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.

rspamd_proxy.c 57KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274
  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. gint parser_from_ref;
  81. gint parser_to_ref;
  82. gboolean local;
  83. gboolean self_scan;
  84. gboolean compress;
  85. };
  86. struct rspamd_http_mirror {
  87. gchar *name;
  88. gchar *settings_id;
  89. struct upstream_list *u;
  90. struct rspamd_cryptobox_pubkey *key;
  91. gdouble prob;
  92. gdouble timeout;
  93. gint parser_from_ref;
  94. gint parser_to_ref;
  95. gboolean local;
  96. gboolean compress;
  97. };
  98. static const guint64 rspamd_rspamd_proxy_magic = 0xcdeb4fd1fc351980ULL;
  99. struct rspamd_proxy_ctx {
  100. guint64 magic;
  101. /* Events base */
  102. struct ev_loop *event_loop;
  103. /* DNS resolver */
  104. struct rspamd_dns_resolver *resolver;
  105. /* Config */
  106. struct rspamd_config *cfg;
  107. /* END OF COMMON PART */
  108. gdouble timeout;
  109. /* Encryption key for clients */
  110. struct rspamd_cryptobox_keypair *key;
  111. /* HTTP context */
  112. struct rspamd_http_context *http_ctx;
  113. /* Upstreams to use */
  114. GHashTable *upstreams;
  115. /* Mirrors to send traffic to */
  116. GPtrArray *mirrors;
  117. /* Default upstream */
  118. struct rspamd_http_upstream *default_upstream;
  119. lua_State *lua_state;
  120. /* Array of callback functions called on end of scan to compare results */
  121. GArray *cmp_refs;
  122. /* Maximum count for retries */
  123. guint max_retries;
  124. /* If we have self_scanning backends, we need to work as a normal worker */
  125. gboolean has_self_scan;
  126. /* It is not HTTP but milter proxy */
  127. gboolean milter;
  128. /* Discard messages instead of rejecting them */
  129. gboolean discard_on_reject;
  130. /* Quarantine messages instead of rejecting them */
  131. gboolean quarantine_on_reject;
  132. /* Milter spam header */
  133. gchar *spam_header;
  134. /* CA name that can be used for client certificates */
  135. gchar *client_ca_name;
  136. /* Milter rejection message */
  137. gchar *reject_message;
  138. /* Sessions cache */
  139. void *sessions_cache;
  140. struct rspamd_milter_context milter_ctx;
  141. /* Language detector */
  142. struct rspamd_lang_detector *lang_det;
  143. };
  144. enum rspamd_backend_flags {
  145. RSPAMD_BACKEND_REPLIED = 1 << 0,
  146. RSPAMD_BACKEND_CLOSED = 1 << 1,
  147. RSPAMD_BACKEND_PARSED = 1 << 2,
  148. };
  149. struct rspamd_proxy_session;
  150. struct rspamd_proxy_backend_connection {
  151. const gchar *name;
  152. struct rspamd_cryptobox_keypair *local_key;
  153. struct rspamd_cryptobox_pubkey *remote_key;
  154. struct upstream *up;
  155. struct rspamd_http_connection *backend_conn;
  156. ucl_object_t *results;
  157. const gchar *err;
  158. struct rspamd_proxy_session *s;
  159. gint backend_sock;
  160. ev_tstamp timeout;
  161. enum rspamd_backend_flags flags;
  162. gint parser_from_ref;
  163. gint parser_to_ref;
  164. struct rspamd_task *task;
  165. };
  166. enum rspamd_proxy_legacy_support {
  167. LEGACY_SUPPORT_NO = 0,
  168. LEGACY_SUPPORT_RSPAMC,
  169. LEGACY_SUPPORT_SPAMC
  170. };
  171. struct rspamd_proxy_session {
  172. struct rspamd_worker *worker;
  173. rspamd_mempool_t *pool;
  174. struct rspamd_proxy_ctx *ctx;
  175. rspamd_inet_addr_t *client_addr;
  176. struct rspamd_http_connection *client_conn;
  177. struct rspamd_milter_session *client_milter_conn;
  178. struct rspamd_http_upstream *backend;
  179. gpointer map;
  180. gchar *fname;
  181. gpointer shmem_ref;
  182. struct rspamd_proxy_backend_connection *master_conn;
  183. struct rspamd_http_message *client_message;
  184. GPtrArray *mirror_conns;
  185. gsize map_len;
  186. gint client_sock;
  187. enum rspamd_proxy_legacy_support legacy_support;
  188. gint retries;
  189. ref_entry_t ref;
  190. };
  191. static gboolean proxy_send_master_message (struct rspamd_proxy_session *session);
  192. static GQuark
  193. rspamd_proxy_quark (void)
  194. {
  195. return g_quark_from_static_string ("rspamd-proxy");
  196. }
  197. static gboolean
  198. rspamd_proxy_parse_lua_parser (lua_State *L, const ucl_object_t *obj,
  199. gint *ref_from, gint *ref_to, GError **err)
  200. {
  201. const gchar *lua_script;
  202. gsize slen;
  203. gint err_idx, ref_idx;
  204. gboolean has_ref = FALSE;
  205. g_assert (obj != NULL);
  206. g_assert (ref_from != NULL);
  207. g_assert (ref_to != NULL);
  208. *ref_from = -1;
  209. *ref_to = -1;
  210. lua_script = ucl_object_tolstring (obj, &slen);
  211. lua_pushcfunction (L, &rspamd_lua_traceback);
  212. err_idx = lua_gettop (L);
  213. /* Load data */
  214. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  215. g_set_error (err,
  216. rspamd_proxy_quark (),
  217. EINVAL,
  218. "cannot load lua parser script: %s",
  219. lua_tostring (L, -1));
  220. lua_settop (L, 0); /* Error function */
  221. return FALSE;
  222. }
  223. /* Now do it */
  224. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  225. g_set_error (err,
  226. rspamd_proxy_quark (),
  227. EINVAL,
  228. "cannot init lua parser script: %s",
  229. lua_tostring (L, -1));
  230. lua_settop (L, 0);
  231. return FALSE;
  232. }
  233. if (lua_istable (L, -1)) {
  234. /*
  235. * We have a table, so we check for two keys:
  236. * 'from' -> function
  237. * 'to' -> function
  238. *
  239. * From converts parent request to a client one
  240. * To converts client request to a parent one
  241. */
  242. lua_pushstring (L, "from");
  243. lua_gettable (L, -2);
  244. if (lua_isfunction (L, -1)) {
  245. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  246. *ref_from = ref_idx;
  247. has_ref = TRUE;
  248. }
  249. lua_pushstring (L, "to");
  250. lua_gettable (L, -2);
  251. if (lua_isfunction (L, -1)) {
  252. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  253. *ref_to = ref_idx;
  254. has_ref = TRUE;
  255. }
  256. }
  257. else if (!lua_isfunction (L, -1)) {
  258. g_set_error (err,
  259. rspamd_proxy_quark (),
  260. EINVAL,
  261. "cannot init lua parser script: "
  262. "must return function");
  263. lua_settop (L, 0);
  264. return FALSE;
  265. }
  266. else {
  267. /* Just parser from protocol */
  268. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  269. *ref_from = ref_idx;
  270. lua_settop (L, 0);
  271. has_ref = TRUE;
  272. }
  273. return has_ref;
  274. }
  275. static gboolean
  276. rspamd_proxy_parse_upstream (rspamd_mempool_t *pool,
  277. const ucl_object_t *obj,
  278. gpointer ud,
  279. struct rspamd_rcl_section *section,
  280. GError **err)
  281. {
  282. const ucl_object_t *elt;
  283. struct rspamd_http_upstream *up = NULL;
  284. struct rspamd_proxy_ctx *ctx;
  285. struct rspamd_rcl_struct_parser *pd = ud;
  286. lua_State *L;
  287. ctx = pd->user_struct;
  288. L = ctx->lua_state;
  289. if (ucl_object_type (obj) != UCL_OBJECT) {
  290. g_set_error (err, rspamd_proxy_quark (), 100,
  291. "upstream option must be an object");
  292. return FALSE;
  293. }
  294. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  295. elt = ucl_object_lookup (obj, "name");
  296. if (elt == NULL) {
  297. if (ucl_object_key (obj)) {
  298. if (strcmp (ucl_object_key (obj), "upstream") == 0) {
  299. /* Iterate over the object and find upstream elements */
  300. ucl_object_iter_t it = NULL;
  301. const ucl_object_t *cur;
  302. gboolean ret = TRUE;
  303. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  304. if (!rspamd_proxy_parse_upstream (pool, cur, ud,
  305. section, err)) {
  306. ret = FALSE;
  307. }
  308. }
  309. return ret;
  310. }
  311. else {
  312. /* Inside upstream */
  313. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  314. }
  315. }
  316. else {
  317. g_set_error (err, rspamd_proxy_quark (), 100,
  318. "upstream option must have some name definition");
  319. return FALSE;
