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 61KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400
  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 "libserver/maps/map.h"
  19. #include "libutil/upstream.h"
  20. #include "libserver/http/http_connection.h"
  21. #include "libserver/http/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. const rspamd_ftok_t *ct)
  739. {
  740. struct ucl_parser *parser;
  741. gint err_idx;
  742. const gchar *in = msg->body_buf.begin;
  743. gsize inlen = msg->body_buf.len;
  744. const rspamd_ftok_t *offset_hdr;
  745. if (inlen == 0 || in == NULL) {
  746. return FALSE;
  747. }
  748. offset_hdr = rspamd_http_message_find_header (msg, MESSAGE_OFFSET_HEADER);
  749. if (offset_hdr) {
  750. gulong val;
  751. if (rspamd_strtoul (offset_hdr->begin, offset_hdr->len, &val)
  752. && val < inlen) {
  753. if (body_offset) {
  754. *body_offset = val;
  755. }
  756. inlen = val;
  757. }
  758. }
  759. if (parser_ref != -1) {
  760. /* Call parser function */
  761. lua_pushcfunction (L, &rspamd_lua_traceback);
  762. err_idx = lua_gettop (L);
  763. lua_rawgeti (L, LUA_REGISTRYINDEX, parser_ref);
  764. /* XXX: copies all data */
  765. lua_pushlstring (L, in, inlen);
  766. if (lua_pcall (L, 1, 1, err_idx) != 0) {
  767. msg_err_session (
  768. "cannot run lua parser script: %s",
  769. lua_tostring (L, -1));
  770. lua_settop (L, 0);
  771. return FALSE;
  772. }
  773. conn->results = ucl_object_lua_import (L, -1);
  774. lua_settop (L, 0);
  775. }
  776. else {
  777. rspamd_ftok_t json_ct;
  778. RSPAMD_FTOK_ASSIGN (&json_ct, "application/json");
  779. if (ct && rspamd_ftok_casecmp (ct, &json_ct) == 0) {
  780. parser = ucl_parser_new (0);
  781. if (!ucl_parser_add_chunk (parser, in, inlen)) {
  782. gchar *encoded;
  783. encoded = rspamd_encode_base64 (in, inlen, 0, NULL);
  784. msg_err_session ("cannot parse input: %s", ucl_parser_get_error (
  785. parser));
  786. msg_err_session ("input encoded: %s", encoded);
  787. ucl_parser_free (parser);
  788. g_free (encoded);
  789. return FALSE;
  790. }
  791. conn->results = ucl_parser_get_object (parser);
  792. ucl_parser_free (parser);
  793. }
  794. }
  795. return TRUE;
  796. }
  797. static void
  798. proxy_call_cmp_script (struct rspamd_proxy_session *session, gint cbref)
  799. {
  800. gint err_idx;
  801. guint i;
  802. struct rspamd_proxy_backend_connection *conn;
  803. lua_State *L;
  804. L = session->ctx->lua_state;
  805. lua_pushcfunction (L, &rspamd_lua_traceback);
  806. err_idx = lua_gettop (L);
  807. lua_rawgeti (L, LUA_REGISTRYINDEX, cbref);
  808. lua_createtable (L, 0, session->mirror_conns->len + 1);
  809. /* Now push master results */
  810. if (session->master_conn && session->master_conn->results) {
  811. lua_pushstring (L, "master");
  812. ucl_object_push_lua (L, session->master_conn->results, true);
  813. lua_settable (L, -3);
  814. }
  815. else {
  816. lua_pushstring (L, "master");
  817. lua_pushstring (L, "no results");
  818. lua_settable (L, -3);
  819. }
  820. for (i = 0; i < session->mirror_conns->len; i ++) {
  821. conn = g_ptr_array_index (session->mirror_conns, i);
  822. if (conn->results) {
  823. lua_pushstring (L, conn->name);
  824. ucl_object_push_lua (L, conn->results, true);
  825. lua_settable (L, -3);
  826. }
  827. else {
  828. lua_pushstring (L, conn->name);
  829. lua_pushstring (L, conn->err ? conn->err : "unknown error");
  830. lua_settable (L, -3);
  831. }
  832. }
  833. if (lua_pcall (L, 1, 0, err_idx) != 0) {
  834. msg_err_session (
  835. "cannot run lua compare script: %s",
  836. lua_tostring (L, -1));
  837. }
  838. lua_settop (L, 0);
  839. }
  840. static void
  841. proxy_session_dtor (struct rspamd_proxy_session *session)
  842. {
  843. guint i;
  844. gint cbref;
  845. struct rspamd_proxy_backend_connection *conn;
  846. if (session->master_conn && session->master_conn->results) {
  847. for (i = 0; i < session->ctx->cmp_refs->len; i++) {
  848. cbref = g_array_index (session->ctx->cmp_refs, gint, i);
  849. proxy_call_cmp_script (session, cbref);
  850. }
  851. }
  852. if (session->master_conn) {
  853. proxy_backend_close_connection (session->master_conn);
  854. }
  855. if (session->client_milter_conn) {
  856. rspamd_milter_session_unref (session->client_milter_conn);
  857. }
  858. else if (session->client_conn) {
  859. rspamd_http_connection_reset (session->client_conn);
  860. rspamd_http_connection_unref (session->client_conn);
  861. }
  862. if (session->map && session->map_len) {
  863. munmap (session->map, session->map_len);
  864. }
  865. for (i = 0; i < session->mirror_conns->len; i ++) {
  866. conn = g_ptr_array_index (session->mirror_conns, i);
  867. if (!(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  868. proxy_backend_close_connection (conn);
  869. }
  870. if (conn->results) {
  871. ucl_object_unref (conn->results);
  872. }
  873. }
  874. if (session->master_conn) {
  875. if (session->master_conn->results) {
  876. ucl_object_unref (session->master_conn->results);
  877. }
  878. if (session->master_conn->task) {
  879. rspamd_session_destroy (session->master_conn->task->s);
  880. }
  881. }
  882. g_ptr_array_free (session->mirror_conns, TRUE);
  883. rspamd_http_message_shmem_unref (session->shmem_ref);
  884. rspamd_http_message_unref (session->client_message);
  885. if (session->client_addr) {
  886. rspamd_inet_address_free (session->client_addr);
  887. }
  888. if (session->client_sock != -1) {
  889. close (session->client_sock);
  890. }
  891. if (session->ctx->sessions_cache) {
  892. rspamd_worker_session_cache_remove (session->ctx->sessions_cache,
  893. session);
  894. }
  895. if (session->pool) {
  896. rspamd_mempool_delete (session->pool);
  897. }
  898. g_free (session);
  899. }
  900. static void
  901. proxy_request_compress (struct rspamd_http_message *msg)
