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 62KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427242824292430243124322433243424352436243724382439244024412442244324442445244624472448244924502451245224532454245524562457245824592460246124622463
  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. };
  78. struct rspamd_http_upstream {
  79. gchar *name;
  80. gchar *settings_id;
  81. struct upstream_list *u;
  82. struct rspamd_cryptobox_pubkey *key;
  83. gdouble timeout;
  84. gint parser_from_ref;
  85. gint parser_to_ref;
  86. gboolean local;
  87. gboolean self_scan;
  88. gboolean compress;
  89. };
  90. struct rspamd_http_mirror {
  91. gchar *name;
  92. gchar *settings_id;
  93. struct upstream_list *u;
  94. struct rspamd_cryptobox_pubkey *key;
  95. gdouble prob;
  96. gdouble timeout;
  97. gint parser_from_ref;
  98. gint parser_to_ref;
  99. gboolean local;
  100. gboolean compress;
  101. };
  102. static const guint64 rspamd_rspamd_proxy_magic = 0xcdeb4fd1fc351980ULL;
  103. struct rspamd_proxy_ctx {
  104. guint64 magic;
  105. /* Events base */
  106. struct ev_loop *event_loop;
  107. /* DNS resolver */
  108. struct rspamd_dns_resolver *resolver;
  109. /* Config */
  110. struct rspamd_config *cfg;
  111. /* END OF COMMON PART */
  112. gdouble timeout;
  113. /* Encryption key for clients */
  114. struct rspamd_cryptobox_keypair *key;
  115. /* HTTP context */
  116. struct rspamd_http_context *http_ctx;
  117. /* Upstreams to use */
  118. GHashTable *upstreams;
  119. /* Mirrors to send traffic to */
  120. GPtrArray *mirrors;
  121. /* Default upstream */
  122. struct rspamd_http_upstream *default_upstream;
  123. lua_State *lua_state;
  124. /* Array of callback functions called on end of scan to compare results */
  125. GArray *cmp_refs;
  126. /* Maximum count for retries */
  127. guint max_retries;
  128. /* If we have self_scanning backends, we need to work as a normal worker */
  129. gboolean has_self_scan;
  130. /* It is not HTTP but milter proxy */
  131. gboolean milter;
  132. /* Discard messages instead of rejecting them */
  133. gboolean discard_on_reject;
  134. /* Quarantine messages instead of rejecting them */
  135. gboolean quarantine_on_reject;
  136. /* Milter spam header */
  137. gchar *spam_header;
  138. /* CA name that can be used for client certificates */
  139. gchar *client_ca_name;
  140. /* Milter rejection message */
  141. gchar *reject_message;
  142. /* Sessions cache */
  143. void *sessions_cache;
  144. struct rspamd_milter_context milter_ctx;
  145. /* Language detector */
  146. struct rspamd_lang_detector *lang_det;
  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") ?
  307. ucl_object_tostring(ucl_object_lookup (obj, "name")) :
  308. ucl_object_key (obj));
  309. return TRUE;
  310. }
  311. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  312. elt = ucl_object_lookup (obj, "name");
  313. if (elt == NULL) {
  314. if (ucl_object_key (obj)) {
  315. if (strcmp (ucl_object_key (obj), "upstream") == 0) {
  316. /* Iterate over the object and find upstream elements */
  317. ucl_object_iter_t it = NULL;
  318. const ucl_object_t *cur;
  319. gboolean ret = TRUE;
  320. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  321. if (!rspamd_proxy_parse_upstream (pool, cur, ud,
  322. section, err)) {
  323. ret = FALSE;
  324. }
  325. }
  326. return ret;
  327. }
  328. else {
  329. /* Inside upstream */
  330. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  331. }
  332. }
  333. else {
  334. g_set_error (err, rspamd_proxy_quark (), 100,
  335. "upstream option must have some name definition");
  336. return FALSE;
  337. }
  338. }
  339. else {
  340. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  341. }
  342. up->parser_from_ref = -1;
  343. up->parser_to_ref = -1;
  344. up->timeout = ctx->timeout;
  345. elt = ucl_object_lookup (obj, "key");
  346. if (elt != NULL) {
  347. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  348. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  349. if (up->key == NULL) {
  350. g_set_error (err, rspamd_proxy_quark (), 100,
  351. "cannot read upstream key");
  352. goto err;
  353. }
  354. rspamd_mempool_add_destructor (pool,
  355. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  356. }
  357. elt = ucl_object_lookup (obj, "self_scan");
  358. if (elt && ucl_object_toboolean (elt)) {
  359. up->self_scan = TRUE;
  360. ctx->has_self_scan = TRUE;
  361. }
  362. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  363. if (elt && ucl_object_toboolean (elt)) {
  364. up->compress = TRUE;
  365. }
  366. elt = ucl_object_lookup (obj, "hosts");
  367. if (elt == NULL && !up->self_scan) {
  368. g_set_error (err, rspamd_proxy_quark (), 100,
  369. "upstream option must have some hosts definition");
  370. goto err;
  371. }
  372. if (elt) {
  373. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  374. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  375. g_set_error (err, rspamd_proxy_quark (), 100,
  376. "upstream has bad hosts definition");
  377. goto err;
  378. }
  379. rspamd_mempool_add_destructor (pool,
  380. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  381. }
  382. elt = ucl_object_lookup (obj, "default");
  383. if (elt) {
  384. if (ucl_object_toboolean (elt)) {
  385. ctx->default_upstream = up;
  386. }
  387. }
  388. else if (up->self_scan) {
  389. ctx->default_upstream = up;
  390. }
  391. elt = ucl_object_lookup (obj, "local");
  392. if (elt && ucl_object_toboolean (elt)) {
  393. up->local = TRUE;
  394. }
  395. elt = ucl_object_lookup (obj, "timeout");
  396. if (elt) {
  397. ucl_object_todouble_safe (elt, &up->timeout);
  398. }
  399. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  400. if (elt && ucl_object_type (elt) == UCL_STRING) {
  401. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  402. }
  403. /*
  404. * Accept lua function here in form
  405. * fun :: String -> UCL
  406. */
  407. elt = ucl_object_lookup (obj, "parser");
  408. if (elt) {
  409. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  410. &up->parser_to_ref, err)) {
  411. goto err;
  412. }
  413. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  414. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  415. }
  416. g_hash_table_insert (ctx->upstreams, up->name, up);
  417. return TRUE;
  418. err:
  419. return FALSE;
  420. }
  421. static gboolean
  422. rspamd_proxy_parse_mirror (rspamd_mempool_t *pool,
  423. const ucl_object_t *obj,
  424. gpointer ud,
  425. struct rspamd_rcl_section *section,
  426. GError **err)
  427. {
  428. const ucl_object_t *elt;
  429. struct rspamd_http_mirror *up = NULL;
  430. struct rspamd_proxy_ctx *ctx;
  431. struct rspamd_rcl_struct_parser *pd = ud;
  432. lua_State *L;
  433. ctx = pd->user_struct;
  434. L = ctx->lua_state;
  435. if (ucl_object_type (obj) != UCL_OBJECT) {
  436. if (ucl_object_type (obj) != UCL_NULL) {
  437. msg_info_pool_check("mirror %s is disabled by setting it to NULL",
  438. ucl_object_key (obj));
  439. return TRUE;
  440. }
  441. g_set_error (err, rspamd_proxy_quark (), 100,
  442. "mirror option must be an object");
  443. return FALSE;
  444. }
  445. if (!rspamd_config_is_enabled_from_ucl (pool, obj)) {
  446. /* Upstream is valid but disabled */
  447. msg_info_pool_check("mirror %s is disabled",
  448. ucl_object_lookup (obj, "name") ?
