Você não pode selecionar mais de 25 tópicos Os tópicos devem começar com uma letra ou um número, podem incluir traços ('-') e podem ter até 35 caracteres.

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427
  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 <math.h>
  40. #ifdef HAVE_NETINET_TCP_H
  41. #include <netinet/tcp.h> /* for TCP_NODELAY */
  42. #endif
  43. #ifdef SYS_ZSTD
  44. #include "zstd.h"
  45. #else
  46. #include "contrib/zstd/zstd.h"
  47. #endif
  48. /* Rotate keys each minute by default */
  49. #define DEFAULT_ROTATION_TIME 60.0
  50. #define DEFAULT_RETRIES 5
  51. #define msg_err_session(...) rspamd_default_log_function(G_LOG_LEVEL_CRITICAL, \
  52. session->pool->tag.tagname, session->pool->tag.uid, \
  53. RSPAMD_LOG_FUNC, \
  54. __VA_ARGS__)
  55. #define msg_warn_session(...) rspamd_default_log_function(G_LOG_LEVEL_WARNING, \
  56. session->pool->tag.tagname, session->pool->tag.uid, \
  57. RSPAMD_LOG_FUNC, \
  58. __VA_ARGS__)
  59. #define msg_info_session(...) rspamd_default_log_function(G_LOG_LEVEL_INFO, \
  60. session->pool->tag.tagname, session->pool->tag.uid, \
  61. RSPAMD_LOG_FUNC, \
  62. __VA_ARGS__)
  63. #define msg_debug_session(...) rspamd_conditional_debug_fast(NULL, session->client_addr, \
  64. rspamd_proxy_log_id, "proxy", session->pool->tag.uid, \
  65. RSPAMD_LOG_FUNC, \
  66. __VA_ARGS__)
  67. INIT_LOG_MODULE(proxy)
  68. gpointer init_rspamd_proxy(struct rspamd_config *cfg);
  69. void start_rspamd_proxy(struct rspamd_worker *worker);
  70. worker_t rspamd_proxy_worker = {
  71. "rspamd_proxy", /* Name */
  72. init_rspamd_proxy, /* Init function */
  73. start_rspamd_proxy, /* Start function */
  74. RSPAMD_WORKER_HAS_SOCKET | RSPAMD_WORKER_KILLABLE | RSPAMD_WORKER_SCANNER,
  75. RSPAMD_WORKER_SOCKET_TCP, /* TCP socket */
  76. RSPAMD_WORKER_VER};
  77. struct rspamd_http_upstream {
  78. gchar *name;
  79. gchar *settings_id;
  80. struct upstream_list *u;
  81. struct rspamd_cryptobox_pubkey *key;
  82. gdouble timeout;
  83. gint parser_from_ref;
  84. gint parser_to_ref;
  85. gboolean local;
  86. gboolean self_scan;
  87. gboolean compress;
  88. };
  89. struct rspamd_http_mirror {
  90. gchar *name;
  91. gchar *settings_id;
  92. struct upstream_list *u;
  93. struct rspamd_cryptobox_pubkey *key;
  94. gdouble prob;
  95. gdouble timeout;
  96. gint parser_from_ref;
  97. gint parser_to_ref;
  98. gboolean local;
  99. gboolean compress;
  100. };
  101. static const guint64 rspamd_rspamd_proxy_magic = 0xcdeb4fd1fc351980ULL;
  102. struct rspamd_proxy_ctx {
  103. guint64 magic;
  104. /* Events base */
  105. struct ev_loop *event_loop;
  106. /* DNS resolver */
  107. struct rspamd_dns_resolver *resolver;
  108. /* Config */
  109. struct rspamd_config *cfg;
  110. /* END OF COMMON PART */
  111. gdouble timeout;
  112. /* Encryption key for clients */
  113. struct rspamd_cryptobox_keypair *key;
  114. /* HTTP context */
  115. struct rspamd_http_context *http_ctx;
  116. /* Upstreams to use */
  117. GHashTable *upstreams;
  118. /* Mirrors to send traffic to */
  119. GPtrArray *mirrors;
  120. /* Default upstream */
  121. struct rspamd_http_upstream *default_upstream;
  122. lua_State *lua_state;
  123. /* Array of callback functions called on end of scan to compare results */
  124. GArray *cmp_refs;
  125. /* Maximum count for retries */
  126. guint max_retries;
  127. /* If we have self_scanning backends, we need to work as a normal worker */
  128. gboolean has_self_scan;
  129. /* It is not HTTP but milter proxy */
  130. gboolean milter;
  131. /* Discard messages instead of rejecting them */
  132. gboolean discard_on_reject;
  133. /* Quarantine messages instead of rejecting them */
  134. gboolean quarantine_on_reject;
  135. /* Milter spam header */
  136. gchar *spam_header;
  137. /* CA name that can be used for client certificates */
  138. gchar *client_ca_name;
  139. /* Milter rejection message */
  140. gchar *reject_message;
  141. /* Sessions cache */
  142. void *sessions_cache;
  143. struct rspamd_milter_context milter_ctx;
  144. /* Language detector */
  145. struct rspamd_lang_detector *lang_det;
  146. gdouble task_timeout;
  147. };
  148. enum rspamd_backend_flags {
  149. RSPAMD_BACKEND_REPLIED = 1 << 0,
  150. RSPAMD_BACKEND_CLOSED = 1 << 1,
  151. RSPAMD_BACKEND_PARSED = 1 << 2,
  152. };
  153. struct rspamd_proxy_session;
  154. struct rspamd_proxy_backend_connection {
  155. const gchar *name;
  156. struct rspamd_cryptobox_keypair *local_key;
  157. struct rspamd_cryptobox_pubkey *remote_key;
  158. struct upstream *up;
  159. struct rspamd_http_connection *backend_conn;
  160. ucl_object_t *results;
  161. const gchar *err;
  162. struct rspamd_proxy_session *s;
  163. gint backend_sock;
  164. ev_tstamp timeout;
  165. enum rspamd_backend_flags flags;
  166. gint parser_from_ref;
  167. gint parser_to_ref;
  168. struct rspamd_task *task;
  169. };
  170. enum rspamd_proxy_legacy_support {
  171. LEGACY_SUPPORT_NO = 0,
  172. LEGACY_SUPPORT_RSPAMC,
  173. LEGACY_SUPPORT_SPAMC
  174. };
  175. struct rspamd_proxy_session {
  176. struct rspamd_worker *worker;
  177. rspamd_mempool_t *pool;
  178. struct rspamd_proxy_ctx *ctx;
  179. rspamd_inet_addr_t *client_addr;
  180. struct rspamd_http_connection *client_conn;
  181. struct rspamd_milter_session *client_milter_conn;
  182. struct rspamd_http_upstream *backend;
  183. gpointer map;
  184. gchar *fname;
  185. gpointer shmem_ref;
  186. struct rspamd_proxy_backend_connection *master_conn;
  187. struct rspamd_http_message *client_message;
  188. GPtrArray *mirror_conns;
  189. gsize map_len;
  190. gint client_sock;
  191. enum rspamd_proxy_legacy_support legacy_support;
  192. gint retries;
  193. ref_entry_t ref;
  194. };
  195. static gboolean proxy_send_master_message(struct rspamd_proxy_session *session);
  196. static GQuark
  197. rspamd_proxy_quark(void)
  198. {
  199. return g_quark_from_static_string("rspamd-proxy");
  200. }
  201. static gboolean
  202. rspamd_proxy_parse_lua_parser(lua_State *L, const ucl_object_t *obj,
  203. gint *ref_from, gint *ref_to, GError **err)
  204. {
  205. const gchar *lua_script;
  206. gsize slen;
  207. gint err_idx, ref_idx;
  208. gboolean has_ref = FALSE;
  209. g_assert(obj != NULL);
  210. g_assert(ref_from != NULL);
  211. g_assert(ref_to != NULL);
  212. *ref_from = -1;
  213. *ref_to = -1;
  214. lua_script = ucl_object_tolstring(obj, &slen);
  215. lua_pushcfunction(L, &rspamd_lua_traceback);
  216. err_idx = lua_gettop(L);
  217. /* Load data */
  218. if (luaL_loadbuffer(L, lua_script, slen, "proxy parser") != 0) {
  219. g_set_error(err,
  220. rspamd_proxy_quark(),
  221. EINVAL,
  222. "cannot load lua parser script: %s",
  223. lua_tostring(L, -1));
  224. lua_settop(L, 0); /* Error function */
  225. return FALSE;
  226. }
  227. /* Now do it */
  228. if (lua_pcall(L, 0, 1, err_idx) != 0) {
  229. g_set_error(err,
  230. rspamd_proxy_quark(),
  231. EINVAL,
  232. "cannot init lua parser script: %s",
  233. lua_tostring(L, -1));
  234. lua_settop(L, 0);
  235. return FALSE;
  236. }
  237. if (lua_istable(L, -1)) {
  238. /*
  239. * We have a table, so we check for two keys:
  240. * 'from' -> function
  241. * 'to' -> function
  242. *
  243. * From converts parent request to a client one
  244. * To converts client request to a parent one
  245. */
  246. lua_pushstring(L, "from");
  247. lua_gettable(L, -2);
  248. if (lua_isfunction(L, -1)) {
  249. ref_idx = luaL_ref(L, LUA_REGISTRYINDEX);
  250. *ref_from = ref_idx;
  251. has_ref = TRUE;
  252. }
  253. lua_pushstring(L, "to");
  254. lua_gettable(L, -2);
  255. if (lua_isfunction(L, -1)) {
  256. ref_idx = luaL_ref(L, LUA_REGISTRYINDEX);
  257. *ref_to = ref_idx;
  258. has_ref = TRUE;
  259. }
  260. }
  261. else if (!lua_isfunction(L, -1)) {
  262. g_set_error(err,
  263. rspamd_proxy_quark(),
  264. EINVAL,
  265. "cannot init lua parser script: "
  266. "must return function");
  267. lua_settop(L, 0);
  268. return FALSE;
  269. }
  270. else {
  271. /* Just parser from protocol */
  272. ref_idx = luaL_ref(L, LUA_REGISTRYINDEX);
  273. *ref_from = ref_idx;
  274. lua_settop(L, 0);
  275. has_ref = TRUE;
  276. }
  277. return has_ref;
  278. }
  279. static gboolean
  280. rspamd_proxy_parse_upstream(rspamd_mempool_t *pool,
  281. const ucl_object_t *obj,
  282. gpointer ud,
  283. struct rspamd_rcl_section *section,
  284. GError **err)
  285. {
  286. const ucl_object_t *elt;
  287. struct rspamd_http_upstream *up = NULL;
  288. struct rspamd_proxy_ctx *ctx;
  289. struct rspamd_rcl_struct_parser *pd = ud;
  290. lua_State *L;
  291. ctx = pd->user_struct;
  292. L = ctx->lua_state;
  293. if (ucl_object_type(obj) != UCL_OBJECT) {
  294. if (ucl_object_type(obj) != UCL_NULL) {
  295. msg_info_pool_check("upstream %s is disabled by setting it to NULL",
  296. ucl_object_key(obj));
  297. return TRUE;
  298. }
