You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

rspamd_proxy.c 60KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374
  1. /*-
  2. * Copyright 2016 Vsevolod Stakhov
  3. *
  4. * Licensed under the Apache License, Version 2.0 (the "License");
  5. * you may not use this file except in compliance with the License.
  6. * You may obtain a copy of the License at
  7. *
  8. * http://www.apache.org/licenses/LICENSE-2.0
  9. *
  10. * Unless required by applicable law or agreed to in writing, software
  11. * distributed under the License is distributed on an "AS IS" BASIS,
  12. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. * See the License for the specific language governing permissions and
  14. * limitations under the License.
  15. */
  16. #include "config.h"
  17. #include "libutil/util.h"
  18. #include "libserver/maps/map.h"
  19. #include "libutil/upstream.h"
  20. #include "libserver/http/http_connection.h"
  21. #include "libserver/http/http_private.h"
  22. #include "libserver/protocol.h"
  23. #include "libserver/protocol_internal.h"
  24. #include "libserver/cfg_file.h"
  25. #include "libserver/url.h"
  26. #include "libserver/dns.h"
  27. #include "libmime/message.h"
  28. #include "rspamd.h"
  29. #include "libserver/worker_util.h"
  30. #include "worker_private.h"
  31. #include "lua/lua_common.h"
  32. #include "keypairs_cache.h"
  33. #include "libstat/stat_api.h"
  34. #include "ottery.h"
  35. #include "unix-std.h"
  36. #include "libserver/milter.h"
  37. #include "libserver/milter_internal.h"
  38. #include "libmime/lang_detection.h"
  39. #include "contrib/zstd/zstd.h"
  40. #include <math.h>
  41. #ifdef HAVE_NETINET_TCP_H
  42. #include <netinet/tcp.h> /* for TCP_NODELAY */
  43. #endif
  44. /* Rotate keys each minute by default */
  45. #define DEFAULT_ROTATION_TIME 60.0
  46. #define DEFAULT_RETRIES 5
  47. #define msg_err_session(...) rspamd_default_log_function (G_LOG_LEVEL_CRITICAL, \
  48. session->pool->tag.tagname, session->pool->tag.uid, \
  49. G_STRFUNC, \
  50. __VA_ARGS__)
  51. #define msg_warn_session(...) rspamd_default_log_function (G_LOG_LEVEL_WARNING, \
  52. session->pool->tag.tagname, session->pool->tag.uid, \
  53. G_STRFUNC, \
  54. __VA_ARGS__)
  55. #define msg_info_session(...) rspamd_default_log_function (G_LOG_LEVEL_INFO, \
  56. session->pool->tag.tagname, session->pool->tag.uid, \
  57. G_STRFUNC, \
  58. __VA_ARGS__)
  59. #define msg_debug_session(...) rspamd_conditional_debug_fast (NULL, session->client_addr, \
  60. rspamd_proxy_log_id, "proxy", session->pool->tag.uid, \
  61. G_STRFUNC, \
  62. __VA_ARGS__)
  63. INIT_LOG_MODULE(proxy)
  64. gpointer init_rspamd_proxy (struct rspamd_config *cfg);
  65. void start_rspamd_proxy (struct rspamd_worker *worker);
  66. worker_t rspamd_proxy_worker = {
  67. "rspamd_proxy", /* Name */
  68. init_rspamd_proxy, /* Init function */
  69. start_rspamd_proxy, /* Start function */
  70. RSPAMD_WORKER_HAS_SOCKET | RSPAMD_WORKER_KILLABLE | RSPAMD_WORKER_SCANNER,
  71. RSPAMD_WORKER_SOCKET_TCP, /* TCP socket */
  72. RSPAMD_WORKER_VER
  73. };
  74. struct rspamd_http_upstream {
  75. gchar *name;
  76. gchar *settings_id;
  77. struct upstream_list *u;
  78. struct rspamd_cryptobox_pubkey *key;
  79. gdouble timeout;
  80. gint parser_from_ref;
  81. gint parser_to_ref;
  82. gboolean local;
  83. gboolean self_scan;
  84. gboolean compress;
  85. };
  86. struct rspamd_http_mirror {
  87. gchar *name;
  88. gchar *settings_id;
  89. struct upstream_list *u;
  90. struct rspamd_cryptobox_pubkey *key;
  91. gdouble prob;
  92. gdouble timeout;
  93. gint parser_from_ref;
  94. gint parser_to_ref;
  95. gboolean local;
  96. gboolean compress;
  97. };
  98. static const guint64 rspamd_rspamd_proxy_magic = 0xcdeb4fd1fc351980ULL;
  99. struct rspamd_proxy_ctx {
