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 60KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375
  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. {
  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", 0);
  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_pretty (
  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, err ? err->message : "unknown");
  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, strerror (errno));
  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. switch (code) {
  1234. case ETIMEDOUT:
  1235. reply->code = 504;
  1236. reply->status = RSPAMD_FSTRING_LIT ("Gateway timeout");
  1237. break;
  1238. case ECONNRESET:
  1239. case ECONNABORTED:
  1240. reply->code = 502;
  1241. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection reset");
  1242. break;
  1243. case ECONNREFUSED:
  1244. reply->code = 502;
  1245. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection refused");
  1246. break;
  1247. default:
  1248. if (code >= 300) {
  1249. /* Likely HTTP error */
  1250. reply->code = code;
  1251. reply->status = rspamd_fstring_new_init (status, strlen (status));
  1252. }
  1253. else {
  1254. reply->code = 502;
  1255. reply->status = RSPAMD_FSTRING_LIT ("Unknown gateway error: ");
  1256. reply->status = rspamd_fstring_append (reply->status,
  1257. status, strlen (status));
  1258. }
  1259. break;
  1260. }
  1261. rspamd_http_connection_write_message (session->client_conn,
  1262. reply, NULL, NULL, session,
  1263. session->ctx->timeout);
  1264. }
  1265. }
  1266. static void
  1267. proxy_backend_master_error_handler (struct rspamd_http_connection *conn, GError *err)
  1268. {
  1269. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1270. struct rspamd_proxy_session *session;
  1271. session = bk_conn->s;
  1272. session->retries ++;
  1273. msg_info_session ("abnormally closing connection from backend: %s, error: %e,"
  1274. " retries left: %d",
  1275. rspamd_inet_address_to_string_pretty (
  1276. rspamd_upstream_addr_cur (session->master_conn->up)),
  1277. err,
  1278. session->ctx->max_retries - session->retries);
  1279. rspamd_upstream_fail (bk_conn->up, FALSE, err ? err->message : "unknown");
  1280. proxy_backend_close_connection (session->master_conn);
  1281. if (session->ctx->max_retries > 0 &&
  1282. session->retries >= session->ctx->max_retries) {
  1283. msg_err_session ("cannot connect to upstream, maximum retries "
  1284. "has been reached: %d", session->retries);
  1285. /* Terminate session immediately */
  1286. proxy_client_write_error (session, err->code, err->message);
  1287. }
  1288. else {
  1289. if (!proxy_send_master_message (session)) {
  1290. proxy_client_write_error (session, err->code, err->message);
  1291. }
  1292. else {
  1293. msg_info_session ("retry connection to: %s"
  1294. " retries left: %d",
  1295. rspamd_inet_address_to_string (
  1296. rspamd_upstream_addr_cur (session->master_conn->up)),
  1297. session->ctx->max_retries - session->retries);
  1298. }
  1299. }
  1300. }
  1301. static gint
  1302. proxy_backend_master_finish_handler (struct rspamd_http_connection *conn,
  1303. struct rspamd_http_message *msg)
  1304. {
  1305. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1306. struct rspamd_proxy_session *session, *nsession;
  1307. rspamd_fstring_t *reply;
  1308. goffset body_offset = -1;
  1309. session = bk_conn->s;
  1310. rspamd_http_connection_steal_msg (session->master_conn->backend_conn);
  1311. proxy_request_decompress (msg);
  1312. rspamd_http_message_remove_header (msg, "Content-Length");
  1313. rspamd_http_message_remove_header (msg, "Key");
  1314. rspamd_http_connection_reset (session->master_conn->backend_conn);
  1315. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1316. bk_conn->parser_from_ref, msg, &body_offset)) {
  1317. msg_warn_session ("cannot parse results from the master backend");
  1318. }
  1319. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1320. /* We need to reformat ucl to fit with legacy spamc protocol */
  1321. if (bk_conn->results) {