  902. {
  903. guint flags;
  904. ZSTD_CCtx *zctx;
  905. rspamd_fstring_t *body;
  906. const gchar *in;
  907. gsize inlen;
  908. flags = rspamd_http_message_get_flags (msg);
  909. if (!rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  910. if ((flags & RSPAMD_HTTP_FLAG_SHMEM) ||
  911. !(flags & RSPAMD_HTTP_FLAG_HAS_BODY)) {
  912. /* Cannot compress shared or empty message */
  913. return;
  914. }
  915. in = rspamd_http_message_get_body (msg, &inlen);
  916. if (in == NULL || inlen == 0) {
  917. return;
  918. }
  919. body = rspamd_fstring_sized_new (ZSTD_compressBound (inlen));
  920. zctx = ZSTD_createCCtx ();
  921. body->len = ZSTD_compressCCtx (zctx, body->str, body->allocated,
  922. in, inlen, 1);
  923. if (ZSTD_isError (body->len)) {
  924. msg_err ("compression error");
  925. rspamd_fstring_free (body);
  926. ZSTD_freeCCtx (zctx);
  927. return;
  928. }
  929. ZSTD_freeCCtx (zctx);
  930. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  931. rspamd_http_message_add_header (msg, COMPRESSION_HEADER, "zstd");
  932. }
  933. }
  934. static void
  935. proxy_request_decompress (struct rspamd_http_message *msg)
  936. {
  937. rspamd_fstring_t *body;
  938. const gchar *in;
  939. gsize inlen, outlen, r;
  940. ZSTD_DStream *zstream;
  941. ZSTD_inBuffer zin;
  942. ZSTD_outBuffer zout;
  943. if (rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  944. in = rspamd_http_message_get_body (msg, &inlen);
  945. if (in == NULL || inlen == 0) {
  946. return;
  947. }
  948. zstream = ZSTD_createDStream ();
  949. ZSTD_initDStream (zstream);
  950. zin.pos = 0;
  951. zin.src = in;
  952. zin.size = inlen;
  953. if ((outlen = ZSTD_getDecompressedSize (zin.src, zin.size)) == 0) {
  954. outlen = ZSTD_DStreamOutSize ();
  955. }
  956. body = rspamd_fstring_sized_new (outlen);
  957. zout.dst = body->str;
  958. zout.pos = 0;
  959. zout.size = outlen;
  960. while (zin.pos < zin.size) {
  961. r = ZSTD_decompressStream (zstream, &zout, &zin);
  962. if (ZSTD_isError (r)) {
  963. msg_err ("Decompression error: %s", ZSTD_getErrorName (r));
  964. ZSTD_freeDStream (zstream);
  965. rspamd_fstring_free (body);
  966. return;
  967. }
  968. if (zout.pos == zout.size) {
  969. /* We need to extend output buffer */
  970. zout.size = zout.size * 2 + 1;
  971. body = rspamd_fstring_grow (body, zout.size);
  972. zout.size = body->allocated;
  973. zout.dst = body->str;
  974. }
  975. }
  976. body->len = zout.pos;
  977. ZSTD_freeDStream (zstream);
  978. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  979. rspamd_http_message_remove_header (msg, COMPRESSION_HEADER);
  980. }
  981. }
  982. static struct rspamd_proxy_session *
  983. proxy_session_refresh (struct rspamd_proxy_session *session)
  984. {
  985. struct rspamd_proxy_session *nsession;
  986. nsession = g_malloc0 (sizeof (*nsession));
  987. nsession->client_milter_conn = session->client_milter_conn;
  988. session->client_milter_conn = NULL;
  989. rspamd_milter_update_userdata (nsession->client_milter_conn,
  990. nsession);
  991. nsession->client_addr = session->client_addr;
  992. session->client_addr = NULL;
  993. nsession->ctx = session->ctx;
  994. nsession->worker = session->worker;
  995. nsession->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (), "proxy", 0);
  996. nsession->client_sock = session->client_sock;
  997. session->client_sock = -1;
  998. nsession->mirror_conns = g_ptr_array_sized_new (nsession->ctx->mirrors->len);
  999. REF_INIT_RETAIN (nsession, proxy_session_dtor);
  1000. if (nsession->ctx->sessions_cache) {
  1001. rspamd_worker_session_cache_add (nsession->ctx->sessions_cache,
  1002. nsession->pool->tag.uid, &nsession->ref.refcount, nsession);
  1003. }
  1004. return nsession;
  1005. }
  1006. static gboolean
  1007. proxy_check_file (struct rspamd_http_message *msg,
  1008. struct rspamd_proxy_session *session)
  1009. {
  1010. const rspamd_ftok_t *tok, *key_tok;
  1011. rspamd_ftok_t srch;
  1012. gchar *file_str;
  1013. GHashTable *query_args;
  1014. GHashTableIter it;
  1015. gpointer k, v;
  1016. struct http_parser_url u;
  1017. rspamd_fstring_t *new_url;
  1018. tok = rspamd_http_message_find_header (msg, "File");
  1019. if (tok) {
  1020. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1021. session->map = rspamd_file_xmap (file_str, PROT_READ, &session->map_len,
  1022. TRUE);
  1023. if (session->map == NULL) {
  1024. if (session->map_len != 0) {
  1025. msg_err_session ("cannot map %s: %s", file_str,
  1026. strerror (errno));
  1027. return FALSE;
  1028. }
  1029. }
  1030. /* Remove header after processing */
  1031. rspamd_http_message_remove_header (msg, "File");
  1032. session->fname = file_str;
  1033. }
  1034. else {
  1035. /* Need to parse query URL */
  1036. if (http_parser_parse_url (RSPAMD_FSTRING_DATA (msg->url),
  1037. RSPAMD_FSTRING_LEN (msg->url), 0, &u) != 0) {
  1038. msg_err_session ("bad request url: %V", msg->url);
  1039. return FALSE;
  1040. }
  1041. if (u.field_set & (1 << UF_QUERY)) {
  1042. /* In case if we have a query, we need to store it somewhere */
  1043. query_args = rspamd_http_message_parse_query (msg);
  1044. srch.begin = "File";
  1045. srch.len = strlen ("File");
  1046. tok = g_hash_table_lookup (query_args, &srch);
  1047. if (tok) {
  1048. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1049. session->map = rspamd_file_xmap (file_str, PROT_READ,
  1050. &session->map_len, TRUE);
  1051. if (session->map == NULL) {
  1052. if (session->map_len != 0) {
  1053. msg_err_session ("cannot map %s: %s", file_str,
  1054. strerror (errno));
  1055. g_hash_table_unref (query_args);
  1056. return FALSE;
  1057. }
  1058. }
  1059. /* We need to create a new URL with file attribute removed */
  1060. new_url = rspamd_fstring_new_init (RSPAMD_FSTRING_DATA (msg->url),
  1061. u.field_data[UF_QUERY].off);
