fuzzy_storage_ctx *ctx, gboolean ip_stat)
  1880. {
  1881. struct fuzzy_key_stat *key_stat;
  1882. GHashTableIter it;
  1883. struct fuzzy_key *key;
  1884. ucl_object_t *obj, *keys_obj, *elt, *ip_elt, *ip_cur;
  1885. gpointer k, v;
  1886. gint i;
  1887. gchar keyname[17];
  1888. obj = ucl_object_typed_new (UCL_OBJECT);
  1889. keys_obj = ucl_object_typed_new (UCL_OBJECT);
  1890. g_hash_table_iter_init (&it, ctx->keys);
  1891. while (g_hash_table_iter_next (&it, &k, &v)) {
  1892. key = v;
  1893. key_stat = key->stat;
  1894. if (key_stat) {
  1895. rspamd_snprintf (keyname, sizeof (keyname), "%8bs", k);
  1896. elt = rspamd_fuzzy_storage_stat_key (key_stat);
  1897. if (key_stat->last_ips && ip_stat) {
  1898. i = 0;
  1899. ip_elt = ucl_object_typed_new (UCL_OBJECT);
  1900. while ((i = rspamd_lru_hash_foreach (key_stat->last_ips,
  1901. i, &k, &v)) != -1) {
  1902. ip_cur = rspamd_fuzzy_storage_stat_key (v);
  1903. ucl_object_insert_key (ip_elt, ip_cur,
  1904. rspamd_inet_address_to_string (k), 0, true);
  1905. }
  1906. ucl_object_insert_key (elt, ip_elt, "ips", 0, false);
  1907. }
  1908. ucl_object_insert_key (keys_obj, elt, keyname, 0, true);
  1909. }
  1910. }
  1911. ucl_object_insert_key (obj, keys_obj, "keys", 0, false);
  1912. /* Now generic stats */
  1913. ucl_object_insert_key (obj,
  1914. ucl_object_fromint (ctx->stat.fuzzy_hashes),
  1915. "fuzzy_stored",
  1916. 0,
  1917. false);
  1918. ucl_object_insert_key (obj,
  1919. ucl_object_fromint (ctx->stat.fuzzy_hashes_expired),
  1920. "fuzzy_expired",
  1921. 0,
  1922. false);
  1923. ucl_object_insert_key (obj,
  1924. ucl_object_fromint (ctx->stat.invalid_requests),
  1925. "invalid_requests",
  1926. 0,
  1927. false);
  1928. if (ctx->errors_ips && ip_stat) {
  1929. i = 0;
  1930. ip_elt = ucl_object_typed_new (UCL_OBJECT);
  1931. while ((i = rspamd_lru_hash_foreach (ctx->errors_ips, i, &k, &v)) != -1) {
  1932. ucl_object_insert_key (ip_elt,
  1933. ucl_object_fromint (*(guint64 *)v),
  1934. rspamd_inet_address_to_string (k), 0, true);
  1935. }
  1936. ucl_object_insert_key (obj,
  1937. ip_elt,
  1938. "errors_ips",
  1939. 0,
  1940. false);
  1941. }
  1942. /* Checked by epoch */
  1943. elt = ucl_object_typed_new (UCL_ARRAY);
  1944. for (i = RSPAMD_FUZZY_EPOCH6; i < RSPAMD_FUZZY_EPOCH_MAX; i++) {
  1945. ucl_array_append (elt,
  1946. ucl_object_fromint (ctx->stat.fuzzy_hashes_checked[i]));
  1947. }
  1948. ucl_object_insert_key (obj, elt, "fuzzy_checked", 0, false);
  1949. /* Shingles by epoch */
  1950. elt = ucl_object_typed_new (UCL_ARRAY);
  1951. for (i = RSPAMD_FUZZY_EPOCH6; i < RSPAMD_FUZZY_EPOCH_MAX; i++) {
  1952. ucl_array_append (elt,
  1953. ucl_object_fromint (ctx->stat.fuzzy_shingles_checked[i]));
  1954. }
  1955. ucl_object_insert_key (obj, elt, "fuzzy_shingles", 0, false);
  1956. /* Matched by epoch */
  1957. elt = ucl_object_typed_new (UCL_ARRAY);
  1958. for (i = RSPAMD_FUZZY_EPOCH6; i < RSPAMD_FUZZY_EPOCH_MAX; i++) {
  1959. ucl_array_append (elt,
  1960. ucl_object_fromint (ctx->stat.fuzzy_hashes_found[i]));
  1961. }
  1962. ucl_object_insert_key (obj, elt, "fuzzy_found", 0, false);
  1963. return obj;
  1964. }
  1965. static gboolean
  1966. rspamd_fuzzy_storage_stat (struct rspamd_main *rspamd_main,
  1967. struct rspamd_worker *worker, gint fd,
  1968. gint attached_fd,
  1969. struct rspamd_control_command *cmd,
  1970. gpointer ud)
  1971. {
  1972. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1973. struct rspamd_control_reply rep;
  1974. ucl_object_t *obj;
  1975. struct ucl_emitter_functions *emit_subr;
  1976. guchar fdspace[CMSG_SPACE(sizeof (int))];
  1977. struct iovec iov;
  1978. struct msghdr msg;
  1979. struct cmsghdr *cmsg;
  1980. gint outfd = -1;
  1981. gchar tmppath[PATH_MAX];
  1982. memset (&rep, 0, sizeof (rep));
  1983. rep.type = RSPAMD_CONTROL_FUZZY_STAT;
  1984. rspamd_snprintf (tmppath, sizeof (tmppath), "%s%c%s-XXXXXXXXXX",
  1985. rspamd_main->cfg->temp_dir, G_DIR_SEPARATOR, "fuzzy-stat");
  1986. if ((outfd = mkstemp (tmppath)) == -1) {
  1987. rep.reply.fuzzy_stat.status = errno;
  1988. msg_info_main ("cannot make temporary stat file for fuzzy stat: %s",