  1322. reply = rspamd_fstring_new ();
  1323. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1324. rspamd_ucl_tospamc_output (bk_conn->results, &reply);
  1325. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1326. }
  1327. else {
  1328. rspamd_ucl_torspamc_output (bk_conn->results, &reply);
  1329. }
  1330. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1331. msg->method = HTTP_SYMBOLS;
  1332. }
  1333. else {
  1334. msg_warn_session ("cannot parse results from the master backend, "
  1335. "return them as is");
  1336. }
  1337. }
  1338. rspamd_upstream_ok (bk_conn->up);
  1339. if (session->client_milter_conn) {
  1340. nsession = proxy_session_refresh (session);
  1341. if (body_offset > 0) {
  1342. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1343. session->master_conn->results,
  1344. msg->body_buf.begin + body_offset,
  1345. msg->body_buf.len - body_offset);
  1346. }
  1347. else {
  1348. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1349. session->master_conn->results, NULL, 0);
  1350. }
  1351. REF_RELEASE (session);
  1352. rspamd_http_message_free (msg);
  1353. }
  1354. else {
  1355. rspamd_http_connection_write_message (session->client_conn,
  1356. msg, NULL, NULL, session,
  1357. bk_conn->timeout);
  1358. }
  1359. return 0;
  1360. }
  1361. static void
  1362. rspamd_proxy_scan_self_reply (struct rspamd_task *task)
  1363. {
  1364. struct rspamd_http_message *msg;
  1365. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1366. ucl_object_t *rep = NULL;
  1367. const char *ctype = "application/json";
  1368. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1369. msg->date = time (NULL);
  1370. msg->code = 200;
  1371. switch (task->cmd) {
  1372. case CMD_CHECK:
  1373. case CMD_SKIP:
  1374. case CMD_CHECK_RSPAMC:
  1375. case CMD_CHECK_SPAMC:
  1376. case CMD_CHECK_V2:
  1377. rspamd_task_set_finish_time (task);
  1378. rspamd_protocol_http_reply (msg, task, &rep);
  1379. rspamd_protocol_write_log_pipe (task);
  1380. break;
  1381. case CMD_PING:
  1382. rspamd_http_message_set_body (msg, "pong" CRLF, 6);
  1383. ctype = "text/plain";
  1384. break;
  1385. default:
  1386. msg_err_task ("BROKEN");
  1387. break;
  1388. }
  1389. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1390. if (rep) {
  1391. session->master_conn->results = ucl_object_ref (rep);
  1392. }
  1393. if (session->client_milter_conn) {
  1394. nsession = proxy_session_refresh (session);
  1395. if (task->flags & RSPAMD_TASK_FLAG_MESSAGE_REWRITE) {
  1396. const gchar *start;
  1397. goffset len, hdr_off;
  1398. start = task->msg.begin;
  1399. len = task->msg.len;
  1400. hdr_off = MESSAGE_FIELD (task, raw_headers_content).len;
  1401. if (hdr_off < len) {
  1402. start += hdr_off;
  1403. len -= hdr_off;
  1404. /* The problem here is that we need not end of headers, we need
  1405. * start of body.
  1406. *
  1407. * Hence, we need to skip one \r\n till there is anything else in
  1408. * a line.
  1409. */
  1410. if (*start == '\r' && len > 0) {
  1411. start++;
  1412. len--;
  1413. }
  1414. if (*start == '\n' && len > 0) {
  1415. start++;
  1416. len--;
  1417. }
  1418. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1419. session->master_conn->results, start, len);
  1420. }
  1421. else {
  1422. /* XXX: should never happen! */
  1423. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1424. session->master_conn->results, NULL, 0);
  1425. }
  1426. }
  1427. else {
  1428. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1429. session->master_conn->results, NULL, 0);
  1430. }
  1431. rspamd_http_message_free (msg);
  1432. REF_RELEASE (session);
  1433. }
  1434. else {
  1435. rspamd_http_connection_reset (session->client_conn);
  1436. rspamd_http_connection_write_message (session->client_conn,
  1437. msg,
  1438. NULL,
  1439. ctype,
  1440. session,
  1441. 0);
  1442. }
  1443. }
  1444. static gboolean
  1445. rspamd_proxy_task_fin (void *ud)
  1446. {
  1447. struct rspamd_task *task = ud;
  1448. msg_debug_task ("finish task");
