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 68KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427242824292430243124322433243424352436243724382439244024412442244324442445244624472448244924502451245224532454245524562457245824592460246124622463246424652466246724682469247024712472247324742475247624772478247924802481248224832484248524862487248824892490249124922493249424952496249724982499250025012502250325042505250625072508250925102511251225132514251525162517251825192520252125222523252425252526252725282529253025312532253325342535253625372538253925402541254225432544254525462547254825492550255125522553255425552556255725582559256025612562256325642565256625672568256925702571257225732574257525762577257825792580258125822583258425852586258725882589259025912592259325942595259625972598259926002601260226032604260526062607260826092610261126122613261426152616261726182619262026212622262326242625262626272628262926302631263226332634263526362637263826392640264126422643264426452646264726482649265026512652265326542655265626572658265926602661266226632664266526662667266826692670267126722673267426752676267726782679268026812682268326842685268626872688268926902691269226932694269526962697
  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 "libserver/maps/map.h"
  24. #include "libserver/maps/map_helpers.h"
  25. #include "libserver/fuzzy_backend/fuzzy_backend.h"
  26. #include "ottery.h"
  27. #include "ref.h"
  28. #include "xxhash.h"
  29. #include "libserver/worker_util.h"
  30. #include "libserver/rspamd_control.h"
  31. #include "libcryptobox/cryptobox.h"
  32. #include "libcryptobox/keypairs_cache.h"
  33. #include "libcryptobox/keypair.h"
  34. #include "libutil/hash.h"
  35. #include "libserver/maps/map_private.h"
  36. #include "contrib/uthash/utlist.h"
  37. #include "lua/lua_common.h"
  38. #include "unix-std.h"
  39. #include <math.h>
  40. /* Resync value in seconds */
  41. #define DEFAULT_SYNC_TIMEOUT 60.0
  42. #define DEFAULT_KEYPAIR_CACHE_SIZE 512
  43. #define DEFAULT_MASTER_TIMEOUT 10.0
  44. #define DEFAULT_UPDATES_MAXFAIL 3
  45. #define COOKIE_SIZE 128
  46. #define DEFAULT_MAX_BUCKETS 2000
  47. #define DEFAULT_BUCKET_TTL 3600
  48. #define DEFAULT_BUCKET_MASK 24
  49. static const gchar *local_db_name = "local";
  50. #define msg_err_fuzzy_update(...) rspamd_default_log_function (G_LOG_LEVEL_CRITICAL, \
  51. session->name, session->uid, \
  52. G_STRFUNC, \
  53. __VA_ARGS__)
  54. #define msg_warn_fuzzy_update(...) rspamd_default_log_function (G_LOG_LEVEL_WARNING, \
  55. session->name, session->uid, \
  56. G_STRFUNC, \
  57. __VA_ARGS__)
  58. #define msg_info_fuzzy_update(...) rspamd_default_log_function (G_LOG_LEVEL_INFO, \
  59. session->name, session->uid, \
  60. G_STRFUNC, \
  61. __VA_ARGS__)
  62. #define msg_err_fuzzy_collection(...) rspamd_default_log_function (G_LOG_LEVEL_CRITICAL, \
  63. "fuzzy_collection", session->uid, \
  64. G_STRFUNC, \
  65. __VA_ARGS__)
  66. #define msg_warn_fuzzy_collection(...) rspamd_default_log_function (G_LOG_LEVEL_WARNING, \
  67. "fuzzy_collection", session->uid, \
  68. G_STRFUNC, \
  69. __VA_ARGS__)
  70. #define msg_info_fuzzy_collection(...) rspamd_default_log_function (G_LOG_LEVEL_INFO, \
  71. "fuzzy_collection", session->uid, \
  72. G_STRFUNC, \
  73. __VA_ARGS__)
  74. /* Init functions */
  75. gpointer init_fuzzy (struct rspamd_config *cfg);
  76. void start_fuzzy (struct rspamd_worker *worker);
  77. worker_t fuzzy_worker = {
  78. "fuzzy", /* Name */
  79. init_fuzzy, /* Init function */
  80. start_fuzzy, /* Start function */
  81. RSPAMD_WORKER_HAS_SOCKET,
  82. RSPAMD_WORKER_SOCKET_UDP, /* UDP socket */
  83. RSPAMD_WORKER_VER /* Version info */
  84. };
  85. struct fuzzy_global_stat {
  86. guint64 fuzzy_hashes;
  87. /**< number of fuzzy hashes stored */
  88. guint64 fuzzy_hashes_expired;
  89. /**< number of fuzzy hashes expired */
  90. guint64 fuzzy_hashes_checked[RSPAMD_FUZZY_EPOCH_MAX];
  91. /**< amount of check requests for each epoch */
  92. guint64 fuzzy_shingles_checked[RSPAMD_FUZZY_EPOCH_MAX];
  93. /**< amount of shingle check requests for each epoch */
  94. guint64 fuzzy_hashes_found[RSPAMD_FUZZY_EPOCH_MAX];
  95. /**< amount of invalid requests */
  96. guint64 invalid_requests;
  97. /**< amount of delayed hashes found */
  98. guint64 delayed_hashes;
  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. ref_entry_t ref;
  108. };
  109. struct rspamd_fuzzy_mirror {
  110. gchar *name;
  111. struct upstream_list *u;
  112. struct rspamd_cryptobox_pubkey *key;
  113. };
  114. struct rspamd_leaky_bucket_elt {
  115. rspamd_inet_addr_t *addr;
  116. gdouble last;
  117. gdouble cur;
  118. };
  119. static const guint64 rspamd_fuzzy_storage_magic = 0x291a3253eb1b3ea5ULL;
  120. struct rspamd_fuzzy_storage_ctx {
  121. guint64 magic;
  122. /* Events base */
  123. struct ev_loop *event_loop;
  124. /* DNS resolver */
  125. struct rspamd_dns_resolver *resolver;
  126. /* Config */
  127. struct rspamd_config *cfg;
  128. /* END OF COMMON PART */
  129. struct fuzzy_global_stat stat;
  130. gdouble expire;
  131. gdouble sync_timeout;
  132. gdouble delay;
  133. struct rspamd_radix_map_helper *update_ips;
  134. struct rspamd_radix_map_helper *blocked_ips;
  135. struct rspamd_radix_map_helper *ratelimit_whitelist;
  136. struct rspamd_radix_map_helper *delay_whitelist;
  137. const ucl_object_t *update_map;
  138. const ucl_object_t *delay_whitelist_map;
  139. const ucl_object_t *blocked_map;
  140. const ucl_object_t *ratelimit_whitelist_map;
  141. guint keypair_cache_size;
  142. ev_timer stat_ev;
  143. ev_io peer_ev;
  144. /* Local keypair */
  145. struct rspamd_cryptobox_keypair *default_keypair; /* Bad clash, need for parse keypair */
  146. struct fuzzy_key *default_key;
  147. GHashTable *keys;
  148. gboolean encrypted_only;
  149. gboolean read_only;
  150. gboolean dedicated_update_worker;
  151. struct rspamd_keypair_cache *keypair_cache;
  152. struct rspamd_http_context *http_ctx;
  153. rspamd_lru_hash_t *errors_ips;
  154. rspamd_lru_hash_t *ratelimit_buckets;
  155. struct rspamd_fuzzy_backend *backend;
  156. GArray *updates_pending;
  157. guint updates_failed;
  158. guint updates_maxfail;
  159. /* Used to send data between workers */
  160. gint peer_fd;
  161. /* Ratelimits */
  162. guint leaky_bucket_ttl;
  163. guint leaky_bucket_mask;
  164. guint max_buckets;
  165. gboolean ratelimit_log_only;
  166. gdouble leaky_bucket_burst;
  167. gdouble leaky_bucket_rate;
  168. struct rspamd_worker *worker;
  169. const ucl_object_t *skip_map;
  170. struct rspamd_hash_map_helper *skip_hashes;
  171. gint lua_pre_handler_cbref;
  172. gint lua_post_handler_cbref;
  173. guchar cookie[COOKIE_SIZE];
  174. };
  175. enum fuzzy_cmd_type {
  176. CMD_NORMAL,
  177. CMD_SHINGLE,
  178. CMD_ENCRYPTED_NORMAL,
  179. CMD_ENCRYPTED_SHINGLE
  180. };
  181. struct fuzzy_session {
  182. struct rspamd_worker *worker;
  183. rspamd_inet_addr_t *addr;
  184. struct rspamd_fuzzy_storage_ctx *ctx;
  185. struct rspamd_fuzzy_shingle_cmd cmd; /* Can handle both shingles and non-shingles */
  186. struct rspamd_fuzzy_encrypted_reply reply; /* Again: contains everything */
  187. struct fuzzy_key_stat *ip_stat;
  188. enum rspamd_fuzzy_epoch epoch;
  189. enum fuzzy_cmd_type cmd_type;
  190. gint fd;
  191. guint64 time;
  192. struct ev_io io;
  193. ref_entry_t ref;
  194. struct fuzzy_key_stat *key_stat;
  195. struct rspamd_fuzzy_cmd_extension *extensions;
  196. guchar nm[rspamd_cryptobox_MAX_NMBYTES];
  197. };
  198. struct fuzzy_peer_request {
  199. ev_io io_ev;
  200. struct fuzzy_peer_cmd cmd;
  201. };
  202. struct fuzzy_key {
  203. struct rspamd_cryptobox_keypair *key;
  204. struct rspamd_cryptobox_pubkey *pk;
  205. struct fuzzy_key_stat *stat;
  206. };
  207. struct rspamd_updates_cbdata {
  208. GArray *updates_pending;
  209. struct rspamd_fuzzy_storage_ctx *ctx;
  210. gchar *source;
  211. gboolean final;
  212. };
  213. static void rspamd_fuzzy_write_reply (struct fuzzy_session *session);
  214. static gboolean rspamd_fuzzy_process_updates_queue (
  215. struct rspamd_fuzzy_storage_ctx *ctx,
  216. const gchar *source, gboolean final);
  217. static gboolean
  218. rspamd_fuzzy_check_ratelimit (struct fuzzy_session *session)
  219. {
  220. rspamd_inet_addr_t *masked;
  221. struct rspamd_leaky_bucket_elt *elt;
  222. ev_tstamp now;
  223. if (session->ctx->ratelimit_whitelist != NULL) {
  224. if (rspamd_match_radix_map_addr (session->ctx->ratelimit_whitelist,
  225. session->addr) != NULL) {
  226. return TRUE;
  227. }
  228. }
  229. /*
  230. if (rspamd_inet_address_is_local (session->addr, TRUE)) {
  231. return TRUE;
  232. }
  233. */
  234. masked = rspamd_inet_address_copy (session->addr);
  235. if (rspamd_inet_address_get_af (masked) == AF_INET) {
  236. rspamd_inet_address_apply_mask (masked,
  237. MIN (session->ctx->leaky_bucket_mask, 32));
  238. }
  239. else {
  240. /* Must be at least /64 */
  241. rspamd_inet_address_apply_mask (masked,
  242. MIN (MAX (session->ctx->leaky_bucket_mask * 4, 64), 128));
  243. }
  244. now = ev_now (session->ctx->event_loop);
  245. elt = rspamd_lru_hash_lookup (session->ctx->ratelimit_buckets, masked,
  246. now);
  247. if (elt) {
  248. gboolean ratelimited = FALSE;
  249. if (isnan (elt->cur)) {
  250. /* Ratelimit exceeded, preserve it for the whole ttl */
  251. ratelimited = TRUE;
  252. }
  253. else {
  254. /* Update bucket */
  255. if (elt->last < now) {
  256. elt->cur -= session->ctx->leaky_bucket_rate * (now - elt->last);
  257. elt->last = now;
  258. if (elt->cur < 0) {
  259. elt->cur = 0;
  260. }
  261. }
  262. else {
  263. elt->last = now;
  264. }
  265. /* Check bucket */
  266. if (elt->cur >= session->ctx->leaky_bucket_burst) {
  267. msg_info ("ratelimiting %s (%s), %.1f max elts",
  268. rspamd_inet_address_to_string (session->addr),
  269. rspamd_inet_address_to_string (masked),
  270. session->ctx->leaky_bucket_burst);
  271. elt->cur = NAN;
  272. }
  273. else {
  274. elt->cur ++; /* Allow one more request */
  275. }
  276. }
  277. rspamd_inet_address_free (masked);
  278. return !ratelimited;
  279. }
  280. else {
  281. /* New bucket */
  282. elt = g_malloc (sizeof (*elt));
  283. elt->addr = masked; /* transfer ownership */
  284. elt->cur = 1;
  285. elt->last = now;
  286. rspamd_lru_hash_insert (session->ctx->ratelimit_buckets,
  287. masked,
  288. elt,
  289. now,
  290. session->ctx->leaky_bucket_ttl);
  291. }
  292. return TRUE;
  293. }
  294. static gboolean
  295. rspamd_fuzzy_check_client (struct rspamd_fuzzy_storage_ctx *ctx,
  296. rspamd_inet_addr_t *addr)
  297. {
  298. if (ctx->blocked_ips != NULL) {
  299. if (rspamd_match_radix_map_addr (ctx->blocked_ips,
  300. addr) != NULL) {
  301. return FALSE;
  302. }
  303. }
  304. return TRUE;
  305. }
  306. static gboolean
  307. rspamd_fuzzy_check_write (struct fuzzy_session *session)
  308. {
  309. if (session->ctx->read_only) {
  310. return FALSE;
  311. }
  312. if (session->ctx->update_ips != NULL) {
  313. if (rspamd_match_radix_map_addr (session->ctx->update_ips,
  314. session->addr) == NULL) {
  315. return FALSE;
  316. }
  317. else {
  318. return TRUE;
  319. }
  320. }
  321. return FALSE;
  322. }
  323. static void
  324. fuzzy_key_stat_dtor (gpointer p)
  325. {
  326. struct fuzzy_key_stat *st = p;
  327. if (st->last_ips) {
  328. rspamd_lru_hash_destroy (st->last_ips);
  329. }
  330. g_free (st);
  331. }
  332. static void
  333. fuzzy_key_stat_unref (gpointer p)
  334. {
  335. struct fuzzy_key_stat *st = p;
  336. REF_RELEASE (st);
  337. }
  338. static void
  339. fuzzy_key_dtor (gpointer p)
  340. {
  341. struct fuzzy_key *key = p;
  342. if (key) {
  343. if (key->stat) {
  344. REF_RELEASE (key->stat);
  345. }
  346. g_free (key);
  347. }
  348. }
  349. static void
  350. fuzzy_count_callback (guint64 count, void *ud)
  351. {
  352. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  353. ctx->stat.fuzzy_hashes = count;
  354. }
  355. static void
  356. fuzzy_rl_bucket_free (gpointer p)
  357. {
  358. struct rspamd_leaky_bucket_elt *elt = (struct rspamd_leaky_bucket_elt *)p;
  359. rspamd_inet_address_free (elt->addr);
  360. g_free (elt);
  361. }
  362. static void
  363. fuzzy_stat_count_callback (guint64 count, void *ud)
  364. {
  365. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  366. ev_timer_again (ctx->event_loop, &ctx->stat_ev);
  367. ctx->stat.fuzzy_hashes = count;
  368. }
  369. static void
  370. rspamd_fuzzy_stat_callback (EV_P_ ev_timer *w, int revents)
  371. {
  372. struct rspamd_fuzzy_storage_ctx *ctx =
  373. (struct rspamd_fuzzy_storage_ctx *)w->data;
  374. rspamd_fuzzy_backend_count (ctx->backend, fuzzy_stat_count_callback, ctx);
  375. }
  376. static void
  377. fuzzy_update_version_callback (guint64 ver, void *ud)
  378. {
  379. }
  380. static void
  381. rspamd_fuzzy_updates_cb (gboolean success,
  382. guint nadded,
  383. guint ndeleted,
  384. guint nextended,
  385. guint nignored,
  386. void *ud)
  387. {
  388. struct rspamd_updates_cbdata *cbdata = ud;
  389. struct rspamd_fuzzy_storage_ctx *ctx;
  390. const gchar *source;
  391. ctx = cbdata->ctx;
  392. source = cbdata->source;
  393. if (success) {
  394. rspamd_fuzzy_backend_count(ctx->backend, fuzzy_count_callback, ctx);
  395. msg_info ("successfully updated fuzzy storage %s: %d updates in queue; "
  396. "%d pending currently; "
  397. "%d added; %d deleted; %d extended; %d duplicates",
  398. ctx->worker->cf->bind_conf ?