  449. ucl_object_tostring(ucl_object_lookup (obj, "name")) :
  450. ucl_object_key (obj));
  451. return TRUE;
  452. }
  453. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  454. elt = ucl_object_lookup (obj, "name");
  455. if (elt == NULL) {
  456. if (ucl_object_key (obj)) {
  457. if (strcmp (ucl_object_key (obj), "mirror") == 0) {
  458. /* Iterate over the object and find upstream elements */
  459. ucl_object_iter_t it = NULL;
  460. const ucl_object_t *cur;
  461. gboolean ret = TRUE;
  462. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  463. if (!rspamd_proxy_parse_mirror (pool, cur, ud,
  464. section, err)) {
  465. ret = FALSE;
  466. }
  467. }
  468. return ret;
  469. }
  470. else {
  471. /* Inside upstream */
  472. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  473. }
  474. }
  475. else {
  476. g_set_error (err, rspamd_proxy_quark (), 100,
  477. "mirror option must have some name definition");
  478. return FALSE;
  479. }
  480. }
  481. else {
  482. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  483. }
  484. up->parser_to_ref = -1;
  485. up->parser_from_ref = -1;
  486. up->timeout = ctx->timeout;
  487. elt = ucl_object_lookup (obj, "key");
  488. if (elt != NULL) {
  489. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  490. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  491. if (up->key == NULL) {
  492. g_set_error (err, rspamd_proxy_quark (), 100,
  493. "cannot read mirror key");
  494. goto err;
  495. }
  496. rspamd_mempool_add_destructor (pool,
  497. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  498. }
  499. elt = ucl_object_lookup (obj, "hosts");
  500. if (elt == NULL) {
  501. g_set_error (err, rspamd_proxy_quark (), 100,
  502. "mirror option must have some hosts definition");
  503. goto err;
  504. }
  505. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  506. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  507. g_set_error (err, rspamd_proxy_quark (), 100,
  508. "mirror has bad hosts definition");
  509. goto err;
  510. }
  511. rspamd_mempool_add_destructor (pool,
  512. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  513. elt = ucl_object_lookup_any (obj, "probability", "prob", NULL);
  514. if (elt) {
  515. up->prob = ucl_object_todouble (elt);
  516. }
  517. else {
  518. up->prob = 1.0;
  519. }
  520. elt = ucl_object_lookup (obj, "local");
  521. if (elt && ucl_object_toboolean (elt)) {
  522. up->local = TRUE;
  523. }
  524. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  525. if (elt && ucl_object_toboolean (elt)) {
  526. up->compress = TRUE;
  527. }
  528. elt = ucl_object_lookup (obj, "timeout");
  529. if (elt) {
  530. ucl_object_todouble_safe (elt, &up->timeout);
  531. }
  532. /*
  533. * Accept lua function here in form
  534. * fun :: String -> UCL
  535. */
  536. elt = ucl_object_lookup (obj, "parser");
  537. if (elt) {
  538. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  539. &up->parser_to_ref, err)) {
  540. goto err;
  541. }
  542. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  543. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  544. }
  545. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  546. if (elt && ucl_object_type (elt) == UCL_STRING) {
  547. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  548. }
  549. g_ptr_array_add (ctx->mirrors, up);
  550. return TRUE;
  551. err:
  552. return FALSE;
  553. }
  554. static gboolean
  555. rspamd_proxy_parse_script (rspamd_mempool_t *pool,
  556. const ucl_object_t *obj,
  557. gpointer ud,
  558. struct rspamd_rcl_section *section,
  559. GError **err)
  560. {
  561. struct rspamd_proxy_ctx *ctx;
  562. struct rspamd_rcl_struct_parser *pd = ud;
  563. lua_State *L;
  564. const gchar *lua_script;
  565. gsize slen;
  566. gint err_idx, ref_idx;
  567. struct stat st;
  568. ctx = pd->user_struct;
  569. L = ctx->lua_state;
  570. if (ucl_object_type (obj) != UCL_STRING) {
  571. g_set_error (err, rspamd_proxy_quark (), 100,
  572. "script option must be a string with file or lua chunk");
  573. return FALSE;
  574. }
  575. lua_script = ucl_object_tolstring (obj, &slen);
  576. lua_pushcfunction (L, &rspamd_lua_traceback);
  577. err_idx = lua_gettop (L);
  578. if (stat (lua_script, &st) != -1) {
  579. /* Load file */
  580. if (luaL_loadfile (L, lua_script) != 0) {
  581. g_set_error (err,
  582. rspamd_proxy_quark (),
  583. EINVAL,
  584. "cannot load lua parser script: %s",
  585. lua_tostring (L, -1));
  586. lua_settop (L, 0); /* Error function */
  587. goto err;
  588. }
  589. }
  590. else {
  591. /* Load data directly */
  592. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  593. g_set_error (err,
  594. rspamd_proxy_quark (),
  595. EINVAL,
  596. "cannot load lua parser script: %s",
  597. lua_tostring (L, -1));
  598. lua_settop (L, 0); /* Error function */
  599. goto err;
  600. }
  601. }
  602. /* Now do it */
  603. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  604. g_set_error (err,
  605. rspamd_proxy_quark (),
  606. EINVAL,
  607. "cannot init lua parser script: %s",
  608. lua_tostring (L, -1));
  609. lua_settop (L, 0);
  610. goto err;
  611. }
  612. if (!lua_isfunction (L, -1)) {
  613. g_set_error (err,
  614. rspamd_proxy_quark (),
  615. EINVAL,
  616. "cannot init lua parser script: "
  617. "must return function, %s returned",
  618. lua_typename (L, lua_type (L, -1)));
  619. lua_settop (L, 0);
  620. goto err;
  621. }
  622. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  623. lua_settop (L, 0);
  624. g_array_append_val (ctx->cmp_refs, ref_idx);
  625. return TRUE;
  626. err:
  627. return FALSE;
  628. }
  629. gpointer
  630. init_rspamd_proxy (struct rspamd_config *cfg)
  631. {
  632. struct rspamd_proxy_ctx *ctx;
  633. GQuark type;
  634. type = g_quark_try_string ("rspamd_proxy");
  635. ctx = rspamd_mempool_alloc0 (cfg->cfg_pool,
  636. sizeof (struct rspamd_proxy_ctx));
  637. ctx->magic = rspamd_rspamd_proxy_magic;
  638. ctx->timeout = 120.0;
  639. ctx->upstreams = g_hash_table_new (rspamd_strcase_hash, rspamd_strcase_equal);
  640. rspamd_mempool_add_destructor (cfg->cfg_pool,
  641. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->upstreams);
  642. ctx->mirrors = g_ptr_array_new ();
  643. rspamd_mempool_add_destructor (cfg->cfg_pool,
  644. (rspamd_mempool_destruct_t)rspamd_ptr_array_free_hard, ctx->mirrors);
  645. ctx->cfg = cfg;
  646. ctx->lua_state = cfg->lua_state;
  647. ctx->cmp_refs = g_array_new (FALSE, FALSE, sizeof (gint));
  648. rspamd_mempool_add_destructor (cfg->cfg_pool,
  649. (rspamd_mempool_destruct_t)rspamd_array_free_hard, ctx->cmp_refs);
  650. ctx->max_retries = DEFAULT_RETRIES;
  651. ctx->spam_header = RSPAMD_MILTER_SPAM_HEADER;
  652. rspamd_rcl_register_worker_option (cfg,
  653. type,
  654. "timeout",
  655. rspamd_rcl_parse_struct_time,
  656. ctx,
  657. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, timeout),
  658. RSPAMD_CL_FLAG_TIME_FLOAT,
  659. "IO timeout");
  660. rspamd_rcl_register_worker_option (cfg,
  661. type,
  662. "keypair",
  663. rspamd_rcl_parse_struct_keypair,
  664. ctx,
  665. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, key),
  666. 0,
  667. "Server's keypair");
  668. rspamd_rcl_register_worker_option (cfg,
  669. type,
  670. "upstream",
  671. rspamd_proxy_parse_upstream,
  672. ctx,
  673. 0,
  674. 0,
  675. "List of upstreams");
  676. rspamd_rcl_register_worker_option (cfg,
  677. type,
  678. "mirror",
  679. rspamd_proxy_parse_mirror,
  680. ctx,
  681. 0,
  682. RSPAMD_CL_FLAG_MULTIPLE,
  683. "List of mirrors");
  684. rspamd_rcl_register_worker_option (cfg,
  685. type,
  686. "script",
  687. rspamd_proxy_parse_script,
  688. ctx,
  689. 0,
  690. RSPAMD_CL_FLAG_MULTIPLE,
  691. "Compare script to be executed");
  692. rspamd_rcl_register_worker_option (cfg,
  693. type,
  694. "max_retries",
  695. rspamd_rcl_parse_struct_integer,
  696. ctx,
  697. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, max_retries),
  698. RSPAMD_CL_FLAG_UINT,
  699. "Maximum number of retries for master connection");
  700. rspamd_rcl_register_worker_option (cfg,
  701. type,
  702. "milter",
  703. rspamd_rcl_parse_struct_boolean,
  704. ctx,
  705. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, milter),
  706. 0,
  707. "Accept milter connections, not HTTP");
