Nevar pievienot vairāk kā 25 tēmas Tēmai ir jāsākas ar burtu vai ciparu, tā var saturēt domu zīmes ('-') un var būt līdz 35 simboliem gara.

rspamd_proxy.c 56KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220
  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. #include "config.h"
  17. #include "libutil/util.h"
  18. #include "libutil/map.h"
  19. #include "libutil/upstream.h"
  20. #include "libutil/http_connection.h"
  21. #include "libutil/http_private.h"
  22. #include "libserver/protocol.h"
  23. #include "libserver/protocol_internal.h"
  24. #include "libserver/cfg_file.h"
  25. #include "libserver/url.h"
  26. #include "libserver/dns.h"
  27. #include "libmime/message.h"
  28. #include "rspamd.h"
  29. #include "libserver/worker_util.h"
  30. #include "worker_private.h"
  31. #include "lua/lua_common.h"
  32. #include "keypairs_cache.h"
  33. #include "libstat/stat_api.h"
  34. #include "ottery.h"
  35. #include "unix-std.h"
  36. #include "libserver/milter.h"
  37. #include "libserver/milter_internal.h"
  38. #include "libmime/lang_detection.h"
  39. #include "contrib/zstd/zstd.h"
  40. #include <math.h>
  41. #ifdef HAVE_NETINET_TCP_H
  42. #include <netinet/tcp.h> /* for TCP_NODELAY */
  43. #endif
  44. /* Rotate keys each minute by default */
  45. #define DEFAULT_ROTATION_TIME 60.0
  46. #define DEFAULT_RETRIES 5
  47. #define msg_err_session(...) rspamd_default_log_function (G_LOG_LEVEL_CRITICAL, \
  48. session->pool->tag.tagname, session->pool->tag.uid, \
  49. G_STRFUNC, \
  50. __VA_ARGS__)
  51. #define msg_warn_session(...) rspamd_default_log_function (G_LOG_LEVEL_WARNING, \
  52. session->pool->tag.tagname, session->pool->tag.uid, \
  53. G_STRFUNC, \
  54. __VA_ARGS__)
  55. #define msg_info_session(...) rspamd_default_log_function (G_LOG_LEVEL_INFO, \
  56. session->pool->tag.tagname, session->pool->tag.uid, \
  57. G_STRFUNC, \
  58. __VA_ARGS__)
  59. #define msg_debug_session(...) rspamd_conditional_debug_fast (NULL, session->client_addr, \
  60. rspamd_proxy_log_id, "proxy", session->pool->tag.uid, \
  61. G_STRFUNC, \
  62. __VA_ARGS__)
  63. INIT_LOG_MODULE(proxy)
  64. gpointer init_rspamd_proxy (struct rspamd_config *cfg);
  65. void start_rspamd_proxy (struct rspamd_worker *worker);
  66. worker_t rspamd_proxy_worker = {
  67. "rspamd_proxy", /* Name */
  68. init_rspamd_proxy, /* Init function */
  69. start_rspamd_proxy, /* Start function */
  70. RSPAMD_WORKER_HAS_SOCKET | RSPAMD_WORKER_KILLABLE | RSPAMD_WORKER_SCANNER,
  71. RSPAMD_WORKER_SOCKET_TCP, /* TCP socket */
  72. RSPAMD_WORKER_VER
  73. };
  74. struct rspamd_http_upstream {
  75. gchar *name;
  76. gchar *settings_id;
  77. struct upstream_list *u;
  78. struct rspamd_cryptobox_pubkey *key;
  79. gdouble timeout;
  80. struct timeval io_tv;
  81. gint parser_from_ref;
  82. gint parser_to_ref;
  83. gboolean local;
  84. gboolean self_scan;
  85. gboolean compress;
  86. };
  87. struct rspamd_http_mirror {
  88. gchar *name;
  89. gchar *settings_id;
  90. struct upstream_list *u;
  91. struct rspamd_cryptobox_pubkey *key;
  92. gdouble prob;
  93. gdouble timeout;
  94. struct timeval io_tv;
  95. gint parser_from_ref;
  96. gint parser_to_ref;
  97. gboolean local;
  98. gboolean compress;
  99. };
  100. static const guint64 rspamd_rspamd_proxy_magic = 0xcdeb4fd1fc351980ULL;
  101. struct rspamd_proxy_ctx {
  102. guint64 magic;
  103. /* Events base */
  104. struct event_base *ev_base;
  105. /* DNS resolver */
  106. struct rspamd_dns_resolver *resolver;
  107. /* Config */
  108. struct rspamd_config *cfg;
  109. /* END OF COMMON PART */
  110. gdouble timeout;
  111. struct timeval io_tv;
  112. /* Encryption key for clients */
  113. struct rspamd_cryptobox_keypair *key;
  114. /* HTTP context */
  115. struct rspamd_http_context *http_ctx;
  116. /* Upstreams to use */
  117. GHashTable *upstreams;
  118. /* Mirrors to send traffic to */
  119. GPtrArray *mirrors;
  120. /* Default upstream */
  121. struct rspamd_http_upstream *default_upstream;
  122. lua_State *lua_state;
  123. /* Array of callback functions called on end of scan to compare results */
  124. GArray *cmp_refs;
  125. /* Maximum count for retries */
  126. guint max_retries;
  127. /* If we have self_scanning backends, we need to work as a normal worker */
  128. gboolean has_self_scan;
  129. /* It is not HTTP but milter proxy */
  130. gboolean milter;
  131. /* Discard messages instead of rejecting them */
  132. gboolean discard_on_reject;
  133. /* Quarantine messages instead of rejecting them */
  134. gboolean quarantine_on_reject;
  135. /* Milter spam header */
  136. gchar *spam_header;
  137. /* CA name that can be used for client certificates */
  138. gchar *client_ca_name;
  139. /* Milter rejection message */
  140. gchar *reject_message;
  141. /* Sessions cache */
  142. void *sessions_cache;
  143. struct rspamd_milter_context milter_ctx;
  144. /* Language detector */
  145. struct rspamd_lang_detector *lang_det;
  146. };
  147. enum rspamd_backend_flags {
  148. RSPAMD_BACKEND_REPLIED = 1 << 0,
  149. RSPAMD_BACKEND_CLOSED = 1 << 1,
  150. RSPAMD_BACKEND_PARSED = 1 << 2,
  151. };
  152. struct rspamd_proxy_session;
  153. struct rspamd_proxy_backend_connection {
  154. const gchar *name;
  155. struct rspamd_cryptobox_keypair *local_key;
  156. struct rspamd_cryptobox_pubkey *remote_key;
  157. struct upstream *up;
  158. struct rspamd_http_connection *backend_conn;
  159. ucl_object_t *results;
  160. const gchar *err;
  161. struct rspamd_proxy_session *s;
  162. struct timeval *io_tv;
  163. gint backend_sock;
  164. enum rspamd_backend_flags flags;
  165. gint parser_from_ref;
  166. gint parser_to_ref;
  167. struct rspamd_task *task;
  168. };
  169. enum rspamd_proxy_legacy_support {
  170. LEGACY_SUPPORT_NO = 0,
  171. LEGACY_SUPPORT_RSPAMC,
  172. LEGACY_SUPPORT_SPAMC
  173. };
  174. struct rspamd_proxy_session {
  175. struct rspamd_worker *worker;
  176. rspamd_mempool_t *pool;
  177. struct rspamd_proxy_ctx *ctx;
  178. rspamd_inet_addr_t *client_addr;
  179. struct rspamd_http_connection *client_conn;
  180. struct rspamd_milter_session *client_milter_conn;
  181. struct rspamd_http_upstream *backend;
  182. gpointer map;
  183. gchar *fname;
  184. gpointer shmem_ref;
  185. struct rspamd_proxy_backend_connection *master_conn;
  186. struct rspamd_http_message *client_message;
  187. GPtrArray *mirror_conns;
  188. gsize map_len;
  189. gint client_sock;
  190. enum rspamd_proxy_legacy_support legacy_support;
  191. gint retries;
  192. ref_entry_t ref;
  193. };
  194. static gboolean proxy_send_master_message (struct rspamd_proxy_session *session);
  195. static GQuark
  196. rspamd_proxy_quark (void)
  197. {
  198. return g_quark_from_static_string ("rspamd-proxy");
  199. }
  200. static gboolean
  201. rspamd_proxy_parse_lua_parser (lua_State *L, const ucl_object_t *obj,
  202. gint *ref_from, gint *ref_to, GError **err)
  203. {
  204. const gchar *lua_script;
  205. gsize slen;
  206. gint err_idx, ref_idx;
  207. GString *tb = NULL;
  208. gboolean has_ref = FALSE;
  209. g_assert (obj != NULL);
  210. g_assert (ref_from != NULL);
  211. g_assert (ref_to != NULL);
  212. *ref_from = -1;
  213. *ref_to = -1;
  214. lua_script = ucl_object_tolstring (obj, &slen);
  215. lua_pushcfunction (L, &rspamd_lua_traceback);
  216. err_idx = lua_gettop (L);
  217. /* Load data */
  218. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  219. g_set_error (err,
  220. rspamd_proxy_quark (),
  221. EINVAL,
  222. "cannot load lua parser script: %s",
  223. lua_tostring (L, -1));
  224. lua_settop (L, 0); /* Error function */
  225. return FALSE;
  226. }
  227. /* Now do it */
  228. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  229. tb = lua_touserdata (L, -1);
  230. g_set_error (err,
  231. rspamd_proxy_quark (),
  232. EINVAL,
  233. "cannot init lua parser script: %s",
  234. tb->str);
  235. g_string_free (tb, TRUE);
  236. lua_settop (L, 0);
  237. return FALSE;
  238. }
  239. if (lua_istable (L, -1)) {
  240. /*
  241. * We have a table, so we check for two keys:
  242. * 'from' -> function
  243. * 'to' -> function
  244. *
  245. * From converts parent request to a client one