  320. }
  321. }
  322. else {
  323. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  324. }
  325. up->parser_from_ref = -1;
  326. up->parser_to_ref = -1;
  327. up->timeout = ctx->timeout;
  328. elt = ucl_object_lookup (obj, "key");
  329. if (elt != NULL) {
  330. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  331. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  332. if (up->key == NULL) {
  333. g_set_error (err, rspamd_proxy_quark (), 100,
  334. "cannot read upstream key");
  335. goto err;
  336. }
  337. rspamd_mempool_add_destructor (pool,
  338. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  339. }
  340. elt = ucl_object_lookup (obj, "self_scan");
  341. if (elt && ucl_object_toboolean (elt)) {
  342. up->self_scan = TRUE;
  343. ctx->has_self_scan = TRUE;
  344. }
  345. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  346. if (elt && ucl_object_toboolean (elt)) {
  347. up->compress = TRUE;
  348. }
  349. elt = ucl_object_lookup (obj, "hosts");
  350. if (elt == NULL && !up->self_scan) {
  351. g_set_error (err, rspamd_proxy_quark (), 100,
  352. "upstream option must have some hosts definition");
  353. goto err;
  354. }
  355. if (elt) {
  356. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  357. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  358. g_set_error (err, rspamd_proxy_quark (), 100,
  359. "upstream has bad hosts definition");
  360. goto err;
  361. }
  362. rspamd_mempool_add_destructor (pool,
  363. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  364. }
  365. elt = ucl_object_lookup (obj, "default");
  366. if (elt) {
  367. if (ucl_object_toboolean (elt)) {
  368. ctx->default_upstream = up;
  369. }
  370. }
  371. else if (up->self_scan) {
  372. ctx->default_upstream = up;
  373. }
  374. elt = ucl_object_lookup (obj, "local");
  375. if (elt && ucl_object_toboolean (elt)) {
  376. up->local = TRUE;
  377. }
  378. elt = ucl_object_lookup (obj, "timeout");
  379. if (elt) {
  380. ucl_object_todouble_safe (elt, &up->timeout);
  381. }
  382. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  383. if (elt && ucl_object_type (elt) == UCL_STRING) {
  384. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  385. }
  386. /*
  387. * Accept lua function here in form
  388. * fun :: String -> UCL
  389. */
  390. elt = ucl_object_lookup (obj, "parser");
  391. if (elt) {
  392. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  393. &up->parser_to_ref, err)) {
  394. goto err;
  395. }
  396. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  397. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  398. }
  399. g_hash_table_insert (ctx->upstreams, up->name, up);
  400. return TRUE;
  401. err:
  402. return FALSE;
  403. }
  404. static gboolean
  405. rspamd_proxy_parse_mirror (rspamd_mempool_t *pool,
  406. const ucl_object_t *obj,
  407. gpointer ud,
  408. struct rspamd_rcl_section *section,
  409. GError **err)
  410. {
  411. const ucl_object_t *elt;
  412. struct rspamd_http_mirror *up = NULL;
  413. struct rspamd_proxy_ctx *ctx;
  414. struct rspamd_rcl_struct_parser *pd = ud;
  415. lua_State *L;
  416. ctx = pd->user_struct;
  417. L = ctx->lua_state;
  418. if (ucl_object_type (obj) != UCL_OBJECT) {
  419. g_set_error (err, rspamd_proxy_quark (), 100,
  420. "mirror option must be an object");
  421. return FALSE;
  422. }
  423. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  424. elt = ucl_object_lookup (obj, "name");
  425. if (elt == NULL) {
  426. if (ucl_object_key (obj)) {
  427. if (strcmp (ucl_object_key (obj), "mirror") == 0) {
  428. /* Iterate over the object and find upstream elements */
  429. ucl_object_iter_t it = NULL;
  430. const ucl_object_t *cur;
  431. gboolean ret = TRUE;
  432. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  433. if (!rspamd_proxy_parse_mirror (pool, cur, ud,
  434. section, err)) {
  435. ret = FALSE;
  436. }
  437. }
  438. return ret;
  439. }
  440. else {
  441. /* Inside upstream */
  442. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  443. }
  444. }
  445. else {
  446. g_set_error (err, rspamd_proxy_quark (), 100,
  447. "mirror option must have some name definition");
  448. return FALSE;
  449. }
  450. }
  451. else {
  452. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  453. }
  454. up->parser_to_ref = -1;
  455. up->parser_from_ref = -1;
  456. up->timeout = ctx->timeout;
  457. elt = ucl_object_lookup (obj, "key");
  458. if (elt != NULL) {
  459. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  460. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  461. if (up->key == NULL) {
  462. g_set_error (err, rspamd_proxy_quark (), 100,
  463. "cannot read mirror key");
  464. goto err;
  465. }
  466. rspamd_mempool_add_destructor (pool,
  467. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  468. }
  469. elt = ucl_object_lookup (obj, "hosts");
  470. if (elt == NULL) {
  471. g_set_error (err, rspamd_proxy_quark (), 100,
  472. "mirror option must have some hosts definition");
  473. goto err;
  474. }
  475. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  476. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  477. g_set_error (err, rspamd_proxy_quark (), 100,
  478. "mirror has bad hosts definition");
  479. goto err;
  480. }
  481. rspamd_mempool_add_destructor (pool,
  482. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  483. elt = ucl_object_lookup_any (obj, "probability", "prob", NULL);
  484. if (elt) {
  485. up->prob = ucl_object_todouble (elt);
  486. }
  487. else {
  488. up->prob = 1.0;
  489. }
  490. elt = ucl_object_lookup (obj, "local");
  491. if (elt && ucl_object_toboolean (elt)) {
  492. up->local = TRUE;
  493. }
  494. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  495. if (elt && ucl_object_toboolean (elt)) {
  496. up->compress = TRUE;
  497. }
  498. elt = ucl_object_lookup (obj, "timeout");
  499. if (elt) {
  500. ucl_object_todouble_safe (elt, &up->timeout);
  501. }
  502. /*
  503. * Accept lua function here in form
  504. * fun :: String -> UCL
  505. */
  506. elt = ucl_object_lookup (obj, "parser");
  507. if (elt) {
  508. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  509. &up->parser_to_ref, err)) {
  510. goto err;
  511. }
  512. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  513. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  514. }
  515. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  516. if (elt && ucl_object_type (elt) == UCL_STRING) {
  517. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  518. }
  519. g_ptr_array_add (ctx->mirrors, up);
  520. return TRUE;
  521. err:
  522. return FALSE;
  523. }
  524. static gboolean
  525. rspamd_proxy_parse_script (rspamd_mempool_t *pool,
  526. const ucl_object_t *obj,
  527. gpointer ud,
  528. struct rspamd_rcl_section *section,
  529. GError **err)
  530. {
  531. struct rspamd_proxy_ctx *ctx;
  532. struct rspamd_rcl_struct_parser *pd = ud;
  533. lua_State *L;
  534. const gchar *lua_script;
  535. gsize slen;
  536. gint err_idx, ref_idx;
  537. struct stat st;
  538. ctx = pd->user_struct;
  539. L = ctx->lua_state;
  540. if (ucl_object_type (obj) != UCL_STRING) {
  541. g_set_error (err, rspamd_proxy_quark (), 100,
  542. "script option must be a string with file or lua chunk");
  543. return FALSE;
  544. }
  545. lua_script = ucl_object_tolstring (obj, &slen);
  546. lua_pushcfunction (L, &rspamd_lua_traceback);
  547. err_idx = lua_gettop (L);
  548. if (stat (lua_script, &st) != -1) {
  549. /* Load file */
  550. if (luaL_loadfile (L, lua_script) != 0) {
  551. g_set_error (err,
  552. rspamd_proxy_quark (),
  553. EINVAL,
  554. "cannot load lua parser script: %s",
  555. lua_tostring (L, -1));
  556. lua_settop (L, 0); /* Error function */
  557. goto err;
  558. }
  559. }
  560. else {
  561. /* Load data directly */
  562. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  563. g_set_error (err,