  1989. strerror (errno));
  1990. }
  1991. else {
  1992. rep.reply.fuzzy_stat.status = 0;
  1993. memcpy (rep.reply.fuzzy_stat.storage_id,
  1994. rspamd_fuzzy_backend_id (ctx->backend),
  1995. sizeof (rep.reply.fuzzy_stat.storage_id));
  1996. obj = rspamd_fuzzy_stat_to_ucl (ctx, TRUE);
  1997. emit_subr = ucl_object_emit_fd_funcs (outfd);
  1998. ucl_object_emit_full (obj, UCL_EMIT_JSON_COMPACT, emit_subr, NULL);
  1999. ucl_object_emit_funcs_free (emit_subr);
  2000. ucl_object_unref (obj);
  2001. /* Rewind output file */
  2002. close (outfd);
  2003. outfd = open (tmppath, O_RDONLY);
  2004. unlink (tmppath);
  2005. }
  2006. /* Now we can send outfd and status message */
  2007. memset (&msg, 0, sizeof (msg));
  2008. /* Attach fd to the message */
  2009. if (outfd != -1) {
  2010. memset (fdspace, 0, sizeof (fdspace));
  2011. msg.msg_control = fdspace;
  2012. msg.msg_controllen = sizeof (fdspace);
  2013. cmsg = CMSG_FIRSTHDR (&msg);
  2014. if (cmsg) {
  2015. cmsg->cmsg_level = SOL_SOCKET;
  2016. cmsg->cmsg_type = SCM_RIGHTS;
  2017. cmsg->cmsg_len = CMSG_LEN (sizeof (int));
  2018. memcpy (CMSG_DATA (cmsg), &outfd, sizeof (int));
  2019. }
  2020. }
  2021. iov.iov_base = &rep;
  2022. iov.iov_len = sizeof (rep);
  2023. msg.msg_iov = &iov;
  2024. msg.msg_iovlen = 1;
  2025. if (sendmsg (fd, &msg, 0) == -1) {
  2026. msg_err_main ("cannot send fuzzy stat: %s", strerror (errno));
  2027. }
  2028. if (outfd != -1) {
  2029. close (outfd);
  2030. }
  2031. return TRUE;
  2032. }
  2033. static gboolean
  2034. fuzzy_storage_parse_mirror (rspamd_mempool_t *pool,
  2035. const ucl_object_t *obj,
  2036. gpointer ud,
  2037. struct rspamd_rcl_section *section,
  2038. GError **err)
  2039. {
  2040. const ucl_object_t *elt;
  2041. struct rspamd_fuzzy_mirror *up = NULL;
  2042. struct rspamd_rcl_struct_parser *pd = ud;
  2043. struct rspamd_fuzzy_storage_ctx *ctx;
  2044. ctx = pd->user_struct;
  2045. if (ucl_object_type (obj) != UCL_OBJECT) {
  2046. g_set_error (err, g_quark_try_string ("fuzzy"), 100,
  2047. "mirror/slave option must be an object");
  2048. return FALSE;
  2049. }
  2050. elt = ucl_object_lookup (obj, "name");
  2051. if (elt == NULL) {
  2052. g_set_error (err, g_quark_try_string ("fuzzy"), 100,
  2053. "mirror option must have some name definition");
  2054. return FALSE;
  2055. }
  2056. up = g_malloc0 (sizeof (*up));
  2057. up->name = g_strdup (ucl_object_tostring (elt));
  2058. elt = ucl_object_lookup (obj, "key");
  2059. if (elt != NULL) {
  2060. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  2061. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  2062. }
  2063. if (up->key == NULL) {
  2064. g_set_error (err, g_quark_try_string ("fuzzy"), 100,
  2065. "cannot read mirror key");
  2066. goto err;
  2067. }
  2068. elt = ucl_object_lookup (obj, "hosts");
  2069. if (elt == NULL) {
  2070. g_set_error (err, g_quark_try_string ("fuzzy"), 100,
  2071. "mirror option must have some hosts definition");
  2072. goto err;
  2073. }
  2074. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  2075. if (!rspamd_upstreams_from_ucl (up->u, elt, 11335, NULL)) {
  2076. g_set_error (err, g_quark_try_string ("fuzzy"), 100,
  2077. "mirror has bad hosts definition");
  2078. goto err;
  2079. }
  2080. g_ptr_array_add (ctx->mirrors, up);
  2081. return TRUE;
  2082. err:
  2083. g_free (up->name);
  2084. rspamd_upstreams_destroy (up->u);
  2085. if (up->key) {
  2086. rspamd_pubkey_unref (up->key);
  2087. }
  2088. g_free (up);
  2089. return FALSE;
  2090. }
  2091. static gboolean
  2092. fuzzy_storage_parse_master_flags (rspamd_mempool_t *pool,
  2093. const ucl_object_t *obj,
  2094. gpointer ud,
  2095. struct rspamd_rcl_section *section,
  2096. GError **err)
  2097. {
  2098. const ucl_object_t *cur;
  2099. struct rspamd_rcl_struct_parser *pd = ud;
  2100. struct rspamd_fuzzy_storage_ctx *ctx;
  2101. ucl_object_iter_t it = NULL;
  2102. gulong remote_flag;
  2103. gint64 local_flag;
  2104. ctx = pd->user_struct;
  2105. if (ucl_object_type (obj) != UCL_OBJECT) {
  2106. g_set_error (err, g_quark_try_string ("fuzzy"), 100,