  1062. new_url = rspamd_fstring_append (new_url, "?", 1);
  1063. g_hash_table_iter_init (&it, query_args);
  1064. while (g_hash_table_iter_next (&it, &k, &v)) {
  1065. key_tok = k;
  1066. tok = v;
  1067. if (!rspamd_ftok_icase_equal (key_tok, &srch)) {
  1068. rspamd_printf_fstring (&new_url, "%T=%T&",
  1069. key_tok, tok);
  1070. }
  1071. }
  1072. /* Erase last character (might be either & or ?) */
  1073. rspamd_fstring_erase (new_url, new_url->len - 1, 1);
  1074. rspamd_fstring_free (msg->url);
  1075. msg->url = new_url;
  1076. session->fname = file_str;
  1077. }
  1078. g_hash_table_unref (query_args);
  1079. }
  1080. }
  1081. return TRUE;
  1082. }
  1083. static void
  1084. proxy_backend_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  1085. {
  1086. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1087. struct rspamd_proxy_session *session;
  1088. session = bk_conn->s;
  1089. msg_info_session ("abnormally closing connection from backend: %s:%s, "
  1090. "error: %e",
  1091. bk_conn->name,
  1092. rspamd_inet_address_to_string_pretty (
  1093. rspamd_upstream_addr_cur (bk_conn->up)),
  1094. err);
  1095. if (err) {
  1096. bk_conn->err = rspamd_mempool_strdup (session->pool, err->message);
  1097. }
  1098. rspamd_upstream_fail (bk_conn->up, FALSE, err ? err->message : "unknown");
  1099. proxy_backend_close_connection (bk_conn);
  1100. REF_RELEASE (bk_conn->s);
  1101. }
  1102. static gint
  1103. proxy_backend_mirror_finish_handler (struct rspamd_http_connection *conn,
  1104. struct rspamd_http_message *msg)
  1105. {
  1106. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1107. struct rspamd_proxy_session *session;
  1108. const rspamd_ftok_t *orig_ct;
  1109. session = bk_conn->s;
  1110. proxy_request_decompress (msg);
  1111. orig_ct = rspamd_http_message_find_header (msg, "Content-Type");
  1112. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1113. bk_conn->parser_from_ref, msg, NULL, orig_ct)) {
  1114. msg_warn_session ("cannot parse results from the mirror backend %s:%s",
  1115. bk_conn->name,
  1116. rspamd_inet_address_to_string (
  1117. rspamd_upstream_addr_cur (bk_conn->up)));
  1118. bk_conn->err = "cannot parse ucl";
  1119. }
  1120. msg_info_session ("finished mirror connection to %s", bk_conn->name);
  1121. rspamd_upstream_ok (bk_conn->up);
  1122. proxy_backend_close_connection (bk_conn);
  1123. REF_RELEASE (bk_conn->s);
  1124. return 0;
  1125. }
  1126. static void
  1127. proxy_open_mirror_connections (struct rspamd_proxy_session *session)
  1128. {
  1129. gdouble coin;
  1130. struct rspamd_http_mirror *m;
  1131. guint i;
  1132. struct rspamd_proxy_backend_connection *bk_conn;
  1133. struct rspamd_http_message *msg;
  1134. GError *err = NULL;
  1135. coin = rspamd_random_double ();
  1136. for (i = 0; i < session->ctx->mirrors->len; i ++) {
  1137. m = g_ptr_array_index (session->ctx->mirrors, i);
  1138. if (m->prob < coin) {
  1139. /* No luck */
  1140. continue;
  1141. }
  1142. bk_conn = rspamd_mempool_alloc0 (session->pool,
  1143. sizeof (*bk_conn));
  1144. bk_conn->s = session;
  1145. bk_conn->name = m->name;
  1146. bk_conn->timeout = m->timeout;
  1147. bk_conn->up = rspamd_upstream_get (m->u,
  1148. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1149. bk_conn->parser_from_ref = m->parser_from_ref;
  1150. bk_conn->parser_to_ref = m->parser_to_ref;
  1151. if (bk_conn->up == NULL) {
  1152. msg_err_session ("cannot select upstream for %s", m->name);
  1153. continue;
  1154. }
  1155. bk_conn->backend_sock = rspamd_inet_address_connect (
  1156. rspamd_upstream_addr_next (bk_conn->up),
  1157. SOCK_STREAM, TRUE);
  1158. if (bk_conn->backend_sock == -1) {
  1159. msg_err_session ("cannot connect upstream for %s", m->name);
  1160. rspamd_upstream_fail (bk_conn->up, TRUE, strerror (errno));
  1161. continue;
  1162. }
  1163. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1164. if (msg == NULL) {
  1165. msg_err_session ("cannot copy message to send to a mirror %s: %e",
  1166. m->name, err);
  1167. if (err) {
  1168. g_error_free (err);
  1169. }
  1170. continue;
  1171. }
  1172. if (msg->url->len == 0) {
  1173. msg->url = rspamd_fstring_append (msg->url, "/check", strlen ("/check"));
  1174. }
  1175. if (m->settings_id != NULL) {
  1176. rspamd_http_message_remove_header (msg, "Settings-ID");
  1177. rspamd_http_message_add_header (msg, "Settings-ID", m->settings_id);
  1178. }
  1179. bk_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1180. session->ctx->http_ctx,
  1181. NULL,
  1182. proxy_backend_mirror_error_handler,
  1183. proxy_backend_mirror_finish_handler,
  1184. RSPAMD_HTTP_CLIENT_SIMPLE,
  1185. bk_conn->backend_sock);
  1186. if (m->key) {
  1187. msg->peer_key = rspamd_pubkey_ref (m->key);
  1188. }
  1189. if (m->local ||
  1190. rspamd_inet_address_is_local (rspamd_upstream_addr_cur (bk_conn->up))) {
  1191. if (session->fname) {
  1192. rspamd_http_message_add_header (msg, "File", session->fname);
  1193. }
  1194. msg->method = HTTP_GET;
  1195. rspamd_http_connection_write_message_shared (bk_conn->backend_conn,
  1196. msg, NULL, NULL, bk_conn,
  1197. bk_conn->timeout);
  1198. }
  1199. else {
  1200. if (session->fname) {
  1201. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1202. rspamd_http_message_set_body (msg, session->map, session->map_len);
  1203. }
  1204. msg->method = HTTP_POST;
  1205. if (m->compress) {
  1206. proxy_request_compress (msg);
  1207. if (session->client_milter_conn) {
  1208. rspamd_http_message_add_header (msg, "Content-Type",
  1209. "application/octet-stream");
  1210. }
  1211. }
  1212. else {
  1213. if (session->client_milter_conn) {
  1214. rspamd_http_message_add_header (msg, "Content-Type",
  1215. "text/plain");
  1216. }
  1217. }
  1218. rspamd_http_connection_write_message (bk_conn->backend_conn,
  1219. msg, NULL, NULL, bk_conn,
  1220. bk_conn->timeout);