  708. rspamd_rcl_register_worker_option (cfg,
  709. type,
  710. "discard_on_reject",
  711. rspamd_rcl_parse_struct_boolean,
  712. ctx,
  713. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, discard_on_reject),
  714. 0,
  715. "Tell MTA to discard rejected messages silently");
  716. rspamd_rcl_register_worker_option (cfg,
  717. type,
  718. "quarantine_on_reject",
  719. rspamd_rcl_parse_struct_boolean,
  720. ctx,
  721. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, quarantine_on_reject),
  722. 0,
  723. "Tell MTA to quarantine rejected messages");
  724. rspamd_rcl_register_worker_option (cfg,
  725. type,
  726. "spam_header",
  727. rspamd_rcl_parse_struct_string,
  728. ctx,
  729. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, spam_header),
  730. 0,
  731. "Use the specific spam header (default: X-Spam)");
  732. rspamd_rcl_register_worker_option (cfg,
  733. type,
  734. "client_ca_name",
  735. rspamd_rcl_parse_struct_string,
  736. ctx,
  737. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, client_ca_name),
  738. 0,
  739. "Allow certificates issued by this CA to be treated as client certificates");
  740. rspamd_rcl_register_worker_option (cfg,
  741. type,
  742. "reject_message",
  743. rspamd_rcl_parse_struct_string,
  744. ctx,
  745. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, reject_message),
  746. 0,
  747. "Use custom rejection message");
  748. return ctx;
  749. }
  750. static void
  751. proxy_backend_close_connection (struct rspamd_proxy_backend_connection *conn)
  752. {
  753. if (conn && !(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  754. if (conn->backend_conn) {
  755. rspamd_http_connection_reset (conn->backend_conn);
  756. rspamd_http_connection_unref (conn->backend_conn);
  757. close (conn->backend_sock);
  758. }
  759. conn->flags |= RSPAMD_BACKEND_CLOSED;
  760. }
  761. }
  762. static gboolean
  763. proxy_backend_parse_results (struct rspamd_proxy_session *session,
  764. struct rspamd_proxy_backend_connection *conn,
  765. lua_State *L, gint parser_ref,
  766. struct rspamd_http_message *msg,
  767. goffset *body_offset,
  768. const rspamd_ftok_t *ct)
  769. {
  770. struct ucl_parser *parser;
  771. gint err_idx;
  772. const gchar *in = msg->body_buf.begin;
  773. gsize inlen = msg->body_buf.len;
  774. const rspamd_ftok_t *offset_hdr;
  775. if (inlen == 0 || in == NULL) {
  776. return FALSE;
  777. }
  778. offset_hdr = rspamd_http_message_find_header (msg, MESSAGE_OFFSET_HEADER);
  779. if (offset_hdr) {
  780. gulong val;
  781. if (rspamd_strtoul (offset_hdr->begin, offset_hdr->len, &val)
  782. && val < inlen) {
  783. if (body_offset) {
  784. *body_offset = val;
  785. }
  786. inlen = val;
  787. }
  788. }
  789. if (parser_ref != -1) {
  790. /* Call parser function */
  791. lua_pushcfunction (L, &rspamd_lua_traceback);
  792. err_idx = lua_gettop (L);
  793. lua_rawgeti (L, LUA_REGISTRYINDEX, parser_ref);
  794. /* XXX: copies all data */
  795. lua_pushlstring (L, in, inlen);
  796. if (lua_pcall (L, 1, 1, err_idx) != 0) {
  797. msg_err_session (
  798. "cannot run lua parser script: %s",
  799. lua_tostring (L, -1));
  800. lua_settop (L, 0);
  801. return FALSE;
  802. }
  803. conn->results = ucl_object_lua_import (L, -1);
  804. lua_settop (L, 0);
  805. }
  806. else {
  807. rspamd_ftok_t json_ct;
  808. RSPAMD_FTOK_ASSIGN (&json_ct, "application/json");
  809. if (ct && rspamd_ftok_casecmp (ct, &json_ct) == 0) {
  810. parser = ucl_parser_new (0);
  811. if (!ucl_parser_add_chunk (parser, in, inlen)) {
  812. gchar *encoded;
  813. encoded = rspamd_encode_base64 (in, inlen, 0, NULL);
  814. msg_err_session ("cannot parse input: %s", ucl_parser_get_error (
  815. parser));
  816. msg_err_session ("input encoded: %s", encoded);
  817. ucl_parser_free (parser);
  818. g_free (encoded);
  819. return FALSE;
  820. }
  821. conn->results = ucl_parser_get_object (parser);
  822. ucl_parser_free (parser);
  823. }
  824. }
  825. return TRUE;
  826. }
  827. static void
  828. proxy_call_cmp_script (struct rspamd_proxy_session *session, gint cbref)
  829. {
  830. gint err_idx;
  831. guint i;
  832. struct rspamd_proxy_backend_connection *conn;
  833. lua_State *L;
  834. L = session->ctx->lua_state;
  835. lua_pushcfunction (L, &rspamd_lua_traceback);
  836. err_idx = lua_gettop (L);
  837. lua_rawgeti (L, LUA_REGISTRYINDEX, cbref);
  838. lua_createtable (L, 0, session->mirror_conns->len + 1);
  839. /* Now push master results */
  840. if (session->master_conn && session->master_conn->results) {
  841. lua_pushstring (L, "master");
  842. ucl_object_push_lua (L, session->master_conn->results, true);
  843. lua_settable (L, -3);
  844. }
  845. else {
  846. lua_pushstring (L, "master");
  847. lua_pushstring (L, "no results");
  848. lua_settable (L, -3);
  849. }
  850. for (i = 0; i < session->mirror_conns->len; i ++) {
  851. conn = g_ptr_array_index (session->mirror_conns, i);
  852. if (conn->results) {
  853. lua_pushstring (L, conn->name);
  854. ucl_object_push_lua (L, conn->results, true);
  855. lua_settable (L, -3);
  856. }
  857. else {
  858. lua_pushstring (L, conn->name);
  859. lua_pushstring (L, conn->err ? conn->err : "unknown error");
  860. lua_settable (L, -3);
  861. }
  862. }
  863. gchar log_tag[RSPAMD_LOG_ID_LEN + 1];
  864. rspamd_strlcpy(log_tag, session->pool->tag.uid, sizeof (session->pool->tag.uid));
  865. lua_pushstring(L, log_tag);
  866. if (lua_pcall (L, 2, 0, err_idx) != 0) {
  867. msg_err_session (
  868. "cannot run lua compare script: %s",
  869. lua_tostring (L, -1));
  870. }
  871. lua_settop (L, err_idx - 1);
  872. }
  873. static void
  874. proxy_session_dtor (struct rspamd_proxy_session *session)
  875. {
  876. guint i;
  877. gint cbref;
  878. struct rspamd_proxy_backend_connection *conn;
  879. if (session->master_conn && session->master_conn->results) {
  880. for (i = 0; i < session->ctx->cmp_refs->len; i++) {
  881. cbref = g_array_index (session->ctx->cmp_refs, gint, i);
  882. proxy_call_cmp_script (session, cbref);
  883. }
  884. }
  885. if (session->master_conn) {
  886. proxy_backend_close_connection (session->master_conn);
  887. }
  888. if (session->client_milter_conn) {
  889. rspamd_milter_session_unref (session->client_milter_conn);
  890. }
  891. else if (session->client_conn) {
  892. rspamd_http_connection_reset (session->client_conn);
  893. rspamd_http_connection_unref (session->client_conn);
  894. }
  895. if (session->map && session->map_len) {
  896. munmap (session->map, session->map_len);
  897. }
  898. for (i = 0; i < session->mirror_conns->len; i ++) {
  899. conn = g_ptr_array_index (session->mirror_conns, i);
  900. if (!(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  901. proxy_backend_close_connection (conn);
  902. }
  903. if (conn->results) {
  904. ucl_object_unref (conn->results);
  905. }
  906. }
  907. if (session->master_conn) {
  908. if (session->master_conn->results) {
  909. ucl_object_unref (session->master_conn->results);
  910. }
  911. if (session->master_conn->task) {
  912. rspamd_session_destroy (session->master_conn->task->s);
  913. }
  914. }
  915. g_ptr_array_free (session->mirror_conns, TRUE);
  916. rspamd_http_message_shmem_unref (session->shmem_ref);
  917. rspamd_http_message_unref (session->client_message);
  918. if (session->client_addr) {
  919. rspamd_inet_address_free (session->client_addr);
  920. }
  921. if (session->client_sock != -1) {
  922. close (session->client_sock);
  923. }
  924. if (session->ctx->sessions_cache) {
  925. rspamd_worker_session_cache_remove (session->ctx->sessions_cache,
  926. session);
  927. }
  928. if (session->pool) {
  929. rspamd_mempool_delete (session->pool);
  930. }
  931. g_free (session);
  932. }
  933. static void
  934. proxy_request_compress (struct rspamd_http_message *msg)
  935. {
  936. guint flags;
  937. ZSTD_CCtx *zctx;
  938. rspamd_fstring_t *body;
  939. const gchar *in;
  940. gsize inlen;
  941. flags = rspamd_http_message_get_flags (msg);
  942. if (!rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  943. if ((flags & RSPAMD_HTTP_FLAG_SHMEM) ||
  944. !(flags & RSPAMD_HTTP_FLAG_HAS_BODY)) {
  945. /* Cannot compress shared or empty message */
  946. return;
  947. }
  948. in = rspamd_http_message_get_body (msg, &inlen);