  1449. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1450. rspamd_proxy_scan_self_reply (task);
  1451. return TRUE;
  1452. }
  1453. if (!rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL)) {
  1454. rspamd_proxy_scan_self_reply (task);
  1455. return TRUE;
  1456. }
  1457. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1458. rspamd_proxy_scan_self_reply (task);
  1459. return TRUE;
  1460. }
  1461. /* One more iteration */
  1462. return FALSE;
  1463. }
  1464. static gboolean
  1465. rspamd_proxy_self_scan (struct rspamd_proxy_session *session)
  1466. {
  1467. struct rspamd_task *task;
  1468. struct rspamd_http_message *msg;
  1469. const gchar *data;
  1470. gsize len;
  1471. msg = session->client_message;
  1472. task = rspamd_task_new (session->worker, session->ctx->cfg,
  1473. session->pool, session->ctx->lang_det,
  1474. session->ctx->event_loop, FALSE);
  1475. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1476. if (session->ctx->milter) {
  1477. task->protocol_flags |= RSPAMD_TASK_PROTOCOL_FLAG_MILTER|
  1478. RSPAMD_TASK_PROTOCOL_FLAG_BODY_BLOCK;
  1479. }
  1480. task->sock = -1;
  1481. if (session->client_milter_conn) {
  1482. task->client_addr = rspamd_inet_address_copy (
  1483. session->client_milter_conn->addr);
  1484. }
  1485. else {
  1486. task->client_addr = rspamd_inet_address_copy (session->client_addr);
  1487. }
  1488. task->fin_arg = session;
  1489. task->resolver = session->ctx->resolver;
  1490. /* TODO: allow to disable autolearn in protocol */
  1491. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1492. task->s = rspamd_session_create (task->task_pool, rspamd_proxy_task_fin,
  1493. NULL, (event_finalizer_t )rspamd_task_free, task);
  1494. data = rspamd_http_message_get_body (msg, &len);
  1495. if (session->backend->settings_id) {
  1496. rspamd_http_message_remove_header (msg, "Settings-ID");
  1497. rspamd_http_message_add_header (msg, "Settings-ID",
  1498. session->backend->settings_id);
  1499. }
  1500. /* Process message */
  1501. if (!rspamd_protocol_handle_request (task, msg)) {
  1502. msg_err_task ("cannot handle request: %e", task->err);
  1503. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1504. }
  1505. else {
  1506. if (task->cmd == CMD_PING) {
  1507. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1508. }
  1509. else {
  1510. if (!rspamd_task_load_message (task, msg, data, len)) {
  1511. msg_err_task ("cannot load message: %e", task->err);
  1512. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1513. }
  1514. }
  1515. }
  1516. /* Set global timeout for the task */
  1517. if (session->ctx->default_upstream->timeout > 0.0) {
  1518. task->timeout_ev.data = task;
  1519. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1520. session->ctx->default_upstream->timeout,
  1521. session->ctx->default_upstream->timeout);
  1522. ev_timer_start (task->event_loop, &task->timeout_ev);
  1523. }
  1524. else if (session->ctx->has_self_scan) {
  1525. if (session->ctx->cfg->task_timeout > 0) {
  1526. task->timeout_ev.data = task;
  1527. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1528. session->ctx->cfg->task_timeout,
  1529. session->ctx->default_upstream->timeout);
  1530. ev_timer_start (task->event_loop, &task->timeout_ev);
  1531. }
  1532. }
  1533. session->master_conn->task = task;
  1534. rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL);
  1535. rspamd_session_pending (task->s);
  1536. return TRUE;
  1537. }
  1538. static gboolean
  1539. proxy_send_master_message (struct rspamd_proxy_session *session)
  1540. {
  1541. struct rspamd_http_message *msg;
  1542. struct rspamd_http_upstream *backend = NULL;
  1543. const rspamd_ftok_t *host;
  1544. GError *err = NULL;
  1545. gchar hostbuf[512];
  1546. host = rspamd_http_message_find_header (session->client_message, "Host");
  1547. if (host == NULL) {
  1548. backend = session->ctx->default_upstream;
  1549. }
  1550. else {
  1551. rspamd_strlcpy (hostbuf, host->begin, MIN(host->len + 1, sizeof (hostbuf)));
  1552. backend = g_hash_table_lookup (session->ctx->upstreams, hostbuf);
  1553. if (backend == NULL) {