  246. * To converts client request to a parent one
  247. */
  248. lua_pushstring (L, "from");
  249. lua_gettable (L, -2);
  250. if (lua_isfunction (L, -1)) {
  251. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  252. *ref_from = ref_idx;
  253. has_ref = TRUE;
  254. }
  255. lua_pushstring (L, "to");
  256. lua_gettable (L, -2);
  257. if (lua_isfunction (L, -1)) {
  258. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  259. *ref_to = ref_idx;
  260. has_ref = TRUE;
  261. }
  262. }
  263. else if (!lua_isfunction (L, -1)) {
  264. g_set_error (err,
  265. rspamd_proxy_quark (),
  266. EINVAL,
  267. "cannot init lua parser script: "
  268. "must return function");
  269. lua_settop (L, 0);
  270. return FALSE;
  271. }
  272. else {
  273. /* Just parser from protocol */
  274. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  275. *ref_from = ref_idx;
  276. lua_settop (L, 0);
  277. has_ref = TRUE;
  278. }
  279. return has_ref;
  280. }
  281. static gboolean
  282. rspamd_proxy_parse_upstream (rspamd_mempool_t *pool,
  283. const ucl_object_t *obj,
  284. gpointer ud,
  285. struct rspamd_rcl_section *section,
  286. GError **err)
  287. {
  288. const ucl_object_t *elt;
  289. struct rspamd_http_upstream *up = NULL;
  290. struct rspamd_proxy_ctx *ctx;
  291. struct rspamd_rcl_struct_parser *pd = ud;
  292. lua_State *L;
  293. ctx = pd->user_struct;
  294. L = ctx->lua_state;
  295. if (ucl_object_type (obj) != UCL_OBJECT) {
  296. g_set_error (err, rspamd_proxy_quark (), 100,
  297. "upstream option must be an object");
  298. return FALSE;
  299. }
  300. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  301. elt = ucl_object_lookup (obj, "name");
  302. if (elt == NULL) {
  303. if (ucl_object_key (obj)) {
  304. if (strcmp (ucl_object_key (obj), "upstream") == 0) {
  305. /* Iterate over the object and find upstream elements */
  306. ucl_object_iter_t it = NULL;
  307. const ucl_object_t *cur;
  308. gboolean ret = TRUE;
  309. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  310. if (!rspamd_proxy_parse_upstream (pool, cur, ud,
  311. section, err)) {
  312. ret = FALSE;
  313. }
  314. }
  315. return ret;
  316. }
  317. else {
  318. /* Inside upstream */
  319. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  320. }
  321. }
  322. else {
  323. g_set_error (err, rspamd_proxy_quark (), 100,
  324. "upstream option must have some name definition");
  325. return FALSE;
  326. }
  327. }
  328. else {
  329. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  330. }
  331. up->parser_from_ref = -1;
  332. up->parser_to_ref = -1;
  333. up->timeout = ctx->timeout;
  334. elt = ucl_object_lookup (obj, "key");
  335. if (elt != NULL) {
  336. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  337. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  338. if (up->key == NULL) {
  339. g_set_error (err, rspamd_proxy_quark (), 100,
  340. "cannot read upstream key");
  341. goto err;
  342. }
  343. rspamd_mempool_add_destructor (pool,
  344. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  345. }
  346. elt = ucl_object_lookup (obj, "self_scan");
  347. if (elt && ucl_object_toboolean (elt)) {
  348. up->self_scan = TRUE;
  349. ctx->has_self_scan = TRUE;
  350. }
  351. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  352. if (elt && ucl_object_toboolean (elt)) {
  353. up->compress = TRUE;
  354. }
  355. elt = ucl_object_lookup (obj, "hosts");
  356. if (elt == NULL && !up->self_scan) {
  357. g_set_error (err, rspamd_proxy_quark (), 100,
  358. "upstream option must have some hosts definition");
  359. goto err;
  360. }
  361. if (elt) {
  362. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  363. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  364. g_set_error (err, rspamd_proxy_quark (), 100,
  365. "upstream has bad hosts definition");
  366. goto err;
  367. }
  368. rspamd_mempool_add_destructor (pool,
  369. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  370. }
  371. elt = ucl_object_lookup (obj, "default");
  372. if (elt) {
  373. if (ucl_object_toboolean (elt)) {
  374. ctx->default_upstream = up;
  375. }
  376. }
  377. else if (up->self_scan) {
  378. ctx->default_upstream = up;
  379. }
  380. elt = ucl_object_lookup (obj, "local");
  381. if (elt && ucl_object_toboolean (elt)) {
  382. up->local = TRUE;
  383. }
  384. elt = ucl_object_lookup (obj, "timeout");
  385. if (elt) {
  386. ucl_object_todouble_safe (elt, &up->timeout);
  387. }
  388. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  389. if (elt && ucl_object_type (elt) == UCL_STRING) {
  390. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  391. }
  392. /*
  393. * Accept lua function here in form
  394. * fun :: String -> UCL
  395. */
  396. elt = ucl_object_lookup (obj, "parser");
  397. if (elt) {
  398. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  399. &up->parser_to_ref, err)) {
  400. goto err;
  401. }
  402. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  403. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  404. }
  405. double_to_tv (up->timeout, &up->io_tv);
  406. g_hash_table_insert (ctx->upstreams, up->name, up);
  407. return TRUE;
  408. err:
  409. return FALSE;
  410. }
  411. static gboolean
  412. rspamd_proxy_parse_mirror (rspamd_mempool_t *pool,
  413. const ucl_object_t *obj,
  414. gpointer ud,
  415. struct rspamd_rcl_section *section,
  416. GError **err)
  417. {
  418. const ucl_object_t *elt;
  419. struct rspamd_http_mirror *up = NULL;
  420. struct rspamd_proxy_ctx *ctx;
  421. struct rspamd_rcl_struct_parser *pd = ud;
  422. lua_State *L;
  423. ctx = pd->user_struct;
  424. L = ctx->lua_state;
  425. if (ucl_object_type (obj) != UCL_OBJECT) {
  426. g_set_error (err, rspamd_proxy_quark (), 100,
  427. "mirror option must be an object");
  428. return FALSE;
  429. }
  430. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  431. elt = ucl_object_lookup (obj, "name");
  432. if (elt == NULL) {
  433. if (ucl_object_key (obj)) {
  434. if (strcmp (ucl_object_key (obj), "mirror") == 0) {
  435. /* Iterate over the object and find upstream elements */
  436. ucl_object_iter_t it = NULL;
  437. const ucl_object_t *cur;
  438. gboolean ret = TRUE;
  439. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  440. if (!rspamd_proxy_parse_mirror (pool, cur, ud,
  441. section, err)) {
  442. ret = FALSE;
  443. }
  444. }
  445. return ret;
  446. }
  447. else {
  448. /* Inside upstream */
  449. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  450. }
  451. }
  452. else {
  453. g_set_error (err, rspamd_proxy_quark (), 100,
  454. "mirror option must have some name definition");
  455. return FALSE;
  456. }
  457. }
  458. else {
  459. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  460. }
  461. up->parser_to_ref = -1;
  462. up->parser_from_ref = -1;
  463. up->timeout = ctx->timeout;
  464. elt = ucl_object_lookup (obj, "key");
  465. if (elt != NULL) {
  466. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  467. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  468. if (up->key == NULL) {
  469. g_set_error (err, rspamd_proxy_quark (), 100,
  470. "cannot read mirror key");
  471. goto err;
  472. }
  473. rspamd_mempool_add_destructor (pool,
  474. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  475. }
  476. elt = ucl_object_lookup (obj, "hosts");
  477. if (elt == NULL) {
  478. g_set_error (err, rspamd_proxy_quark (), 100,
  479. "mirror option must have some hosts definition");
  480. goto err;
  481. }
  482. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  483. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  484. g_set_error (err, rspamd_proxy_quark (), 100,
  485. "mirror has bad hosts definition");
  486. goto err;
  487. }
  488. rspamd_mempool_add_destructor (pool,
  489. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  490. elt = ucl_object_lookup_any (obj, "probability", "prob", NULL);
  491. if (elt) {
  492. up->prob = ucl_object_todouble (elt);
  493. }
  494. else {
  495. up->prob = 1.0;
  496. }
  497. elt = ucl_object_lookup (obj, "local");
  498. if (elt && ucl_object_toboolean (elt)) {
  499. up->local = TRUE;
  500. }