  949. if (in == NULL || inlen == 0) {
  950. return;
  951. }
  952. body = rspamd_fstring_sized_new (ZSTD_compressBound (inlen));
  953. zctx = ZSTD_createCCtx ();
  954. body->len = ZSTD_compressCCtx (zctx, body->str, body->allocated,
  955. in, inlen, 1);
  956. if (ZSTD_isError (body->len)) {
  957. msg_err ("compression error");
  958. rspamd_fstring_free (body);
  959. ZSTD_freeCCtx (zctx);
  960. return;
  961. }
  962. ZSTD_freeCCtx (zctx);
  963. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  964. rspamd_http_message_add_header (msg, COMPRESSION_HEADER, "zstd");
  965. }
  966. }
  967. static void
  968. proxy_request_decompress (struct rspamd_http_message *msg)
  969. {
  970. rspamd_fstring_t *body;
  971. const gchar *in;
  972. gsize inlen, outlen, r;
  973. ZSTD_DStream *zstream;
  974. ZSTD_inBuffer zin;
  975. ZSTD_outBuffer zout;
  976. if (rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  977. in = rspamd_http_message_get_body (msg, &inlen);
  978. if (in == NULL || inlen == 0) {
  979. return;
  980. }
  981. zstream = ZSTD_createDStream ();
  982. ZSTD_initDStream (zstream);
  983. zin.pos = 0;
  984. zin.src = in;
  985. zin.size = inlen;
  986. if ((outlen = ZSTD_getDecompressedSize (zin.src, zin.size)) == 0) {
  987. outlen = ZSTD_DStreamOutSize ();
  988. }
  989. body = rspamd_fstring_sized_new (outlen);
  990. zout.dst = body->str;
  991. zout.pos = 0;
  992. zout.size = outlen;
  993. while (zin.pos < zin.size) {
  994. r = ZSTD_decompressStream (zstream, &zout, &zin);
  995. if (ZSTD_isError (r)) {
  996. msg_err ("Decompression error: %s", ZSTD_getErrorName (r));
  997. ZSTD_freeDStream (zstream);
  998. rspamd_fstring_free (body);
  999. return;
  1000. }
  1001. if (zout.pos == zout.size) {
  1002. /* We need to extend output buffer */
  1003. zout.size = zout.size * 2 + 1;
  1004. body = rspamd_fstring_grow (body, zout.size);
  1005. zout.size = body->allocated;
  1006. zout.dst = body->str;
  1007. }
  1008. }
  1009. body->len = zout.pos;
  1010. ZSTD_freeDStream (zstream);
  1011. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  1012. rspamd_http_message_remove_header (msg, COMPRESSION_HEADER);
  1013. }
  1014. }
  1015. static struct rspamd_proxy_session *
  1016. proxy_session_refresh (struct rspamd_proxy_session *session)
  1017. {
  1018. struct rspamd_proxy_session *nsession;
  1019. nsession = g_malloc0 (sizeof (*nsession));
  1020. nsession->client_milter_conn = session->client_milter_conn;
  1021. session->client_milter_conn = NULL;
  1022. rspamd_milter_update_userdata (nsession->client_milter_conn,
  1023. nsession);
  1024. nsession->client_addr = session->client_addr;
  1025. session->client_addr = NULL;
  1026. nsession->ctx = session->ctx;
  1027. nsession->worker = session->worker;
  1028. nsession->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (), "proxy", 0);
  1029. nsession->client_sock = session->client_sock;
  1030. session->client_sock = -1;
  1031. nsession->mirror_conns = g_ptr_array_sized_new (nsession->ctx->mirrors->len);
  1032. REF_INIT_RETAIN (nsession, proxy_session_dtor);
  1033. if (nsession->ctx->sessions_cache) {
  1034. rspamd_worker_session_cache_add (nsession->ctx->sessions_cache,
  1035. nsession->pool->tag.uid, &nsession->ref.refcount, nsession);
  1036. }
  1037. return nsession;
  1038. }
  1039. static gboolean
  1040. proxy_check_file (struct rspamd_http_message *msg,
  1041. struct rspamd_proxy_session *session)
  1042. {
  1043. const rspamd_ftok_t *tok, *key_tok;
  1044. rspamd_ftok_t srch;
  1045. gchar *file_str;
  1046. GHashTable *query_args;
  1047. GHashTableIter it;
  1048. gpointer k, v;
  1049. struct http_parser_url u;
  1050. rspamd_fstring_t *new_url;
  1051. tok = rspamd_http_message_find_header (msg, "File");
  1052. if (tok) {
  1053. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1054. session->map = rspamd_file_xmap (file_str, PROT_READ, &session->map_len,
  1055. TRUE);
  1056. if (session->map == NULL) {
  1057. if (session->map_len != 0) {
  1058. msg_err_session ("cannot map %s: %s", file_str,
  1059. strerror (errno));
  1060. return FALSE;
  1061. }
  1062. }
  1063. /* Remove header after processing */
  1064. rspamd_http_message_remove_header (msg, "File");
  1065. session->fname = file_str;
  1066. }
  1067. else {
  1068. /* Need to parse query URL */
  1069. if (http_parser_parse_url (RSPAMD_FSTRING_DATA (msg->url),
  1070. RSPAMD_FSTRING_LEN (msg->url), 0, &u) != 0) {
  1071. msg_err_session ("bad request url: %V", msg->url);
  1072. return FALSE;
  1073. }
  1074. if (u.field_set & (1 << UF_QUERY)) {
  1075. /* In case if we have a query, we need to store it somewhere */
  1076. query_args = rspamd_http_message_parse_query (msg);
  1077. srch.begin = "File";
  1078. srch.len = strlen ("File");
  1079. tok = g_hash_table_lookup (query_args, &srch);
  1080. if (tok) {
  1081. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1082. session->map = rspamd_file_xmap (file_str, PROT_READ,
  1083. &session->map_len, TRUE);
  1084. if (session->map == NULL) {
  1085. if (session->map_len != 0) {
  1086. msg_err_session ("cannot map %s: %s", file_str,
  1087. strerror (errno));
  1088. g_hash_table_unref (query_args);
  1089. return FALSE;
  1090. }
  1091. }
  1092. /* We need to create a new URL with file attribute removed */
  1093. new_url = rspamd_fstring_new_init (RSPAMD_FSTRING_DATA (msg->url),
  1094. u.field_data[UF_QUERY].off);
  1095. new_url = rspamd_fstring_append (new_url, "?", 1);
  1096. g_hash_table_iter_init (&it, query_args);
  1097. while (g_hash_table_iter_next (&it, &k, &v)) {
  1098. key_tok = k;
  1099. tok = v;
  1100. if (!rspamd_ftok_icase_equal (key_tok, &srch)) {
  1101. rspamd_printf_fstring (&new_url, "%T=%T&",
  1102. key_tok, tok);
  1103. }
  1104. }
  1105. /* Erase last character (might be either & or ?) */
  1106. rspamd_fstring_erase (new_url, new_url->len - 1, 1);
  1107. rspamd_fstring_free (msg->url);
  1108. msg->url = new_url;
  1109. session->fname = file_str;
  1110. }
  1111. g_hash_table_unref (query_args);
  1112. }
  1113. }
  1114. return TRUE;
  1115. }
  1116. static void
  1117. proxy_backend_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  1118. {
  1119. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1120. struct rspamd_proxy_session *session;
  1121. session = bk_conn->s;
  1122. msg_info_session ("abnormally closing connection from backend: %s:%s, "
  1123. "error: %e",
  1124. bk_conn->name,
  1125. rspamd_inet_address_to_string_pretty (
  1126. rspamd_upstream_addr_cur (bk_conn->up)),
  1127. err);
  1128. if (err) {
  1129. bk_conn->err = rspamd_mempool_strdup (session->pool, err->message);
  1130. }
  1131. rspamd_upstream_fail (bk_conn->up, FALSE, err ? err->message : "unknown");
  1132. proxy_backend_close_connection (bk_conn);
  1133. REF_RELEASE (bk_conn->s);
  1134. }
  1135. static gint
  1136. proxy_backend_mirror_finish_handler (struct rspamd_http_connection *conn,
  1137. struct rspamd_http_message *msg)
  1138. {
  1139. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1140. struct rspamd_proxy_session *session;
  1141. const rspamd_ftok_t *orig_ct;
  1142. session = bk_conn->s;
  1143. proxy_request_decompress (msg);
  1144. orig_ct = rspamd_http_message_find_header (msg, "Content-Type");
  1145. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1146. bk_conn->parser_from_ref, msg, NULL, orig_ct)) {
  1147. msg_warn_session ("cannot parse results from the mirror backend %s:%s",
  1148. bk_conn->name,
  1149. rspamd_inet_address_to_string (
  1150. rspamd_upstream_addr_cur (bk_conn->up)));
  1151. bk_conn->err = "cannot parse ucl";
  1152. }
  1153. msg_info_session ("finished mirror connection to %s", bk_conn->name);
  1154. rspamd_upstream_ok (bk_conn->up);
  1155. proxy_backend_close_connection (bk_conn);
  1156. REF_RELEASE (bk_conn->s);
  1157. return 0;
  1158. }
  1159. static void
  1160. proxy_open_mirror_connections (struct rspamd_proxy_session *session)
  1161. {
  1162. gdouble coin;
  1163. struct rspamd_http_mirror *m;
  1164. guint i;
  1165. struct rspamd_proxy_backend_connection *bk_conn;
  1166. struct rspamd_http_message *msg;
  1167. GError *err = NULL;
  1168. coin = rspamd_random_double ();
  1169. for (i = 0; i < session->ctx->mirrors->len; i ++) {
  1170. m = g_ptr_array_index (session->ctx->mirrors, i);