  399. ctx->worker->cf->bind_conf->bind_line :
  400. "unknown",
  401. cbdata->updates_pending->len,
  402. ctx->updates_pending->len,
  403. nadded, ndeleted, nextended, nignored);
  404. rspamd_fuzzy_backend_version(ctx->backend, source,
  405. fuzzy_update_version_callback, NULL);
  406. ctx->updates_failed = 0;
  407. if (cbdata->final || ctx->worker->state != rspamd_worker_state_running) {
  408. /* Plan exit */
  409. ev_break(ctx->event_loop, EVBREAK_ALL);
  410. }
  411. }
  412. else {
  413. if (++ctx->updates_failed > ctx->updates_maxfail) {
  414. msg_err ("cannot commit update transaction to fuzzy backend %s, discard "
  415. "%ud updates after %d retries",
  416. ctx->worker->cf->bind_conf ?
  417. ctx->worker->cf->bind_conf->bind_line :
  418. "unknown",
  419. cbdata->updates_pending->len,
  420. ctx->updates_maxfail);
  421. ctx->updates_failed = 0;
  422. if (cbdata->final || ctx->worker->state != rspamd_worker_state_running) {
  423. /* Plan exit */
  424. ev_break(ctx->event_loop, EVBREAK_ALL);
  425. }
  426. }
  427. else {
  428. if (ctx->updates_pending) {
  429. msg_err ("cannot commit update transaction to fuzzy backend %s; "
  430. "%ud updates are still left; %ud currently pending;"
  431. " %d retries remaining",
  432. ctx->worker->cf->bind_conf ?
  433. ctx->worker->cf->bind_conf->bind_line : "unknown",
  434. cbdata->updates_pending->len,
  435. ctx->updates_pending->len,
  436. ctx->updates_maxfail - ctx->updates_failed);
  437. /* Move the remaining updates to ctx queue */
  438. g_array_append_vals(ctx->updates_pending,
  439. cbdata->updates_pending->data,
  440. cbdata->updates_pending->len);
  441. if (cbdata->final) {
  442. /* Try one more time */
  443. rspamd_fuzzy_process_updates_queue(cbdata->ctx, cbdata->source,
  444. cbdata->final);
  445. }
  446. }
  447. else {
  448. /* We have lost our ctx, so it is a race condition case */
  449. msg_err ("cannot commit update transaction to fuzzy backend %s; "
  450. "%ud updates are still left; no more retries: a worker is terminated",
  451. ctx->worker->cf->bind_conf ?
  452. ctx->worker->cf->bind_conf->bind_line : "unknown",
  453. cbdata->updates_pending->len);
  454. }
  455. }
  456. }
  457. g_array_free (cbdata->updates_pending, TRUE);
  458. g_free (cbdata->source);
  459. g_free (cbdata);
  460. }
  461. static gboolean
  462. rspamd_fuzzy_process_updates_queue (struct rspamd_fuzzy_storage_ctx *ctx,
  463. const gchar *source, gboolean final)
  464. {
  465. struct rspamd_updates_cbdata *cbdata;
  466. if (ctx->updates_pending->len > 0) {
  467. cbdata = g_malloc (sizeof (*cbdata));
  468. cbdata->ctx = ctx;
  469. cbdata->final = final;
  470. cbdata->updates_pending = ctx->updates_pending;
  471. ctx->updates_pending = g_array_sized_new (FALSE, FALSE,
  472. sizeof (struct fuzzy_peer_cmd),
  473. MAX (cbdata->updates_pending->len, 1024));
  474. cbdata->source = g_strdup (source);
  475. rspamd_fuzzy_backend_process_updates (ctx->backend,
  476. cbdata->updates_pending,
  477. source, rspamd_fuzzy_updates_cb, cbdata);
  478. return TRUE;
  479. }
  480. else if (final) {
  481. /* No need to sync */
  482. ev_break (ctx->event_loop, EVBREAK_ALL);
  483. }
  484. return FALSE;
  485. }
  486. static void
  487. rspamd_fuzzy_reply_io (EV_P_ ev_io *w, int revents)
  488. {
  489. struct fuzzy_session *session = (struct fuzzy_session *)w->data;
  490. ev_io_stop (EV_A_ w);
  491. rspamd_fuzzy_write_reply (session);
  492. REF_RELEASE (session);
  493. }
  494. static void
  495. rspamd_fuzzy_write_reply (struct fuzzy_session *session)
  496. {
  497. gssize r;
  498. gsize len;
  499. gconstpointer data;
  500. if (session->cmd_type == CMD_ENCRYPTED_NORMAL ||
  501. session->cmd_type == CMD_ENCRYPTED_SHINGLE) {
  502. /* Encrypted reply */
  503. data = &session->reply;
  504. if (session->epoch > RSPAMD_FUZZY_EPOCH10) {
  505. len = sizeof (session->reply);
  506. }
  507. else {
  508. len = sizeof (session->reply.hdr) + sizeof (session->reply.rep.v1);
  509. }
  510. }
  511. else {
  512. data = &session->reply.rep;
  513. if (session->epoch > RSPAMD_FUZZY_EPOCH10) {
  514. len = sizeof (session->reply.rep);
  515. }
  516. else {
  517. len = sizeof (session->reply.rep.v1);
  518. }
  519. }
  520. r = rspamd_inet_address_sendto (session->fd, data, len, 0,
  521. session->addr);
  522. if (r == -1) {
  523. if (errno == EINTR || errno == EWOULDBLOCK || errno == EAGAIN) {
  524. /* Grab reference to avoid early destruction */
  525. REF_RETAIN (session);
  526. session->io.data = session;
  527. ev_io_init (&session->io,
  528. rspamd_fuzzy_reply_io, session->fd, EV_WRITE);
  529. ev_io_start (session->ctx->event_loop, &session->io);
  530. }
  531. else {
  532. msg_err ("error while writing reply: %s", strerror (errno));
  533. }
  534. }
  535. }
  536. static void
  537. rspamd_fuzzy_update_stats (struct rspamd_fuzzy_storage_ctx *ctx,
  538. enum rspamd_fuzzy_epoch epoch,
  539. gboolean matched,
  540. gboolean is_shingle,
  541. gboolean is_delayed,
  542. struct fuzzy_key_stat *key_stat,
  543. struct fuzzy_key_stat *ip_stat,
  544. guint cmd, guint reply)
  545. {
  546. ctx->stat.fuzzy_hashes_checked[epoch] ++;
  547. if (matched) {
  548. ctx->stat.fuzzy_hashes_found[epoch]++;
  549. }
  550. if (is_shingle) {
  551. ctx->stat.fuzzy_shingles_checked[epoch]++;
  552. }
  553. if (is_delayed) {
  554. ctx->stat.delayed_hashes ++;
  555. }
  556. if (key_stat) {
  557. if (!matched && reply != 0) {
  558. key_stat->errors ++;
  559. }
  560. else {
  561. if (cmd == FUZZY_CHECK) {
  562. key_stat->checked++;
  563. if (matched) {
  564. key_stat->matched ++;
  565. }
  566. }
  567. else if (cmd == FUZZY_WRITE) {
  568. key_stat->added++;
  569. }
  570. else if (cmd == FUZZY_DEL) {
  571. key_stat->deleted++;
  572. }
  573. }
  574. }
  575. if (ip_stat) {
  576. if (!matched && reply != 0) {
  577. ip_stat->errors++;
  578. }
  579. else {
  580. if (cmd == FUZZY_CHECK) {
  581. ip_stat->checked++;
  582. if (matched) {
  583. ip_stat->matched++;
  584. }
  585. }
  586. else if (cmd == FUZZY_WRITE) {
  587. ip_stat->added++;
  588. }
  589. else if (cmd == FUZZY_DEL) {
  590. ip_stat->deleted++;
  591. }
  592. }
  593. }
  594. }
  595. enum rspamd_fuzzy_reply_flags {
  596. RSPAMD_FUZZY_REPLY_ENCRYPTED = 0x1u << 0u,
  597. RSPAMD_FUZZY_REPLY_SHINGLE = 0x1u << 1u,
  598. RSPAMD_FUZZY_REPLY_DELAY = 0x1u << 2u,
  599. };
  600. static void
  601. rspamd_fuzzy_make_reply (struct rspamd_fuzzy_cmd *cmd,
  602. struct rspamd_fuzzy_reply *result,
  603. struct fuzzy_session *session,
  604. gint flags)
  605. {
  606. gsize len;
  607. if (cmd) {
  608. result->v1.tag = cmd->tag;
  609. memcpy (&session->reply.rep, result, sizeof (*result));
  610. rspamd_fuzzy_update_stats (session->ctx,
  611. session->epoch,
  612. result->v1.prob > 0.5,
  613. flags & RSPAMD_FUZZY_REPLY_SHINGLE,
  614. flags & RSPAMD_FUZZY_REPLY_DELAY,
  615. session->key_stat,
  616. session->ip_stat,
  617. cmd->cmd,
  618. result->v1.value);
  619. if (flags & RSPAMD_FUZZY_REPLY_DELAY) {
  620. /* Hash is too fresh, need to delay it */
  621. session->reply.rep.ts = 0;
  622. session->reply.rep.v1.prob = 0.0;
  623. session->reply.rep.v1.value = 0;
  624. }
  625. if (flags & RSPAMD_FUZZY_REPLY_ENCRYPTED) {
  626. /* We need also to encrypt reply */
  627. ottery_rand_bytes (session->reply.hdr.nonce,
  628. sizeof (session->reply.hdr.nonce));
  629. /*
  630. * For old replies we need to encrypt just old part, otherwise
  631. * decryption would fail due to mac verification mistake
  632. */
  633. if (session->epoch > RSPAMD_FUZZY_EPOCH10) {
  634. len = sizeof (session->reply.rep);
  635. }
  636. else {
  637. len = sizeof (session->reply.rep.v1);
  638. }
  639. rspamd_cryptobox_encrypt_nm_inplace ((guchar *)&session->reply.rep,
  640. len,
  641. session->reply.hdr.nonce,
  642. session->nm,
  643. session->reply.hdr.mac,
  644. RSPAMD_CRYPTOBOX_MODE_25519);
  645. }
  646. }
  647. rspamd_fuzzy_write_reply (session);
  648. }
  649. static gboolean
  650. fuzzy_peer_try_send (gint fd, struct fuzzy_peer_request *up_req)
  651. {
  652. gssize r;
  653. r = write (fd, &up_req->cmd, sizeof (up_req->cmd));
  654. if (r != sizeof (up_req->cmd)) {
  655. return FALSE;
  656. }
  657. return TRUE;
  658. }
  659. static void
  660. fuzzy_peer_send_io (EV_P_ ev_io *w, int revents)
  661. {
  662. struct fuzzy_peer_request *up_req = (struct fuzzy_peer_request *)w->data;
  663. if (!fuzzy_peer_try_send (w->fd, up_req)) {
  664. msg_err ("cannot send update request to the peer: %s", strerror (errno));
  665. }
  666. ev_io_stop (EV_A_ w);
  667. g_free (up_req);
  668. }
  669. static void
  670. rspamd_fuzzy_extensions_tolua (lua_State *L,
  671. struct fuzzy_session *session)
  672. {
  673. struct rspamd_fuzzy_cmd_extension *ext;
  674. rspamd_inet_addr_t *addr;
  675. lua_createtable (L, 0, 0);
  676. LL_FOREACH (session->extensions, ext) {
  677. switch (ext->ext) {
  678. case RSPAMD_FUZZY_EXT_SOURCE_DOMAIN:
  679. lua_pushlstring (L, ext->payload, ext->length);
  680. lua_setfield (L, -2, "domain");
  681. break;
  682. case RSPAMD_FUZZY_EXT_SOURCE_IP4:
  683. addr = rspamd_inet_address_new (AF_INET, ext->payload);
  684. rspamd_lua_ip_push (L, addr);
  685. rspamd_inet_address_free (addr);
  686. lua_setfield (L, -2, "ip");
  687. break;
  688. case RSPAMD_FUZZY_EXT_SOURCE_IP6:
  689. addr = rspamd_inet_address_new (AF_INET6, ext->payload);
  690. rspamd_lua_ip_push (L, addr);
  691. rspamd_inet_address_free (addr);
  692. lua_setfield (L, -2, "ip");
  693. break;
  694. }
  695. }
  696. }
  697. static void
  698. rspamd_fuzzy_check_callback (struct rspamd_fuzzy_reply *result, void *ud)
  699. {
  700. struct fuzzy_session *session = ud;
  701. gboolean is_shingle = FALSE, __attribute__ ((unused)) encrypted = FALSE;