  1221. }
  1222. g_ptr_array_add (session->mirror_conns, bk_conn);
  1223. REF_RETAIN (session);
  1224. msg_info_session ("send request to %s", m->name);
  1225. }
  1226. }
  1227. static void
  1228. proxy_client_write_error (struct rspamd_proxy_session *session, gint code,
  1229. const gchar *status)
  1230. {
  1231. struct rspamd_http_message *reply;
  1232. if (session->client_milter_conn) {
  1233. rspamd_milter_send_action (session->client_milter_conn,
  1234. RSPAMD_MILTER_TEMPFAIL);
  1235. REF_RELEASE (session);
  1236. }
  1237. else {
  1238. reply = rspamd_http_new_message (HTTP_RESPONSE);
  1239. switch (code) {
  1240. case ETIMEDOUT:
  1241. reply->code = 504;
  1242. reply->status = RSPAMD_FSTRING_LIT ("Gateway timeout");
  1243. break;
  1244. case ECONNRESET:
  1245. case ECONNABORTED:
  1246. reply->code = 502;
  1247. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection reset");
  1248. break;
  1249. case ECONNREFUSED:
  1250. reply->code = 502;
  1251. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection refused");
  1252. break;
  1253. default:
  1254. if (code >= 300) {
  1255. /* Likely HTTP error */
  1256. reply->code = code;
  1257. reply->status = rspamd_fstring_new_init (status, strlen (status));
  1258. }
  1259. else {
  1260. reply->code = 502;
  1261. reply->status = RSPAMD_FSTRING_LIT ("Unknown gateway error: ");
  1262. reply->status = rspamd_fstring_append (reply->status,
  1263. status, strlen (status));
  1264. }
  1265. break;
  1266. }
  1267. rspamd_http_connection_write_message (session->client_conn,
  1268. reply, NULL, NULL, session,
  1269. session->ctx->timeout);
  1270. }
  1271. }
  1272. static void
  1273. proxy_backend_master_error_handler (struct rspamd_http_connection *conn, GError *err)
  1274. {
  1275. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1276. struct rspamd_proxy_session *session;
  1277. session = bk_conn->s;
  1278. session->retries ++;
  1279. msg_info_session ("abnormally closing connection from backend: %s, error: %e,"
  1280. " retries left: %d",
  1281. rspamd_inet_address_to_string_pretty (
  1282. rspamd_upstream_addr_cur (session->master_conn->up)),
  1283. err,
  1284. session->ctx->max_retries - session->retries);
  1285. rspamd_upstream_fail (bk_conn->up, FALSE, err ? err->message : "unknown");
  1286. proxy_backend_close_connection (session->master_conn);
  1287. if (session->ctx->max_retries > 0 &&
  1288. session->retries >= session->ctx->max_retries) {
  1289. msg_err_session ("cannot connect to upstream, maximum retries "
  1290. "has been reached: %d", session->retries);
  1291. /* Terminate session immediately */
  1292. proxy_client_write_error (session, err->code, err->message);
  1293. }
  1294. else {
  1295. if (!proxy_send_master_message (session)) {
  1296. proxy_client_write_error (session, err->code, err->message);
  1297. }
  1298. else {
  1299. msg_info_session ("retry connection to: %s"
  1300. " retries left: %d",
  1301. rspamd_inet_address_to_string (
  1302. rspamd_upstream_addr_cur (session->master_conn->up)),
  1303. session->ctx->max_retries - session->retries);
  1304. }
  1305. }
  1306. }
  1307. static gint
  1308. proxy_backend_master_finish_handler (struct rspamd_http_connection *conn,
  1309. struct rspamd_http_message *msg)
  1310. {
  1311. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1312. struct rspamd_proxy_session *session, *nsession;
  1313. rspamd_fstring_t *reply;
  1314. const rspamd_ftok_t *orig_ct;
  1315. goffset body_offset = -1;
  1316. session = bk_conn->s;
  1317. rspamd_http_connection_steal_msg (session->master_conn->backend_conn);
  1318. proxy_request_decompress (msg);
  1319. /*
  1320. * These are likely set by an http library, so we will double these headers
  1321. * if they are not removed
  1322. */
  1323. rspamd_http_message_remove_header (msg, "Content-Length");
  1324. rspamd_http_message_remove_header (msg, "Connection");
  1325. rspamd_http_message_remove_header (msg, "Date");
  1326. rspamd_http_message_remove_header (msg, "Server");
  1327. rspamd_http_message_remove_header (msg, "Key");
  1328. orig_ct = rspamd_http_message_find_header (msg, "Content-Type");
  1329. rspamd_http_connection_reset (session->master_conn->backend_conn);
  1330. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1331. bk_conn->parser_from_ref, msg, &body_offset, orig_ct)) {
  1332. msg_warn_session ("cannot parse results from the master backend");
  1333. }
  1334. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1335. /* We need to reformat ucl to fit with legacy spamc protocol */
  1336. if (bk_conn->results) {
  1337. reply = rspamd_fstring_new ();
  1338. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1339. rspamd_ucl_tospamc_output (bk_conn->results, &reply);
  1340. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1341. }
  1342. else {
  1343. rspamd_ucl_torspamc_output (bk_conn->results, &reply);
  1344. }
  1345. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1346. msg->method = HTTP_SYMBOLS;
  1347. }
  1348. else {
  1349. msg_debug_session ("cannot parse results from the master backend, "
  1350. "return them as is");
  1351. }
  1352. }
  1353. rspamd_upstream_ok (bk_conn->up);
  1354. if (session->client_milter_conn) {
  1355. nsession = proxy_session_refresh (session);
  1356. if (body_offset > 0) {
  1357. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1358. session->master_conn->results,
  1359. msg->body_buf.begin + body_offset,
  1360. msg->body_buf.len - body_offset);
  1361. }
  1362. else {
  1363. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1364. session->master_conn->results, NULL, 0);
  1365. }
  1366. REF_RELEASE (session);
  1367. rspamd_http_message_free (msg);
  1368. }
  1369. else {
  1370. const gchar *passed_ct = NULL;
  1371. if (orig_ct) {
  1372. passed_ct = rspamd_mempool_ftokdup (session->pool, orig_ct);