  100. guint64 magic;
  101. /* Events base */
  102. struct ev_loop *event_loop;
  103. /* DNS resolver */
  104. struct rspamd_dns_resolver *resolver;
  105. /* Config */
  106. struct rspamd_config *cfg;
  107. /* END OF COMMON PART */
  108. gdouble timeout;
  109. /* Encryption key for clients */
  110. struct rspamd_cryptobox_keypair *key;
  111. /* HTTP context */
  112. struct rspamd_http_context *http_ctx;
  113. /* Upstreams to use */
  114. GHashTable *upstreams;
  115. /* Mirrors to send traffic to */
  116. GPtrArray *mirrors;
  117. /* Default upstream */
  118. struct rspamd_http_upstream *default_upstream;
  119. lua_State *lua_state;
  120. /* Array of callback functions called on end of scan to compare results */
  121. GArray *cmp_refs;
  122. /* Maximum count for retries */
  123. guint max_retries;
  124. /* If we have self_scanning backends, we need to work as a normal worker */
  125. gboolean has_self_scan;
  126. /* It is not HTTP but milter proxy */
  127. gboolean milter;
  128. /* Discard messages instead of rejecting them */
  129. gboolean discard_on_reject;
  130. /* Quarantine messages instead of rejecting them */
  131. gboolean quarantine_on_reject;
  132. /* Milter spam header */
  133. gchar *spam_header;
  134. /* CA name that can be used for client certificates */
  135. gchar *client_ca_name;
  136. /* Milter rejection message */
  137. gchar *reject_message;
  138. /* Sessions cache */
  139. void *sessions_cache;
  140. struct rspamd_milter_context milter_ctx;
  141. /* Language detector */
  142. struct rspamd_lang_detector *lang_det;
  143. };
  144. enum rspamd_backend_flags {
  145. RSPAMD_BACKEND_REPLIED = 1 << 0,
  146. RSPAMD_BACKEND_CLOSED = 1 << 1,
  147. RSPAMD_BACKEND_PARSED = 1 << 2,
  148. };
  149. struct rspamd_proxy_session;
  150. struct rspamd_proxy_backend_connection {
  151. const gchar *name;
  152. struct rspamd_cryptobox_keypair *local_key;
  153. struct rspamd_cryptobox_pubkey *remote_key;
  154. struct upstream *up;
  155. struct rspamd_http_connection *backend_conn;
  156. ucl_object_t *results;
  157. const gchar *err;
  158. struct rspamd_proxy_session *s;
  159. gint backend_sock;
  160. ev_tstamp timeout;
  161. enum rspamd_backend_flags flags;
  162. gint parser_from_ref;
  163. gint parser_to_ref;
  164. struct rspamd_task *task;
  165. };
  166. enum rspamd_proxy_legacy_support {
  167. LEGACY_SUPPORT_NO = 0,
  168. LEGACY_SUPPORT_RSPAMC,
  169. LEGACY_SUPPORT_SPAMC
  170. };
  171. struct rspamd_proxy_session {
  172. struct rspamd_worker *worker;
  173. rspamd_mempool_t *pool;
  174. struct rspamd_proxy_ctx *ctx;
  175. rspamd_inet_addr_t *client_addr;
  176. struct rspamd_http_connection *client_conn;
  177. struct rspamd_milter_session *client_milter_conn;
  178. struct rspamd_http_upstream *backend;
  179. gpointer map;
  180. gchar *fname;
  181. gpointer shmem_ref;
  182. struct rspamd_proxy_backend_connection *master_conn;
  183. struct rspamd_http_message *client_message;
  184. GPtrArray *mirror_conns;
  185. gsize map_len;
  186. gint client_sock;
  187. enum rspamd_proxy_legacy_support legacy_support;
  188. gint retries;
  189. ref_entry_t ref;
  190. };
  191. static gboolean proxy_send_master_message (struct rspamd_proxy_session *session);
  192. static GQuark
  193. rspamd_proxy_quark (void)
  194. {
  195. return g_quark_from_static_string ("rspamd-proxy");
  196. }
  197. static gboolean
  198. rspamd_proxy_parse_lua_parser (lua_State *L, const ucl_object_t *obj,
  199. gint *ref_from, gint *ref_to, GError **err)
  200. {
  201. const gchar *lua_script;
  202. gsize slen;
  203. gint err_idx, ref_idx;
  204. gboolean has_ref = FALSE;
  205. g_assert (obj != NULL);
  206. g_assert (ref_from != NULL);
  207. g_assert (ref_to != NULL);
  208. *ref_from = -1;
  209. *ref_to = -1;