  702. struct rspamd_fuzzy_cmd *cmd = NULL;
  703. const struct rspamd_shingle *shingle = NULL;
  704. struct rspamd_shingle sgl_cpy;
  705. gint send_flags = 0;
  706. switch (session->cmd_type) {
  707. case CMD_ENCRYPTED_NORMAL:
  708. encrypted = TRUE;
  709. send_flags |= RSPAMD_FUZZY_REPLY_ENCRYPTED;
  710. /* Fallthrough */
  711. case CMD_NORMAL:
  712. cmd = &session->cmd.basic;
  713. break;
  714. case CMD_ENCRYPTED_SHINGLE:
  715. encrypted = TRUE;
  716. send_flags |= RSPAMD_FUZZY_REPLY_ENCRYPTED;
  717. /* Fallthrough */
  718. case CMD_SHINGLE:
  719. cmd = &session->cmd.basic;
  720. memcpy (&sgl_cpy, &session->cmd.sgl, sizeof (sgl_cpy));
  721. shingle = &sgl_cpy;
  722. is_shingle = TRUE;
  723. send_flags |= RSPAMD_FUZZY_REPLY_SHINGLE;
  724. break;
  725. }
  726. if (session->ctx->lua_post_handler_cbref != -1) {
  727. /* Start lua post handler */
  728. lua_State *L = session->ctx->cfg->lua_state;
  729. gint err_idx, ret;
  730. lua_pushcfunction (L, &rspamd_lua_traceback);
  731. err_idx = lua_gettop (L);
  732. /* Preallocate stack (small opt) */
  733. lua_checkstack (L, err_idx + 9);
  734. /* function */
  735. lua_rawgeti (L, LUA_REGISTRYINDEX, session->ctx->lua_post_handler_cbref);
  736. /* client IP */
  737. rspamd_lua_ip_push (L, session->addr);
  738. /* client command */
  739. lua_pushinteger (L, cmd->cmd);
  740. /* command value (push as rspamd_text) */
  741. (void)lua_new_text (L, result->digest, sizeof (result->digest), FALSE);
  742. /* is shingle */
  743. lua_pushboolean (L, is_shingle);
  744. /* result value */
  745. lua_pushinteger (L, result->v1.value);
  746. /* result probability */
  747. lua_pushnumber (L, result->v1.prob);
  748. /* result flag */
  749. lua_pushinteger (L, result->v1.flag);
  750. /* result timestamp */
  751. lua_pushinteger (L, result->ts);
  752. /* TODO: add additional data maybe (encryption, pubkey, etc) */
  753. rspamd_fuzzy_extensions_tolua (L, session);
  754. if ((ret = lua_pcall (L, 9, LUA_MULTRET, err_idx)) != 0) {
  755. msg_err ("call to lua_post_handler lua "
  756. "script failed (%d): %s", ret, lua_tostring (L, -1));
  757. }
  758. else {
  759. /* Return values order:
  760. * the first reply will be on err_idx + 1
  761. * if it is true, then we need to read the former ones:
  762. * 2-nd will be reply code
  763. * 3-rd will be probability (or 0.0 if missing)
  764. * 4-th value is flag (or default flag if missing)
  765. */
  766. ret = lua_toboolean (L, err_idx + 1);
  767. if (ret) {
  768. /* Artificial reply */
  769. result->v1.value = lua_tointeger (L, err_idx + 2);
  770. if (lua_isnumber (L, err_idx + 3)) {
  771. result->v1.prob = lua_tonumber (L, err_idx + 3);
  772. }
  773. else {
  774. result->v1.prob = 0.0f;
  775. }
  776. if (lua_isnumber (L, err_idx + 4)) {
  777. result->v1.flag = lua_tointeger (L, err_idx + 4);
  778. }
  779. lua_settop (L, 0);
  780. rspamd_fuzzy_make_reply (cmd, result, session, send_flags);
  781. REF_RELEASE (session);
  782. return;
  783. }
  784. }
  785. lua_settop (L, 0);
  786. }
  787. if (!isnan (session->ctx->delay) &&
  788. rspamd_match_radix_map_addr (session->ctx->delay_whitelist,
  789. session->addr) == NULL) {
  790. gdouble hash_age = rspamd_get_calendar_ticks () - result->ts;
  791. gdouble jittered_age = rspamd_time_jitter (session->ctx->delay,
  792. session->ctx->delay / 2.0);
  793. if (hash_age < jittered_age) {
  794. send_flags |= RSPAMD_FUZZY_REPLY_DELAY;
  795. }
  796. }
  797. /* Refresh hash if found with strong confidence */
  798. if (result->v1.prob > 0.9 && !session->ctx->read_only) {
  799. struct fuzzy_peer_cmd up_cmd;
  800. struct fuzzy_peer_request *up_req;
  801. if (session->worker->index == 0) {
  802. /* Just add to the queue */
  803. memset (&up_cmd, 0, sizeof (up_cmd));
  804. up_cmd.is_shingle = is_shingle;
  805. memcpy (up_cmd.cmd.normal.digest, result->digest,
  806. sizeof (up_cmd.cmd.normal.digest));
  807. up_cmd.cmd.normal.flag = result->v1.flag;
  808. up_cmd.cmd.normal.cmd = FUZZY_REFRESH;
  809. up_cmd.cmd.normal.shingles_count = cmd->shingles_count;
  810. if (is_shingle && shingle) {
  811. memcpy (&up_cmd.cmd.shingle.sgl, shingle,
  812. sizeof (up_cmd.cmd.shingle.sgl));
  813. }
  814. g_array_append_val (session->ctx->updates_pending, up_cmd);
  815. }
  816. else {
  817. /* We need to send request to the peer */
  818. up_req = g_malloc0 (sizeof (*up_req));
  819. up_req->cmd.is_shingle = is_shingle;
  820. memcpy (up_req->cmd.cmd.normal.digest, result->digest,
  821. sizeof (up_req->cmd.cmd.normal.digest));
  822. up_req->cmd.cmd.normal.flag = result->v1.flag;
  823. up_req->cmd.cmd.normal.cmd = FUZZY_REFRESH;
  824. up_req->cmd.cmd.normal.shingles_count = cmd->shingles_count;
  825. if (is_shingle && shingle) {
  826. memcpy (&up_req->cmd.cmd.shingle.sgl, shingle,
  827. sizeof (up_req->cmd.cmd.shingle.sgl));
  828. }
  829. if (!fuzzy_peer_try_send (session->ctx->peer_fd, up_req)) {
  830. up_req->io_ev.data = up_req;
  831. ev_io_init (&up_req->io_ev, fuzzy_peer_send_io,
  832. session->ctx->peer_fd, EV_WRITE);
  833. ev_io_start (session->ctx->event_loop, &up_req->io_ev);
  834. }
  835. else {
  836. g_free (up_req);
  837. }
  838. }
  839. }
  840. rspamd_fuzzy_make_reply (cmd, result, session, send_flags);
  841. REF_RELEASE (session);
  842. }
  843. static void
  844. rspamd_fuzzy_process_command (struct fuzzy_session *session)
  845. {
  846. gboolean is_shingle = FALSE, __attribute__ ((unused)) encrypted = FALSE;
  847. struct rspamd_fuzzy_cmd *cmd = NULL;
  848. struct rspamd_fuzzy_reply result;
  849. struct fuzzy_peer_cmd up_cmd;
  850. struct fuzzy_peer_request *up_req;
  851. struct fuzzy_key_stat *ip_stat = NULL;
  852. gchar hexbuf[rspamd_cryptobox_HASHBYTES * 2 + 1];
  853. rspamd_inet_addr_t *naddr;
  854. gpointer ptr;
  855. gsize up_len = 0;
  856. gint send_flags = 0;
  857. cmd = &session->cmd.basic;
  858. switch (session->cmd_type) {
  859. case CMD_NORMAL:
  860. up_len = sizeof (session->cmd.basic);
  861. break;
  862. case CMD_SHINGLE:
  863. up_len = sizeof (session->cmd);
  864. is_shingle = TRUE;
  865. send_flags |= RSPAMD_FUZZY_REPLY_SHINGLE;
  866. break;
  867. case CMD_ENCRYPTED_NORMAL:
  868. up_len = sizeof (session->cmd.basic);
  869. encrypted = TRUE;
  870. send_flags |= RSPAMD_FUZZY_REPLY_ENCRYPTED;
  871. break;
  872. case CMD_ENCRYPTED_SHINGLE:
  873. up_len = sizeof (session->cmd);
  874. encrypted = TRUE;
  875. is_shingle = TRUE;
  876. send_flags |= RSPAMD_FUZZY_REPLY_SHINGLE|RSPAMD_FUZZY_REPLY_ENCRYPTED;
  877. break;
  878. default:
  879. msg_err ("invalid command type: %d", session->cmd_type);
  880. return;
  881. }
  882. memset (&result, 0, sizeof (result));
  883. memcpy (result.digest, cmd->digest, sizeof (result.digest));
  884. result.v1.flag = cmd->flag;
  885. result.v1.tag = cmd->tag;
  886. if (session->ctx->lua_pre_handler_cbref != -1) {
  887. /* Start lua pre handler */
  888. lua_State *L = session->ctx->cfg->lua_state;
  889. gint err_idx, ret;
  890. lua_pushcfunction (L, &rspamd_lua_traceback);
  891. err_idx = lua_gettop (L);
  892. /* Preallocate stack (small opt) */
  893. lua_checkstack (L, err_idx + 5);
  894. /* function */
  895. lua_rawgeti (L, LUA_REGISTRYINDEX, session->ctx->lua_pre_handler_cbref);
  896. /* client IP */
  897. rspamd_lua_ip_push (L, session->addr);
  898. /* client command */
  899. lua_pushinteger (L, cmd->cmd);
  900. /* command value (push as rspamd_text) */
  901. (void)lua_new_text (L, cmd->digest, sizeof (cmd->digest), FALSE);
  902. /* is shingle */
  903. lua_pushboolean (L, is_shingle);
  904. /* TODO: add additional data maybe (encryption, pubkey, etc) */
  905. rspamd_fuzzy_extensions_tolua (L, session);
  906. if ((ret = lua_pcall (L, 5, LUA_MULTRET, err_idx)) != 0) {
  907. msg_err ("call to lua_pre_handler lua "
  908. "script failed (%d): %s", ret, lua_tostring (L, -1));
  909. }
  910. else {
  911. /* Return values order:
  912. * the first reply will be on err_idx + 1
  913. * if it is true, then we need to read the former ones:
  914. * 2-nd will be reply code
  915. * 3-rd will be probability (or 0.0 if missing)
  916. */
  917. ret = lua_toboolean (L, err_idx + 1);
  918. if (ret) {
  919. /* Artificial reply */
  920. result.v1.value = lua_tointeger (L, err_idx + 2);
  921. if (lua_isnumber (L, err_idx + 3)) {
  922. result.v1.prob = lua_tonumber (L, err_idx + 3);
  923. }
  924. else {
  925. result.v1.prob = 0.0f;
  926. }
  927. lua_settop (L, 0);
  928. rspamd_fuzzy_make_reply (cmd, &result, session, send_flags);
  929. return;
  930. }
  931. }
  932. lua_settop (L, 0);
  933. }
  934. if (G_UNLIKELY (cmd == NULL || up_len == 0)) {
  935. result.v1.value = 500;
  936. result.v1.prob = 0.0f;
  937. rspamd_fuzzy_make_reply (cmd, &result, session, send_flags);
  938. return;
  939. }
  940. if (session->ctx->encrypted_only && !encrypted) {
  941. /* Do not accept unencrypted commands */
  942. result.v1.value = 403;
  943. result.v1.prob = 0.0f;
  944. rspamd_fuzzy_make_reply (cmd, &result, session, send_flags);
  945. return;
  946. }
  947. if (session->key_stat) {
  948. ip_stat = rspamd_lru_hash_lookup (session->key_stat->last_ips,
  949. session->addr, -1);
  950. if (ip_stat == NULL) {
  951. naddr = rspamd_inet_address_copy (session->addr);
  952. ip_stat = g_malloc0 (sizeof (*ip_stat));
  953. REF_INIT_RETAIN (ip_stat, fuzzy_key_stat_dtor);
  954. rspamd_lru_hash_insert (session->key_stat->last_ips,
  955. naddr, ip_stat, -1, 0);
  956. }
  957. REF_RETAIN (ip_stat);
  958. session->ip_stat = ip_stat;
  959. }
  960. if (cmd->cmd == FUZZY_CHECK) {
  961. bool can_continue = true;
  962. if (session->ctx->ratelimit_buckets) {
  963. if (session->ctx->ratelimit_log_only) {
  964. (void)rspamd_fuzzy_check_ratelimit (session); /* Check but ignore */
  965. }
  966. else {