  1171. if (m->prob < coin) {
  1172. /* No luck */
  1173. continue;
  1174. }
  1175. bk_conn = rspamd_mempool_alloc0 (session->pool,
  1176. sizeof (*bk_conn));
  1177. bk_conn->s = session;
  1178. bk_conn->name = m->name;
  1179. bk_conn->timeout = m->timeout;
  1180. bk_conn->up = rspamd_upstream_get (m->u,
  1181. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1182. bk_conn->parser_from_ref = m->parser_from_ref;
  1183. bk_conn->parser_to_ref = m->parser_to_ref;
  1184. if (bk_conn->up == NULL) {
  1185. msg_err_session ("cannot select upstream for %s", m->name);
  1186. continue;
  1187. }
  1188. bk_conn->backend_sock = rspamd_inet_address_connect (
  1189. rspamd_upstream_addr_next (bk_conn->up),
  1190. SOCK_STREAM, TRUE);
  1191. if (bk_conn->backend_sock == -1) {
  1192. msg_err_session ("cannot connect upstream for %s", m->name);
  1193. rspamd_upstream_fail (bk_conn->up, TRUE, strerror (errno));
  1194. continue;
  1195. }
  1196. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1197. if (msg == NULL) {
  1198. msg_err_session ("cannot copy message to send to a mirror %s: %e",
  1199. m->name, err);
  1200. if (err) {
  1201. g_error_free (err);
  1202. }
  1203. continue;
  1204. }
  1205. if (msg->url->len == 0) {
  1206. msg->url = rspamd_fstring_append (msg->url, "/check", strlen ("/check"));
  1207. }
  1208. if (m->settings_id != NULL) {
  1209. rspamd_http_message_remove_header (msg, "Settings-ID");
  1210. rspamd_http_message_add_header (msg, "Settings-ID", m->settings_id);
  1211. }
  1212. bk_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1213. session->ctx->http_ctx,
  1214. NULL,
  1215. proxy_backend_mirror_error_handler,
  1216. proxy_backend_mirror_finish_handler,
  1217. RSPAMD_HTTP_CLIENT_SIMPLE,
  1218. bk_conn->backend_sock);
  1219. if (m->key) {
  1220. msg->peer_key = rspamd_pubkey_ref (m->key);
  1221. }
  1222. if (m->local ||
  1223. rspamd_inet_address_is_local (rspamd_upstream_addr_cur (bk_conn->up))) {
  1224. if (session->fname) {
  1225. rspamd_http_message_add_header (msg, "File", session->fname);
  1226. }
  1227. msg->method = HTTP_GET;
  1228. rspamd_http_connection_write_message_shared (bk_conn->backend_conn,
  1229. msg, rspamd_upstream_name(bk_conn->up), NULL, bk_conn,
  1230. bk_conn->timeout);
  1231. }
  1232. else {
  1233. if (session->fname) {
  1234. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1235. rspamd_http_message_set_body (msg, session->map, session->map_len);
  1236. }
  1237. msg->method = HTTP_POST;
  1238. if (m->compress) {
  1239. proxy_request_compress (msg);
  1240. if (session->client_milter_conn) {
  1241. rspamd_http_message_add_header (msg, "Content-Type",
  1242. "application/octet-stream");
  1243. }
  1244. }
  1245. else {
  1246. if (session->client_milter_conn) {
  1247. rspamd_http_message_add_header (msg, "Content-Type",
  1248. "text/plain");
  1249. }
  1250. }
  1251. rspamd_http_connection_write_message (bk_conn->backend_conn,
  1252. msg, rspamd_upstream_name(bk_conn->up), NULL, bk_conn,
  1253. bk_conn->timeout);
  1254. }
  1255. g_ptr_array_add (session->mirror_conns, bk_conn);
  1256. REF_RETAIN (session);
  1257. msg_info_session ("send request to %s", m->name);
  1258. }
  1259. }
  1260. static void
  1261. proxy_client_write_error (struct rspamd_proxy_session *session, gint code,
  1262. const gchar *status)
  1263. {
  1264. struct rspamd_http_message *reply;
  1265. if (session->client_milter_conn) {
  1266. rspamd_milter_send_action (session->client_milter_conn,
  1267. RSPAMD_MILTER_TEMPFAIL);
  1268. REF_RELEASE (session);
  1269. }
  1270. else {
  1271. reply = rspamd_http_new_message (HTTP_RESPONSE);
  1272. switch (code) {
  1273. case ETIMEDOUT:
  1274. reply->code = 504;
  1275. reply->status = RSPAMD_FSTRING_LIT ("Gateway timeout");
  1276. break;
  1277. case ECONNRESET:
  1278. case ECONNABORTED:
  1279. reply->code = 502;
  1280. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection reset");
  1281. break;
  1282. case ECONNREFUSED:
  1283. reply->code = 502;
  1284. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection refused");
  1285. break;
  1286. default:
  1287. if (code >= 300) {
  1288. /* Likely HTTP error */
  1289. reply->code = code;
  1290. reply->status = rspamd_fstring_new_init (status, strlen (status));
  1291. }
  1292. else {
  1293. reply->code = 502;
  1294. reply->status = RSPAMD_FSTRING_LIT ("Unknown gateway error: ");
  1295. reply->status = rspamd_fstring_append (reply->status,
  1296. status, strlen (status));
  1297. }
  1298. break;
  1299. }
  1300. rspamd_http_connection_write_message (session->client_conn,
  1301. reply, NULL, NULL, session,
  1302. session->ctx->timeout);
  1303. }
  1304. }
  1305. static void
  1306. proxy_backend_master_error_handler (struct rspamd_http_connection *conn, GError *err)
  1307. {
  1308. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1309. struct rspamd_proxy_session *session;
  1310. session = bk_conn->s;
  1311. session->retries ++;
  1312. msg_info_session ("abnormally closing connection from backend: %s, error: %e,"
  1313. " retries left: %d",
  1314. rspamd_inet_address_to_string_pretty (
  1315. rspamd_upstream_addr_cur (session->master_conn->up)),
  1316. err,
  1317. session->ctx->max_retries - session->retries);
  1318. rspamd_upstream_fail (bk_conn->up, FALSE, err ? err->message : "unknown");
  1319. proxy_backend_close_connection (session->master_conn);
  1320. if (session->ctx->max_retries > 0 &&
  1321. session->retries >= session->ctx->max_retries) {
  1322. msg_err_session ("cannot connect to upstream, maximum retries "
  1323. "has been reached: %d", session->retries);
  1324. /* Terminate session immediately */
  1325. if (err) {
  1326. proxy_client_write_error(session, err->code, err->message);
  1327. }
  1328. else {
  1329. proxy_client_write_error(session, 503, "Unknown error after no retries left");
  1330. }
  1331. }
  1332. else {
  1333. if (!proxy_send_master_message (session)) {
  1334. if (err) {
  1335. proxy_client_write_error(session, err->code, err->message);
  1336. }
  1337. else {
  1338. proxy_client_write_error(session, 503, "Unknown error on write");
  1339. }
  1340. }
  1341. else {
  1342. msg_info_session ("retry connection to: %s"
  1343. " retries left: %d",
  1344. rspamd_inet_address_to_string (
  1345. rspamd_upstream_addr_cur (session->master_conn->up)),
  1346. session->ctx->max_retries - session->retries);
  1347. }
  1348. }
  1349. }
  1350. static gint
  1351. proxy_backend_master_finish_handler (struct rspamd_http_connection *conn,
  1352. struct rspamd_http_message *msg)
  1353. {
  1354. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1355. struct rspamd_proxy_session *session, *nsession;
  1356. rspamd_fstring_t *reply;
  1357. const rspamd_ftok_t *orig_ct;
  1358. goffset body_offset = -1;
  1359. session = bk_conn->s;
  1360. rspamd_http_connection_steal_msg (session->master_conn->backend_conn);
  1361. proxy_request_decompress (msg);
  1362. /*
  1363. * These are likely set by an http library, so we will double these headers
  1364. * if they are not removed
  1365. */
  1366. rspamd_http_message_remove_header (msg, "Content-Length");
  1367. rspamd_http_message_remove_header (msg, "Connection");
  1368. rspamd_http_message_remove_header (msg, "Date");
  1369. rspamd_http_message_remove_header (msg, "Server");
  1370. rspamd_http_message_remove_header (msg, "Key");
  1371. orig_ct = rspamd_http_message_find_header (msg, "Content-Type");
  1372. rspamd_http_connection_reset (session->master_conn->backend_conn);
  1373. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1374. bk_conn->parser_from_ref, msg, &body_offset, orig_ct)) {
  1375. msg_warn_session ("cannot parse results from the master backend");
  1376. }
  1377. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1378. /* We need to reformat ucl to fit with legacy spamc protocol */
  1379. if (bk_conn->results) {
  1380. reply = rspamd_fstring_new ();
  1381. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1382. rspamd_ucl_tospamc_output (bk_conn->results, &reply);
  1383. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1384. }
  1385. else {
  1386. rspamd_ucl_torspamc_output (bk_conn->results, &reply);
  1387. }
  1388. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1389. msg->method = HTTP_SYMBOLS;