  2107. "master_flags option must be an object");
  2108. return FALSE;
  2109. }
  2110. while ((cur = ucl_iterate_object (obj, &it, true)) != NULL) {
  2111. if (rspamd_strtoul (cur->key, cur->keylen, &remote_flag) &&
  2112. ucl_object_toint_safe (cur, (int64_t *)&local_flag)) {
  2113. g_hash_table_insert (ctx->master_flags, GUINT_TO_POINTER (remote_flag),
  2114. GUINT_TO_POINTER (local_flag));
  2115. }
  2116. }
  2117. return TRUE;
  2118. }
  2119. static gboolean
  2120. fuzzy_parse_keypair (rspamd_mempool_t *pool,
  2121. const ucl_object_t *obj,
  2122. gpointer ud,
  2123. struct rspamd_rcl_section *section,
  2124. GError **err)
  2125. {
  2126. struct rspamd_rcl_struct_parser *pd = ud;
  2127. struct rspamd_fuzzy_storage_ctx *ctx;
  2128. struct rspamd_cryptobox_keypair *kp;
  2129. struct fuzzy_key_stat *keystat;
  2130. struct fuzzy_key *key;
  2131. const ucl_object_t *cur;
  2132. const guchar *pk;
  2133. ucl_object_iter_t it = NULL;
  2134. gboolean ret;
  2135. ctx = pd->user_struct;
  2136. pd->offset = G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, default_keypair);
  2137. /*
  2138. * Single key
  2139. */
  2140. if (ucl_object_type (obj) == UCL_STRING || ucl_object_type (obj)
  2141. == UCL_OBJECT) {
  2142. ret = rspamd_rcl_parse_struct_keypair (pool, obj, pd, section, err);
  2143. if (!ret) {
  2144. return ret;
  2145. }
  2146. /* Insert key to the hash table */
  2147. kp = ctx->default_keypair;
  2148. if (kp == NULL) {
  2149. return FALSE;
  2150. }
  2151. if (rspamd_keypair_alg (kp) != RSPAMD_CRYPTOBOX_MODE_25519 ||
  2152. rspamd_keypair_type (kp) != RSPAMD_KEYPAIR_KEX) {
  2153. return FALSE;
  2154. }
  2155. key = rspamd_mempool_alloc0 (pool, sizeof (*key));
  2156. key->key = kp;
  2157. keystat = rspamd_mempool_alloc0 (pool, sizeof (*keystat));
  2158. /* Hash of ip -> fuzzy_key_stat */
  2159. keystat->last_ips = rspamd_lru_hash_new_full (1024,
  2160. (GDestroyNotify) rspamd_inet_address_free, fuzzy_key_stat_dtor,
  2161. rspamd_inet_address_hash, rspamd_inet_address_equal);
  2162. key->stat = keystat;
  2163. pk = rspamd_keypair_component (kp, RSPAMD_KEYPAIR_COMPONENT_PK,
  2164. NULL);
  2165. g_hash_table_insert (ctx->keys, (gpointer)pk, key);
  2166. ctx->default_key = key;
  2167. msg_info_pool ("loaded keypair %*xs", 8, pk);
  2168. }
  2169. else if (ucl_object_type (obj) == UCL_ARRAY) {
  2170. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  2171. if (!fuzzy_parse_keypair (pool, cur, pd, section, err)) {
  2172. msg_err_pool ("cannot parse keypair");
  2173. }
  2174. }
  2175. }
  2176. return TRUE;
  2177. }
  2178. static guint
  2179. fuzzy_kp_hash (gconstpointer p)
  2180. {
  2181. return *(guint *)p;
  2182. }
  2183. static gboolean
  2184. fuzzy_kp_equal (gconstpointer a, gconstpointer b)
  2185. {
  2186. const guchar *pa = a, *pb = b;
  2187. return (memcmp (pa, pb, RSPAMD_FUZZY_KEYLEN) == 0);
  2188. }
  2189. gpointer
  2190. init_fuzzy (struct rspamd_config *cfg)
  2191. {
  2192. struct rspamd_fuzzy_storage_ctx *ctx;
  2193. GQuark type;
  2194. type = g_quark_try_string ("fuzzy");
  2195. ctx = rspamd_mempool_alloc0 (cfg->cfg_pool,
  2196. sizeof (struct rspamd_fuzzy_storage_ctx));
  2197. ctx->magic = rspamd_fuzzy_storage_magic;
  2198. ctx->sync_timeout = DEFAULT_SYNC_TIMEOUT;
  2199. ctx->master_timeout = DEFAULT_MASTER_TIMEOUT;
  2200. ctx->keypair_cache_size = DEFAULT_KEYPAIR_CACHE_SIZE;
  2201. ctx->keys = g_hash_table_new_full (fuzzy_kp_hash, fuzzy_kp_equal,
  2202. NULL, fuzzy_key_dtor);
  2203. rspamd_mempool_add_destructor (cfg->cfg_pool,
  2204. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->keys);
  2205. ctx->master_flags = g_hash_table_new (g_direct_hash, g_direct_equal);
  2206. rspamd_mempool_add_destructor (cfg->cfg_pool,
  2207. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->master_flags);
  2208. ctx->errors_ips = rspamd_lru_hash_new_full (1024,
  2209. (GDestroyNotify) rspamd_inet_address_free, g_free,
  2210. rspamd_inet_address_hash, rspamd_inet_address_equal);
  2211. rspamd_mempool_add_destructor (cfg->cfg_pool,