  967. can_continue = rspamd_fuzzy_check_ratelimit (session);
  968. }
  969. }
  970. if (can_continue) {
  971. REF_RETAIN (session);
  972. rspamd_fuzzy_backend_check (session->ctx->backend, cmd,
  973. rspamd_fuzzy_check_callback, session);
  974. }
  975. else {
  976. result.v1.value = 403;
  977. result.v1.prob = 0.0f;
  978. result.v1.flag = 0;
  979. rspamd_fuzzy_make_reply (cmd, &result, session, send_flags);
  980. }
  981. }
  982. else if (cmd->cmd == FUZZY_STAT) {
  983. result.v1.prob = 1.0f;
  984. result.v1.value = 0;
  985. result.v1.flag = session->ctx->stat.fuzzy_hashes;
  986. rspamd_fuzzy_make_reply (cmd, &result, session, send_flags);
  987. }
  988. else {
  989. if (rspamd_fuzzy_check_write (session)) {
  990. /* Check whitelist */
  991. if (session->ctx->skip_hashes && cmd->cmd == FUZZY_WRITE) {
  992. rspamd_encode_hex_buf (cmd->digest, sizeof (cmd->digest),
  993. hexbuf, sizeof (hexbuf) - 1);
  994. hexbuf[sizeof (hexbuf) - 1] = '\0';
  995. if (rspamd_match_hash_map (session->ctx->skip_hashes,
  996. hexbuf, sizeof (hexbuf) - 1)) {
  997. result.v1.value = 401;
  998. result.v1.prob = 0.0f;
  999. goto reply;
  1000. }
  1001. }
  1002. if (session->worker->index == 0 || session->ctx->peer_fd == -1) {
  1003. /* Just add to the queue */
  1004. up_cmd.is_shingle = is_shingle;
  1005. ptr = is_shingle ?
  1006. (gpointer)&up_cmd.cmd.shingle :
  1007. (gpointer)&up_cmd.cmd.normal;
  1008. memcpy (ptr, cmd, up_len);
  1009. g_array_append_val (session->ctx->updates_pending, up_cmd);
  1010. }
  1011. else {
  1012. /* We need to send request to the peer */
  1013. up_req = g_malloc0 (sizeof (*up_req));
  1014. up_req->cmd.is_shingle = is_shingle;
  1015. ptr = is_shingle ?
  1016. (gpointer)&up_req->cmd.cmd.shingle :
  1017. (gpointer)&up_req->cmd.cmd.normal;
  1018. memcpy (ptr, cmd, up_len);
  1019. if (!fuzzy_peer_try_send (session->ctx->peer_fd, up_req)) {
  1020. up_req->io_ev.data = up_req;
  1021. ev_io_init (&up_req->io_ev, fuzzy_peer_send_io,
  1022. session->ctx->peer_fd, EV_WRITE);
  1023. ev_io_start (session->ctx->event_loop, &up_req->io_ev);
  1024. }
  1025. else {
  1026. g_free (up_req);
  1027. }
  1028. }
  1029. result.v1.value = 0;
  1030. result.v1.prob = 1.0f;
  1031. }
  1032. else {
  1033. result.v1.value = 403;
  1034. result.v1.prob = 0.0f;
  1035. }
  1036. reply:
  1037. rspamd_fuzzy_make_reply (cmd, &result, session, send_flags);
  1038. }
  1039. }
  1040. static enum rspamd_fuzzy_epoch
  1041. rspamd_fuzzy_command_valid (struct rspamd_fuzzy_cmd *cmd, gint r)
  1042. {
  1043. enum rspamd_fuzzy_epoch ret = RSPAMD_FUZZY_EPOCH_MAX;
  1044. switch (cmd->version) {
  1045. case 4:
  1046. if (cmd->shingles_count > 0) {
  1047. if (r >= sizeof (struct rspamd_fuzzy_shingle_cmd)) {
  1048. ret = RSPAMD_FUZZY_EPOCH11;
  1049. }
  1050. }
  1051. else {
  1052. if (r >= sizeof (*cmd)) {
  1053. ret = RSPAMD_FUZZY_EPOCH11;
  1054. }
  1055. }
  1056. break;
  1057. case 3:
  1058. if (cmd->shingles_count > 0) {
  1059. if (r == sizeof (struct rspamd_fuzzy_shingle_cmd)) {
  1060. ret = RSPAMD_FUZZY_EPOCH10;
  1061. }
  1062. }
  1063. else {
  1064. if (r == sizeof (*cmd)) {
  1065. ret = RSPAMD_FUZZY_EPOCH10;
  1066. }
  1067. }
  1068. break;
  1069. case 2:
  1070. /*
  1071. * rspamd 0.8 has slightly different tokenizer then it might be not
  1072. * 100% compatible
  1073. */
  1074. if (cmd->shingles_count > 0) {
  1075. if (r == sizeof (struct rspamd_fuzzy_shingle_cmd)) {
  1076. ret = RSPAMD_FUZZY_EPOCH8;
  1077. }
  1078. }
  1079. else {
  1080. ret = RSPAMD_FUZZY_EPOCH8;
  1081. }
  1082. break;
  1083. default:
  1084. break;
  1085. }
  1086. return ret;
  1087. }
  1088. static gboolean
  1089. rspamd_fuzzy_decrypt_command (struct fuzzy_session *s, guchar *buf, gsize buflen)
  1090. {
  1091. struct rspamd_fuzzy_encrypted_req_hdr hdr;
  1092. struct rspamd_cryptobox_pubkey *rk;
  1093. struct fuzzy_key *key;
  1094. if (s->ctx->default_key == NULL) {
  1095. msg_warn ("received encrypted request when encryption is not enabled");
  1096. return FALSE;
  1097. }
  1098. if (buflen < sizeof (hdr)) {
  1099. msg_warn ("XXX: should not be reached");
  1100. return FALSE;
  1101. }
  1102. memcpy (&hdr, buf, sizeof (hdr));
  1103. buf += sizeof (hdr);
  1104. buflen -= sizeof (hdr);
  1105. /* Try to find the desired key */
  1106. key = g_hash_table_lookup (s->ctx->keys, hdr.key_id);
  1107. if (key == NULL) {
  1108. /* Unknown key, assume default one */
  1109. key = s->ctx->default_key;
  1110. }
  1111. s->key_stat = key->stat;
  1112. /* Now process keypair */
  1113. rk = rspamd_pubkey_from_bin (hdr.pubkey, sizeof (hdr.pubkey),
  1114. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  1115. if (rk == NULL) {
  1116. msg_err ("bad key; ip=%s",
  1117. rspamd_inet_address_to_string (s->addr));
  1118. return FALSE;
  1119. }
  1120. rspamd_keypair_cache_process (s->ctx->keypair_cache, key->key, rk);
  1121. /* Now decrypt request */
  1122. if (!rspamd_cryptobox_decrypt_nm_inplace (buf, buflen, hdr.nonce,
  1123. rspamd_pubkey_get_nm (rk, key->key),
  1124. hdr.mac, RSPAMD_CRYPTOBOX_MODE_25519)) {
  1125. msg_err ("decryption failed; ip=%s",
  1126. rspamd_inet_address_to_string (s->addr));
  1127. rspamd_pubkey_unref (rk);
  1128. return FALSE;
  1129. }
  1130. memcpy (s->nm, rspamd_pubkey_get_nm (rk, key->key), sizeof (s->nm));
  1131. rspamd_pubkey_unref (rk);
  1132. return TRUE;
  1133. }
  1134. static gboolean
  1135. rspamd_fuzzy_extensions_from_wire (struct fuzzy_session *s, guchar *buf, gsize buflen)
  1136. {
  1137. struct rspamd_fuzzy_cmd_extension *ext, *prev_ext;
  1138. guchar *storage, *p = buf, *end = buf + buflen;
  1139. gsize st_len = 0, n_ext = 0;
  1140. /* Read number of extensions to allocate array */
  1141. while (p < end) {
  1142. guchar cmd = *p++;
  1143. if (p < end) {
  1144. if (cmd == RSPAMD_FUZZY_EXT_SOURCE_DOMAIN) {
  1145. /* Next byte is buf length */
  1146. guchar dom_len = *p++;
  1147. if (dom_len <= (end - p)) {
  1148. st_len += dom_len;
  1149. n_ext ++;
  1150. }
  1151. else {
  1152. /* Truncation */
  1153. return FALSE;
  1154. }
  1155. p += dom_len;
  1156. }
  1157. else if (cmd == RSPAMD_FUZZY_EXT_SOURCE_IP4) {
  1158. if (end - p >= sizeof (in_addr_t)) {
  1159. n_ext ++;
  1160. st_len += sizeof (in_addr_t);
  1161. }
  1162. else {
  1163. /* Truncation */
  1164. return FALSE;
  1165. }
  1166. p += sizeof (in_addr_t);
  1167. }
  1168. else if (cmd == RSPAMD_FUZZY_EXT_SOURCE_IP6) {
  1169. if (end - p >= sizeof (struct in6_addr)) {
  1170. n_ext ++;
  1171. st_len += sizeof (struct in6_addr);
  1172. }
  1173. else {
  1174. /* Truncation */
  1175. return FALSE;
  1176. }
  1177. p += sizeof (struct in6_addr);
  1178. }
  1179. else {
  1180. /* Invalid command */
  1181. return FALSE;
  1182. }
  1183. }
  1184. else {
  1185. /* Truncated extension */
  1186. return FALSE;
  1187. }
  1188. }
  1189. if (n_ext > 0) {
  1190. p = buf;
  1191. /*
  1192. * Memory layout: n_ext of struct rspamd_fuzzy_cmd_extension
  1193. * payload for each extension in a continuous data segment
  1194. */
  1195. storage = g_malloc (n_ext * sizeof (struct rspamd_fuzzy_cmd_extension) +
  1196. st_len);
  1197. guchar *data_buf = storage +
  1198. n_ext * sizeof (struct rspamd_fuzzy_cmd_extension);
  1199. ext = (struct rspamd_fuzzy_cmd_extension *)storage;
  1200. /* All validation has been done, so we can just go further */
  1201. while (p < end) {
  1202. prev_ext = ext;
  1203. guchar cmd = *p++;
  1204. if (cmd == RSPAMD_FUZZY_EXT_SOURCE_DOMAIN) {
  1205. /* Next byte is buf length */
  1206. guchar dom_len = *p++;
  1207. guchar *dest = data_buf;
  1208. ext->ext = RSPAMD_FUZZY_EXT_SOURCE_DOMAIN;
  1209. ext->next = ext + 1;
  1210. ext->length = dom_len;
  1211. ext->payload = dest;
  1212. memcpy (dest, p, dom_len);
  1213. p += dom_len;
  1214. data_buf += dom_len;
  1215. ext = ext->next;
  1216. }
  1217. else if (cmd == RSPAMD_FUZZY_EXT_SOURCE_IP4) {
  1218. guchar *dest = data_buf;
  1219. ext->ext = RSPAMD_FUZZY_EXT_SOURCE_IP4;
  1220. ext->next = ext + 1;
  1221. ext->length = sizeof (in_addr_t);
  1222. ext->payload = dest;
  1223. memcpy (dest, p, sizeof (in_addr_t));
  1224. p += sizeof (in_addr_t);
  1225. data_buf += sizeof (in_addr_t);
  1226. ext = ext->next;
  1227. }
  1228. else if (cmd == RSPAMD_FUZZY_EXT_SOURCE_IP6) {
  1229. guchar *dest = data_buf;
  1230. ext->ext = RSPAMD_FUZZY_EXT_SOURCE_IP6;
  1231. ext->next = ext + 1;
  1232. ext->length = sizeof (struct in6_addr);
  1233. ext->payload = dest;
  1234. memcpy (dest, p, sizeof (struct in6_addr));
  1235. p += sizeof (struct in6_addr);
  1236. data_buf += sizeof (struct in6_addr);
  1237. ext = ext->next;
  1238. }
  1239. else {
  1240. g_assert_not_reached ();
  1241. }
  1242. }
  1243. /* Last next should be NULL */
  1244. prev_ext->next = NULL;
  1245. /* Rewind to the begin */
  1246. ext = (struct rspamd_fuzzy_cmd_extension *)storage;
  1247. s->extensions = ext;
  1248. }
  1249. return TRUE;
  1250. }
  1251. static gboolean
  1252. rspamd_fuzzy_cmd_from_wire (guchar *buf, guint buflen, struct fuzzy_session *s)
  1253. {
  1254. enum rspamd_fuzzy_epoch epoch;
  1255. gboolean encrypted = FALSE;
  1256. if (buflen < sizeof (struct rspamd_fuzzy_cmd)) {
  1257. msg_debug ("truncated fuzzy command of size %d received", buflen);
  1258. return FALSE;
  1259. }
  1260. /* Now check encryption */
  1261. if (buflen >= sizeof (struct rspamd_fuzzy_encrypted_cmd)) {
  1262. if (memcmp (buf, fuzzy_encrypted_magic, sizeof (fuzzy_encrypted_magic)) == 0) {
  1263. /* Encrypted command */
  1264. encrypted = TRUE;
  1265. }
  1266. }
  1267. if (encrypted) {
  1268. /* Decrypt first */
  1269. if (!rspamd_fuzzy_decrypt_command (s, buf, buflen)) {
  1270. return FALSE;
  1271. }
  1272. else {
  1273. /*
  1274. * Advance buffer to skip encrypted header.