  501. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  502. if (elt && ucl_object_toboolean (elt)) {
  503. up->compress = TRUE;
  504. }
  505. elt = ucl_object_lookup (obj, "timeout");
  506. if (elt) {
  507. ucl_object_todouble_safe (elt, &up->timeout);
  508. }
  509. /*
  510. * Accept lua function here in form
  511. * fun :: String -> UCL
  512. */
  513. elt = ucl_object_lookup (obj, "parser");
  514. if (elt) {
  515. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  516. &up->parser_to_ref, err)) {
  517. goto err;
  518. }
  519. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  520. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  521. }
  522. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  523. if (elt && ucl_object_type (elt) == UCL_STRING) {
  524. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  525. }
  526. double_to_tv (up->timeout, &up->io_tv);
  527. g_ptr_array_add (ctx->mirrors, up);
  528. return TRUE;
  529. err:
  530. return FALSE;
  531. }
  532. static gboolean
  533. rspamd_proxy_parse_script (rspamd_mempool_t *pool,
  534. const ucl_object_t *obj,
  535. gpointer ud,
  536. struct rspamd_rcl_section *section,
  537. GError **err)
  538. {
  539. struct rspamd_proxy_ctx *ctx;
  540. struct rspamd_rcl_struct_parser *pd = ud;
  541. lua_State *L;
  542. const gchar *lua_script;
  543. gsize slen;
  544. gint err_idx, ref_idx;
  545. GString *tb = NULL;
  546. struct stat st;
  547. ctx = pd->user_struct;
  548. L = ctx->lua_state;
  549. if (ucl_object_type (obj) != UCL_STRING) {
  550. g_set_error (err, rspamd_proxy_quark (), 100,
  551. "script option must be a string with file or lua chunk");
  552. return FALSE;
  553. }
  554. lua_script = ucl_object_tolstring (obj, &slen);
  555. lua_pushcfunction (L, &rspamd_lua_traceback);
  556. err_idx = lua_gettop (L);
  557. if (stat (lua_script, &st) != -1) {
  558. /* Load file */
  559. if (luaL_loadfile (L, lua_script) != 0) {
  560. g_set_error (err,
  561. rspamd_proxy_quark (),
  562. EINVAL,
  563. "cannot load lua parser script: %s",
  564. lua_tostring (L, -1));
  565. lua_settop (L, 0); /* Error function */
  566. goto err;
  567. }
  568. }
  569. else {
  570. /* Load data directly */
  571. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  572. g_set_error (err,
  573. rspamd_proxy_quark (),
  574. EINVAL,
  575. "cannot load lua parser script: %s",
  576. lua_tostring (L, -1));
  577. lua_settop (L, 0); /* Error function */
  578. goto err;
  579. }
  580. }
  581. /* Now do it */
  582. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  583. tb = lua_touserdata (L, -1);
  584. g_set_error (err,
  585. rspamd_proxy_quark (),
  586. EINVAL,
  587. "cannot init lua parser script: %s",
  588. tb->str);
  589. g_string_free (tb, TRUE);
  590. lua_settop (L, 0);
  591. goto err;
  592. }
  593. if (!lua_isfunction (L, -1)) {
  594. g_set_error (err,
  595. rspamd_proxy_quark (),
  596. EINVAL,
  597. "cannot init lua parser script: "
  598. "must return function, %s returned",
  599. lua_typename (L, lua_type (L, -1)));
  600. lua_settop (L, 0);
  601. goto err;
  602. }
  603. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  604. lua_settop (L, 0);
  605. g_array_append_val (ctx->cmp_refs, ref_idx);
  606. return TRUE;
  607. err:
  608. return FALSE;
  609. }
  610. gpointer
  611. init_rspamd_proxy (struct rspamd_config *cfg)
  612. {
  613. struct rspamd_proxy_ctx *ctx;
  614. GQuark type;
  615. type = g_quark_try_string ("rspamd_proxy");
  616. ctx = rspamd_mempool_alloc0 (cfg->cfg_pool,
  617. sizeof (struct rspamd_proxy_ctx));
  618. ctx->magic = rspamd_rspamd_proxy_magic;
  619. ctx->timeout = 120.0;
  620. ctx->upstreams = g_hash_table_new (rspamd_strcase_hash, rspamd_strcase_equal);
  621. rspamd_mempool_add_destructor (cfg->cfg_pool,
  622. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->upstreams);
  623. ctx->mirrors = g_ptr_array_new ();
  624. rspamd_mempool_add_destructor (cfg->cfg_pool,
  625. (rspamd_mempool_destruct_t)rspamd_ptr_array_free_hard, ctx->mirrors);
  626. ctx->cfg = cfg;
  627. ctx->lua_state = cfg->lua_state;
  628. ctx->cmp_refs = g_array_new (FALSE, FALSE, sizeof (gint));
  629. rspamd_mempool_add_destructor (cfg->cfg_pool,
  630. (rspamd_mempool_destruct_t)rspamd_array_free_hard, ctx->cmp_refs);
  631. ctx->max_retries = DEFAULT_RETRIES;
  632. ctx->spam_header = RSPAMD_MILTER_SPAM_HEADER;
  633. rspamd_rcl_register_worker_option (cfg,
  634. type,
  635. "timeout",
  636. rspamd_rcl_parse_struct_time,
  637. ctx,
  638. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, timeout),
  639. RSPAMD_CL_FLAG_TIME_FLOAT,
  640. "IO timeout");
  641. rspamd_rcl_register_worker_option (cfg,
  642. type,
  643. "keypair",
  644. rspamd_rcl_parse_struct_keypair,
  645. ctx,
  646. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, key),
  647. 0,
  648. "Server's keypair");
  649. rspamd_rcl_register_worker_option (cfg,
  650. type,
  651. "upstream",
  652. rspamd_proxy_parse_upstream,
  653. ctx,
  654. 0,
  655. 0,
  656. "List of upstreams");
  657. rspamd_rcl_register_worker_option (cfg,
  658. type,
  659. "mirror",
  660. rspamd_proxy_parse_mirror,
  661. ctx,
  662. 0,
  663. RSPAMD_CL_FLAG_MULTIPLE,
  664. "List of mirrors");
  665. rspamd_rcl_register_worker_option (cfg,
  666. type,
  667. "script",
  668. rspamd_proxy_parse_script,
  669. ctx,
  670. 0,
  671. RSPAMD_CL_FLAG_MULTIPLE,
  672. "Compare script to be executed");
  673. rspamd_rcl_register_worker_option (cfg,
  674. type,
  675. "max_retries",
  676. rspamd_rcl_parse_struct_integer,
  677. ctx,
  678. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, max_retries),
  679. RSPAMD_CL_FLAG_UINT,
  680. "Maximum number of retries for master connection");
  681. rspamd_rcl_register_worker_option (cfg,
  682. type,
  683. "milter",
  684. rspamd_rcl_parse_struct_boolean,
  685. ctx,
  686. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, milter),
  687. 0,
  688. "Accept milter connections, not HTTP");
  689. rspamd_rcl_register_worker_option (cfg,
  690. type,
  691. "discard_on_reject",
  692. rspamd_rcl_parse_struct_boolean,
  693. ctx,
  694. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, discard_on_reject),
  695. 0,
  696. "Tell MTA to discard rejected messages silently");
  697. rspamd_rcl_register_worker_option (cfg,
  698. type,
  699. "quarantine_on_reject",
  700. rspamd_rcl_parse_struct_boolean,
  701. ctx,
  702. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, quarantine_on_reject),
  703. 0,
  704. "Tell MTA to quarantine rejected messages");
  705. rspamd_rcl_register_worker_option (cfg,
  706. type,
  707. "spam_header",
  708. rspamd_rcl_parse_struct_string,
  709. ctx,
  710. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, spam_header),
  711. 0,
  712. "Use the specific spam header (default: X-Spam)");
  713. rspamd_rcl_register_worker_option (cfg,
  714. type,
  715. "client_ca_name",
  716. rspamd_rcl_parse_struct_string,
  717. ctx,
  718. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, client_ca_name),
  719. 0,
  720. "Allow certificates issued by this CA to be treated as client certificates");
  721. rspamd_rcl_register_worker_option (cfg,
  722. type,
  723. "reject_message",
  724. rspamd_rcl_parse_struct_string,
  725. ctx,
  726. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, reject_message),
  727. 0,
  728. "Use custom rejection message");
  729. return ctx;
  730. }
  731. static void
  732. proxy_backend_close_connection (struct rspamd_proxy_backend_connection *conn)
  733. {
  734. if (conn && !(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  735. if (conn->backend_conn) {
  736. rspamd_http_connection_reset (conn->backend_conn);
  737. rspamd_http_connection_unref (conn->backend_conn);
  738. close (conn->backend_sock);
  739. }
  740. conn->flags |= RSPAMD_BACKEND_CLOSED;
  741. }
  742. }
  743. static gboolean
  744. proxy_backend_parse_results (struct rspamd_proxy_session *session,
  745. struct rspamd_proxy_backend_connection *conn,
  746. lua_State *L, gint parser_ref,
  747. const gchar *in, gsize inlen)
  748. {
  749. struct ucl_parser *parser;
  750. GString *tb = NULL;
  751. gint err_idx;