  299. g_set_error(err, rspamd_proxy_quark(), 100,
  300. "upstream option must be an object");
  301. return FALSE;
  302. }
  303. if (!rspamd_config_is_enabled_from_ucl(pool, obj)) {
  304. /* Upstream is valid but disabled */
  305. msg_info_pool_check("upstream %s is disabled",
  306. ucl_object_lookup(obj, "name") ? ucl_object_tostring(ucl_object_lookup(obj, "name")) : ucl_object_key(obj));
  307. return TRUE;
  308. }
  309. up = rspamd_mempool_alloc0(pool, sizeof(*up));
  310. elt = ucl_object_lookup(obj, "name");
  311. if (elt == NULL) {
  312. if (ucl_object_key(obj)) {
  313. if (strcmp(ucl_object_key(obj), "upstream") == 0) {
  314. /* Iterate over the object and find upstream elements */
  315. ucl_object_iter_t it = NULL;
  316. const ucl_object_t *cur;
  317. gboolean ret = TRUE;
  318. while ((cur = ucl_object_iterate(obj, &it, true)) != NULL) {
  319. if (!rspamd_proxy_parse_upstream(pool, cur, ud,
  320. section, err)) {
  321. ret = FALSE;
  322. }
  323. }
  324. return ret;
  325. }
  326. else {
  327. /* Inside upstream */
  328. up->name = rspamd_mempool_strdup(pool, ucl_object_key(obj));
  329. }
  330. }
  331. else {
  332. g_set_error(err, rspamd_proxy_quark(), 100,
  333. "upstream option must have some name definition");
  334. return FALSE;
  335. }
  336. }
  337. else {
  338. up->name = rspamd_mempool_strdup(pool, ucl_object_tostring(elt));
  339. }
  340. up->parser_from_ref = -1;
  341. up->parser_to_ref = -1;
  342. up->timeout = ctx->timeout;
  343. elt = ucl_object_lookup(obj, "key");
  344. if (elt != NULL) {
  345. up->key = rspamd_pubkey_from_base32(ucl_object_tostring(elt), 0,
  346. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  347. if (up->key == NULL) {
  348. g_set_error(err, rspamd_proxy_quark(), 100,
  349. "cannot read upstream key");
  350. goto err;
  351. }
  352. rspamd_mempool_add_destructor(pool,
  353. (rspamd_mempool_destruct_t) rspamd_pubkey_unref, up->key);
  354. }
  355. elt = ucl_object_lookup(obj, "self_scan");
  356. if (elt && ucl_object_toboolean(elt)) {
  357. up->self_scan = TRUE;
  358. ctx->has_self_scan = TRUE;
  359. }
  360. elt = ucl_object_lookup_any(obj, "compress", "compression", NULL);
  361. if (elt && ucl_object_toboolean(elt)) {
  362. up->compress = TRUE;
  363. }
  364. elt = ucl_object_lookup(obj, "hosts");
  365. if (elt == NULL && !up->self_scan) {
  366. g_set_error(err, rspamd_proxy_quark(), 100,
  367. "upstream option must have some hosts definition");
  368. goto err;
  369. }
  370. if (elt) {
  371. up->u = rspamd_upstreams_create(ctx->cfg->ups_ctx);
  372. if (!rspamd_upstreams_from_ucl(up->u, elt, 11333, NULL)) {
  373. g_set_error(err, rspamd_proxy_quark(), 100,
  374. "upstream has bad hosts definition");
  375. goto err;
  376. }
  377. rspamd_mempool_add_destructor(pool,
  378. (rspamd_mempool_destruct_t) rspamd_upstreams_destroy, up->u);
  379. }
  380. elt = ucl_object_lookup(obj, "default");
  381. if (elt) {
  382. if (ucl_object_toboolean(elt)) {
  383. ctx->default_upstream = up;
  384. }
  385. }
  386. else if (up->self_scan) {
  387. ctx->default_upstream = up;
  388. }
  389. elt = ucl_object_lookup(obj, "local");
  390. if (elt && ucl_object_toboolean(elt)) {
  391. up->local = TRUE;
  392. }
  393. elt = ucl_object_lookup(obj, "timeout");
  394. if (elt) {
  395. ucl_object_todouble_safe(elt, &up->timeout);
  396. }
  397. elt = ucl_object_lookup_any(obj, "settings", "settings_id", NULL);
  398. if (elt && ucl_object_type(elt) == UCL_STRING) {
  399. up->settings_id = rspamd_mempool_strdup(pool, ucl_object_tostring(elt));
  400. }
  401. /*
  402. * Accept lua function here in form
  403. * fun :: String -> UCL
  404. */
  405. elt = ucl_object_lookup(obj, "parser");
  406. if (elt) {
  407. if (!rspamd_proxy_parse_lua_parser(L, elt, &up->parser_from_ref,
  408. &up->parser_to_ref, err)) {
  409. goto err;
  410. }
  411. rspamd_lua_add_ref_dtor(L, pool, up->parser_from_ref);
  412. rspamd_lua_add_ref_dtor(L, pool, up->parser_to_ref);
  413. }
  414. g_hash_table_insert(ctx->upstreams, up->name, up);
  415. return TRUE;
  416. err:
  417. return FALSE;
  418. }
  419. static gboolean
  420. rspamd_proxy_parse_mirror(rspamd_mempool_t *pool,
  421. const ucl_object_t *obj,
  422. gpointer ud,
  423. struct rspamd_rcl_section *section,
  424. GError **err)
  425. {
  426. const ucl_object_t *elt;
  427. struct rspamd_http_mirror *up = NULL;
  428. struct rspamd_proxy_ctx *ctx;
  429. struct rspamd_rcl_struct_parser *pd = ud;
  430. lua_State *L;
  431. ctx = pd->user_struct;
  432. L = ctx->lua_state;
  433. if (ucl_object_type(obj) != UCL_OBJECT) {
  434. if (ucl_object_type(obj) != UCL_NULL) {
  435. msg_info_pool_check("mirror %s is disabled by setting it to NULL",
  436. ucl_object_key(obj));
  437. return TRUE;
  438. }
  439. g_set_error(err, rspamd_proxy_quark(), 100,
  440. "mirror option must be an object");
  441. return FALSE;
  442. }
  443. if (!rspamd_config_is_enabled_from_ucl(pool, obj)) {
  444. /* Upstream is valid but disabled */
  445. msg_info_pool_check("mirror %s is disabled",
  446. ucl_object_lookup(obj, "name") ? ucl_object_tostring(ucl_object_lookup(obj, "name")) : ucl_object_key(obj));
  447. return TRUE;
  448. }
  449. up = rspamd_mempool_alloc0(pool, sizeof(*up));
  450. elt = ucl_object_lookup(obj, "name");
  451. if (elt == NULL) {
  452. if (ucl_object_key(obj)) {
  453. if (strcmp(ucl_object_key(obj), "mirror") == 0) {
  454. /* Iterate over the object and find upstream elements */
  455. ucl_object_iter_t it = NULL;
  456. const ucl_object_t *cur;
  457. gboolean ret = TRUE;
  458. while ((cur = ucl_object_iterate(obj, &it, true)) != NULL) {
  459. if (!rspamd_proxy_parse_mirror(pool, cur, ud,
  460. section, err)) {
  461. ret = FALSE;
  462. }
  463. }
  464. return ret;
  465. }
  466. else {
  467. /* Inside upstream */
  468. up->name = rspamd_mempool_strdup(pool, ucl_object_key(obj));
  469. }
  470. }
  471. else {
  472. g_set_error(err, rspamd_proxy_quark(), 100,
  473. "mirror option must have some name definition");
  474. return FALSE;
  475. }
  476. }
  477. else {
  478. up->name = rspamd_mempool_strdup(pool, ucl_object_tostring(elt));
  479. }
  480. up->parser_to_ref = -1;
  481. up->parser_from_ref = -1;
  482. up->timeout = ctx->timeout;
  483. elt = ucl_object_lookup(obj, "key");
  484. if (elt != NULL) {
  485. up->key = rspamd_pubkey_from_base32(ucl_object_tostring(elt), 0,
  486. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  487. if (up->key == NULL) {
  488. g_set_error(err, rspamd_proxy_quark(), 100,
  489. "cannot read mirror key");
  490. goto err;
  491. }
  492. rspamd_mempool_add_destructor(pool,
  493. (rspamd_mempool_destruct_t) rspamd_pubkey_unref, up->key);
  494. }
  495. elt = ucl_object_lookup(obj, "hosts");
  496. if (elt == NULL) {
  497. g_set_error(err, rspamd_proxy_quark(), 100,
  498. "mirror option must have some hosts definition");
  499. goto err;
  500. }
  501. up->u = rspamd_upstreams_create(ctx->cfg->ups_ctx);
  502. if (!rspamd_upstreams_from_ucl(up->u, elt, 11333, NULL)) {
  503. g_set_error(err, rspamd_proxy_quark(), 100,
  504. "mirror has bad hosts definition");
  505. goto err;
  506. }
  507. rspamd_mempool_add_destructor(pool,
  508. (rspamd_mempool_destruct_t) rspamd_upstreams_destroy, up->u);
  509. elt = ucl_object_lookup_any(obj, "probability", "prob", NULL);
  510. if (elt) {
  511. up->prob = ucl_object_todouble(elt);
  512. }
  513. else {
  514. up->prob = 1.0;
  515. }
  516. elt = ucl_object_lookup(obj, "local");
  517. if (elt && ucl_object_toboolean(elt)) {
  518. up->local = TRUE;
  519. }
  520. elt = ucl_object_lookup_any(obj, "compress", "compression", NULL);
  521. if (elt && ucl_object_toboolean(elt)) {
  522. up->compress = TRUE;
  523. }
  524. elt = ucl_object_lookup(obj, "timeout");
  525. if (elt) {
  526. ucl_object_todouble_safe(elt, &up->timeout);
  527. }
  528. /*
  529. * Accept lua function here in form
  530. * fun :: String -> UCL
  531. */
  532. elt = ucl_object_lookup(obj, "parser");
  533. if (elt) {
  534. if (!rspamd_proxy_parse_lua_parser(L, elt, &up->parser_from_ref,
  535. &up->parser_to_ref, err)) {
  536. goto err;
  537. }
  538. rspamd_lua_add_ref_dtor(L, pool, up->parser_from_ref);
  539. rspamd_lua_add_ref_dtor(L, pool, up->parser_to_ref);
  540. }
  541. elt = ucl_object_lookup_any(obj, "settings", "settings_id", NULL);
  542. if (elt && ucl_object_type(elt) == UCL_STRING) {
  543. up->settings_id = rspamd_mempool_strdup(pool, ucl_object_tostring(elt));
  544. }
  545. g_ptr_array_add(ctx->mirrors, up);
  546. return TRUE;
  547. err:
  548. return FALSE;
  549. }
  550. static gboolean
  551. rspamd_proxy_parse_script(rspamd_mempool_t *pool,
  552. const ucl_object_t *obj,
  553. gpointer ud,
  554. struct rspamd_rcl_section *section,
  555. GError **err)
  556. {
  557. struct rspamd_proxy_ctx *ctx;
  558. struct rspamd_rcl_struct_parser *pd = ud;
  559. lua_State *L;
  560. const gchar *lua_script;