  2212. (rspamd_mempool_destruct_t)rspamd_lru_hash_destroy, ctx->errors_ips);
  2213. ctx->cfg = cfg;
  2214. ctx->mirrors = g_ptr_array_new ();
  2215. rspamd_mempool_add_destructor (cfg->cfg_pool,
  2216. (rspamd_mempool_destruct_t)rspamd_ptr_array_free_hard, ctx->mirrors);
  2217. ctx->updates_maxfail = DEFAULT_UPDATES_MAXFAIL;
  2218. ctx->collection_id_file = RSPAMD_DBDIR "/fuzzy_collection.id";
  2219. ctx->leaky_bucket_mask = DEFAULT_BUCKET_MASK;
  2220. ctx->leaky_bucket_ttl = DEFAULT_BUCKET_TTL;
  2221. ctx->max_buckets = DEFAULT_MAX_BUCKETS;
  2222. ctx->leaky_bucket_burst = NAN;
  2223. ctx->leaky_bucket_rate = NAN;
  2224. rspamd_rcl_register_worker_option (cfg,
  2225. type,
  2226. "sync",
  2227. rspamd_rcl_parse_struct_time,
  2228. ctx,
  2229. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx,
  2230. sync_timeout),
  2231. RSPAMD_CL_FLAG_TIME_FLOAT,
  2232. "Time to perform database sync, default: "
  2233. G_STRINGIFY (DEFAULT_SYNC_TIMEOUT) " seconds");
  2234. rspamd_rcl_register_worker_option (cfg,
  2235. type,
  2236. "expire",
  2237. rspamd_rcl_parse_struct_time,
  2238. ctx,
  2239. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx,
  2240. expire),
  2241. RSPAMD_CL_FLAG_TIME_FLOAT,
  2242. "Default expire time for hashes, default: "
  2243. G_STRINGIFY (DEFAULT_EXPIRE) " seconds");
  2244. rspamd_rcl_register_worker_option (cfg,
  2245. type,
  2246. "allow_update",
  2247. rspamd_rcl_parse_struct_ucl,
  2248. ctx,
  2249. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, update_map),
  2250. 0,
  2251. "Allow modifications from the following IP addresses");
  2252. rspamd_rcl_register_worker_option (cfg,
  2253. type,
  2254. "keypair",
  2255. fuzzy_parse_keypair,
  2256. ctx,
  2257. 0,
  2258. RSPAMD_CL_FLAG_MULTIPLE,
  2259. "Encryption keypair (can be repeated for different keys)");
  2260. rspamd_rcl_register_worker_option (cfg,
  2261. type,
  2262. "keypair_cache_size",
  2263. rspamd_rcl_parse_struct_integer,
  2264. ctx,
  2265. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx,
  2266. keypair_cache_size),
  2267. RSPAMD_CL_FLAG_UINT,
  2268. "Size of keypairs cache, default: "
  2269. G_STRINGIFY (DEFAULT_KEYPAIR_CACHE_SIZE));
  2270. rspamd_rcl_register_worker_option (cfg,
  2271. type,
  2272. "encrypted_only",
  2273. rspamd_rcl_parse_struct_boolean,
  2274. ctx,
  2275. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, encrypted_only),
  2276. 0,
  2277. "Allow encrypted requests only (and forbid all unknown keys or plaintext requests)");
  2278. rspamd_rcl_register_worker_option (cfg,
  2279. type,
  2280. "read_only",
  2281. rspamd_rcl_parse_struct_boolean,
  2282. ctx,
  2283. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, read_only),
  2284. 0,
  2285. "Work in read only mode");
  2286. rspamd_rcl_register_worker_option (cfg,
  2287. type,
  2288. "master_timeout",
  2289. rspamd_rcl_parse_struct_time,
  2290. ctx,
  2291. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, master_timeout),
  2292. RSPAMD_CL_FLAG_TIME_FLOAT,
  2293. "Master protocol IO timeout");
  2294. rspamd_rcl_register_worker_option (cfg,
  2295. type,
  2296. "sync_keypair",
  2297. rspamd_rcl_parse_struct_keypair,
  2298. ctx,
  2299. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, sync_keypair),
  2300. 0,
  2301. "Encryption key for master/slave updates");
  2302. rspamd_rcl_register_worker_option (cfg,
  2303. type,
  2304. "masters",
  2305. rspamd_rcl_parse_struct_ucl,
  2306. ctx,
  2307. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, masters_map),
  2308. 0,
  2309. "Allow master/slave updates from the following IP addresses");
  2310. rspamd_rcl_register_worker_option (cfg,
  2311. type,
  2312. "blocked",
  2313. rspamd_rcl_parse_struct_ucl,
  2314. ctx,
  2315. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, blocked_map),
  2316. 0,
  2317. "Block requests from specific networks");
  2318. rspamd_rcl_register_worker_option (cfg,
  2319. type,
  2320. "master_key",
  2321. rspamd_rcl_parse_struct_pubkey,
  2322. ctx,
  2323. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, master_key),
  2324. 0,
  2325. "Allow master/slave updates merely using the specified key");