  1554. backend = session->ctx->default_upstream;
  1555. }
  1556. }
  1557. if (backend == NULL) {
  1558. /* No backend */
  1559. msg_err_session ("cannot find upstream for %s", host ? hostbuf : "default");
  1560. goto err;
  1561. }
  1562. else {
  1563. session->backend = backend;
  1564. if (backend->self_scan) {
  1565. return rspamd_proxy_self_scan (session);
  1566. }
  1567. retry:
  1568. if (session->ctx->max_retries &&
  1569. session->retries > session->ctx->max_retries) {
  1570. msg_err_session ("cannot connect to upstream, maximum retries "
  1571. "has been reached: %d", session->retries);
  1572. goto err;
  1573. }
  1574. /* Provide hash key if hashing based on source address is desired */
  1575. guint hash_len;
  1576. gpointer hash_key = rspamd_inet_address_get_hash_key (session->client_addr,
  1577. &hash_len);
  1578. if (session->ctx->max_retries > 1 &&
  1579. session->retries == session->ctx->max_retries) {
  1580. session->master_conn->up = rspamd_upstream_get_except (backend->u,
  1581. session->master_conn->up,
  1582. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1583. hash_key, hash_len);
  1584. }
  1585. else {
  1586. session->master_conn->up = rspamd_upstream_get (backend->u,
  1587. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1588. hash_key, hash_len);
  1589. }
  1590. session->master_conn->timeout = backend->timeout;
  1591. if (session->master_conn->up == NULL) {
  1592. msg_err_session ("cannot select upstream for %s",
  1593. host ? hostbuf : "default");
  1594. goto err;
  1595. }
  1596. session->master_conn->backend_sock = rspamd_inet_address_connect (
  1597. rspamd_upstream_addr_next (session->master_conn->up),
  1598. SOCK_STREAM, TRUE);
  1599. if (session->master_conn->backend_sock == -1) {
  1600. msg_err_session ("cannot connect upstream: %s(%s)",
  1601. host ? hostbuf : "default",
  1602. rspamd_inet_address_to_string_pretty (
  1603. rspamd_upstream_addr_cur (
  1604. session->master_conn->up)));
  1605. rspamd_upstream_fail (session->master_conn->up, TRUE,
  1606. strerror (errno));
  1607. session->retries ++;
  1608. goto retry;
  1609. }
  1610. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1611. if (msg == NULL) {
  1612. msg_err_session ("cannot copy message to send it to the upstream: %e",
  1613. err);
  1614. if (err) {
  1615. g_error_free (err);
  1616. }
  1617. goto err; /* No fallback here */
  1618. }
  1619. session->master_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1620. session->ctx->http_ctx,
  1621. NULL,
  1622. proxy_backend_master_error_handler,
  1623. proxy_backend_master_finish_handler,
  1624. RSPAMD_HTTP_CLIENT_SIMPLE,
  1625. session->master_conn->backend_sock);
  1626. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1627. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1628. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1629. if (backend->key) {
  1630. msg->peer_key = rspamd_pubkey_ref (backend->key);
  1631. }
  1632. if (backend->settings_id != NULL) {
  1633. rspamd_http_message_remove_header (msg, "Settings-ID");
  1634. rspamd_http_message_add_header (msg, "Settings-ID",
  1635. backend->settings_id);
  1636. }
  1637. if (backend->local ||
  1638. rspamd_inet_address_is_local (
  1639. rspamd_upstream_addr_cur (
  1640. session->master_conn->up), FALSE)) {
  1641. if (session->fname) {
  1642. rspamd_http_message_add_header (msg, "File", session->fname);
  1643. }
  1644. msg->method = HTTP_GET;
  1645. rspamd_http_connection_write_message_shared (
  1646. session->master_conn->backend_conn,
  1647. msg, NULL, NULL, session->master_conn,
  1648. session->master_conn->timeout);
  1649. }
  1650. else {
  1651. if (session->fname) {
  1652. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1653. rspamd_http_message_set_body (msg,
  1654. session->map, session->map_len);
  1655. }
  1656. msg->method = HTTP_POST;
  1657. if (backend->compress) {
  1658. proxy_request_compress (msg);
  1659. if (session->client_milter_conn) {
  1660. rspamd_http_message_add_header (msg, "Content-Type",
  1661. "application/octet-stream");