  561. gsize slen;
  562. gint err_idx, ref_idx;
  563. struct stat st;
  564. ctx = pd->user_struct;
  565. L = ctx->lua_state;
  566. if (ucl_object_type(obj) != UCL_STRING) {
  567. g_set_error(err, rspamd_proxy_quark(), 100,
  568. "script option must be a string with file or lua chunk");
  569. return FALSE;
  570. }
  571. lua_script = ucl_object_tolstring(obj, &slen);
  572. lua_pushcfunction(L, &rspamd_lua_traceback);
  573. err_idx = lua_gettop(L);
  574. if (stat(lua_script, &st) != -1) {
  575. /* Load file */
  576. if (luaL_loadfile(L, lua_script) != 0) {
  577. g_set_error(err,
  578. rspamd_proxy_quark(),
  579. EINVAL,
  580. "cannot load lua parser script: %s",
  581. lua_tostring(L, -1));
  582. lua_settop(L, 0); /* Error function */
  583. goto err;
  584. }
  585. }
  586. else {
  587. /* Load data directly */
  588. if (luaL_loadbuffer(L, lua_script, slen, "proxy parser") != 0) {
  589. g_set_error(err,
  590. rspamd_proxy_quark(),
  591. EINVAL,
  592. "cannot load lua parser script: %s",
  593. lua_tostring(L, -1));
  594. lua_settop(L, 0); /* Error function */
  595. goto err;
  596. }
  597. }
  598. /* Now do it */
  599. if (lua_pcall(L, 0, 1, err_idx) != 0) {
  600. g_set_error(err,
  601. rspamd_proxy_quark(),
  602. EINVAL,
  603. "cannot init lua parser script: %s",
  604. lua_tostring(L, -1));
  605. lua_settop(L, 0);
  606. goto err;
  607. }
  608. if (!lua_isfunction(L, -1)) {
  609. g_set_error(err,
  610. rspamd_proxy_quark(),
  611. EINVAL,
  612. "cannot init lua parser script: "
  613. "must return function, %s returned",
  614. lua_typename(L, lua_type(L, -1)));
  615. lua_settop(L, 0);
  616. goto err;
  617. }
  618. ref_idx = luaL_ref(L, LUA_REGISTRYINDEX);
  619. lua_settop(L, 0);
  620. g_array_append_val(ctx->cmp_refs, ref_idx);
  621. return TRUE;
  622. err:
  623. return FALSE;
  624. }
  625. gpointer
  626. init_rspamd_proxy(struct rspamd_config *cfg)
  627. {
  628. struct rspamd_proxy_ctx *ctx;
  629. GQuark type;
  630. type = g_quark_try_string("rspamd_proxy");
  631. ctx = rspamd_mempool_alloc0(cfg->cfg_pool,
  632. sizeof(struct rspamd_proxy_ctx));
  633. ctx->magic = rspamd_rspamd_proxy_magic;
  634. ctx->timeout = 120.0;
  635. ctx->upstreams = g_hash_table_new(rspamd_strcase_hash, rspamd_strcase_equal);
  636. rspamd_mempool_add_destructor(cfg->cfg_pool,
  637. (rspamd_mempool_destruct_t) g_hash_table_unref, ctx->upstreams);
  638. ctx->mirrors = g_ptr_array_new();
  639. rspamd_mempool_add_destructor(cfg->cfg_pool,
  640. (rspamd_mempool_destruct_t) rspamd_ptr_array_free_hard, ctx->mirrors);
  641. ctx->cfg = cfg;
  642. ctx->lua_state = cfg->lua_state;
  643. ctx->cmp_refs = g_array_new(FALSE, FALSE, sizeof(gint));
  644. rspamd_mempool_add_destructor(cfg->cfg_pool,
  645. (rspamd_mempool_destruct_t) rspamd_array_free_hard, ctx->cmp_refs);
  646. ctx->max_retries = DEFAULT_RETRIES;
  647. ctx->spam_header = RSPAMD_MILTER_SPAM_HEADER;
  648. rspamd_rcl_register_worker_option(cfg,
  649. type,
  650. "timeout",
  651. rspamd_rcl_parse_struct_time,
  652. ctx,
  653. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, timeout),
  654. RSPAMD_CL_FLAG_TIME_FLOAT,
  655. "IO timeout");
  656. rspamd_rcl_register_worker_option(cfg,
  657. type,
  658. "keypair",
  659. rspamd_rcl_parse_struct_keypair,
  660. ctx,
  661. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, key),
  662. 0,
  663. "Server's keypair");
  664. rspamd_rcl_register_worker_option(cfg,
  665. type,
  666. "upstream",
  667. rspamd_proxy_parse_upstream,
  668. ctx,
  669. 0,
  670. 0,
  671. "List of upstreams");
  672. rspamd_rcl_register_worker_option(cfg,
  673. type,
  674. "mirror",
  675. rspamd_proxy_parse_mirror,
  676. ctx,
  677. 0,
  678. RSPAMD_CL_FLAG_MULTIPLE,
  679. "List of mirrors");
  680. rspamd_rcl_register_worker_option(cfg,
  681. type,
  682. "script",
  683. rspamd_proxy_parse_script,
  684. ctx,
  685. 0,
  686. RSPAMD_CL_FLAG_MULTIPLE,
  687. "Compare script to be executed");
  688. rspamd_rcl_register_worker_option(cfg,
  689. type,
  690. "max_retries",
  691. rspamd_rcl_parse_struct_integer,
  692. ctx,
  693. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, max_retries),
  694. RSPAMD_CL_FLAG_UINT,
  695. "Maximum number of retries for master connection");
  696. rspamd_rcl_register_worker_option(cfg,
  697. type,
  698. "milter",
  699. rspamd_rcl_parse_struct_boolean,
  700. ctx,
  701. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, milter),
  702. 0,
  703. "Accept milter connections, not HTTP");
  704. rspamd_rcl_register_worker_option(cfg,
  705. type,
  706. "discard_on_reject",
  707. rspamd_rcl_parse_struct_boolean,
  708. ctx,
  709. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, discard_on_reject),
  710. 0,
  711. "Tell MTA to discard rejected messages silently");
  712. rspamd_rcl_register_worker_option(cfg,
  713. type,
  714. "quarantine_on_reject",
  715. rspamd_rcl_parse_struct_boolean,
  716. ctx,
  717. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, quarantine_on_reject),
  718. 0,
  719. "Tell MTA to quarantine rejected messages");
  720. rspamd_rcl_register_worker_option(cfg,
  721. type,
  722. "spam_header",
  723. rspamd_rcl_parse_struct_string,
  724. ctx,
  725. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, spam_header),
  726. 0,
  727. "Use the specific spam header (default: X-Spam)");
  728. rspamd_rcl_register_worker_option(cfg,
  729. type,
  730. "client_ca_name",
  731. rspamd_rcl_parse_struct_string,
  732. ctx,
  733. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, client_ca_name),
  734. 0,
  735. "Allow certificates issued by this CA to be treated as client certificates");
  736. rspamd_rcl_register_worker_option(cfg,
  737. type,
  738. "reject_message",
  739. rspamd_rcl_parse_struct_string,
  740. ctx,
  741. G_STRUCT_OFFSET(struct rspamd_proxy_ctx, reject_message),
  742. 0,
  743. "Use custom rejection message");
  744. return ctx;
  745. }
  746. static void
  747. proxy_backend_close_connection(struct rspamd_proxy_backend_connection *conn)
  748. {
  749. if (conn && !(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  750. if (conn->backend_conn) {
  751. rspamd_http_connection_reset(conn->backend_conn);
  752. rspamd_http_connection_unref(conn->backend_conn);
  753. close(conn->backend_sock);
  754. }
  755. conn->flags |= RSPAMD_BACKEND_CLOSED;
  756. }
  757. }
  758. static gboolean
  759. proxy_backend_parse_results(struct rspamd_proxy_session *session,
  760. struct rspamd_proxy_backend_connection *conn,
  761. lua_State *L, gint parser_ref,
  762. struct rspamd_http_message *msg,
  763. goffset *body_offset,
  764. const rspamd_ftok_t *ct)
  765. {
  766. struct ucl_parser *parser;
  767. gint err_idx;
  768. const gchar *in = msg->body_buf.begin;
  769. gsize inlen = msg->body_buf.len;
  770. const rspamd_ftok_t *offset_hdr;
  771. if (inlen == 0 || in == NULL) {
  772. return FALSE;
  773. }
  774. offset_hdr = rspamd_http_message_find_header(msg, MESSAGE_OFFSET_HEADER);
  775. if (offset_hdr) {
  776. gulong val;
  777. if (rspamd_strtoul(offset_hdr->begin, offset_hdr->len, &val) && val < inlen) {
  778. if (body_offset) {
  779. *body_offset = val;
  780. }
  781. inlen = val;
  782. }
  783. }
  784. if (parser_ref != -1) {
  785. /* Call parser function */
  786. lua_pushcfunction(L, &rspamd_lua_traceback);
  787. err_idx = lua_gettop(L);
  788. lua_rawgeti(L, LUA_REGISTRYINDEX, parser_ref);
  789. /* XXX: copies all data */
  790. lua_pushlstring(L, in, inlen);
  791. if (lua_pcall(L, 1, 1, err_idx) != 0) {
  792. msg_err_session(
  793. "cannot run lua parser script: %s",
  794. lua_tostring(L, -1));
  795. lua_settop(L, 0);
  796. return FALSE;
  797. }
  798. conn->results = ucl_object_lua_import(L, -1);
  799. lua_settop(L, 0);
  800. }
  801. else {
  802. rspamd_ftok_t json_ct;
  803. RSPAMD_FTOK_ASSIGN(&json_ct, "application/json");
  804. if (ct && rspamd_ftok_casecmp(ct, &json_ct) == 0) {
  805. parser = ucl_parser_new(0);
  806. if (!ucl_parser_add_chunk(parser, in, inlen)) {
  807. gchar *encoded;
  808. encoded = rspamd_encode_base64(in, inlen, 0, NULL);
  809. msg_err_session("cannot parse input: %s", ucl_parser_get_error(
  810. parser));
  811. msg_err_session("input encoded: %s", encoded);
  812. ucl_parser_free(parser);
  813. g_free(encoded);
  814. return FALSE;
  815. }
  816. conn->results = ucl_parser_get_object(parser);
  817. ucl_parser_free(parser);
  818. }
  819. }
  820. return TRUE;
  821. }
  822. static void
  823. proxy_call_cmp_script(struct rspamd_proxy_session *session, gint cbref)
  824. {
  825. gint err_idx;
  826. guint i;
  827. struct rspamd_proxy_backend_connection *conn;
  828. lua_State *L;
  829. L = session->ctx->lua_state;
  830. lua_pushcfunction(L, &rspamd_lua_traceback);
  831. err_idx = lua_gettop(L);
  832. lua_rawgeti(L, LUA_REGISTRYINDEX, cbref);
  833. lua_createtable(L, 0, session->mirror_conns->len + 1);