  210. lua_script = ucl_object_tolstring (obj, &slen);
  211. lua_pushcfunction (L, &rspamd_lua_traceback);
  212. err_idx = lua_gettop (L);
  213. /* Load data */
  214. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  215. g_set_error (err,
  216. rspamd_proxy_quark (),
  217. EINVAL,
  218. "cannot load lua parser script: %s",
  219. lua_tostring (L, -1));
  220. lua_settop (L, 0); /* Error function */
  221. return FALSE;
  222. }
  223. /* Now do it */
  224. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  225. g_set_error (err,
  226. rspamd_proxy_quark (),
  227. EINVAL,
  228. "cannot init lua parser script: %s",
  229. lua_tostring (L, -1));
  230. lua_settop (L, 0);
  231. return FALSE;
  232. }
  233. if (lua_istable (L, -1)) {
  234. /*
  235. * We have a table, so we check for two keys:
  236. * 'from' -> function
  237. * 'to' -> function
  238. *
  239. * From converts parent request to a client one
  240. * To converts client request to a parent one
  241. */
  242. lua_pushstring (L, "from");
  243. lua_gettable (L, -2);
  244. if (lua_isfunction (L, -1)) {
  245. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  246. *ref_from = ref_idx;
  247. has_ref = TRUE;
  248. }
  249. lua_pushstring (L, "to");
  250. lua_gettable (L, -2);
  251. if (lua_isfunction (L, -1)) {
  252. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  253. *ref_to = ref_idx;
  254. has_ref = TRUE;
  255. }
  256. }
  257. else if (!lua_isfunction (L, -1)) {
  258. g_set_error (err,
  259. rspamd_proxy_quark (),
  260. EINVAL,
  261. "cannot init lua parser script: "
  262. "must return function");
  263. lua_settop (L, 0);
  264. return FALSE;
  265. }
  266. else {
  267. /* Just parser from protocol */
  268. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  269. *ref_from = ref_idx;
  270. lua_settop (L, 0);
  271. has_ref = TRUE;
  272. }
  273. return has_ref;
  274. }
  275. static gboolean
  276. rspamd_proxy_parse_upstream (rspamd_mempool_t *pool,
  277. const ucl_object_t *obj,
  278. gpointer ud,
  279. struct rspamd_rcl_section *section,
  280. GError **err)
  281. {
  282. const ucl_object_t *elt;
  283. struct rspamd_http_upstream *up = NULL;
  284. struct rspamd_proxy_ctx *ctx;
  285. struct rspamd_rcl_struct_parser *pd = ud;
  286. lua_State *L;
  287. ctx = pd->user_struct;
  288. L = ctx->lua_state;
  289. if (ucl_object_type (obj) != UCL_OBJECT) {
  290. g_set_error (err, rspamd_proxy_quark (), 100,
  291. "upstream option must be an object");
  292. return FALSE;
  293. }
  294. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  295. elt = ucl_object_lookup (obj, "name");
  296. if (elt == NULL) {
  297. if (ucl_object_key (obj)) {
  298. if (strcmp (ucl_object_key (obj), "upstream") == 0) {
  299. /* Iterate over the object and find upstream elements */
  300. ucl_object_iter_t it = NULL;
  301. const ucl_object_t *cur;
  302. gboolean ret = TRUE;
  303. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  304. if (!rspamd_proxy_parse_upstream (pool, cur, ud,
  305. section, err)) {
  306. ret = FALSE;
  307. }
  308. }
  309. return ret;
  310. }
  311. else {
  312. /* Inside upstream */
  313. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  314. }
  315. }
  316. else {
  317. g_set_error (err, rspamd_proxy_quark (), 100,
  318. "upstream option must have some name definition");
  319. return FALSE;
  320. }
  321. }
  322. else {
  323. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  324. }
  325. up->parser_from_ref = -1;
  326. up->parser_to_ref = -1;
  327. up->timeout = ctx->timeout;
  328. elt = ucl_object_lookup (obj, "key");
  329. if (elt != NULL) {
  330. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  331. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  332. if (up->key == NULL) {
  333. g_set_error (err, rspamd_proxy_quark (), 100,