  1662. }
  1663. }
  1664. else {
  1665. if (session->client_milter_conn) {
  1666. rspamd_http_message_add_header (msg, "Content-Type",
  1667. "text/plain");
  1668. }
  1669. }
  1670. rspamd_http_connection_write_message (
  1671. session->master_conn->backend_conn,
  1672. msg, NULL, NULL, session->master_conn,
  1673. session->master_conn->timeout);
  1674. }
  1675. }
  1676. return TRUE;
  1677. err:
  1678. if (session->client_milter_conn) {
  1679. rspamd_milter_send_action (session->client_milter_conn,
  1680. RSPAMD_MILTER_TEMPFAIL);
  1681. REF_RELEASE (session);
  1682. }
  1683. else {
  1684. rspamd_http_connection_steal_msg (session->client_conn);
  1685. rspamd_http_connection_reset (session->client_conn);
  1686. proxy_client_write_error (session, 404, "Backend not found");
  1687. }
  1688. return FALSE;
  1689. }
  1690. static void
  1691. proxy_client_error_handler (struct rspamd_http_connection *conn, GError *err)
  1692. {
  1693. struct rspamd_proxy_session *session = conn->ud;
  1694. msg_info_session ("abnormally closing connection from: %s, error: %s",
  1695. rspamd_inet_address_to_string (session->client_addr), err->message);
  1696. /* Terminate session immediately */
  1697. proxy_backend_close_connection (session->master_conn);
  1698. REF_RELEASE (session);
  1699. }
  1700. static gint
  1701. proxy_client_finish_handler (struct rspamd_http_connection *conn,
  1702. struct rspamd_http_message *msg)
  1703. {
  1704. struct rspamd_proxy_session *session = conn->ud;
  1705. if (!session->master_conn) {
  1706. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1707. sizeof (*session->master_conn));
  1708. session->master_conn->s = session;
  1709. session->master_conn->name = "master";
  1710. /* Reset spamc legacy */
  1711. if (msg->method >= HTTP_SYMBOLS) {
  1712. msg->method = HTTP_POST;
  1713. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1714. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1715. msg_info_session ("enabling legacy spamc mode for session");
  1716. }
  1717. else {
  1718. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1719. msg_info_session ("enabling legacy rspamc mode for session");
  1720. }
  1721. }
  1722. if (msg->url->len == 0) {
  1723. msg->url = rspamd_fstring_append (msg->url,
  1724. "/" MSG_CMD_CHECK_V2, strlen ("/" MSG_CMD_CHECK_V2));
  1725. }
  1726. if (!proxy_check_file (msg, session)) {
  1727. goto err;
  1728. }
  1729. session->client_message = rspamd_http_connection_steal_msg (
  1730. session->client_conn);
  1731. session->shmem_ref = rspamd_http_message_shmem_ref (session->client_message);
  1732. rspamd_http_message_remove_header (msg, "Content-Length");
  1733. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1734. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1735. rspamd_http_message_remove_header (msg, "Connection");
  1736. rspamd_http_message_remove_header (msg, "Key");
  1737. proxy_open_mirror_connections (session);
  1738. rspamd_http_connection_reset (session->client_conn);
  1739. proxy_send_master_message (session);
  1740. }
  1741. else {
  1742. msg_info_session ("finished master connection");
  1743. proxy_backend_close_connection (session->master_conn);
  1744. REF_RELEASE (session);
  1745. }
  1746. return 0;
  1747. err:
  1748. rspamd_http_connection_steal_msg (session->client_conn);
  1749. rspamd_http_message_remove_header (msg, "Content-Length");
  1750. rspamd_http_message_remove_header (msg, "Key");
  1751. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1752. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1753. rspamd_http_message_remove_header (msg, "Connection");
  1754. rspamd_http_connection_reset (session->client_conn);
  1755. proxy_client_write_error (session, 404, "Backend not found");
  1756. return 0;
  1757. }
  1758. static void
  1759. proxy_milter_finish_handler (gint fd,
  1760. struct rspamd_milter_session *rms,
  1761. void *ud)
  1762. {
  1763. struct rspamd_proxy_session *session = ud;
  1764. struct rspamd_http_message *msg;
  1765. session->client_milter_conn = rms;