  834. /* Now push master results */
  835. if (session->master_conn && session->master_conn->results) {
  836. lua_pushstring(L, "master");
  837. ucl_object_push_lua(L, session->master_conn->results, true);
  838. lua_settable(L, -3);
  839. }
  840. else {
  841. lua_pushstring(L, "master");
  842. lua_pushstring(L, "no results");
  843. lua_settable(L, -3);
  844. }
  845. for (i = 0; i < session->mirror_conns->len; i++) {
  846. conn = g_ptr_array_index(session->mirror_conns, i);
  847. if (conn->results) {
  848. lua_pushstring(L, conn->name);
  849. ucl_object_push_lua(L, conn->results, true);
  850. lua_settable(L, -3);
  851. }
  852. else {
  853. lua_pushstring(L, conn->name);
  854. lua_pushstring(L, conn->err ? conn->err : "unknown error");
  855. lua_settable(L, -3);
  856. }
  857. }
  858. gchar log_tag[RSPAMD_LOG_ID_LEN + 1];
  859. rspamd_strlcpy(log_tag, session->pool->tag.uid, sizeof(log_tag));
  860. lua_pushstring(L, log_tag);
  861. if (lua_pcall(L, 2, 0, err_idx) != 0) {
  862. msg_err_session(
  863. "cannot run lua compare script: %s",
  864. lua_tostring(L, -1));
  865. }
  866. lua_settop(L, err_idx - 1);
  867. }
  868. static void
  869. proxy_session_dtor(struct rspamd_proxy_session *session)
  870. {
  871. guint i;
  872. gint cbref;
  873. struct rspamd_proxy_backend_connection *conn;
  874. if (session->master_conn && session->master_conn->results) {
  875. for (i = 0; i < session->ctx->cmp_refs->len; i++) {
  876. cbref = g_array_index(session->ctx->cmp_refs, gint, i);
  877. proxy_call_cmp_script(session, cbref);
  878. }
  879. }
  880. if (session->master_conn) {
  881. proxy_backend_close_connection(session->master_conn);
  882. }
  883. if (session->client_milter_conn) {
  884. rspamd_milter_session_unref(session->client_milter_conn);
  885. }
  886. else if (session->client_conn) {
  887. rspamd_http_connection_reset(session->client_conn);
  888. rspamd_http_connection_unref(session->client_conn);
  889. }
  890. if (session->map && session->map_len) {
  891. munmap(session->map, session->map_len);
  892. }
  893. for (i = 0; i < session->mirror_conns->len; i++) {
  894. conn = g_ptr_array_index(session->mirror_conns, i);
  895. if (!(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  896. proxy_backend_close_connection(conn);
  897. }
  898. if (conn->results) {
  899. ucl_object_unref(conn->results);
  900. }
  901. }
  902. if (session->master_conn) {
  903. if (session->master_conn->results) {
  904. ucl_object_unref(session->master_conn->results);
  905. }
  906. if (session->master_conn->task) {
  907. rspamd_session_destroy(session->master_conn->task->s);
  908. }
  909. }
  910. g_ptr_array_free(session->mirror_conns, TRUE);
  911. rspamd_http_message_shmem_unref(session->shmem_ref);
  912. rspamd_http_message_unref(session->client_message);
  913. if (session->client_addr) {
  914. rspamd_inet_address_free(session->client_addr);
  915. }
  916. if (session->client_sock != -1) {
  917. close(session->client_sock);
  918. }
  919. if (session->ctx->sessions_cache) {
  920. rspamd_worker_session_cache_remove(session->ctx->sessions_cache,
  921. session);
  922. }
  923. if (session->pool) {
  924. rspamd_mempool_delete(session->pool);
  925. }
  926. g_free(session);
  927. }
  928. static void
  929. proxy_request_compress(struct rspamd_http_message *msg)
  930. {
  931. guint flags;
  932. ZSTD_CCtx *zctx;
  933. rspamd_fstring_t *body;
  934. const gchar *in;
  935. gsize inlen;
  936. flags = rspamd_http_message_get_flags(msg);
  937. if (!rspamd_http_message_find_header(msg, COMPRESSION_HEADER)) {
  938. if ((flags & RSPAMD_HTTP_FLAG_SHMEM) ||
  939. !(flags & RSPAMD_HTTP_FLAG_HAS_BODY)) {
  940. /* Cannot compress shared or empty message */
  941. return;
  942. }
  943. in = rspamd_http_message_get_body(msg, &inlen);
  944. if (in == NULL || inlen == 0) {
  945. return;
  946. }
  947. body = rspamd_fstring_sized_new(ZSTD_compressBound(inlen));
  948. zctx = ZSTD_createCCtx();
  949. body->len = ZSTD_compressCCtx(zctx, body->str, body->allocated,
  950. in, inlen, 1);
  951. if (ZSTD_isError(body->len)) {
  952. msg_err("compression error");
  953. rspamd_fstring_free(body);
  954. ZSTD_freeCCtx(zctx);
  955. return;
  956. }
  957. ZSTD_freeCCtx(zctx);
  958. rspamd_http_message_set_body_from_fstring_steal(msg, body);
  959. rspamd_http_message_add_header(msg, COMPRESSION_HEADER, "zstd");
  960. }
  961. }
  962. static void
  963. proxy_request_decompress(struct rspamd_http_message *msg)
  964. {
  965. rspamd_fstring_t *body;
  966. const gchar *in;
  967. gsize inlen, outlen, r;
  968. ZSTD_DStream *zstream;
  969. ZSTD_inBuffer zin;
  970. ZSTD_outBuffer zout;
  971. if (rspamd_http_message_find_header(msg, COMPRESSION_HEADER)) {
  972. in = rspamd_http_message_get_body(msg, &inlen);
  973. if (in == NULL || inlen == 0) {
  974. return;
  975. }
  976. zstream = ZSTD_createDStream();
  977. ZSTD_initDStream(zstream);
  978. zin.pos = 0;
  979. zin.src = in;
  980. zin.size = inlen;
  981. if ((outlen = ZSTD_getDecompressedSize(zin.src, zin.size)) == 0) {
  982. outlen = ZSTD_DStreamOutSize();
  983. }
  984. body = rspamd_fstring_sized_new(outlen);
  985. zout.dst = body->str;
  986. zout.pos = 0;
  987. zout.size = outlen;
  988. while (zin.pos < zin.size) {
  989. r = ZSTD_decompressStream(zstream, &zout, &zin);
  990. if (ZSTD_isError(r)) {
  991. msg_err("Decompression error: %s", ZSTD_getErrorName(r));
  992. ZSTD_freeDStream(zstream);
  993. rspamd_fstring_free(body);
  994. return;
  995. }
  996. if (zout.pos == zout.size) {
  997. /* We need to extend output buffer */
  998. zout.size = zout.size * 2 + 1;
  999. body = rspamd_fstring_grow(body, zout.size);
  1000. zout.size = body->allocated;
  1001. zout.dst = body->str;
  1002. }
  1003. }
  1004. body->len = zout.pos;
  1005. ZSTD_freeDStream(zstream);
  1006. rspamd_http_message_set_body_from_fstring_steal(msg, body);
  1007. rspamd_http_message_remove_header(msg, COMPRESSION_HEADER);
  1008. }
  1009. }
  1010. static struct rspamd_proxy_session *
  1011. proxy_session_refresh(struct rspamd_proxy_session *session)
  1012. {
  1013. struct rspamd_proxy_session *nsession;
  1014. nsession = g_malloc0(sizeof(*nsession));
  1015. nsession->client_milter_conn = session->client_milter_conn;
  1016. session->client_milter_conn = NULL;
  1017. rspamd_milter_update_userdata(nsession->client_milter_conn,
  1018. nsession);
  1019. nsession->client_addr = session->client_addr;
  1020. session->client_addr = NULL;
  1021. nsession->ctx = session->ctx;
  1022. nsession->worker = session->worker;
  1023. nsession->pool = rspamd_mempool_new(rspamd_mempool_suggest_size(), "proxy", 0);
  1024. nsession->client_sock = session->client_sock;
  1025. session->client_sock = -1;
  1026. nsession->mirror_conns = g_ptr_array_sized_new(nsession->ctx->mirrors->len);
  1027. REF_INIT_RETAIN(nsession, proxy_session_dtor);
  1028. if (nsession->ctx->sessions_cache) {
  1029. rspamd_worker_session_cache_add(nsession->ctx->sessions_cache,
  1030. nsession->pool->tag.uid, &nsession->ref.refcount, nsession);
  1031. }
  1032. return nsession;
  1033. }
  1034. static gboolean
  1035. proxy_check_file(struct rspamd_http_message *msg,
  1036. struct rspamd_proxy_session *session)
  1037. {
  1038. const rspamd_ftok_t *tok, *key_tok;
  1039. rspamd_ftok_t srch;
  1040. gchar *file_str;
  1041. GHashTable *query_args;
  1042. GHashTableIter it;
  1043. gpointer k, v;
  1044. struct http_parser_url u;
  1045. rspamd_fstring_t *new_url;
  1046. tok = rspamd_http_message_find_header(msg, "File");
  1047. if (tok) {
  1048. file_str = rspamd_mempool_ftokdup(session->pool, tok);
  1049. session->map = rspamd_file_xmap(file_str, PROT_READ, &session->map_len,
  1050. 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. return FALSE;
  1056. }
  1057. }
  1058. /* Remove header after processing */
  1059. rspamd_http_message_remove_header(msg, "File");
  1060. session->fname = file_str;
  1061. }
  1062. else {
  1063. /* Need to parse query URL */
  1064. if (http_parser_parse_url(RSPAMD_FSTRING_DATA(msg->url),
  1065. RSPAMD_FSTRING_LEN(msg->url), 0, &u) != 0) {
  1066. msg_err_session("bad request url: %V", msg->url);
  1067. return FALSE;
  1068. }
  1069. if (u.field_set & (1 << UF_QUERY)) {
  1070. /* In case if we have a query, we need to store it somewhere */
  1071. query_args = rspamd_http_message_parse_query(msg);
  1072. srch.begin = "File";
  1073. srch.len = strlen("File");
  1074. tok = g_hash_table_lookup(query_args, &srch);
  1075. if (tok) {
  1076. file_str = rspamd_mempool_ftokdup(session->pool, tok);
  1077. session->map = rspamd_file_xmap(file_str, PROT_READ,
  1078. &session->map_len, TRUE);
  1079. if (session->map == NULL) {
  1080. if (session->map_len != 0) {
  1081. msg_err_session("cannot map %s: %s", file_str,
  1082. strerror(errno));
  1083. g_hash_table_unref(query_args);
  1084. return FALSE;
  1085. }
  1086. }