  334. "cannot read upstream key");
  335. goto err;
  336. }
  337. rspamd_mempool_add_destructor (pool,
  338. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  339. }
  340. elt = ucl_object_lookup (obj, "self_scan");
  341. if (elt && ucl_object_toboolean (elt)) {
  342. up->self_scan = TRUE;
  343. ctx->has_self_scan = TRUE;
  344. }
  345. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  346. if (elt && ucl_object_toboolean (elt)) {
  347. up->compress = TRUE;
  348. }
  349. elt = ucl_object_lookup (obj, "hosts");
  350. if (elt == NULL && !up->self_scan) {
  351. g_set_error (err, rspamd_proxy_quark (), 100,
  352. "upstream option must have some hosts definition");
  353. goto err;
  354. }
  355. if (elt) {
  356. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  357. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  358. g_set_error (err, rspamd_proxy_quark (), 100,
  359. "upstream has bad hosts definition");
  360. goto err;
  361. }
  362. rspamd_mempool_add_destructor (pool,
  363. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  364. }
  365. elt = ucl_object_lookup (obj, "default");
  366. if (elt) {
  367. if (ucl_object_toboolean (elt)) {
  368. ctx->default_upstream = up;
  369. }
  370. }
  371. else if (up->self_scan) {
  372. ctx->default_upstream = up;
  373. }
  374. elt = ucl_object_lookup (obj, "local");
  375. if (elt && ucl_object_toboolean (elt)) {
  376. up->local = TRUE;
  377. }
  378. elt = ucl_object_lookup (obj, "timeout");
  379. if (elt) {
  380. ucl_object_todouble_safe (elt, &up->timeout);
  381. }
  382. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  383. if (elt && ucl_object_type (elt) == UCL_STRING) {
  384. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  385. }
  386. /*
  387. * Accept lua function here in form
  388. * fun :: String -> UCL
  389. */
  390. elt = ucl_object_lookup (obj, "parser");
  391. if (elt) {
  392. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  393. &up->parser_to_ref, err)) {
  394. goto err;
  395. }
  396. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  397. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  398. }
  399. g_hash_table_insert (ctx->upstreams, up->name, up);
  400. return TRUE;
  401. err:
  402. return FALSE;
  403. }
  404. static gboolean
  405. rspamd_proxy_parse_mirror (rspamd_mempool_t *pool,
  406. const ucl_object_t *obj,
  407. gpointer ud,
  408. struct rspamd_rcl_section *section,
  409. GError **err)
  410. {
  411. const ucl_object_t *elt;
  412. struct rspamd_http_mirror *up = NULL;
  413. struct rspamd_proxy_ctx *ctx;
  414. struct rspamd_rcl_struct_parser *pd = ud;
  415. lua_State *L;
  416. ctx = pd->user_struct;
  417. L = ctx->lua_state;
  418. if (ucl_object_type (obj) != UCL_OBJECT) {
  419. g_set_error (err, rspamd_proxy_quark (), 100,
  420. "mirror option must be an object");
  421. return FALSE;
  422. }
  423. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  424. elt = ucl_object_lookup (obj, "name");
  425. if (elt == NULL) {
  426. if (ucl_object_key (obj)) {
  427. if (strcmp (ucl_object_key (obj), "mirror") == 0) {
  428. /* Iterate over the object and find upstream elements */
  429. ucl_object_iter_t it = NULL;
  430. const ucl_object_t *cur;
  431. gboolean ret = TRUE;
  432. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  433. if (!rspamd_proxy_parse_mirror (pool, cur, ud,
  434. section, err)) {
  435. ret = FALSE;
  436. }
  437. }
  438. return ret;
  439. }
  440. else {
  441. /* Inside upstream */
  442. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  443. }
  444. }
  445. else {
  446. g_set_error (err, rspamd_proxy_quark (), 100,
  447. "mirror option must have some name definition");
  448. return FALSE;
  449. }
  450. }
  451. else {
  452. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  453. }