  2326. rspamd_rcl_register_worker_option (cfg,
  2327. type,
  2328. "mirror",
  2329. fuzzy_storage_parse_mirror,
  2330. ctx,
  2331. 0,
  2332. RSPAMD_CL_FLAG_MULTIPLE,
  2333. "List of slave hosts");
  2334. rspamd_rcl_register_worker_option (cfg,
  2335. type,
  2336. "slave",
  2337. fuzzy_storage_parse_mirror,
  2338. ctx,
  2339. 0,
  2340. RSPAMD_CL_FLAG_MULTIPLE,
  2341. "List of slave hosts");
  2342. rspamd_rcl_register_worker_option (cfg,
  2343. type,
  2344. "master_flags",
  2345. fuzzy_storage_parse_master_flags,
  2346. ctx,
  2347. 0,
  2348. 0,
  2349. "Map of flags in form master_flags = { master_flag = local_flag; ... }; ");
  2350. rspamd_rcl_register_worker_option (cfg,
  2351. type,
  2352. "updates_maxfail",
  2353. rspamd_rcl_parse_struct_integer,
  2354. ctx,
  2355. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, updates_maxfail),
  2356. RSPAMD_CL_FLAG_UINT,
  2357. "Maximum number of updates to be failed before discarding");
  2358. rspamd_rcl_register_worker_option (cfg,
  2359. type,
  2360. "collection_only",
  2361. rspamd_rcl_parse_struct_boolean,
  2362. ctx,
  2363. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, collection_mode),
  2364. 0,
  2365. "Start fuzzy in collection only mode");
  2366. rspamd_rcl_register_worker_option (cfg,
  2367. type,
  2368. "collection_signkey",
  2369. rspamd_rcl_parse_struct_pubkey,
  2370. ctx,
  2371. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, collection_sign_key),
  2372. RSPAMD_CL_FLAG_SIGNKEY,
  2373. "Accept only signed requests with the specified key");
  2374. rspamd_rcl_register_worker_option (cfg,
  2375. type,
  2376. "collection_keypair",
  2377. rspamd_rcl_parse_struct_keypair,
  2378. ctx,
  2379. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, collection_keypair),
  2380. 0,
  2381. "Use the specified keypair to encrypt collection protocol");
  2382. rspamd_rcl_register_worker_option (cfg,
  2383. type,
  2384. "collection_id_file",
  2385. rspamd_rcl_parse_struct_string,
  2386. ctx,
  2387. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, collection_id_file),
  2388. RSPAMD_CL_FLAG_STRING_PATH,
  2389. "Store collection epoch in the desired file");
  2390. rspamd_rcl_register_worker_option (cfg,
  2391. type,
  2392. "skip_hashes",
  2393. rspamd_rcl_parse_struct_ucl,
  2394. ctx,
  2395. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, skip_map),
  2396. 0,
  2397. "Skip specific hashes from the map");
  2398. /* Ratelimits */
  2399. rspamd_rcl_register_worker_option (cfg,
  2400. type,
  2401. "ratelimit_whitelist",
  2402. rspamd_rcl_parse_struct_ucl,
  2403. ctx,
  2404. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, ratelimit_whitelist_map),
  2405. 0,
  2406. "Skip specific addresses from rate limiting");
  2407. rspamd_rcl_register_worker_option (cfg,
  2408. type,
  2409. "ratelimit_max_buckets",
  2410. rspamd_rcl_parse_struct_integer,
  2411. ctx,
  2412. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, max_buckets),
  2413. RSPAMD_CL_FLAG_UINT,
  2414. "Maximum number of leaky buckets (default: " G_STRINGIFY(DEFAULT_MAX_BUCKETS) ")");
  2415. rspamd_rcl_register_worker_option (cfg,
  2416. type,
  2417. "ratelimit_network_mask",
  2418. rspamd_rcl_parse_struct_integer,
  2419. ctx,
  2420. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_mask),
  2421. RSPAMD_CL_FLAG_UINT,
  2422. "Network mask to apply for IPv4 rate addresses (default: " G_STRINGIFY(DEFAULT_BUCKET_MASK) ")");
  2423. rspamd_rcl_register_worker_option (cfg,
  2424. type,
  2425. "ratelimit_bucket_ttl",
  2426. rspamd_rcl_parse_struct_time,
  2427. ctx,
  2428. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_ttl),
  2429. RSPAMD_CL_FLAG_TIME_INTEGER,
  2430. "Time to live for ratelimit element (default: " G_STRINGIFY(DEFAULT_BUCKET_TTL) ")");
  2431. rspamd_rcl_register_worker_option (cfg,
  2432. type,
  2433. "ratelimit_rate",
  2434. rspamd_rcl_parse_struct_double,
  2435. ctx,
  2436. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_rate),
  2437. 0,
  2438. "Leak rate in requests per second");
  2439. rspamd_rcl_register_worker_option (cfg,