  1390. }
  1391. else {
  1392. msg_warn_session ("cannot parse results from the master backend, "
  1393. "return them as is");
  1394. }
  1395. }
  1396. rspamd_upstream_ok (bk_conn->up);
  1397. if (session->client_milter_conn) {
  1398. nsession = proxy_session_refresh (session);
  1399. if (body_offset > 0) {
  1400. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1401. session->master_conn->results,
  1402. msg->body_buf.begin + body_offset,
  1403. msg->body_buf.len - body_offset);
  1404. }
  1405. else {
  1406. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1407. session->master_conn->results, NULL, 0);
  1408. }
  1409. REF_RELEASE (session);
  1410. rspamd_http_message_free (msg);
  1411. }
  1412. else {
  1413. const gchar *passed_ct = NULL;
  1414. if (orig_ct) {
  1415. passed_ct = rspamd_mempool_ftokdup (session->pool, orig_ct);
  1416. /* Remove original */
  1417. rspamd_http_message_remove_header (msg, "Content-Type");
  1418. }
  1419. rspamd_http_connection_write_message (session->client_conn,
  1420. msg, NULL, passed_ct, session,
  1421. bk_conn->timeout);
  1422. }
  1423. return 0;
  1424. }
  1425. static void
  1426. rspamd_proxy_scan_self_reply (struct rspamd_task *task)
  1427. {
  1428. struct rspamd_http_message *msg;
  1429. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1430. ucl_object_t *rep = NULL;
  1431. const char *ctype = "application/json";
  1432. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1433. msg->date = time (NULL);
  1434. msg->code = 200;
  1435. switch (task->cmd) {
  1436. case CMD_CHECK:
  1437. case CMD_SKIP:
  1438. case CMD_CHECK_RSPAMC:
  1439. case CMD_CHECK_SPAMC:
  1440. case CMD_CHECK_V2:
  1441. rspamd_task_set_finish_time (task);
  1442. rspamd_protocol_http_reply (msg, task, &rep);
  1443. rspamd_protocol_write_log_pipe (task);
  1444. break;
  1445. case CMD_PING:
  1446. rspamd_http_message_set_body (msg, "pong" CRLF, 6);
  1447. ctype = "text/plain";
  1448. break;
  1449. default:
  1450. msg_err_task ("BROKEN");
  1451. break;
  1452. }
  1453. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1454. if (rep) {
  1455. session->master_conn->results = ucl_object_ref (rep);
  1456. }
  1457. if (session->client_milter_conn) {
  1458. nsession = proxy_session_refresh (session);
  1459. if (task->flags & RSPAMD_TASK_FLAG_MESSAGE_REWRITE) {
  1460. const gchar *start;
  1461. goffset len, hdr_off;
  1462. start = task->msg.begin;
  1463. len = task->msg.len;
  1464. hdr_off = MESSAGE_FIELD (task, raw_headers_content).len;
  1465. if (hdr_off < len) {
  1466. start += hdr_off;
  1467. len -= hdr_off;
  1468. /* The problem here is that we need not end of headers, we need
  1469. * start of body.
  1470. *
  1471. * Hence, we need to skip one \r\n till there is anything else in
  1472. * a line.
  1473. */
  1474. if (*start == '\r' && len > 0) {
  1475. start++;
  1476. len--;
  1477. }
  1478. if (*start == '\n' && len > 0) {
  1479. start++;
  1480. len--;
  1481. }
  1482. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1483. session->master_conn->results, start, len);
  1484. }
  1485. else {
  1486. /* XXX: should never happen! */
  1487. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1488. session->master_conn->results, NULL, 0);
  1489. }
  1490. }
  1491. else {
  1492. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1493. session->master_conn->results, NULL, 0);
  1494. }
  1495. rspamd_http_message_free (msg);
  1496. REF_RELEASE (session);
  1497. }
  1498. else {
  1499. rspamd_http_connection_reset (session->client_conn);
  1500. rspamd_http_connection_write_message (session->client_conn,
  1501. msg,
  1502. NULL,
  1503. ctype,
  1504. session,
  1505. session->ctx->timeout / 10.0);
  1506. }
  1507. }
  1508. static gboolean
  1509. rspamd_proxy_task_fin (void *ud)
  1510. {
  1511. struct rspamd_task *task = ud;
  1512. msg_debug_task ("finish task");
  1513. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1514. rspamd_proxy_scan_self_reply (task);
  1515. return TRUE;
  1516. }
  1517. if (!rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL)) {
  1518. rspamd_proxy_scan_self_reply (task);
  1519. return TRUE;
  1520. }
  1521. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1522. rspamd_proxy_scan_self_reply (task);
  1523. return TRUE;
  1524. }
  1525. /* One more iteration */
  1526. return FALSE;
  1527. }
  1528. static gboolean
  1529. rspamd_proxy_self_scan (struct rspamd_proxy_session *session)
  1530. {
  1531. struct rspamd_task *task;
  1532. struct rspamd_http_message *msg;
  1533. const gchar *data;
  1534. gsize len;
  1535. msg = session->client_message;
  1536. task = rspamd_task_new (session->worker, session->ctx->cfg,
  1537. session->pool, session->ctx->lang_det,
  1538. session->ctx->event_loop, FALSE);
  1539. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1540. if (session->ctx->milter) {
  1541. task->protocol_flags |= RSPAMD_TASK_PROTOCOL_FLAG_MILTER|
  1542. RSPAMD_TASK_PROTOCOL_FLAG_BODY_BLOCK;
  1543. }
  1544. task->sock = -1;
  1545. if (session->client_milter_conn) {
  1546. task->client_addr = rspamd_inet_address_copy(
  1547. session->client_milter_conn->addr, NULL);
  1548. }
  1549. else {
  1550. task->client_addr = rspamd_inet_address_copy(session->client_addr, NULL);
  1551. }
  1552. task->fin_arg = session;
  1553. task->resolver = session->ctx->resolver;
  1554. /* TODO: allow to disable autolearn in protocol */
  1555. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1556. task->s = rspamd_session_create (task->task_pool, rspamd_proxy_task_fin,
  1557. NULL, (event_finalizer_t )rspamd_task_free, task);
  1558. data = rspamd_http_message_get_body (msg, &len);
  1559. if (session->backend->settings_id) {
  1560. rspamd_http_message_remove_header (msg, "Settings-ID");
  1561. rspamd_http_message_add_header (msg, "Settings-ID",
  1562. session->backend->settings_id);
  1563. }
  1564. /* Process message */
  1565. if (!rspamd_protocol_handle_request (task, msg)) {
  1566. msg_err_task ("cannot handle request: %e", task->err);
  1567. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1568. }
  1569. else {
  1570. if (task->cmd == CMD_PING) {
  1571. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1572. }
  1573. else {
  1574. if (!rspamd_task_load_message (task, msg, data, len)) {
  1575. msg_err_task ("cannot load message: %e", task->err);
  1576. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1577. }
  1578. }
  1579. }
  1580. /* Set global timeout for the task */
  1581. if (session->ctx->default_upstream->timeout > 0.0) {
  1582. task->timeout_ev.data = task;
  1583. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1584. session->ctx->default_upstream->timeout,
  1585. session->ctx->default_upstream->timeout);
  1586. ev_timer_start (task->event_loop, &task->timeout_ev);
  1587. }
  1588. else if (session->ctx->has_self_scan) {
  1589. if (session->ctx->cfg->task_timeout > 0) {
  1590. task->timeout_ev.data = task;
  1591. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1592. session->ctx->cfg->task_timeout,
  1593. session->ctx->default_upstream->timeout);
  1594. ev_timer_start (task->event_loop, &task->timeout_ev);
  1595. }
  1596. }
  1597. session->master_conn->task = task;
  1598. rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL);
  1599. rspamd_session_pending (task->s);
  1600. return TRUE;
  1601. }
  1602. static gboolean
  1603. proxy_send_master_message (struct rspamd_proxy_session *session)
  1604. {
  1605. struct rspamd_http_message *msg;
  1606. struct rspamd_http_upstream *backend = NULL;
  1607. const rspamd_ftok_t *host;
  1608. GError *err = NULL;
  1609. gchar hostbuf[512];
  1610. host = rspamd_http_message_find_header (session->client_message, "Host");
  1611. if (host == NULL) {
  1612. backend = session->ctx->default_upstream;
  1613. }
  1614. else {
  1615. rspamd_strlcpy (hostbuf, host->begin, MIN(host->len + 1, sizeof (hostbuf)));
  1616. backend = g_hash_table_lookup (session->ctx->upstreams, hostbuf);
  1617. if (backend == NULL) {
  1618. backend = session->ctx->default_upstream;
  1619. }
  1620. }
  1621. if (backend == NULL) {
  1622. /* No backend */
  1623. msg_err_session ("cannot find upstream for %s", host ? hostbuf : "default");
  1624. goto err;
  1625. }
  1626. else {