  564. rspamd_proxy_quark (),
  565. EINVAL,
  566. "cannot load lua parser script: %s",
  567. lua_tostring (L, -1));
  568. lua_settop (L, 0); /* Error function */
  569. goto err;
  570. }
  571. }
  572. /* Now do it */
  573. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  574. g_set_error (err,
  575. rspamd_proxy_quark (),
  576. EINVAL,
  577. "cannot init lua parser script: %s",
  578. lua_tostring (L, -1));
  579. lua_settop (L, 0);
  580. goto err;
  581. }
  582. if (!lua_isfunction (L, -1)) {
  583. g_set_error (err,
  584. rspamd_proxy_quark (),
  585. EINVAL,
  586. "cannot init lua parser script: "
  587. "must return function, %s returned",
  588. lua_typename (L, lua_type (L, -1)));
  589. lua_settop (L, 0);
  590. goto err;
  591. }
  592. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  593. lua_settop (L, 0);
  594. g_array_append_val (ctx->cmp_refs, ref_idx);
  595. return TRUE;
  596. err:
  597. return FALSE;
  598. }
  599. gpointer
  600. init_rspamd_proxy (struct rspamd_config *cfg)
  601. {
  602. struct rspamd_proxy_ctx *ctx;
  603. GQuark type;
  604. type = g_quark_try_string ("rspamd_proxy");
  605. ctx = rspamd_mempool_alloc0 (cfg->cfg_pool,
  606. sizeof (struct rspamd_proxy_ctx));
  607. ctx->magic = rspamd_rspamd_proxy_magic;
  608. ctx->timeout = 120.0;
  609. ctx->upstreams = g_hash_table_new (rspamd_strcase_hash, rspamd_strcase_equal);
  610. rspamd_mempool_add_destructor (cfg->cfg_pool,
  611. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->upstreams);
  612. ctx->mirrors = g_ptr_array_new ();
  613. rspamd_mempool_add_destructor (cfg->cfg_pool,
  614. (rspamd_mempool_destruct_t)rspamd_ptr_array_free_hard, ctx->mirrors);
  615. ctx->cfg = cfg;
  616. ctx->lua_state = cfg->lua_state;
  617. ctx->cmp_refs = g_array_new (FALSE, FALSE, sizeof (gint));
  618. rspamd_mempool_add_destructor (cfg->cfg_pool,
  619. (rspamd_mempool_destruct_t)rspamd_array_free_hard, ctx->cmp_refs);
  620. ctx->max_retries = DEFAULT_RETRIES;
  621. ctx->spam_header = RSPAMD_MILTER_SPAM_HEADER;
  622. rspamd_rcl_register_worker_option (cfg,
  623. type,
  624. "timeout",
  625. rspamd_rcl_parse_struct_time,
  626. ctx,
  627. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, timeout),
  628. RSPAMD_CL_FLAG_TIME_FLOAT,
  629. "IO timeout");
  630. rspamd_rcl_register_worker_option (cfg,
  631. type,
  632. "keypair",
  633. rspamd_rcl_parse_struct_keypair,
  634. ctx,
  635. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, key),
  636. 0,
  637. "Server's keypair");
  638. rspamd_rcl_register_worker_option (cfg,
  639. type,
  640. "upstream",
  641. rspamd_proxy_parse_upstream,
  642. ctx,
  643. 0,
  644. 0,
  645. "List of upstreams");
  646. rspamd_rcl_register_worker_option (cfg,
  647. type,
  648. "mirror",
  649. rspamd_proxy_parse_mirror,
  650. ctx,
  651. 0,
  652. RSPAMD_CL_FLAG_MULTIPLE,
  653. "List of mirrors");
  654. rspamd_rcl_register_worker_option (cfg,
  655. type,
  656. "script",
  657. rspamd_proxy_parse_script,
  658. ctx,
  659. 0,
  660. RSPAMD_CL_FLAG_MULTIPLE,
  661. "Compare script to be executed");
  662. rspamd_rcl_register_worker_option (cfg,
  663. type,
  664. "max_retries",
  665. rspamd_rcl_parse_struct_integer,
  666. ctx,
  667. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, max_retries),
  668. RSPAMD_CL_FLAG_UINT,
  669. "Maximum number of retries for master connection");
  670. rspamd_rcl_register_worker_option (cfg,
  671. type,
  672. "milter",
  673. rspamd_rcl_parse_struct_boolean,
  674. ctx,
  675. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, milter),
  676. 0,
  677. "Accept milter connections, not HTTP");
  678. rspamd_rcl_register_worker_option (cfg,
  679. type,
  680. "discard_on_reject",
  681. rspamd_rcl_parse_struct_boolean,
  682. ctx,
  683. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, discard_on_reject),
  684. 0,
  685. "Tell MTA to discard rejected messages silently");
  686. rspamd_rcl_register_worker_option (cfg,
  687. type,
  688. "quarantine_on_reject",
  689. rspamd_rcl_parse_struct_boolean,
  690. ctx,
  691. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, quarantine_on_reject),
  692. 0,
  693. "Tell MTA to quarantine rejected messages");
  694. rspamd_rcl_register_worker_option (cfg,
  695. type,
  696. "spam_header",
  697. rspamd_rcl_parse_struct_string,
  698. ctx,
  699. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, spam_header),
  700. 0,
  701. "Use the specific spam header (default: X-Spam)");
  702. rspamd_rcl_register_worker_option (cfg,
  703. type,
  704. "client_ca_name",
  705. rspamd_rcl_parse_struct_string,
  706. ctx,
  707. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, client_ca_name),
  708. 0,
  709. "Allow certificates issued by this CA to be treated as client certificates");
  710. rspamd_rcl_register_worker_option (cfg,
  711. type,
  712. "reject_message",
  713. rspamd_rcl_parse_struct_string,
  714. ctx,
  715. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, reject_message),
  716. 0,
  717. "Use custom rejection message");
  718. return ctx;
  719. }
  720. static void
  721. proxy_backend_close_connection (struct rspamd_proxy_backend_connection *conn)
  722. {
  723. if (conn && !(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  724. if (conn->backend_conn) {
  725. rspamd_http_connection_reset (conn->backend_conn);
  726. rspamd_http_connection_unref (conn->backend_conn);
  727. close (conn->backend_sock);
  728. }
  729. conn->flags |= RSPAMD_BACKEND_CLOSED;
  730. }
  731. }
  732. static gboolean
  733. proxy_backend_parse_results (struct rspamd_proxy_session *session,
  734. struct rspamd_proxy_backend_connection *conn,
  735. lua_State *L, gint parser_ref,
  736. struct rspamd_http_message *msg,
  737. goffset *body_offset)
  738. {
  739. struct ucl_parser *parser;
  740. gint err_idx;
  741. const gchar *in = msg->body_buf.begin;
  742. gsize inlen = msg->body_buf.len;
  743. const rspamd_ftok_t *offset_hdr;
  744. if (inlen == 0 || in == NULL) {
  745. return FALSE;
  746. }
  747. offset_hdr = rspamd_http_message_find_header (msg, MESSAGE_OFFSET_HEADER);
  748. if (offset_hdr) {
  749. gulong val;
  750. if (rspamd_strtoul (offset_hdr->begin, offset_hdr->len, &val)
  751. && val < inlen) {
  752. if (body_offset) {
  753. *body_offset = val;
  754. }
  755. inlen = val;
  756. }
  757. }
  758. if (parser_ref != -1) {
  759. /* Call parser function */
  760. lua_pushcfunction (L, &rspamd_lua_traceback);
  761. err_idx = lua_gettop (L);
  762. lua_rawgeti (L, LUA_REGISTRYINDEX, parser_ref);
  763. /* XXX: copies all data */
  764. lua_pushlstring (L, in, inlen);
  765. if (lua_pcall (L, 1, 1, err_idx) != 0) {
  766. msg_err_session (
  767. "cannot run lua parser script: %s",
  768. lua_tostring (L, -1));
  769. lua_settop (L, 0);
  770. return FALSE;
  771. }
  772. conn->results = ucl_object_lua_import (L, -1);
  773. lua_settop (L, 0);
  774. }
  775. else {
  776. parser = ucl_parser_new (0);
  777. if (!ucl_parser_add_chunk (parser, in, inlen)) {
  778. gchar *encoded;
  779. encoded = rspamd_encode_base64 (in, inlen, 0, NULL);
  780. msg_err_session ("cannot parse input: %s", ucl_parser_get_error (
  781. parser));
  782. msg_err_session ("input encoded: %s", encoded);
  783. ucl_parser_free (parser);
  784. g_free (encoded);
  785. return FALSE;
  786. }
  787. conn->results = ucl_parser_get_object (parser);
  788. ucl_parser_free (parser);
  789. }
  790. return TRUE;
  791. }
  792. static void
  793. proxy_call_cmp_script (struct rspamd_proxy_session *session, gint cbref)
  794. {
  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. msg_err_session (
  830. "cannot run lua compare script: %s",
  831. lua_tostring (L, -1));
  832. }
  833. lua_settop (L, 0);
  834. }
  835. static void
  836. proxy_session_dtor (struct rspamd_proxy_session *session)
  837. {
  838. guint i;