  2440. type,
  2441. "ratelimit_burst",
  2442. rspamd_rcl_parse_struct_double,
  2443. ctx,
  2444. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_burst),
  2445. 0,
  2446. "Peak value for ratelimit bucket");
  2447. rspamd_rcl_register_worker_option (cfg,
  2448. type,
  2449. "ratelimit_log_only",
  2450. rspamd_rcl_parse_struct_boolean,
  2451. ctx,
  2452. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, ratelimit_log_only),
  2453. 0,
  2454. "Don't really ban on ratelimit reaching, just log");
  2455. return ctx;
  2456. }
  2457. static void
  2458. rspamd_fuzzy_peer_io (gint fd, gshort what, gpointer d)
  2459. {
  2460. struct fuzzy_peer_cmd cmd;
  2461. struct rspamd_fuzzy_storage_ctx *ctx = d;
  2462. gssize r;
  2463. r = read (fd, &cmd, sizeof (cmd));
  2464. if (r != sizeof (cmd)) {
  2465. if (errno == EINTR) {
  2466. rspamd_fuzzy_peer_io (fd, what, d);
  2467. return;
  2468. }
  2469. if (errno != EAGAIN) {
  2470. msg_err ("cannot read command from peers: %s", strerror (errno));
  2471. }
  2472. }
  2473. else {
  2474. g_array_append_val (ctx->updates_pending, cmd);
  2475. }
  2476. }
  2477. static void
  2478. fuzzy_peer_rep (struct rspamd_worker *worker,
  2479. struct rspamd_srv_reply *rep, gint rep_fd,
  2480. gpointer ud)
  2481. {
  2482. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  2483. GList *cur;
  2484. struct rspamd_worker_listen_socket *ls;
  2485. struct event *accept_events;
  2486. ctx->peer_fd = rep_fd;
  2487. if (rep_fd == -1) {
  2488. msg_err ("cannot receive peer fd from the main process");
  2489. exit (EXIT_FAILURE);
  2490. }
  2491. else {
  2492. rspamd_socket_nonblocking (rep_fd);
  2493. }
  2494. msg_info ("got peer fd reply from the main process");
  2495. /* Start listening */
  2496. cur = worker->cf->listen_socks;
  2497. while (cur) {
  2498. ls = cur->data;
  2499. if (ls->fd != -1) {
  2500. msg_info ("start listening on %s",
  2501. rspamd_inet_address_to_string_pretty (ls->addr));
  2502. if (ls->type == RSPAMD_WORKER_SOCKET_UDP) {
  2503. accept_events = g_malloc0 (sizeof (struct event) * 2);
  2504. event_set (&accept_events[0], ls->fd, EV_READ | EV_PERSIST,
  2505. accept_fuzzy_socket, worker);
  2506. event_base_set (ctx->ev_base, &accept_events[0]);
  2507. event_add (&accept_events[0], NULL);
  2508. worker->accept_events = g_list_prepend (worker->accept_events,
  2509. accept_events);
  2510. }
  2511. else if (worker->index == 0) {
  2512. /* We allow TCP listeners only for a update worker */
  2513. accept_events = g_malloc0 (sizeof (struct event) * 2);
  2514. if (ctx->collection_mode) {
  2515. event_set (&accept_events[0], ls->fd, EV_READ | EV_PERSIST,
  2516. accept_fuzzy_collection_socket, worker);
  2517. }
  2518. else {
  2519. event_set (&accept_events[0], ls->fd, EV_READ | EV_PERSIST,
  2520. accept_fuzzy_mirror_socket, worker);
  2521. }
  2522. event_base_set (ctx->ev_base, &accept_events[0]);
  2523. event_add (&accept_events[0], NULL);
  2524. worker->accept_events = g_list_prepend (worker->accept_events,
  2525. accept_events);
  2526. }
  2527. }
  2528. cur = g_list_next (cur);
  2529. }
  2530. if (worker->index == 0 && ctx->peer_fd != -1) {
  2531. /* Listen for peer requests */
  2532. event_set (&ctx->peer_ev, ctx->peer_fd, EV_READ | EV_PERSIST,
  2533. rspamd_fuzzy_peer_io, ctx);
  2534. event_base_set (ctx->ev_base, &ctx->peer_ev);
  2535. event_add (&ctx->peer_ev, NULL);
  2536. }
  2537. }
  2538. /*
  2539. * Start worker process
  2540. */
  2541. void
  2542. start_fuzzy (struct rspamd_worker *worker)
  2543. {
  2544. struct rspamd_fuzzy_storage_ctx *ctx = worker->ctx;
  2545. GError *err = NULL;
  2546. struct rspamd_srv_command srv_cmd;
  2547. struct rspamd_config *cfg = worker->srv->cfg;
  2548. ctx->ev_base = rspamd_prepare_worker (worker,
  2549. "fuzzy",
  2550. NULL);
  2551. ctx->peer_fd = -1;
  2552. ctx->worker = worker;
  2553. ctx->cfg = worker->srv->cfg;
  2554. double_to_tv (ctx->master_timeout, &ctx->master_io_tv);
  2555. ctx->resolver = dns_resolver_init (worker->srv->logger,
  2556. ctx->ev_base,
  2557. worker->srv->cfg);