  752. if (inlen == 0 || in == NULL) {
  753. return FALSE;
  754. }
  755. if (parser_ref != -1) {
  756. /* Call parser function */
  757. lua_pushcfunction (L, &rspamd_lua_traceback);
  758. err_idx = lua_gettop (L);
  759. lua_rawgeti (L, LUA_REGISTRYINDEX, parser_ref);
  760. /* XXX: copies all data */
  761. lua_pushlstring (L, in, inlen);
  762. if (lua_pcall (L, 1, 1, err_idx) != 0) {
  763. tb = lua_touserdata (L, -1);
  764. msg_err_session (
  765. "cannot run lua parser script: %s",
  766. tb->str);
  767. g_string_free (tb, TRUE);
  768. lua_settop (L, 0);
  769. return FALSE;
  770. }
  771. conn->results = ucl_object_lua_import (L, -1);
  772. lua_settop (L, 0);
  773. }
  774. else {
  775. parser = ucl_parser_new (0);
  776. if (!ucl_parser_add_chunk (parser, in, inlen)) {
  777. gchar *encoded;
  778. encoded = rspamd_encode_base64 (in, inlen, 0, NULL);
  779. msg_err_session ("cannot parse input: %s", ucl_parser_get_error (
  780. parser));
  781. msg_err_session ("input encoded: %s", encoded);
  782. ucl_parser_free (parser);
  783. g_free (encoded);
  784. return FALSE;
  785. }
  786. conn->results = ucl_parser_get_object (parser);
  787. ucl_parser_free (parser);
  788. }
  789. return TRUE;
  790. }
  791. static void
  792. proxy_call_cmp_script (struct rspamd_proxy_session *session, gint cbref)
  793. {
  794. GString *tb = NULL;
  795. gint err_idx;
  796. guint i;
  797. struct rspamd_proxy_backend_connection *conn;
  798. lua_State *L;
  799. L = session->ctx->lua_state;
  800. lua_pushcfunction (L, &rspamd_lua_traceback);
  801. err_idx = lua_gettop (L);
  802. lua_rawgeti (L, LUA_REGISTRYINDEX, cbref);
  803. lua_createtable (L, 0, session->mirror_conns->len + 1);
  804. /* Now push master results */
  805. if (session->master_conn && session->master_conn->results) {
  806. lua_pushstring (L, "master");
  807. ucl_object_push_lua (L, session->master_conn->results, true);
  808. lua_settable (L, -3);
  809. }
  810. else {
  811. lua_pushstring (L, "master");
  812. lua_pushstring (L, "no results");
  813. lua_settable (L, -3);
  814. }
  815. for (i = 0; i < session->mirror_conns->len; i ++) {
  816. conn = g_ptr_array_index (session->mirror_conns, i);
  817. if (conn->results) {
  818. lua_pushstring (L, conn->name);
  819. ucl_object_push_lua (L, conn->results, true);
  820. lua_settable (L, -3);
  821. }
  822. else {
  823. lua_pushstring (L, conn->name);
  824. lua_pushstring (L, conn->err ? conn->err : "unknown error");
  825. lua_settable (L, -3);
  826. }
  827. }
  828. if (lua_pcall (L, 1, 0, err_idx) != 0) {
  829. tb = lua_touserdata (L, -1);
  830. msg_err_session (
  831. "cannot run lua compare script: %s",
  832. tb->str);
  833. g_string_free (tb, TRUE);
  834. }
  835. lua_settop (L, 0);
  836. }
  837. static void
  838. proxy_session_dtor (struct rspamd_proxy_session *session)
  839. {
  840. guint i;
  841. gint cbref;
  842. struct rspamd_proxy_backend_connection *conn;
  843. if (session->master_conn && session->master_conn->results) {
  844. for (i = 0; i < session->ctx->cmp_refs->len; i++) {
  845. cbref = g_array_index (session->ctx->cmp_refs, gint, i);
  846. proxy_call_cmp_script (session, cbref);
  847. }
  848. }
  849. if (session->master_conn) {
  850. proxy_backend_close_connection (session->master_conn);
  851. }
  852. if (session->client_milter_conn) {
  853. rspamd_milter_session_unref (session->client_milter_conn);
  854. }
  855. else if (session->client_conn) {
  856. rspamd_http_connection_reset (session->client_conn);
  857. rspamd_http_connection_unref (session->client_conn);
  858. }
  859. if (session->map && session->map_len) {
  860. munmap (session->map, session->map_len);
  861. }
  862. for (i = 0; i < session->mirror_conns->len; i ++) {
  863. conn = g_ptr_array_index (session->mirror_conns, i);
  864. if (!(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  865. proxy_backend_close_connection (conn);
  866. }
  867. if (conn->results) {
  868. ucl_object_unref (conn->results);
  869. }
  870. }
  871. if (session->master_conn) {
  872. if (session->master_conn->results) {
  873. ucl_object_unref (session->master_conn->results);
  874. }
  875. if (session->master_conn->task) {
  876. rspamd_session_destroy (session->master_conn->task->s);
  877. }
  878. }
  879. g_ptr_array_free (session->mirror_conns, TRUE);
  880. rspamd_http_message_shmem_unref (session->shmem_ref);
  881. rspamd_http_message_unref (session->client_message);
  882. if (session->client_addr) {
  883. rspamd_inet_address_free (session->client_addr);
  884. }
  885. if (session->client_sock != -1) {
  886. close (session->client_sock);
  887. }
  888. if (session->ctx->sessions_cache) {
  889. rspamd_worker_session_cache_remove (session->ctx->sessions_cache,
  890. session);
  891. }
  892. if (session->pool) {
  893. rspamd_mempool_delete (session->pool);
  894. }
  895. g_free (session);
  896. }
  897. static void
  898. proxy_request_compress (struct rspamd_http_message *msg)
  899. {
  900. guint flags;
  901. ZSTD_CCtx *zctx;
  902. rspamd_fstring_t *body;
  903. const gchar *in;
  904. gsize inlen;
  905. flags = rspamd_http_message_get_flags (msg);
  906. if (!rspamd_http_message_find_header (msg, "Compression")) {
  907. if ((flags & RSPAMD_HTTP_FLAG_SHMEM) ||
  908. !(flags & RSPAMD_HTTP_FLAG_HAS_BODY)) {
  909. /* Cannot compress shared or empty message */
  910. return;
  911. }
  912. in = rspamd_http_message_get_body (msg, &inlen);
  913. if (in == NULL || inlen == 0) {
  914. return;
  915. }
  916. body = rspamd_fstring_sized_new (ZSTD_compressBound (inlen));
  917. zctx = ZSTD_createCCtx ();
  918. body->len = ZSTD_compressCCtx (zctx, body->str, body->allocated,
  919. in, inlen, 1);
  920. if (ZSTD_isError (body->len)) {
  921. msg_err ("compression error");
  922. rspamd_fstring_free (body);
  923. ZSTD_freeCCtx (zctx);
  924. return;
  925. }
  926. ZSTD_freeCCtx (zctx);
  927. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  928. rspamd_http_message_add_header (msg, "Compression", "zstd");
  929. }
  930. }
  931. static void
  932. proxy_request_decompress (struct rspamd_http_message *msg)
  933. {
  934. rspamd_fstring_t *body;
  935. const gchar *in;
  936. gsize inlen, outlen, r;
  937. ZSTD_DStream *zstream;
  938. ZSTD_inBuffer zin;
  939. ZSTD_outBuffer zout;
  940. if (rspamd_http_message_find_header (msg, "Compression")) {
  941. in = rspamd_http_message_get_body (msg, &inlen);
  942. if (in == NULL || inlen == 0) {
  943. return;
  944. }
  945. zstream = ZSTD_createDStream ();
  946. ZSTD_initDStream (zstream);
  947. zin.pos = 0;
  948. zin.src = in;
  949. zin.size = inlen;
  950. if ((outlen = ZSTD_getDecompressedSize (zin.src, zin.size)) == 0) {
  951. outlen = ZSTD_DStreamOutSize ();
  952. }
  953. body = rspamd_fstring_sized_new (outlen);
  954. zout.dst = body->str;
  955. zout.pos = 0;
  956. zout.size = outlen;
  957. while (zin.pos < zin.size) {
  958. r = ZSTD_decompressStream (zstream, &zout, &zin);
  959. if (ZSTD_isError (r)) {
  960. msg_err ("Decompression error: %s", ZSTD_getErrorName (r));
  961. ZSTD_freeDStream (zstream);
  962. rspamd_fstring_free (body);
  963. return;
  964. }
  965. if (zout.pos == zout.size) {
  966. /* We need to extend output buffer */
  967. zout.size = zout.size * 1.5 + 1.0;
  968. body = rspamd_fstring_grow (body, zout.size);
  969. zout.size = body->allocated;
  970. zout.dst = body->str;
  971. }
  972. }
  973. body->len = zout.pos;
  974. ZSTD_freeDStream (zstream);
  975. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  976. rspamd_http_message_remove_header (msg, "Compression");
  977. }
  978. return;
  979. }
  980. static struct rspamd_proxy_session *
  981. proxy_session_refresh (struct rspamd_proxy_session *session)
  982. {
  983. struct rspamd_proxy_session *nsession;
  984. nsession = g_malloc0 (sizeof (*nsession));
  985. nsession->client_milter_conn = session->client_milter_conn;
  986. session->client_milter_conn = NULL;
  987. rspamd_milter_update_userdata (nsession->client_milter_conn,
  988. nsession);
  989. nsession->client_addr = session->client_addr;
  990. session->client_addr = NULL;
  991. nsession->ctx = session->ctx;
  992. nsession->worker = session->worker;