  839. gint cbref;
  840. struct rspamd_proxy_backend_connection *conn;
  841. if (session->master_conn && session->master_conn->results) {
  842. for (i = 0; i < session->ctx->cmp_refs->len; i++) {
  843. cbref = g_array_index (session->ctx->cmp_refs, gint, i);
  844. proxy_call_cmp_script (session, cbref);
  845. }
  846. }
  847. if (session->master_conn) {
  848. proxy_backend_close_connection (session->master_conn);
  849. }
  850. if (session->client_milter_conn) {
  851. rspamd_milter_session_unref (session->client_milter_conn);
  852. }
  853. else if (session->client_conn) {
  854. rspamd_http_connection_reset (session->client_conn);
  855. rspamd_http_connection_unref (session->client_conn);
  856. }
  857. if (session->map && session->map_len) {
  858. munmap (session->map, session->map_len);
  859. }
  860. for (i = 0; i < session->mirror_conns->len; i ++) {
  861. conn = g_ptr_array_index (session->mirror_conns, i);
  862. if (!(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  863. proxy_backend_close_connection (conn);
  864. }
  865. if (conn->results) {
  866. ucl_object_unref (conn->results);
  867. }
  868. }
  869. if (session->master_conn) {
  870. if (session->master_conn->results) {
  871. ucl_object_unref (session->master_conn->results);
  872. }
  873. if (session->master_conn->task) {
  874. rspamd_session_destroy (session->master_conn->task->s);
  875. }
  876. }
  877. g_ptr_array_free (session->mirror_conns, TRUE);
  878. rspamd_http_message_shmem_unref (session->shmem_ref);
  879. rspamd_http_message_unref (session->client_message);
  880. if (session->client_addr) {
  881. rspamd_inet_address_free (session->client_addr);
  882. }
  883. if (session->client_sock != -1) {
  884. close (session->client_sock);
  885. }
  886. if (session->ctx->sessions_cache) {
  887. rspamd_worker_session_cache_remove (session->ctx->sessions_cache,
  888. session);
  889. }
  890. if (session->pool) {
  891. rspamd_mempool_delete (session->pool);
  892. }
  893. g_free (session);
  894. }
  895. static void
  896. proxy_request_compress (struct rspamd_http_message *msg)
  897. {
  898. guint flags;
  899. ZSTD_CCtx *zctx;
  900. rspamd_fstring_t *body;
  901. const gchar *in;
  902. gsize inlen;
  903. flags = rspamd_http_message_get_flags (msg);
  904. if (!rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  905. if ((flags & RSPAMD_HTTP_FLAG_SHMEM) ||
  906. !(flags & RSPAMD_HTTP_FLAG_HAS_BODY)) {
  907. /* Cannot compress shared or empty message */
  908. return;
  909. }
  910. in = rspamd_http_message_get_body (msg, &inlen);
  911. if (in == NULL || inlen == 0) {
  912. return;
  913. }
  914. body = rspamd_fstring_sized_new (ZSTD_compressBound (inlen));
  915. zctx = ZSTD_createCCtx ();
  916. body->len = ZSTD_compressCCtx (zctx, body->str, body->allocated,
  917. in, inlen, 1);
  918. if (ZSTD_isError (body->len)) {
  919. msg_err ("compression error");
  920. rspamd_fstring_free (body);
  921. ZSTD_freeCCtx (zctx);
  922. return;
  923. }
  924. ZSTD_freeCCtx (zctx);
  925. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  926. rspamd_http_message_add_header (msg, COMPRESSION_HEADER, "zstd");
  927. }
  928. }
  929. static void
  930. proxy_request_decompress (struct rspamd_http_message *msg)
  931. {
  932. rspamd_fstring_t *body;
  933. const gchar *in;
  934. gsize inlen, outlen, r;
  935. ZSTD_DStream *zstream;
  936. ZSTD_inBuffer zin;
  937. ZSTD_outBuffer zout;
  938. if (rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  939. in = rspamd_http_message_get_body (msg, &inlen);
  940. if (in == NULL || inlen == 0) {
  941. return;
  942. }
  943. zstream = ZSTD_createDStream ();
  944. ZSTD_initDStream (zstream);
  945. zin.pos = 0;
  946. zin.src = in;
  947. zin.size = inlen;
  948. if ((outlen = ZSTD_getDecompressedSize (zin.src, zin.size)) == 0) {
  949. outlen = ZSTD_DStreamOutSize ();
  950. }
  951. body = rspamd_fstring_sized_new (outlen);
  952. zout.dst = body->str;
  953. zout.pos = 0;
  954. zout.size = outlen;
  955. while (zin.pos < zin.size) {
  956. r = ZSTD_decompressStream (zstream, &zout, &zin);
  957. if (ZSTD_isError (r)) {
  958. msg_err ("Decompression error: %s", ZSTD_getErrorName (r));
  959. ZSTD_freeDStream (zstream);
  960. rspamd_fstring_free (body);
  961. return;
  962. }
  963. if (zout.pos == zout.size) {
  964. /* We need to extend output buffer */
  965. zout.size = zout.size * 2 + 1;
  966. body = rspamd_fstring_grow (body, zout.size);
  967. zout.size = body->allocated;
  968. zout.dst = body->str;
  969. }
  970. }
  971. body->len = zout.pos;
  972. ZSTD_freeDStream (zstream);
  973. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  974. rspamd_http_message_remove_header (msg, COMPRESSION_HEADER);
  975. }
  976. }
  977. static struct rspamd_proxy_session *
  978. proxy_session_refresh (struct rspamd_proxy_session *session)
  979. {
  980. struct rspamd_proxy_session *nsession;
  981. nsession = g_malloc0 (sizeof (*nsession));
  982. nsession->client_milter_conn = session->client_milter_conn;
  983. session->client_milter_conn = NULL;
  984. rspamd_milter_update_userdata (nsession->client_milter_conn,
  985. nsession);
  986. nsession->client_addr = session->client_addr;
  987. session->client_addr = NULL;
  988. nsession->ctx = session->ctx;
  989. nsession->worker = session->worker;
  990. nsession->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (), "proxy");
  991. nsession->client_sock = session->client_sock;
  992. session->client_sock = -1;
  993. nsession->mirror_conns = g_ptr_array_sized_new (nsession->ctx->mirrors->len);
  994. REF_INIT_RETAIN (nsession, proxy_session_dtor);
  995. if (nsession->ctx->sessions_cache) {
  996. rspamd_worker_session_cache_add (nsession->ctx->sessions_cache,
  997. nsession->pool->tag.uid, &nsession->ref.refcount, nsession);
  998. }
  999. return nsession;
  1000. }
  1001. static gboolean
  1002. proxy_check_file (struct rspamd_http_message *msg,
  1003. struct rspamd_proxy_session *session)
  1004. {
  1005. const rspamd_ftok_t *tok, *key_tok;
  1006. rspamd_ftok_t srch;
  1007. gchar *file_str;
  1008. GHashTable *query_args;
  1009. GHashTableIter it;
  1010. gpointer k, v;
  1011. struct http_parser_url u;
  1012. rspamd_fstring_t *new_url;
  1013. tok = rspamd_http_message_find_header (msg, "File");
  1014. if (tok) {
  1015. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1016. session->map = rspamd_file_xmap (file_str, PROT_READ, &session->map_len,
  1017. TRUE);
  1018. if (session->map == NULL) {
  1019. if (session->map_len != 0) {
  1020. msg_err_session ("cannot map %s: %s", file_str,
  1021. strerror (errno));
  1022. return FALSE;
  1023. }
  1024. }
  1025. /* Remove header after processing */
  1026. rspamd_http_message_remove_header (msg, "File");
  1027. session->fname = file_str;
  1028. }
  1029. else {
  1030. /* Need to parse query URL */
  1031. if (http_parser_parse_url (RSPAMD_FSTRING_DATA (msg->url),
  1032. RSPAMD_FSTRING_LEN (msg->url), 0, &u) != 0) {
  1033. msg_err_session ("bad request url: %V", msg->url);
  1034. return FALSE;
  1035. }
  1036. if (u.field_set & (1 << UF_QUERY)) {
  1037. /* In case if we have a query, we need to store it somewhere */
  1038. query_args = rspamd_http_message_parse_query (msg);
  1039. srch.begin = "File";
  1040. srch.len = strlen ("File");
  1041. tok = g_hash_table_lookup (query_args, &srch);
  1042. if (tok) {
  1043. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1044. session->map = rspamd_file_xmap (file_str, PROT_READ,
  1045. &session->map_len, TRUE);
  1046. if (session->map == NULL) {
  1047. if (session->map_len != 0) {
  1048. msg_err_session ("cannot map %s: %s", file_str,
  1049. strerror (errno));
  1050. g_hash_table_unref (query_args);
  1051. return FALSE;
  1052. }
  1053. }
  1054. /* We need to create a new URL with file attribute removed */
  1055. new_url = rspamd_fstring_new_init (RSPAMD_FSTRING_DATA (msg->url),