  1275. * Note that after rspamd_fuzzy_decrypt_command buf is unencrypted
  1276. */
  1277. buf += sizeof (struct rspamd_fuzzy_encrypted_req_hdr);
  1278. buflen -= sizeof (struct rspamd_fuzzy_encrypted_req_hdr);
  1279. }
  1280. }
  1281. /* Fill the normal command */
  1282. if (buflen < sizeof (s->cmd.basic)) {
  1283. msg_debug ("truncated normal fuzzy command of size %d received", buflen);
  1284. return FALSE;
  1285. }
  1286. memcpy (&s->cmd.basic, buf, sizeof (s->cmd.basic));
  1287. epoch = rspamd_fuzzy_command_valid (&s->cmd.basic, buflen);
  1288. if (epoch == RSPAMD_FUZZY_EPOCH_MAX) {
  1289. msg_debug ("invalid fuzzy command of size %d received", buflen);
  1290. return FALSE;
  1291. }
  1292. s->epoch = epoch;
  1293. /* Advance buf */
  1294. buf += sizeof (s->cmd.basic);
  1295. buflen -= sizeof (s->cmd.basic);
  1296. if (s->cmd.basic.shingles_count > 0) {
  1297. if (buflen >= sizeof (s->cmd.sgl)) {
  1298. /* Copy the shingles part */
  1299. memcpy (&s->cmd.sgl, buf, sizeof (s->cmd.sgl));
  1300. }
  1301. else {
  1302. /* Truncated stuff */
  1303. msg_debug ("truncated fuzzy shingles command of size %d received", buflen);
  1304. return FALSE;
  1305. }
  1306. buf += sizeof (s->cmd.sgl);
  1307. buflen -= sizeof (s->cmd.sgl);
  1308. if (encrypted) {
  1309. s->cmd_type = CMD_ENCRYPTED_SHINGLE;
  1310. }
  1311. else {
  1312. s->cmd_type = CMD_SHINGLE;
  1313. }
  1314. }
  1315. else {
  1316. if (encrypted) {
  1317. s->cmd_type = CMD_ENCRYPTED_NORMAL;
  1318. }
  1319. else {
  1320. s->cmd_type = CMD_NORMAL;
  1321. }
  1322. }
  1323. if (buflen > 0) {
  1324. /* Process possible extensions */
  1325. if (!rspamd_fuzzy_extensions_from_wire (s, buf, buflen)) {
  1326. msg_debug ("truncated fuzzy shingles command of size %d received", buflen);
  1327. return FALSE;
  1328. }
  1329. }
  1330. return TRUE;
  1331. }
  1332. static void
  1333. fuzzy_session_destroy (gpointer d)
  1334. {
  1335. struct fuzzy_session *session = d;
  1336. rspamd_inet_address_free (session->addr);
  1337. rspamd_explicit_memzero (session->nm, sizeof (session->nm));
  1338. session->worker->nconns--;
  1339. if (session->ip_stat) {
  1340. REF_RELEASE (session->ip_stat);
  1341. }
  1342. if (session->extensions) {
  1343. g_free (session->extensions);
  1344. }
  1345. g_free (session);
  1346. }
  1347. #define FUZZY_INPUT_BUFLEN 1024
  1348. #ifdef HAVE_RECVMMSG
  1349. #define MSGVEC_LEN 16
  1350. #else
  1351. #define MSGVEC_LEN 1
  1352. #endif
  1353. /*
  1354. * Accept new connection and construct task
  1355. */
  1356. static void
  1357. accept_fuzzy_socket (EV_P_ ev_io *w, int revents)
  1358. {
  1359. struct rspamd_worker *worker = (struct rspamd_worker *)w->data;
  1360. struct fuzzy_session *session;
  1361. gssize r, msg_len;
  1362. guint64 *nerrors;
  1363. struct iovec iovs[MSGVEC_LEN];
  1364. guint8 bufs[MSGVEC_LEN][FUZZY_INPUT_BUFLEN];
  1365. struct sockaddr_storage peer_sa[MSGVEC_LEN];
  1366. socklen_t salen = sizeof (peer_sa[0]);
  1367. #ifdef HAVE_RECVMMSG
  1368. #define MSG_FIELD(msg, field) msg.msg_hdr.field
  1369. struct mmsghdr msg[MSGVEC_LEN];
  1370. #else
  1371. #define MSG_FIELD(msg, field) msg.field
  1372. struct msghdr msg[MSGVEC_LEN];
  1373. #endif
  1374. memset (msg, 0, sizeof (*msg) * MSGVEC_LEN);
  1375. /* Prepare messages to receive */
  1376. for (int i = 0; i < MSGVEC_LEN; i ++) {
  1377. /* Prepare msghdr structs */
  1378. iovs[i].iov_base = bufs[i];
  1379. iovs[i].iov_len = sizeof (bufs[i]);
  1380. MSG_FIELD(msg[i], msg_name) = (void *)&peer_sa[i];
  1381. MSG_FIELD(msg[i], msg_namelen) = salen;
  1382. MSG_FIELD(msg[i], msg_iov) = &iovs[i];
  1383. MSG_FIELD(msg[i], msg_iovlen) = 1;
  1384. }
  1385. /* Got some data */
  1386. if (revents == EV_READ) {
  1387. for (;;) {
  1388. #ifdef HAVE_RECVMMSG
  1389. r = recvmmsg (w->fd, msg, MSGVEC_LEN, 0, NULL);
  1390. #else
  1391. r = recvmsg (w->fd, msg, 0);
  1392. #endif
  1393. if (r == -1) {
  1394. if (errno == EINTR) {
  1395. continue;
  1396. }
  1397. else if (errno == EAGAIN || errno == EWOULDBLOCK) {
  1398. return;
  1399. }
  1400. msg_err ("got error while reading from socket: %d, %s",
  1401. errno,
  1402. strerror (errno));
  1403. return;
  1404. }
  1405. #ifndef HAVE_RECVMMSG
  1406. msg_len = r; /* Save real length in bytes here */
  1407. r = 1; /* Assume that we have received a single message */
  1408. #endif
  1409. for (int i = 0; i < r; i ++) {
  1410. rspamd_inet_addr_t *client_addr;
  1411. client_addr = rspamd_inet_address_from_sa (MSG_FIELD(msg[i], msg_name),
  1412. MSG_FIELD(msg[i], msg_namelen));
  1413. if (!rspamd_fuzzy_check_client (worker->ctx, client_addr)) {
  1414. /* Disallow forbidden clients silently */
  1415. rspamd_inet_address_free (client_addr);
  1416. continue;
  1417. }
  1418. session = g_malloc0 (sizeof (*session));
  1419. REF_INIT_RETAIN (session, fuzzy_session_destroy);
  1420. session->worker = worker;
  1421. session->fd = w->fd;
  1422. session->ctx = worker->ctx;
  1423. session->time = (guint64) time (NULL);
  1424. session->addr = client_addr;
  1425. worker->nconns++;
  1426. /* Each message can have its length in case of recvmmsg */
  1427. #ifdef HAVE_RECVMMSG
  1428. msg_len = msg[i].msg_len;
  1429. #endif
  1430. if (rspamd_fuzzy_cmd_from_wire (iovs[i].iov_base,
  1431. msg_len, session)) {
  1432. /* Check shingles count sanity */
  1433. rspamd_fuzzy_process_command (session);
  1434. }
  1435. else {
  1436. /* Discard input */
  1437. session->ctx->stat.invalid_requests ++;
  1438. msg_debug ("invalid fuzzy command of size %z received", r);
  1439. nerrors = rspamd_lru_hash_lookup (session->ctx->errors_ips,
  1440. session->addr, -1);
  1441. if (nerrors == NULL) {
  1442. nerrors = g_malloc (sizeof (*nerrors));
  1443. *nerrors = 1;
  1444. rspamd_lru_hash_insert (session->ctx->errors_ips,
  1445. rspamd_inet_address_copy (session->addr),
  1446. nerrors, -1, -1);
  1447. }
  1448. else {
  1449. *nerrors = *nerrors + 1;
  1450. }
  1451. }
  1452. REF_RELEASE (session);
  1453. }
  1454. #ifdef HAVE_RECVMMSG
  1455. /* Stop reading as we are using recvmmsg instead of recvmsg */
  1456. break;
  1457. #endif
  1458. }
  1459. }
  1460. }
  1461. static gboolean
  1462. rspamd_fuzzy_storage_periodic_callback (void *ud)
  1463. {
  1464. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1465. if (ctx->updates_pending->len > 0) {
  1466. rspamd_fuzzy_process_updates_queue (ctx, local_db_name, FALSE);
  1467. return TRUE;
  1468. }
  1469. return FALSE;
  1470. }
  1471. static gboolean
  1472. rspamd_fuzzy_storage_sync (struct rspamd_main *rspamd_main,
  1473. struct rspamd_worker *worker, gint fd,
  1474. gint attached_fd,
  1475. struct rspamd_control_command *cmd,
  1476. gpointer ud)
  1477. {
  1478. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1479. struct rspamd_control_reply rep;
  1480. rep.reply.fuzzy_sync.status = 0;
  1481. rep.type = RSPAMD_CONTROL_FUZZY_SYNC;
  1482. if (ctx->backend && worker->index == 0) {
  1483. rspamd_fuzzy_process_updates_queue (ctx, local_db_name, FALSE);
  1484. rspamd_fuzzy_backend_start_update (ctx->backend, ctx->sync_timeout,
  1485. rspamd_fuzzy_storage_periodic_callback, ctx);
  1486. }
  1487. if (write (fd, &rep, sizeof (rep)) != sizeof (rep)) {
  1488. msg_err ("cannot write reply to the control socket: %s",
  1489. strerror (errno));
  1490. }
  1491. return TRUE;
  1492. }
  1493. static gboolean
  1494. rspamd_fuzzy_storage_reload (struct rspamd_main *rspamd_main,
  1495. struct rspamd_worker *worker, gint fd,
  1496. gint attached_fd,
  1497. struct rspamd_control_command *cmd,
  1498. gpointer ud)
  1499. {
  1500. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1501. GError *err = NULL;
  1502. struct rspamd_control_reply rep;
  1503. msg_info ("reloading fuzzy storage after receiving reload command");
  1504. if (ctx->backend) {
  1505. /* Close backend and reopen it one more time */
  1506. rspamd_fuzzy_backend_close (ctx->backend);
  1507. }
  1508. memset (&rep, 0, sizeof (rep));
  1509. rep.type = RSPAMD_CONTROL_RELOAD;
  1510. if ((ctx->backend = rspamd_fuzzy_backend_create (ctx->event_loop,
  1511. worker->cf->options, rspamd_main->cfg,
  1512. &err)) == NULL) {
  1513. msg_err ("cannot open backend after reload: %e", err);
  1514. rep.reply.reload.status = err->code;
  1515. g_error_free (err);
  1516. }
  1517. else {
  1518. rep.reply.reload.status = 0;
  1519. }
  1520. if (ctx->backend && worker->index == 0) {
  1521. rspamd_fuzzy_backend_start_update (ctx->backend, ctx->sync_timeout,
  1522. rspamd_fuzzy_storage_periodic_callback, ctx);
  1523. }
  1524. if (write (fd, &rep, sizeof (rep)) != sizeof (rep)) {
  1525. msg_err ("cannot write reply to the control socket: %s",
  1526. strerror (errno));
  1527. }
  1528. return TRUE;
  1529. }
  1530. static ucl_object_t *
  1531. rspamd_fuzzy_storage_stat_key (struct fuzzy_key_stat *key_stat)
  1532. {
  1533. ucl_object_t *res;
  1534. res = ucl_object_typed_new (UCL_OBJECT);
  1535. ucl_object_insert_key (res, ucl_object_fromint (key_stat->checked),
  1536. "checked", 0, false);
  1537. ucl_object_insert_key (res, ucl_object_fromint (key_stat->matched),
  1538. "matched", 0, false);
  1539. ucl_object_insert_key (res, ucl_object_fromint (key_stat->added),
  1540. "added", 0, false);
  1541. ucl_object_insert_key (res, ucl_object_fromint (key_stat->deleted),
  1542. "deleted", 0, false);
  1543. ucl_object_insert_key (res, ucl_object_fromint (key_stat->errors),