  993. nsession->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (), "proxy");
  994. nsession->client_sock = session->client_sock;
  995. session->client_sock = -1;
  996. nsession->mirror_conns = g_ptr_array_sized_new (nsession->ctx->mirrors->len);
  997. REF_INIT_RETAIN (nsession, proxy_session_dtor);
  998. if (nsession->ctx->sessions_cache) {
  999. rspamd_worker_session_cache_add (nsession->ctx->sessions_cache,
  1000. nsession->pool->tag.uid, &nsession->ref.refcount, nsession);
  1001. }
  1002. return nsession;
  1003. }
  1004. static gboolean
  1005. proxy_check_file (struct rspamd_http_message *msg,
  1006. struct rspamd_proxy_session *session)
  1007. {
  1008. const rspamd_ftok_t *tok, *key_tok;
  1009. rspamd_ftok_t srch;
  1010. gchar *file_str;
  1011. GHashTable *query_args;
  1012. GHashTableIter it;
  1013. gpointer k, v;
  1014. struct http_parser_url u;
  1015. rspamd_fstring_t *new_url;
  1016. tok = rspamd_http_message_find_header (msg, "File");
  1017. if (tok) {
  1018. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1019. session->map = rspamd_file_xmap (file_str, PROT_READ, &session->map_len,
  1020. TRUE);
  1021. if (session->map == NULL) {
  1022. if (session->map_len != 0) {
  1023. msg_err_session ("cannot map %s: %s", file_str,
  1024. strerror (errno));
  1025. return FALSE;
  1026. }
  1027. }
  1028. /* Remove header after processing */
  1029. rspamd_http_message_remove_header (msg, "File");
  1030. session->fname = file_str;
  1031. }
  1032. else {
  1033. /* Need to parse query URL */
  1034. if (http_parser_parse_url (RSPAMD_FSTRING_DATA (msg->url),
  1035. RSPAMD_FSTRING_LEN (msg->url), 0, &u) != 0) {
  1036. msg_err_session ("bad request url: %V", msg->url);
  1037. return FALSE;
  1038. }
  1039. if (u.field_set & (1 << UF_QUERY)) {
  1040. /* In case if we have a query, we need to store it somewhere */
  1041. query_args = rspamd_http_message_parse_query (msg);
  1042. srch.begin = "File";
  1043. srch.len = strlen ("File");
  1044. tok = g_hash_table_lookup (query_args, &srch);
  1045. if (tok) {
  1046. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1047. session->map = rspamd_file_xmap (file_str, PROT_READ,
  1048. &session->map_len, TRUE);
  1049. if (session->map == NULL) {
  1050. if (session->map_len != 0) {
  1051. msg_err_session ("cannot map %s: %s", file_str,
  1052. strerror (errno));
  1053. g_hash_table_unref (query_args);
  1054. return FALSE;
  1055. }
  1056. }
  1057. /* We need to create a new URL with file attribute removed */
  1058. new_url = rspamd_fstring_new_init (RSPAMD_FSTRING_DATA (msg->url),
  1059. u.field_data[UF_QUERY].off);
  1060. new_url = rspamd_fstring_append (new_url, "?", 1);
  1061. g_hash_table_iter_init (&it, query_args);
  1062. while (g_hash_table_iter_next (&it, &k, &v)) {
  1063. key_tok = k;
  1064. tok = v;
  1065. if (!rspamd_ftok_icase_equal (key_tok, &srch)) {
  1066. rspamd_printf_fstring (&new_url, "%T=%T&",
  1067. key_tok, tok);
  1068. }
  1069. }
  1070. /* Erase last character (might be either & or ?) */
  1071. rspamd_fstring_erase (new_url, new_url->len - 1, 1);
  1072. rspamd_fstring_free (msg->url);
  1073. msg->url = new_url;
  1074. session->fname = file_str;
  1075. }
  1076. g_hash_table_unref (query_args);
  1077. }
  1078. }
  1079. return TRUE;
  1080. }
  1081. static void
  1082. proxy_backend_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  1083. {
  1084. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1085. struct rspamd_proxy_session *session;
  1086. session = bk_conn->s;
  1087. msg_info_session ("abnormally closing connection from backend: %s:%s, "
  1088. "error: %e",
  1089. bk_conn->name,
  1090. rspamd_inet_address_to_string (
  1091. rspamd_upstream_addr_cur (bk_conn->up)),
  1092. err);
  1093. if (err) {
  1094. bk_conn->err = rspamd_mempool_strdup (session->pool, err->message);
  1095. }
  1096. rspamd_upstream_fail (bk_conn->up, FALSE);
  1097. proxy_backend_close_connection (bk_conn);
  1098. REF_RELEASE (bk_conn->s);
  1099. }
  1100. static gint
  1101. proxy_backend_mirror_finish_handler (struct rspamd_http_connection *conn,
  1102. struct rspamd_http_message *msg)
  1103. {
  1104. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1105. struct rspamd_proxy_session *session;
  1106. session = bk_conn->s;
  1107. proxy_request_decompress (msg);
  1108. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1109. bk_conn->parser_from_ref, msg->body_buf.begin, msg->body_buf.len)) {
  1110. msg_warn_session ("cannot parse results from the mirror backend %s:%s",
  1111. bk_conn->name,
  1112. rspamd_inet_address_to_string (
  1113. rspamd_upstream_addr_cur (bk_conn->up)));
  1114. bk_conn->err = "cannot parse ucl";
  1115. }
  1116. msg_info_session ("finished mirror connection to %s", bk_conn->name);
  1117. rspamd_upstream_ok (bk_conn->up);
  1118. proxy_backend_close_connection (bk_conn);
  1119. REF_RELEASE (bk_conn->s);
  1120. return 0;
  1121. }
  1122. static void
  1123. proxy_open_mirror_connections (struct rspamd_proxy_session *session)
  1124. {
  1125. gdouble coin;
  1126. struct rspamd_http_mirror *m;
  1127. guint i;
  1128. struct rspamd_proxy_backend_connection *bk_conn;
  1129. struct rspamd_http_message *msg;
  1130. GError *err = NULL;
  1131. coin = rspamd_random_double ();
  1132. for (i = 0; i < session->ctx->mirrors->len; i ++) {
  1133. m = g_ptr_array_index (session->ctx->mirrors, i);
  1134. if (m->prob < coin) {
  1135. /* No luck */
  1136. continue;
  1137. }
  1138. bk_conn = rspamd_mempool_alloc0 (session->pool,
  1139. sizeof (*bk_conn));
  1140. bk_conn->s = session;
  1141. bk_conn->name = m->name;
  1142. bk_conn->io_tv = &m->io_tv;
  1143. bk_conn->up = rspamd_upstream_get (m->u,
  1144. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1145. bk_conn->parser_from_ref = m->parser_from_ref;
  1146. bk_conn->parser_to_ref = m->parser_to_ref;
  1147. if (bk_conn->up == NULL) {
  1148. msg_err_session ("cannot select upstream for %s", m->name);
  1149. continue;
  1150. }
  1151. bk_conn->backend_sock = rspamd_inet_address_connect (
  1152. rspamd_upstream_addr_next (bk_conn->up),
  1153. SOCK_STREAM, TRUE);
  1154. if (bk_conn->backend_sock == -1) {
  1155. msg_err_session ("cannot connect upstream for %s", m->name);
  1156. rspamd_upstream_fail (bk_conn->up, TRUE);
  1157. continue;
  1158. }
  1159. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1160. if (msg == NULL) {
  1161. msg_err_session ("cannot copy message to send to a mirror %s: %e",
  1162. m->name, err);
  1163. if (err) {
  1164. g_error_free (err);
  1165. }
  1166. continue;
  1167. }
  1168. if (msg->url->len == 0) {
  1169. msg->url = rspamd_fstring_append (msg->url, "/check", strlen ("/check"));
  1170. }
  1171. if (m->settings_id != NULL) {
  1172. rspamd_http_message_remove_header (msg, "Settings-ID");
  1173. rspamd_http_message_add_header (msg, "Settings-ID", m->settings_id);
  1174. }
  1175. bk_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1176. session->ctx->http_ctx,
  1177. NULL,
  1178. proxy_backend_mirror_error_handler,
  1179. proxy_backend_mirror_finish_handler,
  1180. RSPAMD_HTTP_CLIENT_SIMPLE,
  1181. bk_conn->backend_sock);
  1182. if (m->key) {
  1183. msg->peer_key = rspamd_pubkey_ref (m->key);
  1184. }
  1185. if (m->local ||
  1186. rspamd_inet_address_is_local (
  1187. rspamd_upstream_addr_cur (bk_conn->up), FALSE)) {
  1188. if (session->fname) {
  1189. rspamd_http_message_add_header (msg, "File", session->fname);
  1190. }
  1191. msg->method = HTTP_GET;
  1192. rspamd_http_connection_write_message_shared (bk_conn->backend_conn,
  1193. msg, NULL, NULL, bk_conn,
  1194. bk_conn->io_tv);
  1195. }
  1196. else {
  1197. if (session->fname) {
  1198. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1199. rspamd_http_message_set_body (msg, session->map, session->map_len);
  1200. }
  1201. msg->method = HTTP_POST;
  1202. if (m->compress) {
  1203. proxy_request_compress (msg);
  1204. if (session->client_milter_conn) {
  1205. rspamd_http_message_add_header (msg, "Content-Type",
  1206. "application/octet-stream");
  1207. }
  1208. }
  1209. else {
  1210. if (session->client_milter_conn) {