  1056. u.field_data[UF_QUERY].off);
  1057. new_url = rspamd_fstring_append (new_url, "?", 1);
  1058. g_hash_table_iter_init (&it, query_args);
  1059. while (g_hash_table_iter_next (&it, &k, &v)) {
  1060. key_tok = k;
  1061. tok = v;
  1062. if (!rspamd_ftok_icase_equal (key_tok, &srch)) {
  1063. rspamd_printf_fstring (&new_url, "%T=%T&",
  1064. key_tok, tok);
  1065. }
  1066. }
  1067. /* Erase last character (might be either & or ?) */
  1068. rspamd_fstring_erase (new_url, new_url->len - 1, 1);
  1069. rspamd_fstring_free (msg->url);
  1070. msg->url = new_url;
  1071. session->fname = file_str;
  1072. }
  1073. g_hash_table_unref (query_args);
  1074. }
  1075. }
  1076. return TRUE;
  1077. }
  1078. static void
  1079. proxy_backend_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  1080. {
  1081. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1082. struct rspamd_proxy_session *session;
  1083. session = bk_conn->s;
  1084. msg_info_session ("abnormally closing connection from backend: %s:%s, "
  1085. "error: %e",
  1086. bk_conn->name,
  1087. rspamd_inet_address_to_string (
  1088. rspamd_upstream_addr_cur (bk_conn->up)),
  1089. err);
  1090. if (err) {
  1091. bk_conn->err = rspamd_mempool_strdup (session->pool, err->message);
  1092. }
  1093. rspamd_upstream_fail (bk_conn->up, FALSE);
  1094. proxy_backend_close_connection (bk_conn);
  1095. REF_RELEASE (bk_conn->s);
  1096. }
  1097. static gint
  1098. proxy_backend_mirror_finish_handler (struct rspamd_http_connection *conn,
  1099. struct rspamd_http_message *msg)
  1100. {
  1101. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1102. struct rspamd_proxy_session *session;
  1103. session = bk_conn->s;
  1104. proxy_request_decompress (msg);
  1105. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1106. bk_conn->parser_from_ref, msg, NULL)) {
  1107. msg_warn_session ("cannot parse results from the mirror backend %s:%s",
  1108. bk_conn->name,
  1109. rspamd_inet_address_to_string (
  1110. rspamd_upstream_addr_cur (bk_conn->up)));
  1111. bk_conn->err = "cannot parse ucl";
  1112. }
  1113. msg_info_session ("finished mirror connection to %s", bk_conn->name);
  1114. rspamd_upstream_ok (bk_conn->up);
  1115. proxy_backend_close_connection (bk_conn);
  1116. REF_RELEASE (bk_conn->s);
  1117. return 0;
  1118. }
  1119. static void
  1120. proxy_open_mirror_connections (struct rspamd_proxy_session *session)
  1121. {
  1122. gdouble coin;
  1123. struct rspamd_http_mirror *m;
  1124. guint i;
  1125. struct rspamd_proxy_backend_connection *bk_conn;
  1126. struct rspamd_http_message *msg;
  1127. GError *err = NULL;
  1128. coin = rspamd_random_double ();
  1129. for (i = 0; i < session->ctx->mirrors->len; i ++) {
  1130. m = g_ptr_array_index (session->ctx->mirrors, i);
  1131. if (m->prob < coin) {
  1132. /* No luck */
  1133. continue;
  1134. }
  1135. bk_conn = rspamd_mempool_alloc0 (session->pool,
  1136. sizeof (*bk_conn));
  1137. bk_conn->s = session;
  1138. bk_conn->name = m->name;
  1139. bk_conn->timeout = m->timeout;
  1140. bk_conn->up = rspamd_upstream_get (m->u,
  1141. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1142. bk_conn->parser_from_ref = m->parser_from_ref;
  1143. bk_conn->parser_to_ref = m->parser_to_ref;
  1144. if (bk_conn->up == NULL) {
  1145. msg_err_session ("cannot select upstream for %s", m->name);
  1146. continue;
  1147. }
  1148. bk_conn->backend_sock = rspamd_inet_address_connect (
  1149. rspamd_upstream_addr_next (bk_conn->up),
  1150. SOCK_STREAM, TRUE);
  1151. if (bk_conn->backend_sock == -1) {
  1152. msg_err_session ("cannot connect upstream for %s", m->name);
  1153. rspamd_upstream_fail (bk_conn->up, TRUE);
  1154. continue;
  1155. }
  1156. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1157. if (msg == NULL) {
  1158. msg_err_session ("cannot copy message to send to a mirror %s: %e",
  1159. m->name, err);
  1160. if (err) {
  1161. g_error_free (err);
  1162. }
  1163. continue;
  1164. }
  1165. if (msg->url->len == 0) {
  1166. msg->url = rspamd_fstring_append (msg->url, "/check", strlen ("/check"));
  1167. }
  1168. if (m->settings_id != NULL) {
  1169. rspamd_http_message_remove_header (msg, "Settings-ID");
  1170. rspamd_http_message_add_header (msg, "Settings-ID", m->settings_id);
  1171. }
  1172. bk_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1173. session->ctx->http_ctx,
  1174. NULL,
  1175. proxy_backend_mirror_error_handler,
  1176. proxy_backend_mirror_finish_handler,
  1177. RSPAMD_HTTP_CLIENT_SIMPLE,
  1178. bk_conn->backend_sock);
  1179. if (m->key) {
  1180. msg->peer_key = rspamd_pubkey_ref (m->key);
  1181. }
  1182. if (m->local ||
  1183. rspamd_inet_address_is_local (
  1184. rspamd_upstream_addr_cur (bk_conn->up), FALSE)) {
  1185. if (session->fname) {
  1186. rspamd_http_message_add_header (msg, "File", session->fname);
  1187. }
  1188. msg->method = HTTP_GET;
  1189. rspamd_http_connection_write_message_shared (bk_conn->backend_conn,
  1190. msg, NULL, NULL, bk_conn,
  1191. bk_conn->timeout);
  1192. }
  1193. else {
  1194. if (session->fname) {
  1195. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1196. rspamd_http_message_set_body (msg, session->map, session->map_len);
  1197. }
  1198. msg->method = HTTP_POST;
  1199. if (m->compress) {
  1200. proxy_request_compress (msg);
  1201. if (session->client_milter_conn) {
  1202. rspamd_http_message_add_header (msg, "Content-Type",
  1203. "application/octet-stream");
  1204. }
  1205. }
  1206. else {
  1207. if (session->client_milter_conn) {
  1208. rspamd_http_message_add_header (msg, "Content-Type",
  1209. "text/plain");
  1210. }
  1211. }
  1212. rspamd_http_connection_write_message (bk_conn->backend_conn,
  1213. msg, NULL, NULL, bk_conn,
  1214. bk_conn->timeout);
  1215. }
  1216. g_ptr_array_add (session->mirror_conns, bk_conn);
  1217. REF_RETAIN (session);
  1218. msg_info_session ("send request to %s", m->name);
  1219. }
  1220. }
  1221. static void
  1222. proxy_client_write_error (struct rspamd_proxy_session *session, gint code,
  1223. const gchar *status)
  1224. {
  1225. struct rspamd_http_message *reply;
  1226. if (session->client_milter_conn) {
  1227. rspamd_milter_send_action (session->client_milter_conn,
  1228. RSPAMD_MILTER_TEMPFAIL);
  1229. REF_RELEASE (session);
  1230. }
  1231. else {
  1232. reply = rspamd_http_new_message (HTTP_RESPONSE);
  1233. reply->code = code;
  1234. reply->status = rspamd_fstring_new_init (status, strlen (status));
  1235. rspamd_http_connection_write_message (session->client_conn,
  1236. reply, NULL, NULL, session,
  1237. session->ctx->timeout);
  1238. }
  1239. }
  1240. static void
  1241. proxy_backend_master_error_handler (struct rspamd_http_connection *conn, GError *err)
  1242. {
  1243. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1244. struct rspamd_proxy_session *session;
  1245. session = bk_conn->s;
  1246. msg_info_session ("abnormally closing connection from backend: %s, error: %e,"
  1247. " retries left: %d",
  1248. rspamd_inet_address_to_string (
  1249. rspamd_upstream_addr_cur (session->master_conn->up)),
  1250. err,
  1251. session->ctx->max_retries - session->retries);
  1252. session->retries ++;
  1253. rspamd_upstream_fail (bk_conn->up, FALSE);
  1254. proxy_backend_close_connection (session->master_conn);
  1255. if (session->ctx->max_retries &&
  1256. session->retries > session->ctx->max_retries) {
  1257. msg_err_session ("cannot connect to upstream, maximum retries "
  1258. "has been reached: %d", session->retries);
  1259. /* Terminate session immediately */
  1260. proxy_client_write_error (session, err->code, err->message);
  1261. }
  1262. else {
  1263. if (!proxy_send_master_message (session)) {
  1264. proxy_client_write_error (session, err->code, err->message);
  1265. }
  1266. else {
  1267. msg_info_session ("retry connection to: %s"
  1268. " retries left: %d",