  1627. session->backend = backend;
  1628. if (backend->self_scan) {
  1629. return rspamd_proxy_self_scan (session);
  1630. }
  1631. retry:
  1632. if (session->ctx->max_retries &&
  1633. session->retries > session->ctx->max_retries) {
  1634. msg_err_session ("cannot connect to upstream, maximum retries "
  1635. "has been reached: %d", session->retries);
  1636. goto err;
  1637. }
  1638. /* Provide hash key if hashing based on source address is desired */
  1639. guint hash_len;
  1640. gpointer hash_key = rspamd_inet_address_get_hash_key (session->client_addr,
  1641. &hash_len);
  1642. if (session->ctx->max_retries > 1 &&
  1643. session->retries == session->ctx->max_retries) {
  1644. session->master_conn->up = rspamd_upstream_get_except (backend->u,
  1645. session->master_conn->up,
  1646. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1647. hash_key, hash_len);
  1648. }
  1649. else {
  1650. session->master_conn->up = rspamd_upstream_get (backend->u,
  1651. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1652. hash_key, hash_len);
  1653. }
  1654. session->master_conn->timeout = backend->timeout;
  1655. if (session->master_conn->up == NULL) {
  1656. msg_err_session ("cannot select upstream for %s",
  1657. host ? hostbuf : "default");
  1658. goto err;
  1659. }
  1660. session->master_conn->backend_sock = rspamd_inet_address_connect (
  1661. rspamd_upstream_addr_next (session->master_conn->up),
  1662. SOCK_STREAM, TRUE);
  1663. if (session->master_conn->backend_sock == -1) {
  1664. msg_err_session ("cannot connect upstream: %s(%s)",
  1665. host ? hostbuf : "default",
  1666. rspamd_inet_address_to_string_pretty (
  1667. rspamd_upstream_addr_cur (
  1668. session->master_conn->up)));
  1669. rspamd_upstream_fail (session->master_conn->up, TRUE,
  1670. strerror (errno));
  1671. session->retries ++;
  1672. goto retry;
  1673. }
  1674. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1675. if (msg == NULL) {
  1676. msg_err_session ("cannot copy message to send it to the upstream: %e",
  1677. err);
  1678. if (err) {
  1679. g_error_free (err);
  1680. }
  1681. goto err; /* No fallback here */
  1682. }
  1683. session->master_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1684. session->ctx->http_ctx,
  1685. NULL,
  1686. proxy_backend_master_error_handler,
  1687. proxy_backend_master_finish_handler,
  1688. RSPAMD_HTTP_CLIENT_SIMPLE,
  1689. session->master_conn->backend_sock);
  1690. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1691. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1692. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1693. if (backend->key) {
  1694. msg->peer_key = rspamd_pubkey_ref (backend->key);
  1695. }
  1696. if (backend->settings_id != NULL) {
  1697. rspamd_http_message_remove_header (msg, "Settings-ID");
  1698. rspamd_http_message_add_header (msg, "Settings-ID",
  1699. backend->settings_id);
  1700. }
  1701. if (backend->local ||
  1702. rspamd_inet_address_is_local (
  1703. rspamd_upstream_addr_cur (
  1704. session->master_conn->up))) {
  1705. if (session->fname) {
  1706. rspamd_http_message_add_header (msg, "File", session->fname);
  1707. }
  1708. msg->method = HTTP_GET;
  1709. rspamd_http_connection_write_message_shared (
  1710. session->master_conn->backend_conn,
  1711. msg, rspamd_upstream_name(session->master_conn->up),
  1712. NULL, session->master_conn,
  1713. session->master_conn->timeout);
  1714. }
  1715. else {
  1716. if (session->fname) {
  1717. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1718. rspamd_http_message_set_body (msg,
  1719. session->map, session->map_len);
  1720. }
  1721. msg->method = HTTP_POST;
  1722. if (backend->compress) {
  1723. proxy_request_compress (msg);
  1724. if (session->client_milter_conn) {
  1725. rspamd_http_message_add_header (msg, "Content-Type",
  1726. "application/octet-stream");
  1727. }
  1728. }
  1729. else {
  1730. if (session->client_milter_conn) {
  1731. rspamd_http_message_add_header (msg, "Content-Type",
  1732. "text/plain");
  1733. }
  1734. }
  1735. rspamd_http_connection_write_message (
  1736. session->master_conn->backend_conn,
  1737. msg, rspamd_upstream_name(session->master_conn->up),
  1738. NULL, session->master_conn,
  1739. session->master_conn->timeout);
  1740. }
  1741. }
  1742. return TRUE;
  1743. err:
  1744. if (session->client_milter_conn) {
  1745. rspamd_milter_send_action (session->client_milter_conn,
  1746. RSPAMD_MILTER_TEMPFAIL);
  1747. REF_RELEASE (session);
  1748. }
  1749. else {
  1750. rspamd_http_connection_steal_msg (session->client_conn);
  1751. rspamd_http_connection_reset (session->client_conn);
  1752. proxy_client_write_error (session, 404, "Backend not found");
  1753. }
  1754. return FALSE;
  1755. }
  1756. static void
  1757. proxy_client_error_handler (struct rspamd_http_connection *conn, GError *err)
  1758. {
  1759. struct rspamd_proxy_session *session = conn->ud;
  1760. msg_info_session ("abnormally closing connection from: %s, error: %s",
  1761. rspamd_inet_address_to_string (session->client_addr), err->message);
  1762. /* Terminate session immediately */
  1763. proxy_backend_close_connection (session->master_conn);
  1764. REF_RELEASE (session);
  1765. }
  1766. static gint
  1767. proxy_client_finish_handler (struct rspamd_http_connection *conn,
  1768. struct rspamd_http_message *msg)
  1769. {
  1770. struct rspamd_proxy_session *session = conn->ud;
  1771. if (!session->master_conn) {
  1772. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1773. sizeof (*session->master_conn));
  1774. session->master_conn->s = session;
  1775. session->master_conn->name = "master";
  1776. /* Reset spamc legacy */
  1777. if (msg->method >= HTTP_SYMBOLS) {
  1778. msg->method = HTTP_POST;
  1779. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1780. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1781. msg_info_session ("enabling legacy spamc mode for session");
  1782. }
  1783. else {
  1784. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1785. msg_info_session ("enabling legacy rspamc mode for session");
  1786. }
  1787. }
  1788. if (msg->url->len == 0) {
  1789. msg->url = rspamd_fstring_append (msg->url,
  1790. "/" MSG_CMD_CHECK_V2, strlen ("/" MSG_CMD_CHECK_V2));
  1791. }
  1792. if (!proxy_check_file (msg, session)) {
  1793. goto err;
  1794. }
  1795. session->client_message = rspamd_http_connection_steal_msg (
  1796. session->client_conn);
  1797. session->shmem_ref = rspamd_http_message_shmem_ref (session->client_message);
  1798. rspamd_http_message_remove_header (msg, "Content-Length");
  1799. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1800. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1801. rspamd_http_message_remove_header (msg, "Connection");
  1802. rspamd_http_message_remove_header (msg, "Key");
  1803. proxy_open_mirror_connections (session);
  1804. rspamd_http_connection_reset (session->client_conn);
  1805. proxy_send_master_message (session);
  1806. }
  1807. else {
  1808. msg_info_session ("finished master connection");
  1809. proxy_backend_close_connection (session->master_conn);
  1810. REF_RELEASE (session);
  1811. }
  1812. return 0;
  1813. err:
  1814. rspamd_http_connection_steal_msg (session->client_conn);
  1815. rspamd_http_message_remove_header (msg, "Content-Length");
  1816. rspamd_http_message_remove_header (msg, "Key");
  1817. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1818. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1819. rspamd_http_message_remove_header (msg, "Connection");
  1820. rspamd_http_connection_reset (session->client_conn);
  1821. proxy_client_write_error (session, 404, "Backend not found");
  1822. return 0;
  1823. }
  1824. static void
  1825. proxy_milter_finish_handler (gint fd,
  1826. struct rspamd_milter_session *rms,
  1827. void *ud)
  1828. {
  1829. struct rspamd_proxy_session *session = ud;
  1830. struct rspamd_http_message *msg;
  1831. session->client_milter_conn = rms;
  1832. if (rms->message == NULL || rms->message->len == 0) {
  1833. msg_info_session ("finished milter connection");
  1834. proxy_backend_close_connection (session->master_conn);
  1835. REF_RELEASE (session);
  1836. }
  1837. else {
  1838. if (!session->master_conn) {