  1211. rspamd_http_message_add_header (msg, "Content-Type",
  1212. "text/plain");
  1213. }
  1214. }
  1215. rspamd_http_connection_write_message (bk_conn->backend_conn,
  1216. msg, NULL, NULL, bk_conn,
  1217. bk_conn->io_tv);
  1218. }
  1219. g_ptr_array_add (session->mirror_conns, bk_conn);
  1220. REF_RETAIN (session);
  1221. msg_info_session ("send request to %s", m->name);
  1222. }
  1223. }
  1224. static void
  1225. proxy_client_write_error (struct rspamd_proxy_session *session, gint code,
  1226. const gchar *status)
  1227. {
  1228. struct rspamd_http_message *reply;
  1229. if (session->client_milter_conn) {
  1230. rspamd_milter_send_action (session->client_milter_conn,
  1231. RSPAMD_MILTER_TEMPFAIL);
  1232. REF_RELEASE (session);
  1233. }
  1234. else {
  1235. reply = rspamd_http_new_message (HTTP_RESPONSE);
  1236. reply->code = code;
  1237. reply->status = rspamd_fstring_new_init (status, strlen (status));
  1238. rspamd_http_connection_write_message (session->client_conn,
  1239. reply, NULL, NULL, session,
  1240. &session->ctx->io_tv);
  1241. }
  1242. }
  1243. static void
  1244. proxy_backend_master_error_handler (struct rspamd_http_connection *conn, GError *err)
  1245. {
  1246. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1247. struct rspamd_proxy_session *session;
  1248. session = bk_conn->s;
  1249. msg_info_session ("abnormally closing connection from backend: %s, error: %e,"
  1250. " retries left: %d",
  1251. rspamd_inet_address_to_string (
  1252. rspamd_upstream_addr_cur (session->master_conn->up)),
  1253. err,
  1254. session->ctx->max_retries - session->retries);
  1255. session->retries ++;
  1256. rspamd_upstream_fail (bk_conn->up, FALSE);
  1257. proxy_backend_close_connection (session->master_conn);
  1258. if (session->ctx->max_retries &&
  1259. session->retries > session->ctx->max_retries) {
  1260. msg_err_session ("cannot connect to upstream, maximum retries "
  1261. "has been reached: %d", session->retries);
  1262. /* Terminate session immediately */
  1263. proxy_client_write_error (session, err->code, err->message);
  1264. }
  1265. else {
  1266. if (!proxy_send_master_message (session)) {
  1267. proxy_client_write_error (session, err->code, err->message);
  1268. }
  1269. else {
  1270. msg_info_session ("retry connection to: %s"
  1271. " retries left: %d",
  1272. rspamd_inet_address_to_string (
  1273. rspamd_upstream_addr_cur (session->master_conn->up)),
  1274. session->ctx->max_retries - session->retries);
  1275. }
  1276. }
  1277. }
  1278. static gint
  1279. proxy_backend_master_finish_handler (struct rspamd_http_connection *conn,
  1280. struct rspamd_http_message *msg)
  1281. {
  1282. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1283. struct rspamd_proxy_session *session, *nsession;
  1284. rspamd_fstring_t *reply;
  1285. session = bk_conn->s;
  1286. rspamd_http_connection_steal_msg (session->master_conn->backend_conn);
  1287. proxy_request_decompress (msg);
  1288. rspamd_http_message_remove_header (msg, "Content-Length");
  1289. rspamd_http_message_remove_header (msg, "Key");
  1290. rspamd_http_connection_reset (session->master_conn->backend_conn);
  1291. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1292. bk_conn->parser_from_ref, msg->body_buf.begin, msg->body_buf.len)) {
  1293. msg_warn_session ("cannot parse results from the master backend");
  1294. }
  1295. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1296. /* We need to reformat ucl to fit with legacy spamc protocol */
  1297. if (bk_conn->results) {
  1298. reply = rspamd_fstring_new ();
  1299. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1300. rspamd_ucl_tospamc_output (bk_conn->results, &reply);
  1301. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1302. }
  1303. else {
  1304. rspamd_ucl_torspamc_output (bk_conn->results, &reply);
  1305. }
  1306. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1307. msg->method = HTTP_SYMBOLS;
  1308. }
  1309. else {
  1310. msg_warn_session ("cannot parse results from the master backend, "
  1311. "return them as is");
  1312. }
  1313. }
  1314. rspamd_upstream_ok (bk_conn->up);
  1315. if (session->client_milter_conn) {
  1316. nsession = proxy_session_refresh (session);
  1317. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1318. session->master_conn->results);
  1319. REF_RELEASE (session);
  1320. rspamd_http_message_free (msg);
  1321. }
  1322. else {
  1323. rspamd_http_connection_write_message (session->client_conn,
  1324. msg, NULL, NULL, session,
  1325. bk_conn->io_tv);
  1326. }
  1327. return 0;
  1328. }
  1329. static void
  1330. rspamd_proxy_scan_self_reply (struct rspamd_task *task)
  1331. {
  1332. struct rspamd_http_message *msg;
  1333. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1334. ucl_object_t *rep = NULL;
  1335. const char *ctype = "application/json";
  1336. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1337. msg->date = time (NULL);
  1338. msg->code = 200;
  1339. switch (task->cmd) {
  1340. case CMD_REPORT_IFSPAM:
  1341. case CMD_REPORT:
  1342. case CMD_CHECK:
  1343. case CMD_SYMBOLS:
  1344. case CMD_PROCESS:
  1345. case CMD_SKIP:
  1346. case CMD_CHECK_V2:
  1347. rspamd_task_set_finish_time (task);
  1348. rspamd_protocol_http_reply (msg, task, &rep);
  1349. rspamd_protocol_write_log_pipe (task);
  1350. break;
  1351. case CMD_PING:
  1352. rspamd_http_message_set_body (msg, "pong" CRLF, 6);
  1353. ctype = "text/plain";
  1354. break;
  1355. case CMD_OTHER:
  1356. msg_err_task ("BROKEN");
  1357. break;
  1358. }
  1359. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1360. if (rep) {
  1361. session->master_conn->results = ucl_object_ref (rep);
  1362. }
  1363. if (session->client_milter_conn) {
  1364. nsession = proxy_session_refresh (session);
  1365. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1366. session->master_conn->results);
  1367. rspamd_http_message_free (msg);
  1368. REF_RELEASE (session);
  1369. }
  1370. else {
  1371. rspamd_http_connection_reset (session->client_conn);
  1372. rspamd_http_connection_write_message (session->client_conn,
  1373. msg,
  1374. NULL,
  1375. ctype,
  1376. session,
  1377. NULL);
  1378. }
  1379. }
  1380. static gboolean
  1381. rspamd_proxy_task_fin (void *ud)
  1382. {
  1383. struct rspamd_task *task = ud;
  1384. msg_debug_task ("finish task");
  1385. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1386. rspamd_proxy_scan_self_reply (task);
  1387. return TRUE;
  1388. }
  1389. if (!rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL)) {
  1390. rspamd_proxy_scan_self_reply (task);
  1391. return TRUE;
  1392. }
  1393. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1394. rspamd_proxy_scan_self_reply (task);
  1395. return TRUE;
  1396. }
  1397. /* One more iteration */
  1398. return FALSE;
  1399. }
  1400. static gboolean
  1401. rspamd_proxy_self_scan (struct rspamd_proxy_session *session)
  1402. {
  1403. struct rspamd_task *task;
  1404. struct rspamd_http_message *msg;
  1405. const gchar *data;
  1406. gsize len;
  1407. msg = session->client_message;
  1408. task = rspamd_task_new (session->worker, session->ctx->cfg,
  1409. session->pool, session->ctx->lang_det,
  1410. session->ctx->ev_base);
  1411. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1412. task->sock = -1;
  1413. if (session->client_milter_conn) {
  1414. task->client_addr = rspamd_inet_address_copy (
  1415. session->client_milter_conn->addr);
  1416. }
  1417. else {
  1418. task->client_addr = rspamd_inet_address_copy (session->client_addr);
  1419. }
  1420. task->fin_arg = session;
  1421. task->resolver = session->ctx->resolver;
  1422. /* TODO: allow to disable autolearn in protocol */
  1423. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1424. task->s = rspamd_session_create (task->task_pool, rspamd_proxy_task_fin,
  1425. NULL, (event_finalizer_t )rspamd_task_free, task);
  1426. data = rspamd_http_message_get_body (msg, &len);
  1427. if (session->backend->settings_id) {
  1428. rspamd_http_message_remove_header (msg, "Settings-ID");
  1429. rspamd_http_message_add_header (msg, "Settings-ID",
  1430. session->backend->settings_id);
  1431. }