  1544. "errors", 0, false);
  1545. return res;
  1546. }
  1547. static ucl_object_t *
  1548. rspamd_fuzzy_stat_to_ucl (struct rspamd_fuzzy_storage_ctx *ctx, gboolean ip_stat)
  1549. {
  1550. struct fuzzy_key_stat *key_stat;
  1551. GHashTableIter it;
  1552. struct fuzzy_key *key;
  1553. ucl_object_t *obj, *keys_obj, *elt, *ip_elt, *ip_cur;
  1554. gpointer k, v;
  1555. gint i;
  1556. gchar keyname[17];
  1557. obj = ucl_object_typed_new (UCL_OBJECT);
  1558. keys_obj = ucl_object_typed_new (UCL_OBJECT);
  1559. g_hash_table_iter_init (&it, ctx->keys);
  1560. while (g_hash_table_iter_next (&it, &k, &v)) {
  1561. key = v;
  1562. key_stat = key->stat;
  1563. if (key_stat) {
  1564. rspamd_snprintf (keyname, sizeof (keyname), "%8bs", k);
  1565. elt = rspamd_fuzzy_storage_stat_key (key_stat);
  1566. if (key_stat->last_ips && ip_stat) {
  1567. i = 0;
  1568. ip_elt = ucl_object_typed_new (UCL_OBJECT);
  1569. while ((i = rspamd_lru_hash_foreach (key_stat->last_ips,
  1570. i, &k, &v)) != -1) {
  1571. ip_cur = rspamd_fuzzy_storage_stat_key (v);
  1572. ucl_object_insert_key (ip_elt, ip_cur,
  1573. rspamd_inet_address_to_string (k), 0, true);
  1574. }
  1575. ucl_object_insert_key (elt, ip_elt, "ips", 0, false);
  1576. }
  1577. ucl_object_insert_key (keys_obj, elt, keyname, 0, true);
  1578. }
  1579. }
  1580. ucl_object_insert_key (obj, keys_obj, "keys", 0, false);
  1581. /* Now generic stats */
  1582. ucl_object_insert_key (obj,
  1583. ucl_object_fromint (ctx->stat.fuzzy_hashes),
  1584. "fuzzy_stored",
  1585. 0,
  1586. false);
  1587. ucl_object_insert_key (obj,
  1588. ucl_object_fromint (ctx->stat.fuzzy_hashes_expired),
  1589. "fuzzy_expired",
  1590. 0,
  1591. false);
  1592. ucl_object_insert_key (obj,
  1593. ucl_object_fromint (ctx->stat.invalid_requests),
  1594. "invalid_requests",
  1595. 0,
  1596. false);
  1597. ucl_object_insert_key (obj,
  1598. ucl_object_fromint (ctx->stat.delayed_hashes),
  1599. "delayed_hashes",
  1600. 0,
  1601. false);
  1602. if (ctx->errors_ips && ip_stat) {
  1603. i = 0;
  1604. ip_elt = ucl_object_typed_new (UCL_OBJECT);
  1605. while ((i = rspamd_lru_hash_foreach (ctx->errors_ips, i, &k, &v)) != -1) {
  1606. ucl_object_insert_key (ip_elt,
  1607. ucl_object_fromint (*(guint64 *)v),
  1608. rspamd_inet_address_to_string (k), 0, true);
  1609. }
  1610. ucl_object_insert_key (obj,
  1611. ip_elt,
  1612. "errors_ips",
  1613. 0,
  1614. false);
  1615. }
  1616. /* Checked by epoch */
  1617. elt = ucl_object_typed_new (UCL_ARRAY);
  1618. for (i = RSPAMD_FUZZY_EPOCH6; i < RSPAMD_FUZZY_EPOCH_MAX; i++) {
  1619. ucl_array_append (elt,
  1620. ucl_object_fromint (ctx->stat.fuzzy_hashes_checked[i]));
  1621. }
  1622. ucl_object_insert_key (obj, elt, "fuzzy_checked", 0, false);
  1623. /* Shingles by epoch */
  1624. elt = ucl_object_typed_new (UCL_ARRAY);
  1625. for (i = RSPAMD_FUZZY_EPOCH6; i < RSPAMD_FUZZY_EPOCH_MAX; i++) {
  1626. ucl_array_append (elt,
  1627. ucl_object_fromint (ctx->stat.fuzzy_shingles_checked[i]));
  1628. }
  1629. ucl_object_insert_key (obj, elt, "fuzzy_shingles", 0, false);
  1630. /* Matched by epoch */
  1631. elt = ucl_object_typed_new (UCL_ARRAY);
  1632. for (i = RSPAMD_FUZZY_EPOCH6; i < RSPAMD_FUZZY_EPOCH_MAX; i++) {
  1633. ucl_array_append (elt,
  1634. ucl_object_fromint (ctx->stat.fuzzy_hashes_found[i]));
  1635. }
  1636. ucl_object_insert_key (obj, elt, "fuzzy_found", 0, false);
  1637. return obj;
  1638. }
  1639. static int
  1640. lua_fuzzy_add_pre_handler (lua_State *L)
  1641. {
  1642. struct rspamd_worker *wrk, **pwrk = (struct rspamd_worker **)
  1643. rspamd_lua_check_udata (L, 1, "rspamd{worker}");
  1644. struct rspamd_fuzzy_storage_ctx *ctx;
  1645. if (!pwrk) {
  1646. return luaL_error (L, "invalid arguments, worker + function are expected");
  1647. }
  1648. wrk = *pwrk;
  1649. if (wrk && lua_isfunction (L, 2)) {
  1650. ctx = (struct rspamd_fuzzy_storage_ctx *)wrk->ctx;
  1651. if (ctx->lua_pre_handler_cbref != -1) {
  1652. /* Should not happen */
  1653. luaL_unref (L, LUA_REGISTRYINDEX, ctx->lua_pre_handler_cbref);
  1654. }
  1655. lua_pushvalue (L, 2);
  1656. ctx->lua_pre_handler_cbref = luaL_ref (L, LUA_REGISTRYINDEX);
  1657. }
  1658. else {
  1659. return luaL_error (L, "invalid arguments, worker + function are expected");
  1660. }
  1661. return 0;
  1662. }
  1663. static int
  1664. lua_fuzzy_add_post_handler (lua_State *L)
  1665. {
  1666. struct rspamd_worker *wrk, **pwrk = (struct rspamd_worker **)
  1667. rspamd_lua_check_udata (L, 1, "rspamd{worker}");
  1668. struct rspamd_fuzzy_storage_ctx *ctx;
  1669. if (!pwrk) {
  1670. return luaL_error (L, "invalid arguments, worker + function are expected");
  1671. }
  1672. wrk = *pwrk;
  1673. if (wrk && lua_isfunction (L, 2)) {
  1674. ctx = (struct rspamd_fuzzy_storage_ctx *)wrk->ctx;
  1675. if (ctx->lua_post_handler_cbref != -1) {
  1676. /* Should not happen */
  1677. luaL_unref (L, LUA_REGISTRYINDEX, ctx->lua_post_handler_cbref);
  1678. }
  1679. lua_pushvalue (L, 2);
  1680. ctx->lua_post_handler_cbref = luaL_ref (L, LUA_REGISTRYINDEX);
  1681. }
  1682. else {
  1683. return luaL_error (L, "invalid arguments, worker + function are expected");
  1684. }
  1685. return 0;
  1686. }
  1687. static gboolean
  1688. rspamd_fuzzy_storage_stat (struct rspamd_main *rspamd_main,
  1689. struct rspamd_worker *worker, gint fd,
  1690. gint attached_fd,
  1691. struct rspamd_control_command *cmd,
  1692. gpointer ud)
  1693. {
  1694. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  1695. struct rspamd_control_reply rep;
  1696. ucl_object_t *obj;
  1697. struct ucl_emitter_functions *emit_subr;
  1698. guchar fdspace[CMSG_SPACE(sizeof (int))];
  1699. struct iovec iov;
  1700. struct msghdr msg;
  1701. struct cmsghdr *cmsg;
  1702. gint outfd = -1;
  1703. gchar tmppath[PATH_MAX];
  1704. memset (&rep, 0, sizeof (rep));
  1705. rep.type = RSPAMD_CONTROL_FUZZY_STAT;
  1706. rspamd_snprintf (tmppath, sizeof (tmppath), "%s%c%s-XXXXXXXXXX",
  1707. rspamd_main->cfg->temp_dir, G_DIR_SEPARATOR, "fuzzy-stat");
  1708. if ((outfd = mkstemp (tmppath)) == -1) {
  1709. rep.reply.fuzzy_stat.status = errno;
  1710. msg_info_main ("cannot make temporary stat file for fuzzy stat: %s",
  1711. strerror (errno));
  1712. }
  1713. else {
  1714. rep.reply.fuzzy_stat.status = 0;
  1715. memcpy (rep.reply.fuzzy_stat.storage_id,
  1716. rspamd_fuzzy_backend_id (ctx->backend),
  1717. sizeof (rep.reply.fuzzy_stat.storage_id));
  1718. obj = rspamd_fuzzy_stat_to_ucl (ctx, TRUE);
  1719. emit_subr = ucl_object_emit_fd_funcs (outfd);
  1720. ucl_object_emit_full (obj, UCL_EMIT_JSON_COMPACT, emit_subr, NULL);
  1721. ucl_object_emit_funcs_free (emit_subr);
  1722. ucl_object_unref (obj);
  1723. /* Rewind output file */
  1724. close (outfd);
  1725. outfd = open (tmppath, O_RDONLY);
  1726. unlink (tmppath);
  1727. }
  1728. /* Now we can send outfd and status message */
  1729. memset (&msg, 0, sizeof (msg));
  1730. /* Attach fd to the message */
  1731. if (outfd != -1) {
  1732. memset (fdspace, 0, sizeof (fdspace));
  1733. msg.msg_control = fdspace;
  1734. msg.msg_controllen = sizeof (fdspace);
  1735. cmsg = CMSG_FIRSTHDR (&msg);
  1736. if (cmsg) {
  1737. cmsg->cmsg_level = SOL_SOCKET;
  1738. cmsg->cmsg_type = SCM_RIGHTS;
  1739. cmsg->cmsg_len = CMSG_LEN (sizeof (int));
  1740. memcpy (CMSG_DATA (cmsg), &outfd, sizeof (int));
  1741. }
  1742. }
  1743. iov.iov_base = &rep;
  1744. iov.iov_len = sizeof (rep);
  1745. msg.msg_iov = &iov;
  1746. msg.msg_iovlen = 1;
  1747. if (sendmsg (fd, &msg, 0) == -1) {
  1748. msg_err_main ("cannot send fuzzy stat: %s", strerror (errno));
  1749. }
  1750. if (outfd != -1) {
  1751. close (outfd);
  1752. }
  1753. return TRUE;
  1754. }
  1755. static gboolean
  1756. fuzzy_parse_keypair (rspamd_mempool_t *pool,
  1757. const ucl_object_t *obj,
  1758. gpointer ud,
  1759. struct rspamd_rcl_section *section,
  1760. GError **err)
  1761. {
  1762. struct rspamd_rcl_struct_parser *pd = ud;
  1763. struct rspamd_fuzzy_storage_ctx *ctx;
  1764. struct rspamd_cryptobox_keypair *kp;
  1765. struct fuzzy_key_stat *keystat;
  1766. struct fuzzy_key *key;
  1767. const ucl_object_t *cur;
  1768. const guchar *pk;
  1769. ucl_object_iter_t it = NULL;
  1770. gboolean ret;
  1771. ctx = pd->user_struct;
  1772. pd->offset = G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, default_keypair);
  1773. /*
  1774. * Single key
  1775. */
  1776. if (ucl_object_type (obj) == UCL_STRING || ucl_object_type (obj)
  1777. == UCL_OBJECT) {
  1778. ret = rspamd_rcl_parse_struct_keypair (pool, obj, pd, section, err);
  1779. if (!ret) {
  1780. return ret;
  1781. }
  1782. /* Insert key to the hash table */
  1783. kp = ctx->default_keypair;
  1784. if (kp == NULL) {
  1785. return FALSE;
  1786. }
  1787. if (rspamd_keypair_alg (kp) != RSPAMD_CRYPTOBOX_MODE_25519 ||
  1788. rspamd_keypair_type (kp) != RSPAMD_KEYPAIR_KEX) {
  1789. return FALSE;
  1790. }
  1791. key = g_malloc0 (sizeof (*key));
  1792. key->key = kp;
  1793. keystat = g_malloc0 (sizeof (*keystat));
  1794. REF_INIT_RETAIN (keystat, fuzzy_key_stat_dtor);
  1795. /* Hash of ip -> fuzzy_key_stat */
  1796. keystat->last_ips = rspamd_lru_hash_new_full (1024,
  1797. (GDestroyNotify) rspamd_inet_address_free,
  1798. fuzzy_key_stat_unref,
  1799. rspamd_inet_address_hash, rspamd_inet_address_equal);
  1800. key->stat = keystat;