  1269. rspamd_inet_address_to_string (
  1270. rspamd_upstream_addr_cur (session->master_conn->up)),
  1271. session->ctx->max_retries - session->retries);
  1272. }
  1273. }
  1274. }
  1275. static gint
  1276. proxy_backend_master_finish_handler (struct rspamd_http_connection *conn,
  1277. struct rspamd_http_message *msg)
  1278. {
  1279. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1280. struct rspamd_proxy_session *session, *nsession;
  1281. rspamd_fstring_t *reply;
  1282. goffset body_offset = -1;
  1283. session = bk_conn->s;
  1284. rspamd_http_connection_steal_msg (session->master_conn->backend_conn);
  1285. proxy_request_decompress (msg);
  1286. rspamd_http_message_remove_header (msg, "Content-Length");
  1287. rspamd_http_message_remove_header (msg, "Key");
  1288. rspamd_http_connection_reset (session->master_conn->backend_conn);
  1289. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1290. bk_conn->parser_from_ref, msg, &body_offset)) {
  1291. msg_warn_session ("cannot parse results from the master backend");
  1292. }
  1293. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1294. /* We need to reformat ucl to fit with legacy spamc protocol */
  1295. if (bk_conn->results) {
  1296. reply = rspamd_fstring_new ();
  1297. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1298. rspamd_ucl_tospamc_output (bk_conn->results, &reply);
  1299. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1300. }
  1301. else {
  1302. rspamd_ucl_torspamc_output (bk_conn->results, &reply);
  1303. }
  1304. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1305. msg->method = HTTP_SYMBOLS;
  1306. }
  1307. else {
  1308. msg_warn_session ("cannot parse results from the master backend, "
  1309. "return them as is");
  1310. }
  1311. }
  1312. rspamd_upstream_ok (bk_conn->up);
  1313. if (session->client_milter_conn) {
  1314. nsession = proxy_session_refresh (session);
  1315. if (body_offset > 0) {
  1316. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1317. session->master_conn->results,
  1318. msg->body_buf.begin + body_offset,
  1319. msg->body_buf.len - body_offset);
  1320. }
  1321. else {
  1322. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1323. session->master_conn->results, NULL, 0);
  1324. }
  1325. REF_RELEASE (session);
  1326. rspamd_http_message_free (msg);
  1327. }
  1328. else {
  1329. rspamd_http_connection_write_message (session->client_conn,
  1330. msg, NULL, NULL, session,
  1331. bk_conn->timeout);
  1332. }
  1333. return 0;
  1334. }
  1335. static void
  1336. rspamd_proxy_scan_self_reply (struct rspamd_task *task)
  1337. {
  1338. struct rspamd_http_message *msg;
  1339. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1340. ucl_object_t *rep = NULL;
  1341. const char *ctype = "application/json";
  1342. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1343. msg->date = time (NULL);
  1344. msg->code = 200;
  1345. switch (task->cmd) {
  1346. case CMD_CHECK:
  1347. case CMD_SKIP:
  1348. case CMD_CHECK_RSPAMC:
  1349. case CMD_CHECK_SPAMC:
  1350. case CMD_CHECK_V2:
  1351. rspamd_task_set_finish_time (task);
  1352. rspamd_protocol_http_reply (msg, task, &rep);
  1353. rspamd_protocol_write_log_pipe (task);
  1354. break;
  1355. case CMD_PING:
  1356. rspamd_http_message_set_body (msg, "pong" CRLF, 6);
  1357. ctype = "text/plain";
  1358. break;
  1359. default:
  1360. msg_err_task ("BROKEN");
  1361. break;
  1362. }
  1363. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1364. if (rep) {
  1365. session->master_conn->results = ucl_object_ref (rep);
  1366. }
  1367. if (session->client_milter_conn) {
  1368. nsession = proxy_session_refresh (session);
  1369. if (task->flags & RSPAMD_TASK_FLAG_MESSAGE_REWRITE) {
  1370. const gchar *start;
  1371. goffset len, hdr_off;
  1372. start = task->msg.begin;
  1373. len = task->msg.len;
  1374. hdr_off = MESSAGE_FIELD (task, raw_headers_content).len;
  1375. if (hdr_off < len) {
  1376. start += hdr_off;
  1377. len -= hdr_off;
  1378. /* The problem here is that we need not end of headers, we need
  1379. * start of body.
  1380. *
  1381. * Hence, we need to skip one \r\n till there is anything else in
  1382. * a line.
  1383. */
  1384. if (*start == '\r' && len > 0) {
  1385. start++;
  1386. len--;
  1387. }
  1388. if (*start == '\n' && len > 0) {
  1389. start++;
  1390. len--;
  1391. }
  1392. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1393. session->master_conn->results, start, len);
  1394. }
  1395. else {
  1396. /* XXX: should never happen! */
  1397. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1398. session->master_conn->results, NULL, 0);
  1399. }
  1400. }
  1401. else {
  1402. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1403. session->master_conn->results, NULL, 0);
  1404. }
  1405. rspamd_http_message_free (msg);
  1406. REF_RELEASE (session);
  1407. }
  1408. else {
  1409. rspamd_http_connection_reset (session->client_conn);
  1410. rspamd_http_connection_write_message (session->client_conn,
  1411. msg,
  1412. NULL,
  1413. ctype,
  1414. session,
  1415. 0);
  1416. }
  1417. }
  1418. static gboolean
  1419. rspamd_proxy_task_fin (void *ud)
  1420. {
  1421. struct rspamd_task *task = ud;
  1422. msg_debug_task ("finish task");
  1423. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1424. rspamd_proxy_scan_self_reply (task);
  1425. return TRUE;
  1426. }
  1427. if (!rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL)) {
  1428. rspamd_proxy_scan_self_reply (task);
  1429. return TRUE;
  1430. }
  1431. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1432. rspamd_proxy_scan_self_reply (task);
  1433. return TRUE;
  1434. }
  1435. /* One more iteration */
  1436. return FALSE;
  1437. }
  1438. static gboolean
  1439. rspamd_proxy_self_scan (struct rspamd_proxy_session *session)
  1440. {
  1441. struct rspamd_task *task;
  1442. struct rspamd_http_message *msg;
  1443. const gchar *data;
  1444. gsize len;
  1445. msg = session->client_message;
  1446. task = rspamd_task_new (session->worker, session->ctx->cfg,
  1447. session->pool, session->ctx->lang_det,
  1448. session->ctx->event_loop);
  1449. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1450. if (session->ctx->milter) {
  1451. task->protocol_flags |= RSPAMD_TASK_PROTOCOL_FLAG_MILTER|
  1452. RSPAMD_TASK_PROTOCOL_FLAG_BODY_BLOCK;
  1453. }
  1454. task->sock = -1;
  1455. if (session->client_milter_conn) {
  1456. task->client_addr = rspamd_inet_address_copy (
  1457. session->client_milter_conn->addr);
  1458. }
  1459. else {
  1460. task->client_addr = rspamd_inet_address_copy (session->client_addr);
  1461. }
  1462. task->fin_arg = session;
  1463. task->resolver = session->ctx->resolver;
  1464. /* TODO: allow to disable autolearn in protocol */
  1465. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1466. task->s = rspamd_session_create (task->task_pool, rspamd_proxy_task_fin,
  1467. NULL, (event_finalizer_t )rspamd_task_free, task);
  1468. data = rspamd_http_message_get_body (msg, &len);
  1469. if (session->backend->settings_id) {
  1470. rspamd_http_message_remove_header (msg, "Settings-ID");
  1471. rspamd_http_message_add_header (msg, "Settings-ID",
  1472. session->backend->settings_id);
  1473. }
  1474. /* Process message */
  1475. if (!rspamd_protocol_handle_request (task, msg)) {
  1476. msg_err_task ("cannot handle request: %e", task->err);
  1477. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1478. }
  1479. else {
  1480. if (task->cmd == CMD_PING) {
  1481. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1482. }
  1483. else {
  1484. if (!rspamd_task_load_message (task, msg, data, len)) {
  1485. msg_err_task ("cannot load message: %e", task->err);
  1486. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1487. }
  1488. }
  1489. }
  1490. /* Set global timeout for the task */
  1491. if (session->ctx->default_upstream->timeout > 0.0) {