  1373. /* Remove original */
  1374. rspamd_http_message_remove_header (msg, "Content-Type");
  1375. }
  1376. rspamd_http_connection_write_message (session->client_conn,
  1377. msg, NULL, passed_ct, session,
  1378. bk_conn->timeout);
  1379. }
  1380. return 0;
  1381. }
  1382. static void
  1383. rspamd_proxy_scan_self_reply (struct rspamd_task *task)
  1384. {
  1385. struct rspamd_http_message *msg;
  1386. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1387. ucl_object_t *rep = NULL;
  1388. const char *ctype = "application/json";
  1389. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1390. msg->date = time (NULL);
  1391. msg->code = 200;
  1392. switch (task->cmd) {
  1393. case CMD_CHECK:
  1394. case CMD_SKIP:
  1395. case CMD_CHECK_RSPAMC:
  1396. case CMD_CHECK_SPAMC:
  1397. case CMD_CHECK_V2:
  1398. rspamd_task_set_finish_time (task);
  1399. rspamd_protocol_http_reply (msg, task, &rep);
  1400. rspamd_protocol_write_log_pipe (task);
  1401. break;
  1402. case CMD_PING:
  1403. rspamd_http_message_set_body (msg, "pong" CRLF, 6);
  1404. ctype = "text/plain";
  1405. break;
  1406. default:
  1407. msg_err_task ("BROKEN");
  1408. break;
  1409. }
  1410. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1411. if (rep) {
  1412. session->master_conn->results = ucl_object_ref (rep);
  1413. }
  1414. if (session->client_milter_conn) {
  1415. nsession = proxy_session_refresh (session);
  1416. if (task->flags & RSPAMD_TASK_FLAG_MESSAGE_REWRITE) {
  1417. const gchar *start;
  1418. goffset len, hdr_off;
  1419. start = task->msg.begin;
  1420. len = task->msg.len;
  1421. hdr_off = MESSAGE_FIELD (task, raw_headers_content).len;
  1422. if (hdr_off < len) {
  1423. start += hdr_off;
  1424. len -= hdr_off;
  1425. /* The problem here is that we need not end of headers, we need
  1426. * start of body.
  1427. *
  1428. * Hence, we need to skip one \r\n till there is anything else in
  1429. * a line.
  1430. */
  1431. if (*start == '\r' && len > 0) {
  1432. start++;
  1433. len--;
  1434. }
  1435. if (*start == '\n' && len > 0) {
  1436. start++;
  1437. len--;
  1438. }
  1439. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1440. session->master_conn->results, start, len);
  1441. }
  1442. else {
  1443. /* XXX: should never happen! */
  1444. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1445. session->master_conn->results, NULL, 0);
  1446. }
  1447. }
  1448. else {
  1449. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1450. session->master_conn->results, NULL, 0);
  1451. }
  1452. rspamd_http_message_free (msg);
  1453. REF_RELEASE (session);
  1454. }
  1455. else {
  1456. rspamd_http_connection_reset (session->client_conn);
  1457. rspamd_http_connection_write_message (session->client_conn,
  1458. msg,
  1459. NULL,
  1460. ctype,
  1461. session,
  1462. 0);
  1463. }
  1464. }
  1465. static gboolean
  1466. rspamd_proxy_task_fin (void *ud)
  1467. {
  1468. struct rspamd_task *task = ud;
  1469. msg_debug_task ("finish task");
  1470. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1471. rspamd_proxy_scan_self_reply (task);
  1472. return TRUE;
  1473. }
  1474. if (!rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL)) {
  1475. rspamd_proxy_scan_self_reply (task);
  1476. return TRUE;
  1477. }
  1478. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1479. rspamd_proxy_scan_self_reply (task);
  1480. return TRUE;
  1481. }
  1482. /* One more iteration */
  1483. return FALSE;
  1484. }
  1485. static gboolean
  1486. rspamd_proxy_self_scan (struct rspamd_proxy_session *session)
  1487. {
  1488. struct rspamd_task *task;
  1489. struct rspamd_http_message *msg;
  1490. const gchar *data;
  1491. gsize len;
  1492. msg = session->client_message;
  1493. task = rspamd_task_new (session->worker, session->ctx->cfg,
  1494. session->pool, session->ctx->lang_det,
  1495. session->ctx->event_loop, FALSE);
  1496. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1497. if (session->ctx->milter) {
  1498. task->protocol_flags |= RSPAMD_TASK_PROTOCOL_FLAG_MILTER|
  1499. RSPAMD_TASK_PROTOCOL_FLAG_BODY_BLOCK;
  1500. }
  1501. task->sock = -1;
  1502. if (session->client_milter_conn) {
  1503. task->client_addr = rspamd_inet_address_copy (
  1504. session->client_milter_conn->addr);
  1505. }
  1506. else {
  1507. task->client_addr = rspamd_inet_address_copy (session->client_addr);
  1508. }
  1509. task->fin_arg = session;
  1510. task->resolver = session->ctx->resolver;
  1511. /* TODO: allow to disable autolearn in protocol */
  1512. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1513. task->s = rspamd_session_create (task->task_pool, rspamd_proxy_task_fin,
  1514. NULL, (event_finalizer_t )rspamd_task_free, task);
  1515. data = rspamd_http_message_get_body (msg, &len);
  1516. if (session->backend->settings_id) {
  1517. rspamd_http_message_remove_header (msg, "Settings-ID");
  1518. rspamd_http_message_add_header (msg, "Settings-ID",
  1519. session->backend->settings_id);
  1520. }
  1521. /* Process message */
  1522. if (!rspamd_protocol_handle_request (task, msg)) {
  1523. msg_err_task ("cannot handle request: %e", task->err);
  1524. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1525. }
  1526. else {
  1527. if (task->cmd == CMD_PING) {
  1528. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1529. }
  1530. else {
  1531. if (!rspamd_task_load_message (task, msg, data, len)) {
  1532. msg_err_task ("cannot load message: %e", task->err);
  1533. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1534. }
  1535. }
  1536. }
  1537. /* Set global timeout for the task */
  1538. if (session->ctx->default_upstream->timeout > 0.0) {
  1539. task->timeout_ev.data = task;
  1540. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1541. session->ctx->default_upstream->timeout,
  1542. session->ctx->default_upstream->timeout);