  1432. /* Process message */
  1433. if (!rspamd_protocol_handle_request (task, msg)) {
  1434. msg_err_task ("cannot handle request: %e", task->err);
  1435. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1436. }
  1437. else {
  1438. if (task->cmd == CMD_PING) {
  1439. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1440. }
  1441. else {
  1442. if (!rspamd_task_load_message (task, msg, data, len)) {
  1443. msg_err_task ("cannot load message: %e", task->err);
  1444. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1445. }
  1446. }
  1447. }
  1448. /* Set global timeout for the task */
  1449. if (session->ctx->default_upstream->timeout > 0.0) {
  1450. struct timeval task_tv;
  1451. event_set (&task->timeout_ev, -1, EV_TIMEOUT, rspamd_task_timeout,
  1452. task);
  1453. event_base_set (session->ctx->ev_base, &task->timeout_ev);
  1454. double_to_tv (session->ctx->default_upstream->timeout, &task_tv);
  1455. event_add (&task->timeout_ev, &task_tv);
  1456. }
  1457. else if (session->ctx->has_self_scan) {
  1458. if (session->ctx->cfg->task_timeout > 0) {
  1459. struct timeval task_tv;
  1460. event_set (&task->timeout_ev, -1, EV_TIMEOUT, rspamd_task_timeout,
  1461. task);
  1462. event_base_set (session->ctx->ev_base, &task->timeout_ev);
  1463. double_to_tv (session->ctx->cfg->task_timeout, &task_tv);
  1464. event_add (&task->timeout_ev, &task_tv);
  1465. }
  1466. }
  1467. session->master_conn->task = task;
  1468. rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL);
  1469. rspamd_session_pending (task->s);
  1470. return TRUE;
  1471. }
  1472. static gboolean
  1473. proxy_send_master_message (struct rspamd_proxy_session *session)
  1474. {
  1475. struct rspamd_http_message *msg;
  1476. struct rspamd_http_upstream *backend = NULL;
  1477. const rspamd_ftok_t *host;
  1478. GError *err = NULL;
  1479. gchar hostbuf[512];
  1480. host = rspamd_http_message_find_header (session->client_message, "Host");
  1481. if (host == NULL) {
  1482. backend = session->ctx->default_upstream;
  1483. }
  1484. else {
  1485. rspamd_strlcpy (hostbuf, host->begin, MIN(host->len + 1, sizeof (hostbuf)));
  1486. backend = g_hash_table_lookup (session->ctx->upstreams, hostbuf);
  1487. if (backend == NULL) {
  1488. backend = session->ctx->default_upstream;
  1489. }
  1490. }
  1491. if (backend == NULL) {
  1492. /* No backend */
  1493. msg_err_session ("cannot find upstream for %s", host ? hostbuf : "default");
  1494. goto err;
  1495. }
  1496. else {
  1497. session->backend = backend;
  1498. if (backend->self_scan) {
  1499. return rspamd_proxy_self_scan (session);
  1500. }
  1501. retry:
  1502. if (session->ctx->max_retries &&
  1503. session->retries > session->ctx->max_retries) {
  1504. msg_err_session ("cannot connect to upstream, maximum retries "
  1505. "has been reached: %d", session->retries);
  1506. goto err;
  1507. }
  1508. session->master_conn->up = rspamd_upstream_get (backend->u,
  1509. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1510. session->master_conn->io_tv = &backend->io_tv;
  1511. if (session->master_conn->up == NULL) {
  1512. msg_err_session ("cannot select upstream for %s",
  1513. host ? hostbuf : "default");
  1514. goto err;
  1515. }
  1516. session->master_conn->backend_sock = rspamd_inet_address_connect (
  1517. rspamd_upstream_addr_next (session->master_conn->up),
  1518. SOCK_STREAM, TRUE);
  1519. if (session->master_conn->backend_sock == -1) {
  1520. msg_err_session ("cannot connect upstream: %s(%s)",
  1521. host ? hostbuf : "default",
  1522. rspamd_inet_address_to_string (
  1523. rspamd_upstream_addr_cur (
  1524. session->master_conn->up)));
  1525. rspamd_upstream_fail (session->master_conn->up, TRUE);
  1526. session->retries ++;
  1527. goto retry;
  1528. }
  1529. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1530. if (msg == NULL) {
  1531. msg_err_session ("cannot copy message to send it to the upstream: %e",
  1532. err);
  1533. if (err) {
  1534. g_error_free (err);
  1535. }
  1536. goto err; /* No fallback here */
  1537. }
  1538. session->master_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1539. session->ctx->http_ctx,
  1540. NULL,
  1541. proxy_backend_master_error_handler,
  1542. proxy_backend_master_finish_handler,
  1543. RSPAMD_HTTP_CLIENT_SIMPLE,
  1544. session->master_conn->backend_sock);
  1545. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1546. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1547. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1548. if (backend->key) {
  1549. msg->peer_key = rspamd_pubkey_ref (backend->key);
  1550. }
  1551. if (backend->settings_id != NULL) {
  1552. rspamd_http_message_remove_header (msg, "Settings-ID");
  1553. rspamd_http_message_add_header (msg, "Settings-ID",
  1554. backend->settings_id);
  1555. }
  1556. if (backend->local ||
  1557. rspamd_inet_address_is_local (
  1558. rspamd_upstream_addr_cur (
  1559. session->master_conn->up), FALSE)) {
  1560. if (session->fname) {
  1561. rspamd_http_message_add_header (msg, "File", session->fname);
  1562. }
  1563. msg->method = HTTP_GET;
  1564. rspamd_http_connection_write_message_shared (
  1565. session->master_conn->backend_conn,
  1566. msg, NULL, NULL, session->master_conn,
  1567. session->master_conn->io_tv);
  1568. }
  1569. else {
  1570. if (session->fname) {
  1571. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1572. rspamd_http_message_set_body (msg,
  1573. session->map, session->map_len);
  1574. }
  1575. msg->method = HTTP_POST;
  1576. if (backend->compress) {
  1577. proxy_request_compress (msg);
  1578. if (session->client_milter_conn) {
  1579. rspamd_http_message_add_header (msg, "Content-Type",
  1580. "application/octet-stream");
  1581. }
  1582. }
  1583. else {
  1584. if (session->client_milter_conn) {
  1585. rspamd_http_message_add_header (msg, "Content-Type",
  1586. "text/plain");
  1587. }
  1588. }
  1589. rspamd_http_connection_write_message (
  1590. session->master_conn->backend_conn,
  1591. msg, NULL, NULL, session->master_conn,
  1592. session->master_conn->io_tv);
  1593. }
  1594. }
  1595. return TRUE;
  1596. err:
  1597. if (session->client_milter_conn) {
  1598. rspamd_milter_send_action (session->client_milter_conn,
  1599. RSPAMD_MILTER_TEMPFAIL);
  1600. REF_RELEASE (session);
  1601. }
  1602. else {
  1603. rspamd_http_connection_steal_msg (session->client_conn);
  1604. rspamd_http_connection_reset (session->client_conn);
  1605. proxy_client_write_error (session, 404, "Backend not found");
  1606. }
  1607. return FALSE;
  1608. }
  1609. static void
  1610. proxy_client_error_handler (struct rspamd_http_connection *conn, GError *err)
  1611. {
  1612. struct rspamd_proxy_session *session = conn->ud;
  1613. msg_info_session ("abnormally closing connection from: %s, error: %s",
  1614. rspamd_inet_address_to_string (session->client_addr), err->message);
  1615. /* Terminate session immediately */
  1616. proxy_backend_close_connection (session->master_conn);
  1617. REF_RELEASE (session);
  1618. }
  1619. static gint
  1620. proxy_client_finish_handler (struct rspamd_http_connection *conn,
  1621. struct rspamd_http_message *msg)
  1622. {
  1623. struct rspamd_proxy_session *session = conn->ud;
  1624. if (!session->master_conn) {
  1625. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1626. sizeof (*session->master_conn));
  1627. session->master_conn->s = session;
  1628. session->master_conn->name = "master";
  1629. /* Reset spamc legacy */
  1630. if (msg->method >= HTTP_SYMBOLS) {
  1631. msg->method = HTTP_POST;
  1632. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1633. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1634. msg_info_session ("enabling legacy spamc mode for session");
  1635. }
  1636. else {
  1637. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1638. msg_info_session ("enabling legacy rspamc mode for session");
  1639. }
  1640. }
  1641. if (msg->url->len == 0) {
  1642. msg->url = rspamd_fstring_append (msg->url,
  1643. "/" MSG_CMD_CHECK_V2, strlen ("/" MSG_CMD_CHECK_V2));
  1644. }
  1645. if (!proxy_check_file (msg, session)) {
  1646. goto err;
  1647. }