  1492. task->timeout_ev.data = task;
  1493. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1494. session->ctx->default_upstream->timeout,
  1495. session->ctx->default_upstream->timeout);
  1496. ev_timer_start (task->event_loop, &task->timeout_ev);
  1497. }
  1498. else if (session->ctx->has_self_scan) {
  1499. if (session->ctx->cfg->task_timeout > 0) {
  1500. task->timeout_ev.data = task;
  1501. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1502. session->ctx->cfg->task_timeout,
  1503. session->ctx->default_upstream->timeout);
  1504. ev_timer_start (task->event_loop, &task->timeout_ev);
  1505. }
  1506. }
  1507. session->master_conn->task = task;
  1508. rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL);
  1509. rspamd_session_pending (task->s);
  1510. return TRUE;
  1511. }
  1512. static gboolean
  1513. proxy_send_master_message (struct rspamd_proxy_session *session)
  1514. {
  1515. struct rspamd_http_message *msg;
  1516. struct rspamd_http_upstream *backend = NULL;
  1517. const rspamd_ftok_t *host;
  1518. GError *err = NULL;
  1519. gchar hostbuf[512];
  1520. host = rspamd_http_message_find_header (session->client_message, "Host");
  1521. if (host == NULL) {
  1522. backend = session->ctx->default_upstream;
  1523. }
  1524. else {
  1525. rspamd_strlcpy (hostbuf, host->begin, MIN(host->len + 1, sizeof (hostbuf)));
  1526. backend = g_hash_table_lookup (session->ctx->upstreams, hostbuf);
  1527. if (backend == NULL) {
  1528. backend = session->ctx->default_upstream;
  1529. }
  1530. }
  1531. if (backend == NULL) {
  1532. /* No backend */
  1533. msg_err_session ("cannot find upstream for %s", host ? hostbuf : "default");
  1534. goto err;
  1535. }
  1536. else {
  1537. session->backend = backend;
  1538. if (backend->self_scan) {
  1539. return rspamd_proxy_self_scan (session);
  1540. }
  1541. retry:
  1542. if (session->ctx->max_retries &&
  1543. session->retries > session->ctx->max_retries) {
  1544. msg_err_session ("cannot connect to upstream, maximum retries "
  1545. "has been reached: %d", session->retries);
  1546. goto err;
  1547. }
  1548. session->master_conn->up = rspamd_upstream_get (backend->u,
  1549. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1550. session->master_conn->timeout = backend->timeout;
  1551. if (session->master_conn->up == NULL) {
  1552. msg_err_session ("cannot select upstream for %s",
  1553. host ? hostbuf : "default");
  1554. goto err;
  1555. }
  1556. session->master_conn->backend_sock = rspamd_inet_address_connect (
  1557. rspamd_upstream_addr_next (session->master_conn->up),
  1558. SOCK_STREAM, TRUE);
  1559. if (session->master_conn->backend_sock == -1) {
  1560. msg_err_session ("cannot connect upstream: %s(%s)",
  1561. host ? hostbuf : "default",
  1562. rspamd_inet_address_to_string (
  1563. rspamd_upstream_addr_cur (
  1564. session->master_conn->up)));
  1565. rspamd_upstream_fail (session->master_conn->up, TRUE);
  1566. session->retries ++;
  1567. goto retry;
  1568. }
  1569. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1570. if (msg == NULL) {
  1571. msg_err_session ("cannot copy message to send it to the upstream: %e",
  1572. err);
  1573. if (err) {
  1574. g_error_free (err);
  1575. }
  1576. goto err; /* No fallback here */
  1577. }
  1578. session->master_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1579. session->ctx->http_ctx,
  1580. NULL,
  1581. proxy_backend_master_error_handler,
  1582. proxy_backend_master_finish_handler,
  1583. RSPAMD_HTTP_CLIENT_SIMPLE,
  1584. session->master_conn->backend_sock);
  1585. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1586. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1587. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1588. if (backend->key) {
  1589. msg->peer_key = rspamd_pubkey_ref (backend->key);
  1590. }
  1591. if (backend->settings_id != NULL) {
  1592. rspamd_http_message_remove_header (msg, "Settings-ID");
  1593. rspamd_http_message_add_header (msg, "Settings-ID",
  1594. backend->settings_id);
  1595. }
  1596. if (backend->local ||
  1597. rspamd_inet_address_is_local (
  1598. rspamd_upstream_addr_cur (
  1599. session->master_conn->up), FALSE)) {
  1600. if (session->fname) {
  1601. rspamd_http_message_add_header (msg, "File", session->fname);
  1602. }
  1603. msg->method = HTTP_GET;
  1604. rspamd_http_connection_write_message_shared (
  1605. session->master_conn->backend_conn,
  1606. msg, NULL, NULL, session->master_conn,
  1607. session->master_conn->timeout);
  1608. }
  1609. else {
  1610. if (session->fname) {
  1611. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1612. rspamd_http_message_set_body (msg,
  1613. session->map, session->map_len);
  1614. }
  1615. msg->method = HTTP_POST;
  1616. if (backend->compress) {
  1617. proxy_request_compress (msg);
  1618. if (session->client_milter_conn) {
  1619. rspamd_http_message_add_header (msg, "Content-Type",
  1620. "application/octet-stream");
  1621. }
  1622. }
  1623. else {
  1624. if (session->client_milter_conn) {
  1625. rspamd_http_message_add_header (msg, "Content-Type",
  1626. "text/plain");
  1627. }
  1628. }
  1629. rspamd_http_connection_write_message (
  1630. session->master_conn->backend_conn,
  1631. msg, NULL, NULL, session->master_conn,
  1632. session->master_conn->timeout);
  1633. }
  1634. }
  1635. return TRUE;
  1636. err:
  1637. if (session->client_milter_conn) {
  1638. rspamd_milter_send_action (session->client_milter_conn,
  1639. RSPAMD_MILTER_TEMPFAIL);
  1640. REF_RELEASE (session);
  1641. }
  1642. else {
  1643. rspamd_http_connection_steal_msg (session->client_conn);
  1644. rspamd_http_connection_reset (session->client_conn);
  1645. proxy_client_write_error (session, 404, "Backend not found");
  1646. }
  1647. return FALSE;
  1648. }
  1649. static void
  1650. proxy_client_error_handler (struct rspamd_http_connection *conn, GError *err)
  1651. {
  1652. struct rspamd_proxy_session *session = conn->ud;
  1653. msg_info_session ("abnormally closing connection from: %s, error: %s",
  1654. rspamd_inet_address_to_string (session->client_addr), err->message);
  1655. /* Terminate session immediately */
  1656. proxy_backend_close_connection (session->master_conn);
  1657. REF_RELEASE (session);
  1658. }
  1659. static gint
  1660. proxy_client_finish_handler (struct rspamd_http_connection *conn,
  1661. struct rspamd_http_message *msg)
  1662. {
  1663. struct rspamd_proxy_session *session = conn->ud;
  1664. if (!session->master_conn) {
  1665. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1666. sizeof (*session->master_conn));
  1667. session->master_conn->s = session;
  1668. session->master_conn->name = "master";
  1669. /* Reset spamc legacy */
  1670. if (msg->method >= HTTP_SYMBOLS) {
  1671. msg->method = HTTP_POST;
  1672. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1673. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1674. msg_info_session ("enabling legacy spamc mode for session");
  1675. }
  1676. else {
  1677. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1678. msg_info_session ("enabling legacy rspamc mode for session");
  1679. }
  1680. }
  1681. if (msg->url->len == 0) {
  1682. msg->url = rspamd_fstring_append (msg->url,
  1683. "/" MSG_CMD_CHECK_V2, strlen ("/" MSG_CMD_CHECK_V2));
  1684. }
  1685. if (!proxy_check_file (msg, session)) {
  1686. goto err;
  1687. }
  1688. session->client_message = rspamd_http_connection_steal_msg (
  1689. session->client_conn);
  1690. session->shmem_ref = rspamd_http_message_shmem_ref (session->client_message);
  1691. rspamd_http_message_remove_header (msg, "Content-Length");
  1692. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1693. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1694. rspamd_http_message_remove_header (msg, "Connection");
  1695. rspamd_http_message_remove_header (msg, "Key");