  1543. ev_timer_start (task->event_loop, &task->timeout_ev);
  1544. }
  1545. else if (session->ctx->has_self_scan) {
  1546. if (session->ctx->cfg->task_timeout > 0) {
  1547. task->timeout_ev.data = task;
  1548. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1549. session->ctx->cfg->task_timeout,
  1550. session->ctx->default_upstream->timeout);
  1551. ev_timer_start (task->event_loop, &task->timeout_ev);
  1552. }
  1553. }
  1554. session->master_conn->task = task;
  1555. rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL);
  1556. rspamd_session_pending (task->s);
  1557. return TRUE;
  1558. }
  1559. static gboolean
  1560. proxy_send_master_message (struct rspamd_proxy_session *session)
  1561. {
  1562. struct rspamd_http_message *msg;
  1563. struct rspamd_http_upstream *backend = NULL;
  1564. const rspamd_ftok_t *host;
  1565. GError *err = NULL;
  1566. gchar hostbuf[512];
  1567. host = rspamd_http_message_find_header (session->client_message, "Host");
  1568. if (host == NULL) {
  1569. backend = session->ctx->default_upstream;
  1570. }
  1571. else {
  1572. rspamd_strlcpy (hostbuf, host->begin, MIN(host->len + 1, sizeof (hostbuf)));
  1573. backend = g_hash_table_lookup (session->ctx->upstreams, hostbuf);
  1574. if (backend == NULL) {
  1575. backend = session->ctx->default_upstream;
  1576. }
  1577. }
  1578. if (backend == NULL) {
  1579. /* No backend */
  1580. msg_err_session ("cannot find upstream for %s", host ? hostbuf : "default");
  1581. goto err;
  1582. }
  1583. else {
  1584. session->backend = backend;
  1585. if (backend->self_scan) {
  1586. return rspamd_proxy_self_scan (session);
  1587. }
  1588. retry:
  1589. if (session->ctx->max_retries &&
  1590. session->retries > session->ctx->max_retries) {
  1591. msg_err_session ("cannot connect to upstream, maximum retries "
  1592. "has been reached: %d", session->retries);
  1593. goto err;
  1594. }
  1595. /* Provide hash key if hashing based on source address is desired */
  1596. guint hash_len;
  1597. gpointer hash_key = rspamd_inet_address_get_hash_key (session->client_addr,
  1598. &hash_len);
  1599. if (session->ctx->max_retries > 1 &&
  1600. session->retries == session->ctx->max_retries) {
  1601. session->master_conn->up = rspamd_upstream_get_except (backend->u,
  1602. session->master_conn->up,
  1603. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1604. hash_key, hash_len);
  1605. }
  1606. else {
  1607. session->master_conn->up = rspamd_upstream_get (backend->u,
  1608. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1609. hash_key, hash_len);
  1610. }
  1611. session->master_conn->timeout = backend->timeout;
  1612. if (session->master_conn->up == NULL) {
  1613. msg_err_session ("cannot select upstream for %s",
  1614. host ? hostbuf : "default");
  1615. goto err;
  1616. }
  1617. session->master_conn->backend_sock = rspamd_inet_address_connect (
  1618. rspamd_upstream_addr_next (session->master_conn->up),
  1619. SOCK_STREAM, TRUE);
  1620. if (session->master_conn->backend_sock == -1) {
  1621. msg_err_session ("cannot connect upstream: %s(%s)",
  1622. host ? hostbuf : "default",
  1623. rspamd_inet_address_to_string_pretty (
  1624. rspamd_upstream_addr_cur (
  1625. session->master_conn->up)));
  1626. rspamd_upstream_fail (session->master_conn->up, TRUE,
  1627. strerror (errno));
  1628. session->retries ++;
  1629. goto retry;
  1630. }
  1631. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1632. if (msg == NULL) {
  1633. msg_err_session ("cannot copy message to send it to the upstream: %e",
  1634. err);
  1635. if (err) {
  1636. g_error_free (err);
  1637. }
  1638. goto err; /* No fallback here */
  1639. }
  1640. session->master_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1641. session->ctx->http_ctx,
  1642. NULL,
  1643. proxy_backend_master_error_handler,
  1644. proxy_backend_master_finish_handler,
  1645. RSPAMD_HTTP_CLIENT_SIMPLE,
  1646. session->master_conn->backend_sock);
  1647. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1648. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1649. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1650. if (backend->key) {
  1651. msg->peer_key = rspamd_pubkey_ref (backend->key);
  1652. }
  1653. if (backend->settings_id != NULL) {
  1654. rspamd_http_message_remove_header (msg, "Settings-ID");
  1655. rspamd_http_message_add_header (msg, "Settings-ID",
  1656. backend->settings_id);
  1657. }
  1658. if (backend->local ||
  1659. rspamd_inet_address_is_local (
  1660. rspamd_upstream_addr_cur (
  1661. session->master_conn->up))) {
  1662. if (session->fname) {
  1663. rspamd_http_message_add_header (msg, "File", session->fname);
  1664. }
  1665. msg->method = HTTP_GET;
  1666. rspamd_http_connection_write_message_shared (
  1667. session->master_conn->backend_conn,
  1668. msg, NULL, NULL, session->master_conn,
  1669. session->master_conn->timeout);
  1670. }
  1671. else {
  1672. if (session->fname) {
  1673. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1674. rspamd_http_message_set_body (msg,
  1675. session->map, session->map_len);
  1676. }
  1677. msg->method = HTTP_POST;
  1678. if (backend->compress) {
  1679. proxy_request_compress (msg);
  1680. if (session->client_milter_conn) {
  1681. rspamd_http_message_add_header (msg, "Content-Type",
  1682. "application/octet-stream");
  1683. }
  1684. }
  1685. else {
  1686. if (session->client_milter_conn) {
  1687. rspamd_http_message_add_header (msg, "Content-Type",
  1688. "text/plain");
  1689. }
  1690. }
  1691. rspamd_http_connection_write_message (
  1692. session->master_conn->backend_conn,
  1693. msg, NULL, NULL, session->master_conn,
  1694. session->master_conn->timeout);
  1695. }
  1696. }
  1697. return TRUE;
  1698. err:
  1699. if (session->client_milter_conn) {
  1700. rspamd_milter_send_action (session->client_milter_conn,