  1766. if (rms->message == NULL || rms->message->len == 0) {
  1767. msg_info_session ("finished milter connection");
  1768. proxy_backend_close_connection (session->master_conn);
  1769. REF_RELEASE (session);
  1770. }
  1771. else {
  1772. if (!session->master_conn) {
  1773. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1774. sizeof (*session->master_conn));
  1775. }
  1776. msg = rspamd_milter_to_http (rms);
  1777. session->master_conn->s = session;
  1778. session->master_conn->name = "master";
  1779. session->client_message = msg;
  1780. proxy_open_mirror_connections (session);
  1781. proxy_send_master_message (session);
  1782. }
  1783. }
  1784. static void
  1785. proxy_milter_error_handler (gint fd,
  1786. struct rspamd_milter_session *rms,
  1787. void *ud, GError *err)
  1788. {
  1789. struct rspamd_proxy_session *session = ud;
  1790. msg_info_session ("abnormally closing milter connection from: %s, "
  1791. "error: %e",
  1792. rspamd_inet_address_to_string_pretty (session->client_addr),
  1793. err);
  1794. /* Terminate session immediately */
  1795. proxy_backend_close_connection (session->master_conn);
  1796. REF_RELEASE (session);
  1797. }
  1798. static void
  1799. proxy_accept_socket (EV_P_ ev_io *w, int revents)
  1800. {
  1801. struct rspamd_worker *worker = (struct rspamd_worker *)w->data;
  1802. struct rspamd_proxy_ctx *ctx;
  1803. rspamd_inet_addr_t *addr;
  1804. struct rspamd_proxy_session *session;
  1805. gint nfd;
  1806. ctx = worker->ctx;
  1807. if ((nfd =
  1808. rspamd_accept_from_socket (w->fd, &addr,
  1809. rspamd_worker_throttle_accept_events, worker->accept_events)) == -1) {
  1810. msg_warn ("accept failed: %s", strerror (errno));
  1811. return;
  1812. }
  1813. /* Check for EAGAIN */
  1814. if (nfd == 0) {
  1815. return;
  1816. }
  1817. session = g_malloc0 (sizeof (*session));
  1818. REF_INIT_RETAIN (session, proxy_session_dtor);
  1819. session->client_sock = nfd;
  1820. session->client_addr = addr;
  1821. session->mirror_conns = g_ptr_array_sized_new (ctx->mirrors->len);
  1822. session->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (),
  1823. "proxy", 0);
  1824. session->ctx = ctx;
  1825. session->worker = worker;
  1826. if (ctx->sessions_cache) {
  1827. rspamd_worker_session_cache_add (ctx->sessions_cache,
  1828. session->pool->tag.uid, &session->ref.refcount, session);
  1829. }
  1830. if (!ctx->milter) {
  1831. session->client_conn = rspamd_http_connection_new_server (
  1832. ctx->http_ctx,
  1833. nfd,
  1834. NULL,
  1835. proxy_client_error_handler,
  1836. proxy_client_finish_handler,
  1837. 0);
  1838. if (ctx->key) {
  1839. rspamd_http_connection_set_key (session->client_conn, ctx->key);
  1840. }
  1841. msg_info_session ("accepted http connection from %s port %d",
  1842. rspamd_inet_address_to_string (addr),
  1843. rspamd_inet_address_get_port (addr));
  1844. rspamd_http_connection_read_message_shared (session->client_conn,
  1845. session,
  1846. session->ctx->timeout);
  1847. }
  1848. else {
  1849. msg_info_session ("accepted milter connection from %s port %d",
  1850. rspamd_inet_address_to_string (addr),
  1851. rspamd_inet_address_get_port (addr));
  1852. #ifdef TCP_NODELAY
  1853. #ifndef SOL_TCP
  1854. #define SOL_TCP IPPROTO_TCP
  1855. #endif
  1856. if (rspamd_inet_address_get_af (addr) != AF_UNIX) {
  1857. gint sopt = 1;
  1858. if (setsockopt (nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof (sopt)) ==
  1859. -1) {
  1860. msg_warn_session ("cannot set TCP_NODELAY: %s",
  1861. strerror (errno));
  1862. }
  1863. }
  1864. #endif
  1865. rspamd_milter_handle_socket (nfd, 0.0,
  1866. session->pool,
  1867. ctx->event_loop,
  1868. proxy_milter_finish_handler,
  1869. proxy_milter_error_handler,
  1870. session);
  1871. }
  1872. }
  1873. static void
  1874. adjust_upstreams_limits (struct rspamd_proxy_ctx *ctx)
  1875. {
  1876. struct rspamd_http_upstream *backend;
  1877. gpointer k, v;
  1878. GHashTableIter it;
  1879. /*
  1880. * We set error time equal to max_retries * backend_timeout and max_errors