  454. up->parser_to_ref = -1;
  455. up->parser_from_ref = -1;
  456. up->timeout = ctx->timeout;
  457. elt = ucl_object_lookup (obj, "key");
  458. if (elt != NULL) {
  459. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  460. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  461. if (up->key == NULL) {
  462. g_set_error (err, rspamd_proxy_quark (), 100,
  463. "cannot read mirror key");
  464. goto err;
  465. }
  466. rspamd_mempool_add_destructor (pool,
  467. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  468. }
  469. elt = ucl_object_lookup (obj, "hosts");
  470. if (elt == NULL) {
  471. g_set_error (err, rspamd_proxy_quark (), 100,
  472. "mirror option must have some hosts definition");
  473. goto err;
  474. }
  475. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  476. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  477. g_set_error (err, rspamd_proxy_quark (), 100,
  478. "mirror has bad hosts definition");
  479. goto err;
  480. }
  481. rspamd_mempool_add_destructor (pool,
  482. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  483. elt = ucl_object_lookup_any (obj, "probability", "prob", NULL);
  484. if (elt) {
  485. up->prob = ucl_object_todouble (elt);
  486. }
  487. else {
  488. up->prob = 1.0;
  489. }
  490. elt = ucl_object_lookup (obj, "local");
  491. if (elt && ucl_object_toboolean (elt)) {
  492. up->local = TRUE;
  493. }
  494. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  495. if (elt && ucl_object_toboolean (elt)) {
  496. up->compress = TRUE;
  497. }
  498. elt = ucl_object_lookup (obj, "timeout");
  499. if (elt) {
  500. ucl_object_todouble_safe (elt, &up->timeout);
  501. }
  502. /*
  503. * Accept lua function here in form
  504. * fun :: String -> UCL
  505. */
  506. elt = ucl_object_lookup (obj, "parser");
  507. if (elt) {
  508. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  509. &up->parser_to_ref, err)) {
  510. goto err;
  511. }
  512. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  513. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  514. }
  515. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  516. if (elt && ucl_object_type (elt) == UCL_STRING) {
  517. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  518. }
  519. g_ptr_array_add (ctx->mirrors, up);
  520. return TRUE;
  521. err:
  522. return FALSE;
  523. }
  524. static gboolean
  525. rspamd_proxy_parse_script (rspamd_mempool_t *pool,
  526. const ucl_object_t *obj,
  527. gpointer ud,
  528. struct rspamd_rcl_section *section,
  529. GError **err)
  530. {
  531. struct rspamd_proxy_ctx *ctx;
  532. struct rspamd_rcl_struct_parser *pd = ud;
  533. lua_State *L;
  534. const gchar *lua_script;
  535. gsize slen;
  536. gint err_idx, ref_idx;
  537. struct stat st;
  538. ctx = pd->user_struct;
  539. L = ctx->lua_state;
  540. if (ucl_object_type (obj) != UCL_STRING) {
  541. g_set_error (err, rspamd_proxy_quark (), 100,
  542. "script option must be a string with file or lua chunk");
  543. return FALSE;
  544. }
  545. lua_script = ucl_object_tolstring (obj, &slen);
  546. lua_pushcfunction (L, &rspamd_lua_traceback);
  547. err_idx = lua_gettop (L);
  548. if (stat (lua_script, &st) != -1) {
  549. /* Load file */
  550. if (luaL_loadfile (L, lua_script) != 0) {
  551. g_set_error (err,
  552. rspamd_proxy_quark (),
  553. EINVAL,
  554. "cannot load lua parser script: %s",
  555. lua_tostring (L, -1));
  556. lua_settop (L, 0); /* Error function */
  557. goto err;
  558. }
  559. }
  560. else {
  561. /* Load data directly */
  562. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  563. g_set_error (err,
  564. rspamd_proxy_quark (),
  565. EINVAL,
  566. "cannot load lua parser script: %s",
  567. lua_tostring (L, -1));
  568. lua_settop (L, 0); /* Error function */