  1701. RSPAMD_MILTER_TEMPFAIL);
  1702. REF_RELEASE (session);
  1703. }
  1704. else {
  1705. rspamd_http_connection_steal_msg (session->client_conn);
  1706. rspamd_http_connection_reset (session->client_conn);
  1707. proxy_client_write_error (session, 404, "Backend not found");
  1708. }
  1709. return FALSE;
  1710. }
  1711. static void
  1712. proxy_client_error_handler (struct rspamd_http_connection *conn, GError *err)
  1713. {
  1714. struct rspamd_proxy_session *session = conn->ud;
  1715. msg_info_session ("abnormally closing connection from: %s, error: %s",
  1716. rspamd_inet_address_to_string (session->client_addr), err->message);
  1717. /* Terminate session immediately */
  1718. proxy_backend_close_connection (session->master_conn);
  1719. REF_RELEASE (session);
  1720. }
  1721. static gint
  1722. proxy_client_finish_handler (struct rspamd_http_connection *conn,
  1723. struct rspamd_http_message *msg)
  1724. {
  1725. struct rspamd_proxy_session *session = conn->ud;
  1726. if (!session->master_conn) {
  1727. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1728. sizeof (*session->master_conn));
  1729. session->master_conn->s = session;
  1730. session->master_conn->name = "master";
  1731. /* Reset spamc legacy */
  1732. if (msg->method >= HTTP_SYMBOLS) {
  1733. msg->method = HTTP_POST;
  1734. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1735. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1736. msg_info_session ("enabling legacy spamc mode for session");
  1737. }
  1738. else {
  1739. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1740. msg_info_session ("enabling legacy rspamc mode for session");
  1741. }
  1742. }
  1743. if (msg->url->len == 0) {
  1744. msg->url = rspamd_fstring_append (msg->url,
  1745. "/" MSG_CMD_CHECK_V2, strlen ("/" MSG_CMD_CHECK_V2));
  1746. }
  1747. if (!proxy_check_file (msg, session)) {
  1748. goto err;
  1749. }
  1750. session->client_message = rspamd_http_connection_steal_msg (
  1751. session->client_conn);
  1752. session->shmem_ref = rspamd_http_message_shmem_ref (session->client_message);
  1753. rspamd_http_message_remove_header (msg, "Content-Length");
  1754. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1755. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1756. rspamd_http_message_remove_header (msg, "Connection");
  1757. rspamd_http_message_remove_header (msg, "Key");
  1758. proxy_open_mirror_connections (session);
  1759. rspamd_http_connection_reset (session->client_conn);
  1760. proxy_send_master_message (session);
  1761. }
  1762. else {
  1763. msg_info_session ("finished master connection");
  1764. proxy_backend_close_connection (session->master_conn);
  1765. REF_RELEASE (session);
  1766. }
  1767. return 0;
  1768. err:
  1769. rspamd_http_connection_steal_msg (session->client_conn);
  1770. rspamd_http_message_remove_header (msg, "Content-Length");
  1771. rspamd_http_message_remove_header (msg, "Key");
  1772. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1773. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1774. rspamd_http_message_remove_header (msg, "Connection");
  1775. rspamd_http_connection_reset (session->client_conn);
  1776. proxy_client_write_error (session, 404, "Backend not found");
  1777. return 0;
  1778. }
  1779. static void
  1780. proxy_milter_finish_handler (gint fd,
  1781. struct rspamd_milter_session *rms,
  1782. void *ud)
  1783. {
  1784. struct rspamd_proxy_session *session = ud;
  1785. struct rspamd_http_message *msg;
  1786. session->client_milter_conn = rms;
  1787. if (rms->message == NULL || rms->message->len == 0) {
  1788. msg_info_session ("finished milter connection");
  1789. proxy_backend_close_connection (session->master_conn);
  1790. REF_RELEASE (session);
  1791. }
  1792. else {
  1793. if (!session->master_conn) {
  1794. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1795. sizeof (*session->master_conn));
  1796. }
  1797. msg = rspamd_milter_to_http (rms);
  1798. session->master_conn->s = session;
  1799. session->master_conn->name = "master";
  1800. session->client_message = msg;
  1801. proxy_open_mirror_connections (session);
  1802. proxy_send_master_message (session);
  1803. }
  1804. }
  1805. static void
  1806. proxy_milter_error_handler (gint fd,
  1807. struct rspamd_milter_session *rms,
  1808. void *ud, GError *err)
  1809. {
  1810. struct rspamd_proxy_session *session = ud;
  1811. msg_info_session ("abnormally closing milter connection from: %s, "
  1812. "error: %e",
  1813. rspamd_inet_address_to_string_pretty (session->client_addr),
  1814. err);
  1815. /* Terminate session immediately */
  1816. proxy_backend_close_connection (session->master_conn);
  1817. REF_RELEASE (session);
  1818. }
  1819. static void
  1820. proxy_accept_socket (EV_P_ ev_io *w, int revents)
  1821. {
  1822. struct rspamd_worker *worker = (struct rspamd_worker *)w->data;
  1823. struct rspamd_proxy_ctx *ctx;
  1824. rspamd_inet_addr_t *addr;
  1825. struct rspamd_proxy_session *session;
  1826. gint nfd;
  1827. ctx = worker->ctx;
  1828. if ((nfd =
  1829. rspamd_accept_from_socket (w->fd, &addr,
  1830. rspamd_worker_throttle_accept_events, worker->accept_events)) == -1) {
  1831. msg_warn ("accept failed: %s", strerror (errno));
  1832. return;
  1833. }
  1834. /* Check for EAGAIN */
  1835. if (nfd == 0) {
  1836. return;
  1837. }
  1838. session = g_malloc0 (sizeof (*session));
  1839. REF_INIT_RETAIN (session, proxy_session_dtor);
  1840. session->client_sock = nfd;
  1841. session->client_addr = addr;
  1842. session->mirror_conns = g_ptr_array_sized_new (ctx->mirrors->len);
  1843. session->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (),
  1844. "proxy", 0);
  1845. session->ctx = ctx;
  1846. session->worker = worker;
  1847. if (ctx->sessions_cache) {
  1848. rspamd_worker_session_cache_add (ctx->sessions_cache,