  1087. /* We need to create a new URL with file attribute removed */
  1088. new_url = rspamd_fstring_new_init(RSPAMD_FSTRING_DATA(msg->url),
  1089. u.field_data[UF_QUERY].off);
  1090. new_url = rspamd_fstring_append(new_url, "?", 1);
  1091. g_hash_table_iter_init(&it, query_args);
  1092. while (g_hash_table_iter_next(&it, &k, &v)) {
  1093. key_tok = k;
  1094. tok = v;
  1095. if (!rspamd_ftok_icase_equal(key_tok, &srch)) {
  1096. rspamd_printf_fstring(&new_url, "%T=%T&",
  1097. key_tok, tok);
  1098. }
  1099. }
  1100. /* Erase last character (might be either & or ?) */
  1101. rspamd_fstring_erase(new_url, new_url->len - 1, 1);
  1102. rspamd_fstring_free(msg->url);
  1103. msg->url = new_url;
  1104. session->fname = file_str;
  1105. }
  1106. g_hash_table_unref(query_args);
  1107. }
  1108. }
  1109. return TRUE;
  1110. }
  1111. static void
  1112. proxy_backend_mirror_error_handler(struct rspamd_http_connection *conn, GError *err)
  1113. {
  1114. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1115. struct rspamd_proxy_session *session;
  1116. session = bk_conn->s;
  1117. msg_info_session("abnormally closing connection from backend: %s:%s, "
  1118. "error: %e",
  1119. bk_conn->name,
  1120. rspamd_inet_address_to_string_pretty(
  1121. rspamd_upstream_addr_cur(bk_conn->up)),
  1122. err);
  1123. if (err) {
  1124. bk_conn->err = rspamd_mempool_strdup(session->pool, err->message);
  1125. }
  1126. rspamd_upstream_fail(bk_conn->up, FALSE, err ? err->message : "unknown");
  1127. proxy_backend_close_connection(bk_conn);
  1128. REF_RELEASE(bk_conn->s);
  1129. }
  1130. static gint
  1131. proxy_backend_mirror_finish_handler(struct rspamd_http_connection *conn,
  1132. struct rspamd_http_message *msg)
  1133. {
  1134. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1135. struct rspamd_proxy_session *session;
  1136. const rspamd_ftok_t *orig_ct;
  1137. session = bk_conn->s;
  1138. proxy_request_decompress(msg);
  1139. orig_ct = rspamd_http_message_find_header(msg, "Content-Type");
  1140. if (!proxy_backend_parse_results(session, bk_conn, session->ctx->lua_state,
  1141. bk_conn->parser_from_ref, msg, NULL, orig_ct)) {
  1142. msg_warn_session("cannot parse results from the mirror backend %s:%s",
  1143. bk_conn->name,
  1144. rspamd_inet_address_to_string(
  1145. rspamd_upstream_addr_cur(bk_conn->up)));
  1146. bk_conn->err = "cannot parse ucl";
  1147. }
  1148. msg_info_session("finished mirror connection to %s", bk_conn->name);
  1149. rspamd_upstream_ok(bk_conn->up);
  1150. proxy_backend_close_connection(bk_conn);
  1151. REF_RELEASE(bk_conn->s);
  1152. return 0;
  1153. }
  1154. static void
  1155. proxy_open_mirror_connections(struct rspamd_proxy_session *session)
  1156. {
  1157. gdouble coin;
  1158. struct rspamd_http_mirror *m;
  1159. guint i;
  1160. struct rspamd_proxy_backend_connection *bk_conn;
  1161. struct rspamd_http_message *msg;
  1162. GError *err = NULL;
  1163. coin = rspamd_random_double();
  1164. for (i = 0; i < session->ctx->mirrors->len; i++) {
  1165. m = g_ptr_array_index(session->ctx->mirrors, i);
  1166. if (m->prob < coin) {
  1167. /* No luck */
  1168. continue;
  1169. }
  1170. bk_conn = rspamd_mempool_alloc0(session->pool,
  1171. sizeof(*bk_conn));
  1172. bk_conn->s = session;
  1173. bk_conn->name = m->name;
  1174. bk_conn->timeout = m->timeout;
  1175. bk_conn->up = rspamd_upstream_get(m->u,
  1176. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1177. bk_conn->parser_from_ref = m->parser_from_ref;
  1178. bk_conn->parser_to_ref = m->parser_to_ref;
  1179. if (bk_conn->up == NULL) {
  1180. msg_err_session("cannot select upstream for %s", m->name);
  1181. continue;
  1182. }
  1183. bk_conn->backend_sock = rspamd_inet_address_connect(
  1184. rspamd_upstream_addr_next(bk_conn->up),
  1185. SOCK_STREAM, TRUE);
  1186. if (bk_conn->backend_sock == -1) {
  1187. msg_err_session("cannot connect upstream for %s", m->name);
  1188. rspamd_upstream_fail(bk_conn->up, TRUE, strerror(errno));
  1189. continue;
  1190. }
  1191. msg = rspamd_http_connection_copy_msg(session->client_message, &err);
  1192. if (msg == NULL) {
  1193. msg_err_session("cannot copy message to send to a mirror %s: %e",
  1194. m->name, err);
  1195. if (err) {
  1196. g_error_free(err);
  1197. }
  1198. continue;
  1199. }
  1200. if (msg->url->len == 0) {
  1201. msg->url = rspamd_fstring_append(msg->url, "/check", strlen("/check"));
  1202. }
  1203. if (m->settings_id != NULL) {
  1204. rspamd_http_message_remove_header(msg, "Settings-ID");
  1205. rspamd_http_message_add_header(msg, "Settings-ID", m->settings_id);
  1206. }
  1207. bk_conn->backend_conn = rspamd_http_connection_new_client_socket(
  1208. session->ctx->http_ctx,
  1209. NULL,
  1210. proxy_backend_mirror_error_handler,
  1211. proxy_backend_mirror_finish_handler,
  1212. RSPAMD_HTTP_CLIENT_SIMPLE,
  1213. bk_conn->backend_sock);
  1214. if (m->key) {
  1215. msg->peer_key = rspamd_pubkey_ref(m->key);
  1216. }
  1217. if (m->local ||
  1218. rspamd_inet_address_is_local(rspamd_upstream_addr_cur(bk_conn->up))) {
  1219. if (session->fname) {
  1220. rspamd_http_message_add_header(msg, "File", session->fname);
  1221. }
  1222. msg->method = HTTP_GET;
  1223. rspamd_http_connection_write_message_shared(bk_conn->backend_conn,
  1224. msg, rspamd_upstream_name(bk_conn->up), NULL, bk_conn,
  1225. bk_conn->timeout);
  1226. }
  1227. else {
  1228. if (session->fname) {
  1229. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1230. rspamd_http_message_set_body(msg, session->map, session->map_len);
  1231. }
  1232. msg->method = HTTP_POST;
  1233. if (m->compress) {
  1234. proxy_request_compress(msg);
  1235. if (session->client_milter_conn) {
  1236. rspamd_http_message_add_header(msg, "Content-Type",
  1237. "application/octet-stream");
  1238. }
  1239. }
  1240. else {
  1241. if (session->client_milter_conn) {
  1242. rspamd_http_message_add_header(msg, "Content-Type",
  1243. "text/plain");
  1244. }
  1245. }
  1246. rspamd_http_connection_write_message(bk_conn->backend_conn,
  1247. msg, rspamd_upstream_name(bk_conn->up), NULL, bk_conn,
  1248. bk_conn->timeout);
  1249. }
  1250. g_ptr_array_add(session->mirror_conns, bk_conn);
  1251. REF_RETAIN(session);
  1252. msg_info_session("send request to %s", m->name);
  1253. }
  1254. }
  1255. static void
  1256. proxy_client_write_error(struct rspamd_proxy_session *session, gint code,
  1257. const gchar *status)
  1258. {
  1259. struct rspamd_http_message *reply;
  1260. if (session->client_milter_conn) {
  1261. rspamd_milter_send_action(session->client_milter_conn,
  1262. RSPAMD_MILTER_TEMPFAIL);
  1263. REF_RELEASE(session);
  1264. }
  1265. else {
  1266. reply = rspamd_http_new_message(HTTP_RESPONSE);
  1267. switch (code) {
  1268. case ETIMEDOUT:
  1269. reply->code = 504;
  1270. reply->status = RSPAMD_FSTRING_LIT("Gateway timeout");
  1271. break;
  1272. case ECONNRESET:
  1273. case ECONNABORTED:
  1274. reply->code = 502;
  1275. reply->status = RSPAMD_FSTRING_LIT("Gateway connection reset");
  1276. break;
  1277. case ECONNREFUSED:
  1278. reply->code = 502;
  1279. reply->status = RSPAMD_FSTRING_LIT("Gateway connection refused");
  1280. break;
  1281. default:
  1282. if (code >= 300) {
  1283. /* Likely HTTP error */
  1284. reply->code = code;
  1285. reply->status = rspamd_fstring_new_init(status, strlen(status));
  1286. }
  1287. else {
  1288. reply->code = 502;
  1289. reply->status = RSPAMD_FSTRING_LIT("Unknown gateway error: ");
  1290. reply->status = rspamd_fstring_append(reply->status,
  1291. status, strlen(status));
  1292. }
  1293. break;
  1294. }
  1295. rspamd_http_connection_write_message(session->client_conn,
  1296. reply, NULL, NULL, session,
  1297. session->ctx->timeout);
  1298. }
  1299. }
  1300. static void
  1301. proxy_backend_master_error_handler(struct rspamd_http_connection *conn, GError *err)
  1302. {
  1303. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1304. struct rspamd_proxy_session *session;
  1305. session = bk_conn->s;
  1306. session->retries++;
  1307. msg_info_session("abnormally closing connection from backend: %s, error: %e,"
  1308. " retries left: %d",
  1309. rspamd_inet_address_to_string_pretty(
  1310. rspamd_upstream_addr_cur(session->master_conn->up)),
  1311. err,
  1312. session->ctx->max_retries - session->retries);
  1313. rspamd_upstream_fail(bk_conn->up, FALSE, err ? err->message : "unknown");
  1314. proxy_backend_close_connection(session->master_conn);
  1315. if (session->ctx->max_retries > 0 &&
  1316. session->retries >= session->ctx->max_retries) {
  1317. msg_err_session("cannot connect to upstream, maximum retries "
  1318. "has been reached: %d",
  1319. session->retries);
  1320. /* Terminate session immediately */
  1321. if (err) {
  1322. proxy_client_write_error(session, err->code, err->message);
  1323. }
  1324. else {
  1325. proxy_client_write_error(session, 503, "Unknown error after no retries left");