  1648. session->client_message = rspamd_http_connection_steal_msg (
  1649. session->client_conn);
  1650. session->shmem_ref = rspamd_http_message_shmem_ref (session->client_message);
  1651. rspamd_http_message_remove_header (msg, "Content-Length");
  1652. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1653. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1654. rspamd_http_message_remove_header (msg, "Connection");
  1655. rspamd_http_message_remove_header (msg, "Key");
  1656. proxy_open_mirror_connections (session);
  1657. rspamd_http_connection_reset (session->client_conn);
  1658. proxy_send_master_message (session);
  1659. }
  1660. else {
  1661. msg_info_session ("finished master connection");
  1662. proxy_backend_close_connection (session->master_conn);
  1663. REF_RELEASE (session);
  1664. }
  1665. return 0;
  1666. err:
  1667. rspamd_http_connection_steal_msg (session->client_conn);
  1668. rspamd_http_message_remove_header (msg, "Content-Length");
  1669. rspamd_http_message_remove_header (msg, "Key");
  1670. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1671. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1672. rspamd_http_message_remove_header (msg, "Connection");
  1673. rspamd_http_connection_reset (session->client_conn);
  1674. proxy_client_write_error (session, 404, "Backend not found");
  1675. return 0;
  1676. }
  1677. static void
  1678. proxy_milter_finish_handler (gint fd,
  1679. struct rspamd_milter_session *rms,
  1680. void *ud)
  1681. {
  1682. struct rspamd_proxy_session *session = ud;
  1683. struct rspamd_http_message *msg;
  1684. session->client_milter_conn = rms;
  1685. if (rms->message == NULL || rms->message->len == 0) {
  1686. msg_info_session ("finished milter connection");
  1687. proxy_backend_close_connection (session->master_conn);
  1688. REF_RELEASE (session);
  1689. }
  1690. else {
  1691. if (!session->master_conn) {
  1692. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1693. sizeof (*session->master_conn));
  1694. }
  1695. msg = rspamd_milter_to_http (rms);
  1696. session->master_conn->s = session;
  1697. session->master_conn->name = "master";
  1698. session->client_message = msg;
  1699. proxy_open_mirror_connections (session);
  1700. proxy_send_master_message (session);
  1701. }
  1702. }
  1703. static void
  1704. proxy_milter_error_handler (gint fd,
  1705. struct rspamd_milter_session *rms,
  1706. void *ud, GError *err)
  1707. {
  1708. struct rspamd_proxy_session *session = ud;
  1709. msg_info_session ("abnormally closing milter connection from: %s, "
  1710. "error: %s", rspamd_inet_address_to_string (session->client_addr),
  1711. err->message);
  1712. /* Terminate session immediately */
  1713. proxy_backend_close_connection (session->master_conn);
  1714. REF_RELEASE (session);
  1715. }
  1716. static void
  1717. proxy_accept_socket (gint fd, short what, void *arg)
  1718. {
  1719. struct rspamd_worker *worker = (struct rspamd_worker *) arg;
  1720. struct rspamd_proxy_ctx *ctx;
  1721. rspamd_inet_addr_t *addr;
  1722. struct rspamd_proxy_session *session;
  1723. gint nfd;
  1724. ctx = worker->ctx;
  1725. if ((nfd =
  1726. rspamd_accept_from_socket (fd, &addr, worker->accept_events)) == -1) {
  1727. msg_warn ("accept failed: %s", strerror (errno));
  1728. return;
  1729. }
  1730. /* Check for EAGAIN */
  1731. if (nfd == 0) {
  1732. return;
  1733. }
  1734. session = g_malloc0 (sizeof (*session));
  1735. REF_INIT_RETAIN (session, proxy_session_dtor);
  1736. session->client_sock = nfd;
  1737. session->client_addr = addr;
  1738. session->mirror_conns = g_ptr_array_sized_new (ctx->mirrors->len);
  1739. session->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (),
  1740. "proxy");
  1741. session->ctx = ctx;
  1742. session->worker = worker;
  1743. if (ctx->sessions_cache) {
  1744. rspamd_worker_session_cache_add (ctx->sessions_cache,
  1745. session->pool->tag.uid, &session->ref.refcount, session);
  1746. }
  1747. if (!ctx->milter) {
  1748. session->client_conn = rspamd_http_connection_new_server (
  1749. ctx->http_ctx,
  1750. nfd,
  1751. NULL,
  1752. proxy_client_error_handler,
  1753. proxy_client_finish_handler,
  1754. 0);
  1755. if (ctx->key) {
  1756. rspamd_http_connection_set_key (session->client_conn, ctx->key);
  1757. }
  1758. msg_info_session ("accepted http connection from %s port %d",
  1759. rspamd_inet_address_to_string (addr),
  1760. rspamd_inet_address_get_port (addr));
  1761. rspamd_http_connection_read_message_shared (session->client_conn,
  1762. session,
  1763. &ctx->io_tv);
  1764. }
  1765. else {
  1766. msg_info_session ("accepted milter connection from %s port %d",
  1767. rspamd_inet_address_to_string (addr),
  1768. rspamd_inet_address_get_port (addr));
  1769. #ifdef TCP_NODELAY
  1770. #ifndef SOL_TCP
  1771. #define SOL_TCP IPPROTO_TCP
  1772. #endif
  1773. if (rspamd_inet_address_get_af (addr) != AF_UNIX) {
  1774. gint sopt = 1;
  1775. if (setsockopt (nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof (sopt)) ==
  1776. -1) {
  1777. msg_warn_session ("cannot set TCP_NODELAY: %s",
  1778. strerror (errno));
  1779. }
  1780. }
  1781. #endif
  1782. rspamd_milter_handle_socket (nfd, NULL,
  1783. session->pool,
  1784. ctx->ev_base,
  1785. proxy_milter_finish_handler,
  1786. proxy_milter_error_handler,
  1787. session);
  1788. }
  1789. }
  1790. static void
  1791. adjust_upstreams_limits (struct rspamd_proxy_ctx *ctx)
  1792. {
  1793. struct rspamd_http_upstream *backend;
  1794. gpointer k, v;
  1795. GHashTableIter it;
  1796. /*
  1797. * We set error time equal to max_retries * backend_timeout and max_errors
  1798. * to max_retries - 1
  1799. *
  1800. * So if we failed to scan a message on a backend for some reasons, we
  1801. * will try to re-resolve it faster
  1802. */
  1803. g_hash_table_iter_init (&it, ctx->upstreams);
  1804. while (g_hash_table_iter_next (&it, &k, &v)) {
  1805. backend = (struct rspamd_http_upstream *)v;
  1806. if (!backend->self_scan && backend->u) {
  1807. rspamd_upstreams_set_limits (backend->u,
  1808. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1809. ctx->max_retries - 1, 0);
  1810. }
  1811. }
  1812. }
  1813. void
  1814. start_rspamd_proxy (struct rspamd_worker *worker)
  1815. {
  1816. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1817. ctx->cfg = worker->srv->cfg;
  1818. ctx->ev_base = rspamd_prepare_worker (worker, "rspamd_proxy",
  1819. proxy_accept_socket);
  1820. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  1821. ctx->ev_base,
  1822. worker->srv->cfg);
  1823. double_to_tv (ctx->timeout, &ctx->io_tv);
  1824. rspamd_map_watch (worker->srv->cfg, ctx->ev_base, ctx->resolver, worker, 0);
  1825. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  1826. ctx->ev_base, ctx->resolver->r);
  1827. ctx->http_ctx = rspamd_http_context_create (ctx->cfg, ctx->ev_base,
  1828. ctx->cfg->ups_ctx);
  1829. if (ctx->has_self_scan) {
  1830. /* Additional initialisation needed */
  1831. rspamd_worker_init_scanner (worker, ctx->ev_base, ctx->resolver,
  1832. &ctx->lang_det);
  1833. }
  1834. if (worker->srv->cfg->enable_sessions_cache) {
  1835. ctx->sessions_cache = rspamd_worker_session_cache_new (worker,
  1836. ctx->ev_base);
  1837. }
  1838. ctx->milter_ctx.spam_header = ctx->spam_header;
  1839. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  1840. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  1841. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  1842. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  1843. ctx->milter_ctx.reject_message = ctx->reject_message;
  1844. ctx->milter_ctx.cfg = ctx->cfg;
  1845. rspamd_milter_init_library (&ctx->milter_ctx);
  1846. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->ev_base,
  1847. worker);
  1848. adjust_upstreams_limits (ctx);
  1849. event_base_loop (ctx->ev_base, 0);
  1850. rspamd_worker_block_signals ();
  1851. if (ctx->has_self_scan) {
  1852. rspamd_stat_close ();
  1853. }
  1854. struct rspamd_http_context *http_ctx = ctx->http_ctx;
  1855. REF_RELEASE (ctx->cfg);
  1856. rspamd_http_context_free (http_ctx);
  1857. rspamd_log_close (worker->srv->logger, TRUE);
  1858. exit (EXIT_SUCCESS);
  1859. }