  1696. proxy_open_mirror_connections (session);
  1697. rspamd_http_connection_reset (session->client_conn);
  1698. proxy_send_master_message (session);
  1699. }
  1700. else {
  1701. msg_info_session ("finished master connection");
  1702. proxy_backend_close_connection (session->master_conn);
  1703. REF_RELEASE (session);
  1704. }
  1705. return 0;
  1706. err:
  1707. rspamd_http_connection_steal_msg (session->client_conn);
  1708. rspamd_http_message_remove_header (msg, "Content-Length");
  1709. rspamd_http_message_remove_header (msg, "Key");
  1710. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1711. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1712. rspamd_http_message_remove_header (msg, "Connection");
  1713. rspamd_http_connection_reset (session->client_conn);
  1714. proxy_client_write_error (session, 404, "Backend not found");
  1715. return 0;
  1716. }
  1717. static void
  1718. proxy_milter_finish_handler (gint fd,
  1719. struct rspamd_milter_session *rms,
  1720. void *ud)
  1721. {
  1722. struct rspamd_proxy_session *session = ud;
  1723. struct rspamd_http_message *msg;
  1724. session->client_milter_conn = rms;
  1725. if (rms->message == NULL || rms->message->len == 0) {
  1726. msg_info_session ("finished milter connection");
  1727. proxy_backend_close_connection (session->master_conn);
  1728. REF_RELEASE (session);
  1729. }
  1730. else {
  1731. if (!session->master_conn) {
  1732. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1733. sizeof (*session->master_conn));
  1734. }
  1735. msg = rspamd_milter_to_http (rms);
  1736. session->master_conn->s = session;
  1737. session->master_conn->name = "master";
  1738. session->client_message = msg;
  1739. proxy_open_mirror_connections (session);
  1740. proxy_send_master_message (session);
  1741. }
  1742. }
  1743. static void
  1744. proxy_milter_error_handler (gint fd,
  1745. struct rspamd_milter_session *rms,
  1746. void *ud, GError *err)
  1747. {
  1748. struct rspamd_proxy_session *session = ud;
  1749. msg_info_session ("abnormally closing milter connection from: %s, "
  1750. "error: %s", rspamd_inet_address_to_string (session->client_addr),
  1751. err->message);
  1752. /* Terminate session immediately */
  1753. proxy_backend_close_connection (session->master_conn);
  1754. REF_RELEASE (session);
  1755. }
  1756. static void
  1757. proxy_accept_socket (EV_P_ ev_io *w, int revents)
  1758. {
  1759. struct rspamd_worker *worker = (struct rspamd_worker *)w->data;
  1760. struct rspamd_proxy_ctx *ctx;
  1761. rspamd_inet_addr_t *addr;
  1762. struct rspamd_proxy_session *session;
  1763. gint nfd;
  1764. ctx = worker->ctx;
  1765. if ((nfd =
  1766. rspamd_accept_from_socket (w->fd, &addr,
  1767. rspamd_worker_throttle_accept_events, worker->accept_events)) == -1) {
  1768. msg_warn ("accept failed: %s", strerror (errno));
  1769. return;
  1770. }
  1771. /* Check for EAGAIN */
  1772. if (nfd == 0) {
  1773. return;
  1774. }
  1775. session = g_malloc0 (sizeof (*session));
  1776. REF_INIT_RETAIN (session, proxy_session_dtor);
  1777. session->client_sock = nfd;
  1778. session->client_addr = addr;
  1779. session->mirror_conns = g_ptr_array_sized_new (ctx->mirrors->len);
  1780. session->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (),
  1781. "proxy");
  1782. session->ctx = ctx;
  1783. session->worker = worker;
  1784. if (ctx->sessions_cache) {
  1785. rspamd_worker_session_cache_add (ctx->sessions_cache,
  1786. session->pool->tag.uid, &session->ref.refcount, session);
  1787. }
  1788. if (!ctx->milter) {
  1789. session->client_conn = rspamd_http_connection_new_server (
  1790. ctx->http_ctx,
  1791. nfd,
  1792. NULL,
  1793. proxy_client_error_handler,
  1794. proxy_client_finish_handler,
  1795. 0);
  1796. if (ctx->key) {
  1797. rspamd_http_connection_set_key (session->client_conn, ctx->key);
  1798. }
  1799. msg_info_session ("accepted http connection from %s port %d",
  1800. rspamd_inet_address_to_string (addr),
  1801. rspamd_inet_address_get_port (addr));
  1802. rspamd_http_connection_read_message_shared (session->client_conn,
  1803. session,
  1804. session->ctx->timeout);
  1805. }
  1806. else {
  1807. msg_info_session ("accepted milter connection from %s port %d",
  1808. rspamd_inet_address_to_string (addr),
  1809. rspamd_inet_address_get_port (addr));
  1810. #ifdef TCP_NODELAY
  1811. #ifndef SOL_TCP
  1812. #define SOL_TCP IPPROTO_TCP
  1813. #endif
  1814. if (rspamd_inet_address_get_af (addr) != AF_UNIX) {
  1815. gint sopt = 1;
  1816. if (setsockopt (nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof (sopt)) ==
  1817. -1) {
  1818. msg_warn_session ("cannot set TCP_NODELAY: %s",
  1819. strerror (errno));
  1820. }
  1821. }
  1822. #endif
  1823. rspamd_milter_handle_socket (nfd, 0.0,
  1824. session->pool,
  1825. ctx->event_loop,
  1826. proxy_milter_finish_handler,
  1827. proxy_milter_error_handler,
  1828. session);
  1829. }
  1830. }
  1831. static void
  1832. adjust_upstreams_limits (struct rspamd_proxy_ctx *ctx)
  1833. {
  1834. struct rspamd_http_upstream *backend;
  1835. gpointer k, v;
  1836. GHashTableIter it;
  1837. /*
  1838. * We set error time equal to max_retries * backend_timeout and max_errors
  1839. * to max_retries - 1
  1840. *
  1841. * So if we failed to scan a message on a backend for some reasons, we
  1842. * will try to re-resolve it faster
  1843. */
  1844. g_hash_table_iter_init (&it, ctx->upstreams);
  1845. while (g_hash_table_iter_next (&it, &k, &v)) {
  1846. backend = (struct rspamd_http_upstream *)v;
  1847. if (!backend->self_scan && backend->u) {
  1848. rspamd_upstreams_set_limits (backend->u,
  1849. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1850. ctx->max_retries - 1, 0);
  1851. }
  1852. }
  1853. }
  1854. void
  1855. start_rspamd_proxy (struct rspamd_worker *worker)
  1856. {
  1857. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1858. g_assert (rspamd_worker_check_context (worker->ctx, rspamd_rspamd_proxy_magic));
  1859. ctx->cfg = worker->srv->cfg;
  1860. ctx->event_loop = rspamd_prepare_worker (worker, "rspamd_proxy",
  1861. proxy_accept_socket);
  1862. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  1863. ctx->event_loop,
  1864. worker->srv->cfg);
  1865. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver, worker, 0);
  1866. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  1867. ctx->event_loop, ctx->resolver->r);
  1868. ctx->http_ctx = rspamd_http_context_create (ctx->cfg, ctx->event_loop,
  1869. ctx->cfg->ups_ctx);
  1870. rspamd_mempool_add_destructor (ctx->cfg->cfg_pool,
  1871. (rspamd_mempool_destruct_t)rspamd_http_context_free,
  1872. ctx->http_ctx);
  1873. if (ctx->has_self_scan) {
  1874. /* Additional initialisation needed */
  1875. rspamd_worker_init_scanner (worker, ctx->event_loop, ctx->resolver,
  1876. &ctx->lang_det);
  1877. }
  1878. if (worker->srv->cfg->enable_sessions_cache) {
  1879. ctx->sessions_cache = rspamd_worker_session_cache_new (worker,
  1880. ctx->event_loop);
  1881. }
  1882. ctx->milter_ctx.spam_header = ctx->spam_header;
  1883. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  1884. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  1885. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  1886. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  1887. ctx->milter_ctx.reject_message = ctx->reject_message;
  1888. ctx->milter_ctx.cfg = ctx->cfg;
  1889. rspamd_milter_init_library (&ctx->milter_ctx);
  1890. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->event_loop,
  1891. worker);
  1892. adjust_upstreams_limits (ctx);
  1893. ev_loop (ctx->event_loop, 0);
  1894. rspamd_worker_block_signals ();
  1895. if (ctx->has_self_scan) {
  1896. rspamd_stat_close ();
  1897. }
  1898. REF_RELEASE (ctx->cfg);
  1899. rspamd_log_close (worker->srv->logger, TRUE);
  1900. exit (EXIT_SUCCESS);
  1901. }