  1326. }
  1327. }
  1328. else {
  1329. if (!proxy_send_master_message(session)) {
  1330. if (err) {
  1331. proxy_client_write_error(session, err->code, err->message);
  1332. }
  1333. else {
  1334. proxy_client_write_error(session, 503, "Unknown error on write");
  1335. }
  1336. }
  1337. else {
  1338. msg_info_session("retry connection to: %s"
  1339. " retries left: %d",
  1340. rspamd_inet_address_to_string(
  1341. rspamd_upstream_addr_cur(session->master_conn->up)),
  1342. session->ctx->max_retries - session->retries);
  1343. }
  1344. }
  1345. }
  1346. static gint
  1347. proxy_backend_master_finish_handler(struct rspamd_http_connection *conn,
  1348. struct rspamd_http_message *msg)
  1349. {
  1350. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1351. struct rspamd_proxy_session *session, *nsession;
  1352. rspamd_fstring_t *reply;
  1353. const rspamd_ftok_t *orig_ct;
  1354. goffset body_offset = -1;
  1355. session = bk_conn->s;
  1356. rspamd_http_connection_steal_msg(session->master_conn->backend_conn);
  1357. proxy_request_decompress(msg);
  1358. /*
  1359. * These are likely set by an http library, so we will double these headers
  1360. * if they are not removed
  1361. */
  1362. rspamd_http_message_remove_header(msg, "Content-Length");
  1363. rspamd_http_message_remove_header(msg, "Connection");
  1364. rspamd_http_message_remove_header(msg, "Date");
  1365. rspamd_http_message_remove_header(msg, "Server");
  1366. rspamd_http_message_remove_header(msg, "Key");
  1367. orig_ct = rspamd_http_message_find_header(msg, "Content-Type");
  1368. rspamd_http_connection_reset(session->master_conn->backend_conn);
  1369. if (!proxy_backend_parse_results(session, bk_conn, session->ctx->lua_state,
  1370. bk_conn->parser_from_ref, msg, &body_offset, orig_ct)) {
  1371. msg_warn_session("cannot parse results from the master backend");
  1372. }
  1373. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1374. /* We need to reformat ucl to fit with legacy spamc protocol */
  1375. if (bk_conn->results) {
  1376. reply = rspamd_fstring_new();
  1377. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1378. rspamd_ucl_tospamc_output(bk_conn->results, &reply);
  1379. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1380. }
  1381. else {
  1382. rspamd_ucl_torspamc_output(bk_conn->results, &reply);
  1383. }
  1384. rspamd_http_message_set_body_from_fstring_steal(msg, reply);
  1385. msg->method = HTTP_SYMBOLS;
  1386. }
  1387. else {
  1388. msg_warn_session("cannot parse results from the master backend, "
  1389. "return them as is");
  1390. }
  1391. }
  1392. rspamd_upstream_ok(bk_conn->up);
  1393. if (session->client_milter_conn) {
  1394. nsession = proxy_session_refresh(session);
  1395. if (body_offset > 0) {
  1396. rspamd_milter_send_task_results(nsession->client_milter_conn,
  1397. session->master_conn->results,
  1398. msg->body_buf.begin + body_offset,
  1399. msg->body_buf.len - body_offset);
  1400. }
  1401. else {
  1402. rspamd_milter_send_task_results(nsession->client_milter_conn,
  1403. session->master_conn->results, NULL, 0);
  1404. }
  1405. REF_RELEASE(session);
  1406. rspamd_http_message_free(msg);
  1407. }
  1408. else {
  1409. const gchar *passed_ct = NULL;
  1410. if (orig_ct) {
  1411. passed_ct = rspamd_mempool_ftokdup(session->pool, orig_ct);
  1412. /* Remove original */
  1413. rspamd_http_message_remove_header(msg, "Content-Type");
  1414. }
  1415. rspamd_http_connection_write_message(session->client_conn,
  1416. msg, NULL, passed_ct, session,
  1417. bk_conn->timeout);
  1418. }
  1419. return 0;
  1420. }
  1421. static void
  1422. rspamd_proxy_scan_self_reply(struct rspamd_task *task)
  1423. {
  1424. struct rspamd_http_message *msg;
  1425. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1426. ucl_object_t *rep = NULL;
  1427. const char *ctype = "application/json";
  1428. msg = rspamd_http_new_message(HTTP_RESPONSE);
  1429. msg->date = time(NULL);
  1430. msg->code = 200;
  1431. switch (task->cmd) {
  1432. case CMD_CHECK:
  1433. case CMD_SKIP:
  1434. case CMD_CHECK_RSPAMC:
  1435. case CMD_CHECK_SPAMC:
  1436. case CMD_CHECK_V2:
  1437. rspamd_task_set_finish_time(task);
  1438. rspamd_protocol_http_reply(msg, task, &rep);
  1439. rspamd_protocol_write_log_pipe(task);
  1440. break;
  1441. case CMD_PING:
  1442. rspamd_http_message_set_body(msg, "pong" CRLF, 6);
  1443. ctype = "text/plain";
  1444. break;
  1445. default:
  1446. msg_err_task("BROKEN");
  1447. break;
  1448. }
  1449. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1450. if (rep) {
  1451. session->master_conn->results = ucl_object_ref(rep);
  1452. }
  1453. if (session->client_milter_conn) {
  1454. nsession = proxy_session_refresh(session);
  1455. if (task->flags & RSPAMD_TASK_FLAG_MESSAGE_REWRITE) {
  1456. const gchar *start;
  1457. goffset len, hdr_off;
  1458. start = task->msg.begin;
  1459. len = task->msg.len;
  1460. hdr_off = MESSAGE_FIELD(task, raw_headers_content).len;
  1461. if (hdr_off < len) {
  1462. start += hdr_off;
  1463. len -= hdr_off;
  1464. /* The problem here is that we need not end of headers, we need
  1465. * start of body.
  1466. *
  1467. * Hence, we need to skip one \r\n till there is anything else in
  1468. * a line.
  1469. */
  1470. if (*start == '\r' && len > 0) {
  1471. start++;
  1472. len--;
  1473. }
  1474. if (*start == '\n' && len > 0) {
  1475. start++;
  1476. len--;
  1477. }
  1478. rspamd_milter_send_task_results(nsession->client_milter_conn,
  1479. session->master_conn->results, start, len);
  1480. }
  1481. else {
  1482. /* XXX: should never happen! */
  1483. rspamd_milter_send_task_results(nsession->client_milter_conn,
  1484. session->master_conn->results, NULL, 0);
  1485. }
  1486. }
  1487. else {
  1488. rspamd_milter_send_task_results(nsession->client_milter_conn,
  1489. session->master_conn->results, NULL, 0);
  1490. }
  1491. rspamd_http_message_free(msg);
  1492. REF_RELEASE(session);
  1493. }
  1494. else {
  1495. rspamd_http_connection_reset(session->client_conn);
  1496. rspamd_http_connection_write_message(session->client_conn,
  1497. msg,
  1498. NULL,
  1499. ctype,
  1500. session,
  1501. session->ctx->timeout / 10.0);
  1502. }
  1503. }
  1504. static gboolean
  1505. rspamd_proxy_task_fin(void *ud)
  1506. {
  1507. struct rspamd_task *task = ud;
  1508. msg_debug_task("finish task");
  1509. if (RSPAMD_TASK_IS_PROCESSED(task)) {
  1510. rspamd_proxy_scan_self_reply(task);
  1511. return TRUE;
  1512. }
  1513. if (!rspamd_task_process(task, RSPAMD_TASK_PROCESS_ALL)) {
  1514. rspamd_proxy_scan_self_reply(task);
  1515. return TRUE;
  1516. }
  1517. if (RSPAMD_TASK_IS_PROCESSED(task)) {
  1518. rspamd_proxy_scan_self_reply(task);
  1519. return TRUE;
  1520. }
  1521. /* One more iteration */
  1522. return FALSE;
  1523. }
  1524. static gboolean
  1525. rspamd_proxy_self_scan(struct rspamd_proxy_session *session)
  1526. {
  1527. struct rspamd_task *task;
  1528. struct rspamd_http_message *msg;
  1529. const gchar *data;
  1530. gsize len;
  1531. msg = session->client_message;
  1532. task = rspamd_task_new(session->worker, session->ctx->cfg,
  1533. session->pool, session->ctx->lang_det,
  1534. session->ctx->event_loop, FALSE);
  1535. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1536. if (session->ctx->milter) {
  1537. task->protocol_flags |= RSPAMD_TASK_PROTOCOL_FLAG_MILTER |
  1538. RSPAMD_TASK_PROTOCOL_FLAG_BODY_BLOCK;
  1539. }
  1540. task->sock = -1;
  1541. if (session->client_milter_conn) {
  1542. task->client_addr = rspamd_inet_address_copy(
  1543. session->client_milter_conn->addr, NULL);
  1544. }
  1545. else {
  1546. task->client_addr = rspamd_inet_address_copy(session->client_addr, NULL);
  1547. }
  1548. task->fin_arg = session;
  1549. task->resolver = session->ctx->resolver;
  1550. /* TODO: allow to disable autolearn in protocol */
  1551. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1552. task->s = rspamd_session_create(task->task_pool, rspamd_proxy_task_fin,
  1553. NULL, (event_finalizer_t) rspamd_task_free, task);
  1554. data = rspamd_http_message_get_body(msg, &len);
  1555. if (session->backend->settings_id) {
  1556. rspamd_http_message_remove_header(msg, "Settings-ID");
  1557. rspamd_http_message_add_header(msg, "Settings-ID",
  1558. session->backend->settings_id);
  1559. }
  1560. /* Process message */
  1561. if (!rspamd_protocol_handle_request(task, msg)) {
  1562. msg_err_task("cannot handle request: %e", task->err);
  1563. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1564. }
  1565. else {
  1566. if (task->cmd == CMD_PING) {
  1567. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1568. }
  1569. else {
  1570. if (!rspamd_task_load_message(task, msg, data, len)) {
  1571. msg_err_task("cannot load message: %e", task->err);
  1572. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1573. }