  2558. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  2559. ctx->ev_base, ctx->resolver->r);
  2560. if (ctx->keypair_cache_size > 0) {
  2561. /* Create keypairs cache */
  2562. ctx->keypair_cache = rspamd_keypair_cache_new (ctx->keypair_cache_size);
  2563. }
  2564. if (!ctx->collection_mode) {
  2565. /*
  2566. * Open DB and perform VACUUM
  2567. */
  2568. if ((ctx->backend = rspamd_fuzzy_backend_create (ctx->ev_base,
  2569. worker->cf->options, cfg, &err)) == NULL) {
  2570. msg_err ("cannot open backend: %e", err);
  2571. if (err) {
  2572. g_error_free (err);
  2573. }
  2574. exit (EXIT_SUCCESS);
  2575. }
  2576. rspamd_fuzzy_backend_count (ctx->backend, fuzzy_count_callback, ctx);
  2577. if (worker->index == 0) {
  2578. ctx->updates_pending = g_array_sized_new (FALSE, FALSE,
  2579. sizeof (struct fuzzy_peer_cmd), 1024);
  2580. rspamd_fuzzy_backend_start_update (ctx->backend, ctx->sync_timeout,
  2581. rspamd_fuzzy_storage_periodic_callback, ctx);
  2582. }
  2583. double_to_tv (ctx->sync_timeout, &ctx->stat_tv);
  2584. event_set (&ctx->stat_ev, -1, EV_TIMEOUT, rspamd_fuzzy_stat_callback, ctx);
  2585. event_base_set (ctx->ev_base, &ctx->stat_ev);
  2586. event_add (&ctx->stat_ev, &ctx->stat_tv);
  2587. /* Register custom reload and stat commands for the control socket */
  2588. rspamd_control_worker_add_cmd_handler (worker, RSPAMD_CONTROL_RELOAD,
  2589. rspamd_fuzzy_storage_reload, ctx);
  2590. rspamd_control_worker_add_cmd_handler (worker, RSPAMD_CONTROL_FUZZY_STAT,
  2591. rspamd_fuzzy_storage_stat, ctx);
  2592. rspamd_control_worker_add_cmd_handler (worker, RSPAMD_CONTROL_FUZZY_SYNC,
  2593. rspamd_fuzzy_storage_sync, ctx);
  2594. }
  2595. else {
  2596. /*
  2597. * In collection mode we do a different thing:
  2598. * we collect fuzzy hashes in the updates queue and ignore all read commands
  2599. */
  2600. if (worker->index == 0) {
  2601. ctx->updates_pending = g_array_sized_new (FALSE, FALSE,
  2602. sizeof (struct fuzzy_peer_cmd), 1024);
  2603. double_to_tv (ctx->sync_timeout, &ctx->stat_tv);
  2604. event_set (&ctx->stat_ev, -1, EV_TIMEOUT|EV_PERSIST,
  2605. rspamd_fuzzy_collection_periodic, ctx);
  2606. event_base_set (ctx->ev_base, &ctx->stat_ev);
  2607. event_add (&ctx->stat_ev, &ctx->stat_tv);
  2608. ctx->collection_rt = rspamd_http_router_new (
  2609. rspamd_fuzzy_collection_error_handler,
  2610. rspamd_fuzzy_collection_finish_handler,
  2611. &ctx->stat_tv,
  2612. ctx->ev_base,
  2613. NULL, ctx->keypair_cache);
  2614. if (ctx->collection_keypair) {
  2615. rspamd_http_router_set_key (ctx->collection_rt,
  2616. ctx->collection_keypair);
  2617. }
  2618. /* Try to load collection id */
  2619. if (ctx->collection_id_file) {
  2620. gint fd;
  2621. fd = rspamd_file_xopen (ctx->collection_id_file, O_RDONLY, 0,
  2622. FALSE);
  2623. if (fd == -1) {
  2624. if (errno != ENOENT) {
  2625. msg_err ("cannot open collection id from %s: %s",
  2626. ctx->collection_id_file, strerror (errno));
  2627. }
  2628. ctx->collection_id = 0;
  2629. }
  2630. else {
  2631. if (read (fd, &ctx->collection_id,
  2632. sizeof (ctx->collection_id)) == -1) {
  2633. msg_err ("cannot read collection id from %s: %s",
  2634. ctx->collection_id_file, strerror (errno));
  2635. ctx->collection_id = 0;
  2636. }
  2637. close (fd);
  2638. }
  2639. }
  2640. /* Generate new cookie */
  2641. ottery_rand_bytes (ctx->cookie, sizeof (ctx->cookie));
  2642. /* Register paths */
  2643. rspamd_http_router_add_path (ctx->collection_rt,
  2644. "/cookie",
  2645. rspamd_fuzzy_collection_cookie);
  2646. rspamd_http_router_add_path (ctx->collection_rt,
  2647. "/data",
  2648. rspamd_fuzzy_collection_data);
  2649. }
  2650. }
  2651. if (ctx->mirrors && ctx->mirrors->len != 0) {
  2652. if (ctx->sync_keypair == NULL) {
  2653. GString *pk_str = NULL;
  2654. ctx->sync_keypair = rspamd_keypair_new (RSPAMD_KEYPAIR_KEX,
  2655. RSPAMD_CRYPTOBOX_MODE_25519);
  2656. pk_str = rspamd_keypair_print (ctx->sync_keypair,
  2657. RSPAMD_KEYPAIR_COMPONENT_PK|RSPAMD_KEYPAIR_BASE32);
  2658. msg_warn_config ("generating new temporary keypair for communicating"