  1881. * to max_retries - 1
  1882. *
  1883. * So if we failed to scan a message on a backend for some reasons, we
  1884. * will try to re-resolve it faster
  1885. */
  1886. g_hash_table_iter_init (&it, ctx->upstreams);
  1887. while (g_hash_table_iter_next (&it, &k, &v)) {
  1888. backend = (struct rspamd_http_upstream *)v;
  1889. if (!backend->self_scan && backend->u) {
  1890. rspamd_upstreams_set_limits (backend->u,
  1891. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1892. ctx->max_retries - 1, 0);
  1893. }
  1894. }
  1895. }
  1896. void
  1897. start_rspamd_proxy (struct rspamd_worker *worker)
  1898. {
  1899. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1900. gboolean is_controller = FALSE;
  1901. g_assert (rspamd_worker_check_context (worker->ctx, rspamd_rspamd_proxy_magic));
  1902. ctx->cfg = worker->srv->cfg;
  1903. ctx->event_loop = rspamd_prepare_worker (worker, "rspamd_proxy",
  1904. proxy_accept_socket);
  1905. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  1906. ctx->event_loop,
  1907. worker->srv->cfg);
  1908. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  1909. ctx->event_loop, ctx->resolver->r);
  1910. ctx->http_ctx = rspamd_http_context_create (ctx->cfg, ctx->event_loop,
  1911. ctx->cfg->ups_ctx);
  1912. rspamd_mempool_add_destructor (ctx->cfg->cfg_pool,
  1913. (rspamd_mempool_destruct_t)rspamd_http_context_free,
  1914. ctx->http_ctx);
  1915. if (ctx->has_self_scan) {
  1916. /* Additional initialisation needed */
  1917. rspamd_worker_init_scanner (worker, ctx->event_loop, ctx->resolver,
  1918. &ctx->lang_det);
  1919. if (worker->index == 0) {
  1920. /*
  1921. * If there are no controllers and no normal workers,
  1922. * then pretend that we are a controller
  1923. */
  1924. gboolean controller_seen = FALSE;
  1925. GList *cur;
  1926. cur = worker->srv->cfg->workers;
  1927. while (cur) {
  1928. struct rspamd_worker_conf *cf;
  1929. cf = (struct rspamd_worker_conf *)cur->data;
  1930. if ((cf->type == g_quark_from_static_string ("controller")) ||
  1931. (cf->type == g_quark_from_static_string ("normal"))) {
  1932. if (cf->enabled && cf->count >= 0) {
  1933. controller_seen = TRUE;
  1934. break;
  1935. }
  1936. }
  1937. cur = g_list_next (cur);
  1938. }
  1939. if (!controller_seen) {
  1940. msg_info ("no controller or normal workers defined, execute "
  1941. "controller periodics in this worker");
  1942. worker->flags |= RSPAMD_WORKER_CONTROLLER;
  1943. is_controller = TRUE;
  1944. }
  1945. }
  1946. }
  1947. else {
  1948. worker->flags &= ~RSPAMD_WORKER_SCANNER;
  1949. }
  1950. if (worker->srv->cfg->enable_sessions_cache) {
  1951. ctx->sessions_cache = rspamd_worker_session_cache_new (worker,
  1952. ctx->event_loop);
  1953. }
  1954. ctx->milter_ctx.spam_header = ctx->spam_header;
  1955. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  1956. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  1957. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  1958. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  1959. ctx->milter_ctx.reject_message = ctx->reject_message;
  1960. ctx->milter_ctx.cfg = ctx->cfg;
  1961. rspamd_milter_init_library (&ctx->milter_ctx);
  1962. if (is_controller) {
  1963. rspamd_worker_init_controller (worker, NULL);
  1964. }
  1965. else {
  1966. if (ctx->has_self_scan) {
  1967. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  1968. worker, RSPAMD_MAP_WATCH_SCANNER);
  1969. }
  1970. else {
  1971. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  1972. worker, RSPAMD_MAP_WATCH_WORKER);
  1973. }
  1974. }
  1975. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->event_loop,
  1976. worker);
  1977. adjust_upstreams_limits (ctx);
  1978. ev_loop (ctx->event_loop, 0);
  1979. rspamd_worker_block_signals ();
  1980. if (ctx->has_self_scan) {
  1981. rspamd_stat_close ();
  1982. }
  1983. if (is_controller) {
  1984. rspamd_controller_on_terminate (worker, NULL);
  1985. }
  1986. REF_RELEASE (ctx->cfg);
  1987. rspamd_log_close (worker->srv->logger);
  1988. exit (EXIT_SUCCESS);
  1989. }