  1839. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1840. sizeof (*session->master_conn));
  1841. }
  1842. msg = rspamd_milter_to_http (rms);
  1843. session->master_conn->s = session;
  1844. session->master_conn->name = "master";
  1845. session->client_message = msg;
  1846. proxy_open_mirror_connections (session);
  1847. proxy_send_master_message (session);
  1848. }
  1849. }
  1850. static void
  1851. proxy_milter_error_handler (gint fd,
  1852. struct rspamd_milter_session *rms, /* unused */
  1853. void *ud, GError *err)
  1854. {
  1855. struct rspamd_proxy_session *session = ud;
  1856. if (err && err->code != 0) {
  1857. msg_info_session ("abnormally closing milter connection from: %s, "
  1858. "error: %e",
  1859. rspamd_inet_address_to_string_pretty (session->client_addr),
  1860. err);
  1861. /* Terminate session immediately */
  1862. proxy_backend_close_connection (session->master_conn);
  1863. REF_RELEASE (session);
  1864. }
  1865. else {
  1866. msg_info_session ("normally closing milter connection from: %s, "
  1867. "%e",
  1868. rspamd_inet_address_to_string_pretty (session->client_addr),
  1869. err);
  1870. /* Terminate session immediately */
  1871. proxy_backend_close_connection (session->master_conn);
  1872. REF_RELEASE (session);
  1873. }
  1874. }
  1875. static void
  1876. proxy_accept_socket (EV_P_ ev_io *w, int revents)
  1877. {
  1878. struct rspamd_worker *worker = (struct rspamd_worker *)w->data;
  1879. struct rspamd_proxy_ctx *ctx;
  1880. rspamd_inet_addr_t *addr = NULL;
  1881. struct rspamd_proxy_session *session;
  1882. gint nfd;
  1883. ctx = worker->ctx;
  1884. if ((nfd =
  1885. rspamd_accept_from_socket (w->fd, &addr,
  1886. rspamd_worker_throttle_accept_events, worker->accept_events)) == -1) {
  1887. msg_warn ("accept failed: %s", strerror (errno));
  1888. return;
  1889. }
  1890. /* Check for EAGAIN */
  1891. if (nfd == 0) {
  1892. rspamd_inet_address_free (addr);
  1893. return;
  1894. }
  1895. session = g_malloc0 (sizeof (*session));
  1896. REF_INIT_RETAIN (session, proxy_session_dtor);
  1897. session->client_sock = nfd;
  1898. session->client_addr = addr;
  1899. session->mirror_conns = g_ptr_array_sized_new (ctx->mirrors->len);
  1900. session->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (),
  1901. "proxy", 0);
  1902. session->ctx = ctx;
  1903. session->worker = worker;
  1904. if (ctx->sessions_cache) {
  1905. rspamd_worker_session_cache_add (ctx->sessions_cache,
  1906. session->pool->tag.uid, &session->ref.refcount, session);
  1907. }
  1908. if (!ctx->milter) {
  1909. session->client_conn = rspamd_http_connection_new_server (
  1910. ctx->http_ctx,
  1911. nfd,
  1912. NULL,
  1913. proxy_client_error_handler,
  1914. proxy_client_finish_handler,
  1915. 0);
  1916. if (ctx->key) {
  1917. rspamd_http_connection_set_key (session->client_conn, ctx->key);
  1918. }
  1919. msg_info_session ("accepted http connection from %s port %d",
  1920. rspamd_inet_address_to_string (addr),
  1921. rspamd_inet_address_get_port (addr));
  1922. rspamd_http_connection_read_message_shared (session->client_conn,
  1923. session,
  1924. session->ctx->timeout);
  1925. }
  1926. else {
  1927. msg_info_session ("accepted milter connection from %s port %d",
  1928. rspamd_inet_address_to_string (addr),
  1929. rspamd_inet_address_get_port (addr));
  1930. #ifdef TCP_NODELAY
  1931. #ifndef SOL_TCP
  1932. #define SOL_TCP IPPROTO_TCP
  1933. #endif
  1934. if (rspamd_inet_address_get_af (addr) != AF_UNIX) {
  1935. gint sopt = 1;
  1936. if (setsockopt (nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof (sopt)) ==
  1937. -1) {
  1938. msg_warn_session ("cannot set TCP_NODELAY: %s",
  1939. strerror (errno));
  1940. }
  1941. }
  1942. #endif
  1943. rspamd_milter_handle_socket (nfd, 0.0,
  1944. session->pool,
  1945. ctx->event_loop,
  1946. proxy_milter_finish_handler,
  1947. proxy_milter_error_handler,
  1948. session);
  1949. }
  1950. }
  1951. static void
  1952. adjust_upstreams_limits (struct rspamd_proxy_ctx *ctx)
  1953. {
  1954. struct rspamd_http_upstream *backend;
  1955. gpointer k, v;
  1956. GHashTableIter it;
  1957. /*
  1958. * We set error time equal to max_retries * backend_timeout and max_errors
  1959. * to max_retries - 1
  1960. *
  1961. * So if we failed to scan a message on a backend for some reasons, we
  1962. * will try to re-resolve it faster
  1963. */
  1964. g_hash_table_iter_init (&it, ctx->upstreams);
  1965. while (g_hash_table_iter_next (&it, &k, &v)) {
  1966. backend = (struct rspamd_http_upstream *)v;
  1967. if (!backend->self_scan && backend->u) {
  1968. rspamd_upstreams_set_limits (backend->u,
  1969. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1970. ctx->max_retries - 1, 0);
  1971. }
  1972. }
  1973. }
  1974. __attribute__((noreturn))
  1975. void
  1976. start_rspamd_proxy (struct rspamd_worker *worker)
  1977. {
  1978. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1979. gboolean is_controller = FALSE;
  1980. g_assert (rspamd_worker_check_context (worker->ctx, rspamd_rspamd_proxy_magic));
  1981. ctx->cfg = worker->srv->cfg;
  1982. ctx->event_loop = rspamd_prepare_worker (worker, "rspamd_proxy",
  1983. proxy_accept_socket);
  1984. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  1985. ctx->event_loop,
  1986. worker->srv->cfg);
  1987. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  1988. ctx->event_loop, ctx->resolver->r);
  1989. ctx->http_ctx = rspamd_http_context_create (ctx->cfg, ctx->event_loop,
  1990. ctx->cfg->ups_ctx);
  1991. rspamd_mempool_add_destructor (ctx->cfg->cfg_pool,
  1992. (rspamd_mempool_destruct_t)rspamd_http_context_free,
  1993. ctx->http_ctx);
  1994. if (ctx->has_self_scan) {
  1995. /* Additional initialisation needed */
  1996. rspamd_worker_init_scanner (worker, ctx->event_loop, ctx->resolver,
  1997. &ctx->lang_det);
  1998. if (worker->index == 0) {
  1999. /*
  2000. * If there are no controllers and no normal workers,
  2001. * then pretend that we are a controller
  2002. */
  2003. gboolean controller_seen = FALSE;
  2004. GList *cur;
  2005. cur = worker->srv->cfg->workers;
  2006. while (cur) {
  2007. struct rspamd_worker_conf *cf;
  2008. cf = (struct rspamd_worker_conf *)cur->data;
  2009. if ((cf->type == g_quark_from_static_string ("controller")) ||
  2010. (cf->type == g_quark_from_static_string ("normal"))) {
  2011. if (cf->enabled && cf->count >= 0) {
  2012. controller_seen = TRUE;
  2013. break;
  2014. }
  2015. }
  2016. cur = g_list_next (cur);
  2017. }
  2018. if (!controller_seen) {
  2019. msg_info ("no controller or normal workers defined, execute "
  2020. "controller periodics in this worker");
  2021. worker->flags |= RSPAMD_WORKER_CONTROLLER;
  2022. is_controller = TRUE;
  2023. }
  2024. }
  2025. }
  2026. else {
  2027. worker->flags &= ~RSPAMD_WORKER_SCANNER;
  2028. }
  2029. if (worker->srv->cfg->enable_sessions_cache) {
  2030. ctx->sessions_cache = rspamd_worker_session_cache_new (worker,
  2031. ctx->event_loop);
  2032. }
  2033. ctx->milter_ctx.spam_header = ctx->spam_header;
  2034. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  2035. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  2036. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  2037. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  2038. ctx->milter_ctx.reject_message = ctx->reject_message;
  2039. ctx->milter_ctx.cfg = ctx->cfg;
  2040. rspamd_milter_init_library (&ctx->milter_ctx);
  2041. if (is_controller) {
  2042. rspamd_worker_init_controller (worker, NULL);
  2043. }
  2044. else {
  2045. if (ctx->has_self_scan) {
  2046. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  2047. worker, RSPAMD_MAP_WATCH_SCANNER);
  2048. }
  2049. else {
  2050. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  2051. worker, RSPAMD_MAP_WATCH_WORKER);
  2052. }
  2053. }
  2054. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->event_loop,
  2055. worker);
  2056. adjust_upstreams_limits (ctx);
  2057. ev_loop (ctx->event_loop, 0);
  2058. rspamd_worker_block_signals ();
  2059. if (ctx->has_self_scan) {
  2060. rspamd_stat_close ();
  2061. }
  2062. if (is_controller) {
  2063. rspamd_controller_on_terminate (worker, NULL);
  2064. }
  2065. REF_RELEASE (ctx->cfg);
  2066. rspamd_log_close (worker->srv->logger);
  2067. exit (EXIT_SUCCESS);
  2068. }