  1801. pk = rspamd_keypair_component (kp, RSPAMD_KEYPAIR_COMPONENT_PK,
  1802. NULL);
  1803. g_hash_table_insert (ctx->keys, (gpointer)pk, key);
  1804. ctx->default_key = key;
  1805. msg_info_pool ("loaded keypair %*xs", 8, pk);
  1806. }
  1807. else if (ucl_object_type (obj) == UCL_ARRAY) {
  1808. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  1809. if (!fuzzy_parse_keypair (pool, cur, pd, section, err)) {
  1810. msg_err_pool ("cannot parse keypair");
  1811. }
  1812. }
  1813. }
  1814. return TRUE;
  1815. }
  1816. static guint
  1817. fuzzy_kp_hash (gconstpointer p)
  1818. {
  1819. return *(guint *)p;
  1820. }
  1821. static gboolean
  1822. fuzzy_kp_equal (gconstpointer a, gconstpointer b)
  1823. {
  1824. const guchar *pa = a, *pb = b;
  1825. return (memcmp (pa, pb, RSPAMD_FUZZY_KEYLEN) == 0);
  1826. }
  1827. gpointer
  1828. init_fuzzy (struct rspamd_config *cfg)
  1829. {
  1830. struct rspamd_fuzzy_storage_ctx *ctx;
  1831. GQuark type;
  1832. type = g_quark_try_string ("fuzzy");
  1833. ctx = rspamd_mempool_alloc0 (cfg->cfg_pool,
  1834. sizeof (struct rspamd_fuzzy_storage_ctx));
  1835. ctx->magic = rspamd_fuzzy_storage_magic;
  1836. ctx->sync_timeout = DEFAULT_SYNC_TIMEOUT;
  1837. ctx->keypair_cache_size = DEFAULT_KEYPAIR_CACHE_SIZE;
  1838. ctx->lua_pre_handler_cbref = -1;
  1839. ctx->lua_post_handler_cbref = -1;
  1840. ctx->keys = g_hash_table_new_full (fuzzy_kp_hash, fuzzy_kp_equal,
  1841. NULL, fuzzy_key_dtor);
  1842. rspamd_mempool_add_destructor (cfg->cfg_pool,
  1843. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->keys);
  1844. ctx->errors_ips = rspamd_lru_hash_new_full (1024,
  1845. (GDestroyNotify) rspamd_inet_address_free, g_free,
  1846. rspamd_inet_address_hash, rspamd_inet_address_equal);
  1847. rspamd_mempool_add_destructor (cfg->cfg_pool,
  1848. (rspamd_mempool_destruct_t)rspamd_lru_hash_destroy, ctx->errors_ips);
  1849. ctx->cfg = cfg;
  1850. ctx->updates_maxfail = DEFAULT_UPDATES_MAXFAIL;
  1851. ctx->leaky_bucket_mask = DEFAULT_BUCKET_MASK;
  1852. ctx->leaky_bucket_ttl = DEFAULT_BUCKET_TTL;
  1853. ctx->max_buckets = DEFAULT_MAX_BUCKETS;
  1854. ctx->leaky_bucket_burst = NAN;
  1855. ctx->leaky_bucket_rate = NAN;
  1856. ctx->delay = NAN;
  1857. rspamd_rcl_register_worker_option (cfg,
  1858. type,
  1859. "sync",
  1860. rspamd_rcl_parse_struct_time,
  1861. ctx,
  1862. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx,
  1863. sync_timeout),
  1864. RSPAMD_CL_FLAG_TIME_FLOAT,
  1865. "Time to perform database sync, default: "
  1866. G_STRINGIFY (DEFAULT_SYNC_TIMEOUT) " seconds");
  1867. rspamd_rcl_register_worker_option (cfg,
  1868. type,
  1869. "expire",
  1870. rspamd_rcl_parse_struct_time,
  1871. ctx,
  1872. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx,
  1873. expire),
  1874. RSPAMD_CL_FLAG_TIME_FLOAT,
  1875. "Default expire time for hashes, default: "
  1876. G_STRINGIFY (DEFAULT_EXPIRE) " seconds");
  1877. rspamd_rcl_register_worker_option (cfg,
  1878. type,
  1879. "delay",
  1880. rspamd_rcl_parse_struct_time,
  1881. ctx,
  1882. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx,
  1883. delay),
  1884. RSPAMD_CL_FLAG_TIME_FLOAT,
  1885. "Default delay time for hashes, default: not enabled");
  1886. rspamd_rcl_register_worker_option (cfg,
  1887. type,
  1888. "allow_update",
  1889. rspamd_rcl_parse_struct_ucl,
  1890. ctx,
  1891. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, update_map),
  1892. 0,
  1893. "Allow modifications from the following IP addresses");
  1894. rspamd_rcl_register_worker_option (cfg,
  1895. type,
  1896. "delay_whitelist",
  1897. rspamd_rcl_parse_struct_ucl,
  1898. ctx,
  1899. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, delay_whitelist_map),
  1900. 0,
  1901. "Disable delay check for the following IP addresses");
  1902. rspamd_rcl_register_worker_option (cfg,
  1903. type,
  1904. "keypair",
  1905. fuzzy_parse_keypair,
  1906. ctx,
  1907. 0,
  1908. RSPAMD_CL_FLAG_MULTIPLE,
  1909. "Encryption keypair (can be repeated for different keys)");
  1910. rspamd_rcl_register_worker_option (cfg,
  1911. type,
  1912. "keypair_cache_size",
  1913. rspamd_rcl_parse_struct_integer,
  1914. ctx,
  1915. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx,
  1916. keypair_cache_size),
  1917. RSPAMD_CL_FLAG_UINT,
  1918. "Size of keypairs cache, default: "
  1919. G_STRINGIFY (DEFAULT_KEYPAIR_CACHE_SIZE));
  1920. rspamd_rcl_register_worker_option (cfg,
  1921. type,
  1922. "encrypted_only",
  1923. rspamd_rcl_parse_struct_boolean,
  1924. ctx,
  1925. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, encrypted_only),
  1926. 0,
  1927. "Allow encrypted requests only (and forbid all unknown keys or plaintext requests)");
  1928. rspamd_rcl_register_worker_option (cfg,
  1929. type,
  1930. "dedicated_update_worker",
  1931. rspamd_rcl_parse_struct_boolean,
  1932. ctx,
  1933. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, dedicated_update_worker),
  1934. 0,
  1935. "Use worker 0 for updates only");
  1936. rspamd_rcl_register_worker_option (cfg,
  1937. type,
  1938. "read_only",
  1939. rspamd_rcl_parse_struct_boolean,
  1940. ctx,
  1941. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, read_only),
  1942. 0,
  1943. "Work in read only mode");
  1944. rspamd_rcl_register_worker_option (cfg,
  1945. type,
  1946. "blocked",
  1947. rspamd_rcl_parse_struct_ucl,
  1948. ctx,
  1949. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, blocked_map),
  1950. 0,
  1951. "Block requests from specific networks");
  1952. rspamd_rcl_register_worker_option (cfg,
  1953. type,
  1954. "updates_maxfail",
  1955. rspamd_rcl_parse_struct_integer,
  1956. ctx,
  1957. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, updates_maxfail),
  1958. RSPAMD_CL_FLAG_UINT,
  1959. "Maximum number of updates to be failed before discarding");
  1960. rspamd_rcl_register_worker_option (cfg,
  1961. type,
  1962. "skip_hashes",
  1963. rspamd_rcl_parse_struct_ucl,
  1964. ctx,
  1965. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, skip_map),
  1966. 0,
  1967. "Skip specific hashes from the map");
  1968. /* Ratelimits */
  1969. rspamd_rcl_register_worker_option (cfg,
  1970. type,
  1971. "ratelimit_whitelist",
  1972. rspamd_rcl_parse_struct_ucl,
  1973. ctx,
  1974. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, ratelimit_whitelist_map),
  1975. 0,
  1976. "Skip specific addresses from rate limiting");
  1977. rspamd_rcl_register_worker_option (cfg,
  1978. type,
  1979. "ratelimit_max_buckets",
  1980. rspamd_rcl_parse_struct_integer,
  1981. ctx,
  1982. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, max_buckets),
  1983. RSPAMD_CL_FLAG_UINT,
  1984. "Maximum number of leaky buckets (default: " G_STRINGIFY(DEFAULT_MAX_BUCKETS) ")");
  1985. rspamd_rcl_register_worker_option (cfg,
  1986. type,
  1987. "ratelimit_network_mask",
  1988. rspamd_rcl_parse_struct_integer,
  1989. ctx,
  1990. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_mask),
  1991. RSPAMD_CL_FLAG_UINT,
  1992. "Network mask to apply for IPv4 rate addresses (default: " G_STRINGIFY(DEFAULT_BUCKET_MASK) ")");
  1993. rspamd_rcl_register_worker_option (cfg,
  1994. type,
  1995. "ratelimit_bucket_ttl",
  1996. rspamd_rcl_parse_struct_time,
  1997. ctx,
  1998. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_ttl),
  1999. RSPAMD_CL_FLAG_TIME_INTEGER,
  2000. "Time to live for ratelimit element (default: " G_STRINGIFY(DEFAULT_BUCKET_TTL) ")");
  2001. rspamd_rcl_register_worker_option (cfg,
  2002. type,
  2003. "ratelimit_rate",
  2004. rspamd_rcl_parse_struct_double,
  2005. ctx,
  2006. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_rate),
  2007. 0,
  2008. "Leak rate in requests per second");
  2009. rspamd_rcl_register_worker_option (cfg,
  2010. type,
  2011. "ratelimit_burst",
  2012. rspamd_rcl_parse_struct_double,
  2013. ctx,
  2014. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, leaky_bucket_burst),
  2015. 0,
  2016. "Peak value for ratelimit bucket");
  2017. rspamd_rcl_register_worker_option (cfg,
  2018. type,
  2019. "ratelimit_log_only",
  2020. rspamd_rcl_parse_struct_boolean,
  2021. ctx,
  2022. G_STRUCT_OFFSET (struct rspamd_fuzzy_storage_ctx, ratelimit_log_only),
  2023. 0,
  2024. "Don't really ban on ratelimit reaching, just log");
  2025. return ctx;
  2026. }
  2027. static void
  2028. rspamd_fuzzy_peer_io (EV_P_ ev_io *w, int revents)
  2029. {
  2030. struct fuzzy_peer_cmd cmd;
  2031. struct rspamd_fuzzy_storage_ctx *ctx =
  2032. (struct rspamd_fuzzy_storage_ctx *)w->data;
  2033. gssize r;
  2034. for (;;) {
  2035. r = read (w->fd, &cmd, sizeof (cmd));
  2036. if (r != sizeof (cmd)) {
  2037. if (errno == EINTR) {
  2038. continue;
  2039. }
  2040. if (errno != EAGAIN) {
  2041. msg_err ("cannot read command from peers: %s", strerror (errno));
  2042. }
  2043. break;
  2044. }
  2045. else {
  2046. g_array_append_val (ctx->updates_pending, cmd);
  2047. }
  2048. }
  2049. }
  2050. static void
  2051. fuzzy_peer_rep (struct rspamd_worker *worker,
  2052. struct rspamd_srv_reply *rep, gint rep_fd,
  2053. gpointer ud)
  2054. {
  2055. struct rspamd_fuzzy_storage_ctx *ctx = ud;
  2056. GList *cur;
  2057. struct rspamd_worker_listen_socket *ls;
  2058. struct rspamd_worker_accept_event *ac_ev;
  2059. ctx->peer_fd = rep_fd;
  2060. if (rep_fd == -1) {
  2061. msg_err ("cannot receive peer fd from the main process");
  2062. exit (EXIT_FAILURE);
  2063. }
  2064. else {