  1849. session->pool->tag.uid, &session->ref.refcount, session);
  1850. }
  1851. if (!ctx->milter) {
  1852. session->client_conn = rspamd_http_connection_new_server (
  1853. ctx->http_ctx,
  1854. nfd,
  1855. NULL,
  1856. proxy_client_error_handler,
  1857. proxy_client_finish_handler,
  1858. 0);
  1859. if (ctx->key) {
  1860. rspamd_http_connection_set_key (session->client_conn, ctx->key);
  1861. }
  1862. msg_info_session ("accepted http connection from %s port %d",
  1863. rspamd_inet_address_to_string (addr),
  1864. rspamd_inet_address_get_port (addr));
  1865. rspamd_http_connection_read_message_shared (session->client_conn,
  1866. session,
  1867. session->ctx->timeout);
  1868. }
  1869. else {
  1870. msg_info_session ("accepted milter connection from %s port %d",
  1871. rspamd_inet_address_to_string (addr),
  1872. rspamd_inet_address_get_port (addr));
  1873. #ifdef TCP_NODELAY
  1874. #ifndef SOL_TCP
  1875. #define SOL_TCP IPPROTO_TCP
  1876. #endif
  1877. if (rspamd_inet_address_get_af (addr) != AF_UNIX) {
  1878. gint sopt = 1;
  1879. if (setsockopt (nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof (sopt)) ==
  1880. -1) {
  1881. msg_warn_session ("cannot set TCP_NODELAY: %s",
  1882. strerror (errno));
  1883. }
  1884. }
  1885. #endif
  1886. rspamd_milter_handle_socket (nfd, 0.0,
  1887. session->pool,
  1888. ctx->event_loop,
  1889. proxy_milter_finish_handler,
  1890. proxy_milter_error_handler,
  1891. session);
  1892. }
  1893. }
  1894. static void
  1895. adjust_upstreams_limits (struct rspamd_proxy_ctx *ctx)
  1896. {
  1897. struct rspamd_http_upstream *backend;
  1898. gpointer k, v;
  1899. GHashTableIter it;
  1900. /*
  1901. * We set error time equal to max_retries * backend_timeout and max_errors
  1902. * to max_retries - 1
  1903. *
  1904. * So if we failed to scan a message on a backend for some reasons, we
  1905. * will try to re-resolve it faster
  1906. */
  1907. g_hash_table_iter_init (&it, ctx->upstreams);
  1908. while (g_hash_table_iter_next (&it, &k, &v)) {
  1909. backend = (struct rspamd_http_upstream *)v;
  1910. if (!backend->self_scan && backend->u) {
  1911. rspamd_upstreams_set_limits (backend->u,
  1912. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1913. ctx->max_retries - 1, 0);
  1914. }
  1915. }
  1916. }
  1917. __attribute__((noreturn))
  1918. void
  1919. start_rspamd_proxy (struct rspamd_worker *worker)
  1920. {
  1921. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1922. gboolean is_controller = FALSE;
  1923. g_assert (rspamd_worker_check_context (worker->ctx, rspamd_rspamd_proxy_magic));
  1924. ctx->cfg = worker->srv->cfg;
  1925. ctx->event_loop = rspamd_prepare_worker (worker, "rspamd_proxy",
  1926. proxy_accept_socket);
  1927. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  1928. ctx->event_loop,
  1929. worker->srv->cfg);
  1930. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  1931. ctx->event_loop, ctx->resolver->r);
  1932. ctx->http_ctx = rspamd_http_context_create (ctx->cfg, ctx->event_loop,
  1933. ctx->cfg->ups_ctx);
  1934. rspamd_mempool_add_destructor (ctx->cfg->cfg_pool,
  1935. (rspamd_mempool_destruct_t)rspamd_http_context_free,
  1936. ctx->http_ctx);
  1937. if (ctx->has_self_scan) {
  1938. /* Additional initialisation needed */
  1939. rspamd_worker_init_scanner (worker, ctx->event_loop, ctx->resolver,
  1940. &ctx->lang_det);
  1941. if (worker->index == 0) {
  1942. /*
  1943. * If there are no controllers and no normal workers,
  1944. * then pretend that we are a controller
  1945. */
  1946. gboolean controller_seen = FALSE;
  1947. GList *cur;
  1948. cur = worker->srv->cfg->workers;
  1949. while (cur) {
  1950. struct rspamd_worker_conf *cf;
  1951. cf = (struct rspamd_worker_conf *)cur->data;
  1952. if ((cf->type == g_quark_from_static_string ("controller")) ||
  1953. (cf->type == g_quark_from_static_string ("normal"))) {
  1954. if (cf->enabled && cf->count >= 0) {
  1955. controller_seen = TRUE;
  1956. break;
  1957. }
  1958. }
  1959. cur = g_list_next (cur);
  1960. }
  1961. if (!controller_seen) {
  1962. msg_info ("no controller or normal workers defined, execute "
  1963. "controller periodics in this worker");
  1964. worker->flags |= RSPAMD_WORKER_CONTROLLER;
  1965. is_controller = TRUE;
  1966. }
  1967. }
  1968. }
  1969. else {
  1970. worker->flags &= ~RSPAMD_WORKER_SCANNER;
  1971. }
  1972. if (worker->srv->cfg->enable_sessions_cache) {
  1973. ctx->sessions_cache = rspamd_worker_session_cache_new (worker,
  1974. ctx->event_loop);
  1975. }
  1976. ctx->milter_ctx.spam_header = ctx->spam_header;
  1977. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  1978. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  1979. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  1980. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  1981. ctx->milter_ctx.reject_message = ctx->reject_message;
  1982. ctx->milter_ctx.cfg = ctx->cfg;
  1983. rspamd_milter_init_library (&ctx->milter_ctx);
  1984. if (is_controller) {
  1985. rspamd_worker_init_controller (worker, NULL);
  1986. }
  1987. else {
  1988. if (ctx->has_self_scan) {
  1989. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  1990. worker, RSPAMD_MAP_WATCH_SCANNER);
  1991. }
  1992. else {
  1993. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  1994. worker, RSPAMD_MAP_WATCH_WORKER);
  1995. }
  1996. }
  1997. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->event_loop,
  1998. worker);
  1999. adjust_upstreams_limits (ctx);
  2000. ev_loop (ctx->event_loop, 0);
  2001. rspamd_worker_block_signals ();
  2002. if (ctx->has_self_scan) {
  2003. rspamd_stat_close ();
  2004. }
  2005. if (is_controller) {
  2006. rspamd_controller_on_terminate (worker, NULL);
  2007. }
  2008. REF_RELEASE (ctx->cfg);
  2009. rspamd_log_close (worker->srv->logger);
  2010. exit (EXIT_SUCCESS);
  2011. }