  569. goto err;
  570. }
  571. }
  572. /* Now do it */
  573. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  574. g_set_error (err,
  575. rspamd_proxy_quark (),
  576. EINVAL,
  577. "cannot init lua parser script: %s",
  578. lua_tostring (L, -1));
  579. lua_settop (L, 0);
  580. goto err;
  581. }
  582. if (!lua_isfunction (L, -1)) {
  583. g_set_error (err,
  584. rspamd_proxy_quark (),
  585. EINVAL,
  586. "cannot init lua parser script: "
  587. "must return function, %s returned",
  588. lua_typename (L, lua_type (L, -1)));
  589. lua_settop (L, 0);
  590. goto err;
  591. }
  592. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  593. lua_settop (L, 0);
  594. g_array_append_val (ctx->cmp_refs, ref_idx);
  595. return TRUE;
  596. err:
  597. return FALSE;
  598. }
  599. gpointer
  600. init_rspamd_proxy (struct rspamd_config *cfg)
  601. {
  602. struct rspamd_proxy_ctx *ctx;
  603. GQuark type;
  604. type = g_quark_try_string ("rspamd_proxy");
  605. ctx = rspamd_mempool_alloc0 (cfg->cfg_pool,
  606. sizeof (struct rspamd_proxy_ctx));
  607. ctx->magic = rspamd_rspamd_proxy_magic;
  608. ctx->timeout = 120.0;
  609. ctx->upstreams = g_hash_table_new (rspamd_strcase_hash, rspamd_strcase_equal);
  610. rspamd_mempool_add_destructor (cfg->cfg_pool,
  611. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->upstreams);
  612. ctx->mirrors = g_ptr_array_new ();
  613. rspamd_mempool_add_destructor (cfg->cfg_pool,
  614. (rspamd_mempool_destruct_t)rspamd_ptr_array_free_hard, ctx->mirrors);
  615. ctx->cfg = cfg;
  616. ctx->lua_state = cfg->lua_state;
  617. ctx->cmp_refs = g_array_new (FALSE, FALSE, sizeof (gint));
  618. rspamd_mempool_add_destructor (cfg->cfg_pool,
  619. (rspamd_mempool_destruct_t)rspamd_array_free_hard, ctx->cmp_refs);
  620. ctx->max_retries = DEFAULT_RETRIES;
  621. ctx->spam_header = RSPAMD_MILTER_SPAM_HEADER;
  622. rspamd_rcl_register_worker_option (cfg,
  623. type,
  624. "timeout",
  625. rspamd_rcl_parse_struct_time,
  626. ctx,
  627. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, timeout),
  628. RSPAMD_CL_FLAG_TIME_FLOAT,
  629. "IO timeout");
  630. rspamd_rcl_register_worker_option (cfg,
  631. type,
  632. "keypair",
  633. rspamd_rcl_parse_struct_keypair,
  634. ctx,
  635. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, key),
  636. 0,
  637. "Server's keypair");
  638. rspamd_rcl_register_worker_option (cfg,
  639. type,
  640. "upstream",
  641. rspamd_proxy_parse_upstream,
  642. ctx,
  643. 0,
  644. 0,
  645. "List of upstreams");
  646. rspamd_rcl_register_worker_option (cfg,
  647. type,
  648. "mirror",
  649. rspamd_proxy_parse_mirror,
  650. ctx,
  651. 0,
  652. RSPAMD_CL_FLAG_MULTIPLE,
  653. "List of mirrors");
  654. rspamd_rcl_register_worker_option (cfg,
  655. type,
  656. "script",
  657. rspamd_proxy_parse_script,
  658. ctx,
  659. 0,
  660. RSPAMD_CL_FLAG_MULTIPLE,
  661. "Compare script to be executed");
  662. rspamd_rcl_register_worker_option (cfg,
  663. type,
  664. "max_retries",
  665. rspamd_rcl_parse_struct_integer,
  666. ctx,
  667. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, max_retries),
  668. RSPAMD_CL_FLAG_UINT,
  669. "Maximum number of retries for master connection");
  670. rspamd_rcl_register_worker_option (cfg,
  671. type,
  672. "milter",
  673. rspamd_rcl_parse_struct_boolean,
  674. ctx,
  675. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, milter),
  676. 0,
  677. "Accept milter connections, not HTTP");
  678. rspamd_rcl_register_worker_option (cfg,
  679. type,
  680. "discard_on_reject",
  681. rspamd_rcl_parse_struct_boolean,
  682. ctx,
  683. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, discard_on_reject),
  684. 0,
  685. "Tell MTA to discard rejected messages silently");
  686. rspamd_rcl_register_worker_option (cfg,
  687. type,