  2065. rspamd_socket_nonblocking (rep_fd);
  2066. }
  2067. msg_info ("got peer fd reply from the main process");
  2068. /* Start listening */
  2069. cur = worker->cf->listen_socks;
  2070. while (cur) {
  2071. ls = cur->data;
  2072. if (ls->fd != -1) {
  2073. msg_info ("start listening on %s",
  2074. rspamd_inet_address_to_string_pretty (ls->addr));
  2075. if (ls->type == RSPAMD_WORKER_SOCKET_UDP) {
  2076. ac_ev = g_malloc0 (sizeof (*ac_ev));
  2077. ac_ev->accept_ev.data = worker;
  2078. ac_ev->event_loop = ctx->event_loop;
  2079. ev_io_init (&ac_ev->accept_ev, accept_fuzzy_socket, ls->fd,
  2080. EV_READ);
  2081. ev_io_start (ctx->event_loop, &ac_ev->accept_ev);
  2082. DL_APPEND (worker->accept_events, ac_ev);
  2083. }
  2084. else {
  2085. /* We allow TCP listeners only for a update worker */
  2086. g_assert_not_reached ();
  2087. }
  2088. }
  2089. cur = g_list_next (cur);
  2090. }
  2091. if (ctx->peer_fd != -1) {
  2092. if (worker->index == 0) {
  2093. /* Listen for peer requests */
  2094. shutdown (ctx->peer_fd, SHUT_WR);
  2095. ctx->peer_ev.data = ctx;
  2096. ev_io_init (&ctx->peer_ev, rspamd_fuzzy_peer_io, ctx->peer_fd, EV_READ);
  2097. ev_io_start (ctx->event_loop, &ctx->peer_ev);
  2098. }
  2099. else {
  2100. shutdown (ctx->peer_fd, SHUT_RD);
  2101. }
  2102. }
  2103. }
  2104. /*
  2105. * Start worker process
  2106. */
  2107. __attribute__((noreturn))
  2108. void
  2109. start_fuzzy (struct rspamd_worker *worker)
  2110. {
  2111. struct rspamd_fuzzy_storage_ctx *ctx = worker->ctx;
  2112. GError *err = NULL;
  2113. struct rspamd_srv_command srv_cmd;
  2114. struct rspamd_config *cfg = worker->srv->cfg;
  2115. g_assert (rspamd_worker_check_context (worker->ctx, rspamd_fuzzy_storage_magic));
  2116. ctx->event_loop = rspamd_prepare_worker (worker,
  2117. "fuzzy",
  2118. NULL);
  2119. ctx->peer_fd = -1;
  2120. ctx->worker = worker;
  2121. ctx->cfg = worker->srv->cfg;
  2122. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  2123. ctx->event_loop,
  2124. worker->srv->cfg);
  2125. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  2126. ctx->event_loop, ctx->resolver->r);
  2127. /* Since this worker uses maps it needs a valid HTTP context */
  2128. ctx->http_ctx = rspamd_http_context_create (ctx->cfg, ctx->event_loop,
  2129. ctx->cfg->ups_ctx);
  2130. if (ctx->keypair_cache_size > 0) {
  2131. /* Create keypairs cache */
  2132. ctx->keypair_cache = rspamd_keypair_cache_new (ctx->keypair_cache_size);
  2133. }
  2134. if ((ctx->backend = rspamd_fuzzy_backend_create (ctx->event_loop,
  2135. worker->cf->options, cfg, &err)) == NULL) {
  2136. msg_err ("cannot open backend: %e", err);
  2137. if (err) {
  2138. g_error_free (err);
  2139. }
  2140. exit (EXIT_SUCCESS);
  2141. }
  2142. rspamd_fuzzy_backend_count (ctx->backend, fuzzy_count_callback, ctx);
  2143. if (worker->index == 0) {
  2144. ctx->updates_pending = g_array_sized_new (FALSE, FALSE,
  2145. sizeof (struct fuzzy_peer_cmd), 1024);
  2146. rspamd_fuzzy_backend_start_update (ctx->backend, ctx->sync_timeout,
  2147. rspamd_fuzzy_storage_periodic_callback, ctx);
  2148. if (ctx->dedicated_update_worker && worker->cf->count > 1) {
  2149. msg_info_config ("stop serving clients request in dedicated update mode");
  2150. rspamd_worker_stop_accept (worker);
  2151. GList *cur = worker->cf->listen_socks;
  2152. while (cur) {
  2153. struct rspamd_worker_listen_socket *ls =
  2154. (struct rspamd_worker_listen_socket *)cur->data;
  2155. if (ls->fd != -1) {
  2156. close (ls->fd);
  2157. }
  2158. cur = g_list_next (cur);
  2159. }
  2160. }
  2161. }
  2162. ctx->stat_ev.data = ctx;
  2163. ev_timer_init (&ctx->stat_ev, rspamd_fuzzy_stat_callback, ctx->sync_timeout,
  2164. ctx->sync_timeout);
  2165. ev_timer_start (ctx->event_loop, &ctx->stat_ev);
  2166. /* Register custom reload and stat commands for the control socket */
  2167. rspamd_control_worker_add_cmd_handler (worker, RSPAMD_CONTROL_RELOAD,
  2168. rspamd_fuzzy_storage_reload, ctx);
  2169. rspamd_control_worker_add_cmd_handler (worker, RSPAMD_CONTROL_FUZZY_STAT,
  2170. rspamd_fuzzy_storage_stat, ctx);
  2171. rspamd_control_worker_add_cmd_handler (worker, RSPAMD_CONTROL_FUZZY_SYNC,
  2172. rspamd_fuzzy_storage_sync, ctx);
  2173. /* Create radix trees */
  2174. if (ctx->update_map != NULL) {
  2175. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->update_map,
  2176. "Allow fuzzy updates from specified addresses",
  2177. &ctx->update_ips, NULL, worker, "fuzzy update");
  2178. }
  2179. if (ctx->skip_map != NULL) {
  2180. struct rspamd_map *m;
  2181. if ((m = rspamd_map_add_from_ucl (cfg, ctx->skip_map,
  2182. "Skip hashes",
  2183. rspamd_kv_list_read,
  2184. rspamd_kv_list_fin,
  2185. rspamd_kv_list_dtor,
  2186. (void **)&ctx->skip_hashes,
  2187. worker, RSPAMD_MAP_DEFAULT)) == NULL) {
  2188. msg_warn_config ("cannot load hashes list from %s",
  2189. ucl_object_tostring (ctx->skip_map));
  2190. }
  2191. else {
  2192. m->active_http = TRUE;
  2193. }
  2194. }
  2195. if (ctx->blocked_map != NULL) {
  2196. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->blocked_map,
  2197. "Block fuzzy requests from the specific IPs",
  2198. &ctx->blocked_ips,
  2199. NULL,
  2200. worker, "fuzzy blocked");
  2201. }
  2202. /* Create radix trees */
  2203. if (ctx->ratelimit_whitelist_map != NULL) {
  2204. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->ratelimit_whitelist_map,
  2205. "Skip ratelimits from specific ip addresses/networks",
  2206. &ctx->ratelimit_whitelist,
  2207. NULL,
  2208. worker, "fuzzy ratelimit whitelist");
  2209. }
  2210. if (!isnan (ctx->delay) && ctx->delay_whitelist_map != NULL) {
  2211. rspamd_config_radix_from_ucl (worker->srv->cfg, ctx->delay_whitelist_map,
  2212. "Skip delay from the following ips",
  2213. &ctx->delay_whitelist, NULL, worker,
  2214. "fuzzy delayed whitelist");
  2215. }
  2216. /* Ratelimits */
  2217. if (!isnan (ctx->leaky_bucket_rate) && !isnan (ctx->leaky_bucket_burst)) {
  2218. ctx->ratelimit_buckets = rspamd_lru_hash_new_full (ctx->max_buckets,
  2219. NULL, fuzzy_rl_bucket_free,
  2220. rspamd_inet_address_hash, rspamd_inet_address_equal);
  2221. }
  2222. /* Maps events */
  2223. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  2224. ctx->event_loop,
  2225. worker->srv->cfg);
  2226. rspamd_map_watch (worker->srv->cfg, ctx->event_loop,
  2227. ctx->resolver, worker, RSPAMD_MAP_WATCH_WORKER);
  2228. /* Get peer pipe */
  2229. memset (&srv_cmd, 0, sizeof (srv_cmd));
  2230. srv_cmd.type = RSPAMD_SRV_SOCKETPAIR;
  2231. srv_cmd.cmd.spair.af = SOCK_DGRAM;
  2232. srv_cmd.cmd.spair.pair_num = worker->index;
  2233. memset (srv_cmd.cmd.spair.pair_id, 0, sizeof (srv_cmd.cmd.spair.pair_id));
  2234. /* 6 bytes of id (including \0) and bind_conf id */
  2235. G_STATIC_ASSERT (sizeof (srv_cmd.cmd.spair.pair_id) >=
  2236. sizeof ("fuzzy") + sizeof (guint64));
  2237. memcpy (srv_cmd.cmd.spair.pair_id, "fuzzy", sizeof ("fuzzy"));
  2238. /* Distinguish workers from each others... */
  2239. if (worker->cf->bind_conf && worker->cf->bind_conf->bind_line) {
  2240. guint64 bind_hash = rspamd_cryptobox_fast_hash (worker->cf->bind_conf->bind_line,
  2241. strlen (worker->cf->bind_conf->bind_line), 0xdeadbabe);
  2242. /* 8 more bytes */
  2243. memcpy (srv_cmd.cmd.spair.pair_id + sizeof ("fuzzy"), &bind_hash,
  2244. sizeof (bind_hash));
  2245. }
  2246. rspamd_srv_send_command (worker, ctx->event_loop, &srv_cmd, -1,
  2247. fuzzy_peer_rep, ctx);
  2248. luaL_Reg fuzzy_lua_reg = {
  2249. .name = "add_fuzzy_pre_handler",
  2250. .func = lua_fuzzy_add_pre_handler,
  2251. };
  2252. rspamd_lua_add_metamethod (ctx->cfg->lua_state, "rspamd{worker}", &fuzzy_lua_reg);
  2253. fuzzy_lua_reg = (luaL_Reg){
  2254. .name = "add_fuzzy_post_handler",
  2255. .func = lua_fuzzy_add_post_handler,
  2256. };
  2257. rspamd_lua_add_metamethod (ctx->cfg->lua_state, "rspamd{worker}", &fuzzy_lua_reg);
  2258. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->event_loop,
  2259. worker);
  2260. ev_loop (ctx->event_loop, 0);
  2261. rspamd_worker_block_signals ();
  2262. if (ctx->peer_fd != -1) {
  2263. if (worker->index == 0) {
  2264. ev_io_stop (ctx->event_loop, &ctx->peer_ev);
  2265. }
  2266. close (ctx->peer_fd);
  2267. }
  2268. if (worker->index == 0 && ctx->updates_pending->len > 0) {
  2269. msg_info_config ("start another event loop to sync fuzzy storage");
  2270. if (rspamd_fuzzy_process_updates_queue (ctx, local_db_name, TRUE)) {
  2271. ev_loop (ctx->event_loop, 0);
  2272. msg_info_config ("sync cycle is done");
  2273. }
  2274. else {
  2275. msg_info_config ("no need to sync");
  2276. }
  2277. }
  2278. rspamd_fuzzy_backend_close (ctx->backend);
  2279. if (worker->index == 0) {
  2280. g_array_free (ctx->updates_pending, TRUE);
  2281. ctx->updates_pending = NULL;
  2282. }
  2283. if (ctx->keypair_cache) {
  2284. rspamd_keypair_cache_destroy (ctx->keypair_cache);
  2285. }
  2286. if (ctx->ratelimit_buckets) {
  2287. rspamd_lru_hash_destroy (ctx->ratelimit_buckets);
  2288. }
  2289. if (ctx->lua_pre_handler_cbref != -1) {
  2290. luaL_unref (ctx->cfg->lua_state, LUA_REGISTRYINDEX, ctx->lua_pre_handler_cbref);
  2291. }
  2292. if (ctx->lua_post_handler_cbref != -1) {
  2293. luaL_unref (ctx->cfg->lua_state, LUA_REGISTRYINDEX, ctx->lua_post_handler_cbref);
  2294. }
  2295. REF_RELEASE (ctx->cfg);
  2296. rspamd_log_close (worker->srv->logger);
  2297. exit (EXIT_SUCCESS);
  2298. }