  1574. }
  1575. }
  1576. /* Set global timeout for the task */
  1577. if (session->ctx->default_upstream->timeout > 0.0) {
  1578. task->timeout_ev.data = task;
  1579. ev_timer_init(&task->timeout_ev, rspamd_task_timeout,
  1580. session->ctx->default_upstream->timeout,
  1581. session->ctx->default_upstream->timeout);
  1582. ev_timer_start(task->event_loop, &task->timeout_ev);
  1583. }
  1584. else if (session->ctx->has_self_scan) {
  1585. if (!isnan(session->ctx->task_timeout) && session->ctx->task_timeout > 0) {
  1586. task->timeout_ev.data = task;
  1587. ev_timer_init(&task->timeout_ev, rspamd_task_timeout,
  1588. session->ctx->cfg->task_timeout,
  1589. session->ctx->default_upstream->timeout);
  1590. ev_timer_start(task->event_loop, &task->timeout_ev);
  1591. }
  1592. }
  1593. session->master_conn->task = task;
  1594. rspamd_task_process(task, RSPAMD_TASK_PROCESS_ALL);
  1595. rspamd_session_pending(task->s);
  1596. return TRUE;
  1597. }
  1598. static gboolean
  1599. proxy_send_master_message(struct rspamd_proxy_session *session)
  1600. {
  1601. struct rspamd_http_message *msg;
  1602. struct rspamd_http_upstream *backend = NULL;
  1603. const rspamd_ftok_t *host;
  1604. GError *err = NULL;
  1605. gchar hostbuf[512];
  1606. host = rspamd_http_message_find_header(session->client_message, "Host");
  1607. if (host == NULL) {
  1608. backend = session->ctx->default_upstream;
  1609. }
  1610. else {
  1611. rspamd_strlcpy(hostbuf, host->begin, MIN(host->len + 1, sizeof(hostbuf)));
  1612. backend = g_hash_table_lookup(session->ctx->upstreams, hostbuf);
  1613. if (backend == NULL) {
  1614. backend = session->ctx->default_upstream;
  1615. }
  1616. }
  1617. if (backend == NULL) {
  1618. /* No backend */
  1619. msg_err_session("cannot find upstream for %s", host ? hostbuf : "default");
  1620. goto err;
  1621. }
  1622. else {
  1623. session->backend = backend;
  1624. if (backend->self_scan) {
  1625. return rspamd_proxy_self_scan(session);
  1626. }
  1627. retry:
  1628. if (session->ctx->max_retries &&
  1629. session->retries > session->ctx->max_retries) {
  1630. msg_err_session("cannot connect to upstream, maximum retries "
  1631. "has been reached: %d",
  1632. session->retries);
  1633. goto err;
  1634. }
  1635. /* Provide hash key if hashing based on source address is desired */
  1636. guint hash_len;
  1637. gpointer hash_key = rspamd_inet_address_get_hash_key(session->client_addr,
  1638. &hash_len);
  1639. if (session->ctx->max_retries > 1 &&
  1640. session->retries == session->ctx->max_retries) {
  1641. session->master_conn->up = rspamd_upstream_get_except(backend->u,
  1642. session->master_conn->up,
  1643. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1644. hash_key, hash_len);
  1645. }
  1646. else {
  1647. session->master_conn->up = rspamd_upstream_get(backend->u,
  1648. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1649. hash_key, hash_len);
  1650. }
  1651. session->master_conn->timeout = backend->timeout;
  1652. if (session->master_conn->up == NULL) {
  1653. msg_err_session("cannot select upstream for %s",
  1654. host ? hostbuf : "default");
  1655. goto err;
  1656. }
  1657. session->master_conn->backend_sock = rspamd_inet_address_connect(
  1658. rspamd_upstream_addr_next(session->master_conn->up),
  1659. SOCK_STREAM, TRUE);
  1660. if (session->master_conn->backend_sock == -1) {
  1661. msg_err_session("cannot connect upstream: %s(%s)",
  1662. host ? hostbuf : "default",
  1663. rspamd_inet_address_to_string_pretty(
  1664. rspamd_upstream_addr_cur(
  1665. session->master_conn->up)));
  1666. rspamd_upstream_fail(session->master_conn->up, TRUE,
  1667. strerror(errno));
  1668. session->retries++;
  1669. goto retry;
  1670. }
  1671. msg = rspamd_http_connection_copy_msg(session->client_message, &err);
  1672. if (msg == NULL) {
  1673. msg_err_session("cannot copy message to send it to the upstream: %e",
  1674. err);
  1675. if (err) {
  1676. g_error_free(err);
  1677. }
  1678. goto err; /* No fallback here */
  1679. }
  1680. session->master_conn->backend_conn = rspamd_http_connection_new_client_socket(
  1681. session->ctx->http_ctx,
  1682. NULL,
  1683. proxy_backend_master_error_handler,
  1684. proxy_backend_master_finish_handler,
  1685. RSPAMD_HTTP_CLIENT_SIMPLE,
  1686. session->master_conn->backend_sock);
  1687. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1688. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1689. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1690. if (backend->key) {
  1691. msg->peer_key = rspamd_pubkey_ref(backend->key);
  1692. }
  1693. if (backend->settings_id != NULL) {
  1694. rspamd_http_message_remove_header(msg, "Settings-ID");
  1695. rspamd_http_message_add_header(msg, "Settings-ID",
  1696. backend->settings_id);
  1697. }
  1698. if (backend->local ||
  1699. rspamd_inet_address_is_local(
  1700. rspamd_upstream_addr_cur(
  1701. session->master_conn->up))) {
  1702. if (session->fname) {
  1703. rspamd_http_message_add_header(msg, "File", session->fname);
  1704. }
  1705. msg->method = HTTP_GET;
  1706. rspamd_http_connection_write_message_shared(
  1707. session->master_conn->backend_conn,
  1708. msg, rspamd_upstream_name(session->master_conn->up),
  1709. NULL, session->master_conn,
  1710. session->master_conn->timeout);
  1711. }
  1712. else {
  1713. if (session->fname) {
  1714. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1715. rspamd_http_message_set_body(msg,
  1716. session->map, session->map_len);
  1717. }
  1718. msg->method = HTTP_POST;
  1719. if (backend->compress) {
  1720. proxy_request_compress(msg);
  1721. if (session->client_milter_conn) {
  1722. rspamd_http_message_add_header(msg, "Content-Type",
  1723. "application/octet-stream");
  1724. }
  1725. }
  1726. else {
  1727. if (session->client_milter_conn) {
  1728. rspamd_http_message_add_header(msg, "Content-Type",
  1729. "text/plain");
  1730. }
  1731. }
  1732. rspamd_http_connection_write_message(
  1733. session->master_conn->backend_conn,
  1734. msg, rspamd_upstream_name(session->master_conn->up),
  1735. NULL, session->master_conn,
  1736. session->master_conn->timeout);
  1737. }
  1738. }
  1739. return TRUE;
  1740. err:
  1741. if (session->client_milter_conn) {
  1742. rspamd_milter_send_action(session->client_milter_conn,
  1743. RSPAMD_MILTER_TEMPFAIL);
  1744. REF_RELEASE(session);
  1745. }
  1746. else {
  1747. rspamd_http_connection_steal_msg(session->client_conn);
  1748. rspamd_http_connection_reset(session->client_conn);
  1749. proxy_client_write_error(session, 404, "Backend not found");
  1750. }
  1751. return FALSE;
  1752. }
  1753. static void
  1754. proxy_client_error_handler(struct rspamd_http_connection *conn, GError *err)
  1755. {
  1756. struct rspamd_proxy_session *session = conn->ud;
  1757. msg_info_session("abnormally closing connection from: %s, error: %s",
  1758. rspamd_inet_address_to_string(session->client_addr), err->message);
  1759. /* Terminate session immediately */
  1760. proxy_backend_close_connection(session->master_conn);
  1761. REF_RELEASE(session);
  1762. }
  1763. static gint
  1764. proxy_client_finish_handler(struct rspamd_http_connection *conn,
  1765. struct rspamd_http_message *msg)
  1766. {
  1767. struct rspamd_proxy_session *session = conn->ud;
  1768. if (!session->master_conn) {
  1769. session->master_conn = rspamd_mempool_alloc0(session->pool,
  1770. sizeof(*session->master_conn));
  1771. session->master_conn->s = session;
  1772. session->master_conn->name = "master";
  1773. /* Reset spamc legacy */
  1774. if (msg->method >= HTTP_SYMBOLS) {
  1775. msg->method = HTTP_POST;
  1776. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1777. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1778. msg_info_session("enabling legacy spamc mode for session");
  1779. }
  1780. else {
  1781. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1782. msg_info_session("enabling legacy rspamc mode for session");
  1783. }
  1784. }
  1785. if (msg->url->len == 0) {
  1786. msg->url = rspamd_fstring_append(msg->url,
  1787. "/" MSG_CMD_CHECK_V2, strlen("/" MSG_CMD_CHECK_V2));
  1788. }
  1789. if (!proxy_check_file(msg, session)) {
  1790. goto err;
  1791. }
  1792. session->client_message = rspamd_http_connection_steal_msg(
  1793. session->client_conn);
  1794. session->shmem_ref = rspamd_http_message_shmem_ref(session->client_message);
  1795. rspamd_http_message_remove_header(msg, "Content-Length");
  1796. rspamd_http_message_remove_header(msg, "Transfer-Encoding");
  1797. rspamd_http_message_remove_header(msg, "Keep-Alive");
  1798. rspamd_http_message_remove_header(msg, "Connection");
  1799. rspamd_http_message_remove_header(msg, "Key");
  1800. proxy_open_mirror_connections(session);
  1801. rspamd_http_connection_reset(session->client_conn);
  1802. proxy_send_master_message(session);
  1803. }