  2659. " with slave hosts, pk is %s", pk_str->str);
  2660. g_string_free (pk_str, TRUE);
  2661. }
  2662. }
  2663. /* Create radix trees */
  2664. if (ctx->update_map != NULL) {
  2665. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->update_map,
  2666. "Allow fuzzy updates from specified addresses",
  2667. &ctx->update_ips, NULL);
  2668. }
  2669. if (ctx->masters_map != NULL) {
  2670. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->masters_map,
  2671. "Allow fuzzy master/slave updates from specified addresses",
  2672. &ctx->master_ips, NULL);
  2673. }
  2674. if (ctx->skip_map != NULL) {
  2675. struct rspamd_map *m;
  2676. if ((m = rspamd_map_add_from_ucl (cfg, ctx->skip_map,
  2677. "Skip hashes",
  2678. rspamd_kv_list_read,
  2679. rspamd_kv_list_fin,
  2680. rspamd_kv_list_dtor,
  2681. (void **)&ctx->skip_hashes)) == NULL) {
  2682. msg_warn_config ("cannot load hashes list from %s",
  2683. ucl_object_tostring (ctx->skip_map));
  2684. }
  2685. else {
  2686. m->active_http = TRUE;
  2687. }
  2688. }
  2689. if (ctx->blocked_map != NULL) {
  2690. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->blocked_map,
  2691. "Block fuzzy requests from the specific IPs",
  2692. &ctx->blocked_ips, NULL);
  2693. }
  2694. /* Create radix trees */
  2695. if (ctx->ratelimit_whitelist_map != NULL) {
  2696. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->ratelimit_whitelist_map,
  2697. "Skip ratelimits from specific ip addresses/networks",
  2698. &ctx->ratelimit_whitelist, NULL);
  2699. }
  2700. /* Ratelimits */
  2701. if (!isnan (ctx->leaky_bucket_rate) && !isnan (ctx->leaky_bucket_burst)) {
  2702. ctx->ratelimit_buckets = rspamd_lru_hash_new_full (ctx->max_buckets,
  2703. NULL, fuzzy_rl_bucket_free,
  2704. rspamd_inet_address_hash, rspamd_inet_address_equal);
  2705. }
  2706. /* Maps events */
  2707. ctx->resolver = dns_resolver_init (worker->srv->logger,
  2708. ctx->ev_base,
  2709. worker->srv->cfg);
  2710. rspamd_map_watch (worker->srv->cfg, ctx->ev_base, ctx->resolver, worker, 0);
  2711. /* Get peer pipe */
  2712. memset (&srv_cmd, 0, sizeof (srv_cmd));
  2713. srv_cmd.type = RSPAMD_SRV_SOCKETPAIR;
  2714. srv_cmd.cmd.spair.af = SOCK_DGRAM;
  2715. srv_cmd.cmd.spair.pair_num = worker->index;
  2716. memset (srv_cmd.cmd.spair.pair_id, 0, sizeof (srv_cmd.cmd.spair.pair_id));
  2717. memcpy (srv_cmd.cmd.spair.pair_id, "fuzzy", sizeof ("fuzzy"));
  2718. rspamd_srv_send_command (worker, ctx->ev_base, &srv_cmd, -1,
  2719. fuzzy_peer_rep, ctx);
  2720. event_base_loop (ctx->ev_base, 0);
  2721. rspamd_worker_block_signals ();
  2722. if (worker->index == 0 && ctx->updates_pending->len > 0) {
  2723. if (!ctx->collection_mode) {
  2724. rspamd_fuzzy_process_updates_queue (ctx, local_db_name, FALSE);
  2725. event_base_loop (ctx->ev_base, 0);
  2726. }
  2727. }
  2728. if (!ctx->collection_mode) {
  2729. rspamd_fuzzy_backend_close (ctx->backend);
  2730. }
  2731. else if (worker->index == 0) {
  2732. gint fd;
  2733. /* Steal keypairs cache... */
  2734. ctx->collection_rt->cache = NULL;
  2735. rspamd_http_router_free (ctx->collection_rt);
  2736. /* Try to save collection id */
  2737. fd = rspamd_file_xopen (ctx->collection_id_file,
  2738. O_WRONLY | O_CREAT | O_TRUNC, 00644, 0);
  2739. if (fd == -1) {
  2740. msg_err ("cannot open collection id to store in %s: %s",
  2741. ctx->collection_id_file, strerror (errno));
  2742. }
  2743. else {
  2744. if (write (fd, &ctx->collection_id,
  2745. sizeof (ctx->collection_id)) == -1) {
  2746. msg_err ("cannot store collection id in %s: %s",
  2747. ctx->collection_id_file, strerror (errno));
  2748. }
  2749. close (fd);
  2750. }
  2751. }
  2752. if (worker->index == 0) {
  2753. g_array_free (ctx->updates_pending, TRUE);
  2754. }
  2755. if (ctx->peer_fd != -1) {
  2756. if (worker->index == 0) {
  2757. event_del (&ctx->peer_ev);
  2758. }
  2759. close (ctx->peer_fd);
  2760. }
  2761. if (ctx->keypair_cache) {
  2762. rspamd_keypair_cache_destroy (ctx->keypair_cache);
  2763. }
  2764. REF_RELEASE (ctx->cfg);
  2765. rspamd_log_close (worker->srv->logger, TRUE);
  2766. exit (EXIT_SUCCESS);
  2767. }