  1804. else {
  1805. msg_info_session("finished master connection");
  1806. proxy_backend_close_connection(session->master_conn);
  1807. REF_RELEASE(session);
  1808. }
  1809. return 0;
  1810. err:
  1811. rspamd_http_connection_steal_msg(session->client_conn);
  1812. rspamd_http_message_remove_header(msg, "Content-Length");
  1813. rspamd_http_message_remove_header(msg, "Key");
  1814. rspamd_http_message_remove_header(msg, "Transfer-Encoding");
  1815. rspamd_http_message_remove_header(msg, "Keep-Alive");
  1816. rspamd_http_message_remove_header(msg, "Connection");
  1817. rspamd_http_connection_reset(session->client_conn);
  1818. proxy_client_write_error(session, 404, "Backend not found");
  1819. return 0;
  1820. }
  1821. static void
  1822. proxy_milter_finish_handler(gint fd,
  1823. struct rspamd_milter_session *rms,
  1824. void *ud)
  1825. {
  1826. struct rspamd_proxy_session *session = ud;
  1827. struct rspamd_http_message *msg;
  1828. session->client_milter_conn = rms;
  1829. if (rms->message == NULL || rms->message->len == 0) {
  1830. msg_info_session("finished milter connection");
  1831. proxy_backend_close_connection(session->master_conn);
  1832. REF_RELEASE(session);
  1833. }
  1834. else {
  1835. if (!session->master_conn) {
  1836. session->master_conn = rspamd_mempool_alloc0(session->pool,
  1837. sizeof(*session->master_conn));
  1838. }
  1839. msg = rspamd_milter_to_http(rms);
  1840. session->master_conn->s = session;
  1841. session->master_conn->name = "master";
  1842. session->client_message = msg;
  1843. proxy_open_mirror_connections(session);
  1844. proxy_send_master_message(session);
  1845. }
  1846. }
  1847. static void
  1848. proxy_milter_error_handler(gint fd,
  1849. struct rspamd_milter_session *rms, /* unused */
  1850. void *ud, GError *err)
  1851. {
  1852. struct rspamd_proxy_session *session = ud;
  1853. if (err && err->code != 0) {
  1854. msg_info_session("abnormally closing milter connection from: %s, "
  1855. "error: %e",
  1856. rspamd_inet_address_to_string_pretty(session->client_addr),
  1857. err);
  1858. /* Terminate session immediately */
  1859. proxy_backend_close_connection(session->master_conn);
  1860. REF_RELEASE(session);
  1861. }
  1862. else {
  1863. msg_info_session("normally closing milter connection from: %s, "
  1864. "%e",
  1865. rspamd_inet_address_to_string_pretty(session->client_addr),
  1866. err);
  1867. /* Terminate session immediately */
  1868. proxy_backend_close_connection(session->master_conn);
  1869. REF_RELEASE(session);
  1870. }
  1871. }
  1872. static void
  1873. proxy_accept_socket(EV_P_ ev_io *w, int revents)
  1874. {
  1875. struct rspamd_worker *worker = (struct rspamd_worker *) w->data;
  1876. struct rspamd_proxy_ctx *ctx;
  1877. rspamd_inet_addr_t *addr = NULL;
  1878. struct rspamd_proxy_session *session;
  1879. gint nfd;
  1880. ctx = worker->ctx;
  1881. if ((nfd =
  1882. rspamd_accept_from_socket(w->fd, &addr,
  1883. rspamd_worker_throttle_accept_events, worker->accept_events)) == -1) {
  1884. msg_warn("accept failed: %s", strerror(errno));
  1885. return;
  1886. }
  1887. /* Check for EAGAIN */
  1888. if (nfd == 0) {
  1889. rspamd_inet_address_free(addr);
  1890. return;
  1891. }
  1892. session = g_malloc0(sizeof(*session));
  1893. REF_INIT_RETAIN(session, proxy_session_dtor);
  1894. session->client_sock = nfd;
  1895. session->client_addr = addr;
  1896. session->mirror_conns = g_ptr_array_sized_new(ctx->mirrors->len);
  1897. session->pool = rspamd_mempool_new(rspamd_mempool_suggest_size(),
  1898. "proxy", 0);
  1899. session->ctx = ctx;
  1900. session->worker = worker;
  1901. if (ctx->sessions_cache) {
  1902. rspamd_worker_session_cache_add(ctx->sessions_cache,
  1903. session->pool->tag.uid, &session->ref.refcount, session);
  1904. }
  1905. if (!ctx->milter) {
  1906. session->client_conn = rspamd_http_connection_new_server(
  1907. ctx->http_ctx,
  1908. nfd,
  1909. NULL,
  1910. proxy_client_error_handler,
  1911. proxy_client_finish_handler,
  1912. 0);
  1913. if (ctx->key) {
  1914. rspamd_http_connection_set_key(session->client_conn, ctx->key);
  1915. }
  1916. msg_info_session("accepted http connection from %s port %d",
  1917. rspamd_inet_address_to_string(addr),
  1918. rspamd_inet_address_get_port(addr));
  1919. rspamd_http_connection_read_message_shared(session->client_conn,
  1920. session,
  1921. session->ctx->timeout);
  1922. }
  1923. else {
  1924. msg_info_session("accepted milter connection from %s port %d",
  1925. rspamd_inet_address_to_string(addr),
  1926. rspamd_inet_address_get_port(addr));
  1927. #ifdef TCP_NODELAY
  1928. #ifndef SOL_TCP
  1929. #define SOL_TCP IPPROTO_TCP
  1930. #endif
  1931. if (rspamd_inet_address_get_af(addr) != AF_UNIX) {
  1932. gint sopt = 1;
  1933. if (setsockopt(nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof(sopt)) ==
  1934. -1) {
  1935. msg_warn_session("cannot set TCP_NODELAY: %s",
  1936. strerror(errno));
  1937. }
  1938. }
  1939. #endif
  1940. rspamd_milter_handle_socket(nfd, 0.0,
  1941. session->pool,
  1942. ctx->event_loop,
  1943. proxy_milter_finish_handler,
  1944. proxy_milter_error_handler,
  1945. session);
  1946. }
  1947. }
  1948. static void
  1949. adjust_upstreams_limits(struct rspamd_proxy_ctx *ctx)
  1950. {
  1951. struct rspamd_http_upstream *backend;
  1952. gpointer k, v;
  1953. GHashTableIter it;
  1954. /*
  1955. * We set error time equal to max_retries * backend_timeout and max_errors
  1956. * to max_retries - 1
  1957. *
  1958. * So if we failed to scan a message on a backend for some reasons, we
  1959. * will try to re-resolve it faster
  1960. */
  1961. g_hash_table_iter_init(&it, ctx->upstreams);
  1962. while (g_hash_table_iter_next(&it, &k, &v)) {
  1963. backend = (struct rspamd_http_upstream *) v;
  1964. if (!backend->self_scan && backend->u) {
  1965. rspamd_upstreams_set_limits(backend->u,
  1966. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1967. ctx->max_retries - 1, 0);
  1968. }
  1969. }
  1970. }
  1971. __attribute__((noreturn)) void
  1972. start_rspamd_proxy(struct rspamd_worker *worker)
  1973. {
  1974. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1975. gboolean is_controller = FALSE;
  1976. g_assert(rspamd_worker_check_context(worker->ctx, rspamd_rspamd_proxy_magic));
  1977. ctx->cfg = worker->srv->cfg;
  1978. ctx->event_loop = rspamd_prepare_worker(worker, "rspamd_proxy",
  1979. proxy_accept_socket);
  1980. ctx->resolver = rspamd_dns_resolver_init(worker->srv->logger,
  1981. ctx->event_loop,
  1982. worker->srv->cfg);
  1983. rspamd_upstreams_library_config(worker->srv->cfg, ctx->cfg->ups_ctx,
  1984. ctx->event_loop, ctx->resolver->r);
  1985. ctx->http_ctx = rspamd_http_context_create(ctx->cfg, ctx->event_loop,
  1986. ctx->cfg->ups_ctx);
  1987. rspamd_mempool_add_destructor(ctx->cfg->cfg_pool,
  1988. (rspamd_mempool_destruct_t) rspamd_http_context_free,
  1989. ctx->http_ctx);
  1990. if (ctx->has_self_scan) {
  1991. /* Additional initialisation needed */
  1992. rspamd_worker_init_scanner(worker, ctx->event_loop, ctx->resolver,
  1993. &ctx->lang_det);
  1994. ctx->task_timeout = rspamd_worker_check_and_adjust_timeout(ctx->cfg, NAN);
  1995. is_controller = rspamd_worker_check_controller_presence(worker);
  1996. }
  1997. else {
  1998. worker->flags &= ~RSPAMD_WORKER_SCANNER;
  1999. }
  2000. if (worker->srv->cfg->enable_sessions_cache) {
  2001. ctx->sessions_cache = rspamd_worker_session_cache_new(worker,
  2002. ctx->event_loop);
  2003. }
  2004. ctx->milter_ctx.spam_header = ctx->spam_header;
  2005. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  2006. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  2007. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  2008. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  2009. ctx->milter_ctx.reject_message = ctx->reject_message;
  2010. ctx->milter_ctx.cfg = ctx->cfg;
  2011. rspamd_milter_init_library(&ctx->milter_ctx);
  2012. if (is_controller) {
  2013. rspamd_worker_init_controller(worker, NULL);
  2014. }
  2015. else {
  2016. if (ctx->has_self_scan) {
  2017. rspamd_map_watch(worker->srv->cfg, ctx->event_loop, ctx->resolver,
  2018. worker, RSPAMD_MAP_WATCH_SCANNER);
  2019. }
  2020. else {
  2021. rspamd_map_watch(worker->srv->cfg, ctx->event_loop, ctx->resolver,
  2022. worker, RSPAMD_MAP_WATCH_WORKER);
  2023. }
  2024. }
  2025. rspamd_lua_run_postloads(ctx->cfg->lua_state, ctx->cfg, ctx->event_loop,
  2026. worker);
  2027. adjust_upstreams_limits(ctx);
  2028. ev_loop(ctx->event_loop, 0);
  2029. rspamd_worker_block_signals();
  2030. if (ctx->has_self_scan) {
  2031. rspamd_stat_close();
  2032. }
  2033. if (is_controller) {
  2034. rspamd_controller_on_terminate(worker, NULL);
  2035. }
  2036. REF_RELEASE(ctx->cfg);
  2037. rspamd_log_close(worker->srv->logger);
  2038. rspamd_unset_crash_handler(worker->srv);
  2039. exit(EXIT_SUCCESS);
  2040. }