  688. "quarantine_on_reject",
  689. rspamd_rcl_parse_struct_boolean,
  690. ctx,
  691. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, quarantine_on_reject),
  692. 0,
  693. "Tell MTA to quarantine rejected messages");
  694. rspamd_rcl_register_worker_option (cfg,
  695. type,
  696. "spam_header",
  697. rspamd_rcl_parse_struct_string,
  698. ctx,
  699. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, spam_header),
  700. 0,
  701. "Use the specific spam header (default: X-Spam)");
  702. rspamd_rcl_register_worker_option (cfg,
  703. type,
  704. "client_ca_name",
  705. rspamd_rcl_parse_struct_string,
  706. ctx,
  707. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, client_ca_name),
  708. 0,
  709. "Allow certificates issued by this CA to be treated as client certificates");
  710. rspamd_rcl_register_worker_option (cfg,
  711. type,
  712. "reject_message",
  713. rspamd_rcl_parse_struct_string,
  714. ctx,
  715. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, reject_message),
  716. 0,
  717. "Use custom rejection message");
  718. return ctx;
  719. }
  720. static void
  721. proxy_backend_close_connection (struct rspamd_proxy_backend_connection *conn)
  722. {
  723. if (conn && !(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  724. if (conn->backend_conn) {
  725. rspamd_http_connection_reset (conn->backend_conn);
  726. rspamd_http_connection_unref (conn->backend_conn);
  727. close (conn->backend_sock);
  728. }
  729. conn->flags |= RSPAMD_BACKEND_CLOSED;
  730. }
  731. }
  732. static gboolean
  733. proxy_backend_parse_results (struct rspamd_proxy_session *session,
  734. struct rspamd_proxy_backend_connection *conn,
  735. lua_State *L, gint parser_ref,
  736. struct rspamd_http_message *msg,
  737. goffset *body_offset)
  738. {
  739. struct ucl_parser *parser;
  740. gint err_idx;
  741. const gchar *in = msg->body_buf.begin;
  742. gsize inlen = msg->body_buf.len;
  743. const rspamd_ftok_t *offset_hdr;
  744. if (inlen == 0 || in == NULL) {
  745. return FALSE;
  746. }
  747. offset_hdr = rspamd_http_message_find_header (msg, MESSAGE_OFFSET_HEADER);
  748. if (offset_hdr) {
  749. gulong val;
  750. if (rspamd_strtoul (offset_hdr->begin, offset_hdr->len, &val)
  751. && val < inlen) {
  752. if (body_offset) {
  753. *body_offset = val;
  754. }
  755. inlen = val;
  756. }
  757. }
  758. if (parser_ref != -1) {
  759. /* Call parser function */
  760. lua_pushcfunction (L, &rspamd_lua_traceback);
  761. err_idx = lua_gettop (L);
  762. lua_rawgeti (L, LUA_REGISTRYINDEX, parser_ref);
  763. /* XXX: copies all data */
  764. lua_pushlstring (L, in, inlen);
  765. if (lua_pcall (L, 1, 1, err_idx) != 0) {
  766. msg_err_session (
  767. "cannot run lua parser script: %s",
  768. lua_tostring (L, -1));
  769. lua_settop (L, 0);
  770. return FALSE;
  771. }
  772. conn->results = ucl_object_lua_import (L, -1);
  773. lua_settop (L, 0);
  774. }
  775. else {
  776. parser = ucl_parser_new (0);
  777. if (!ucl_parser_add_chunk (parser, in, inlen)) {
  778. gchar *encoded;
  779. encoded = rspamd_encode_base64 (in, inlen, 0, NULL);
  780. msg_err_session ("cannot parse input: %s", ucl_parser_get_error (
  781. parser));
  782. msg_err_session ("input encoded: %s", encoded);
  783. ucl_parser_free (parser);
  784. g_free (encoded);
  785. return FALSE;
  786. }
  787. conn->results = ucl_parser_get_object (parser);
  788. ucl_parser_free (parser);
  789. }
  790. return TRUE;
  791. }
  792. static void
  793. proxy_call_cmp_script (struct rspamd_proxy_session *session, gint cbref)
  794. {
  795. gint err_idx;
  796. guint i;
  797. struct rspamd_proxy_backend_connection *conn;
  798. lua_State *L;
  799. L = session->ctx->lua_state;
  800. lua_pushcfunction (L, &rspamd_lua_traceback);
  801. err_idx = lua_gettop (L);
  802. lua_rawgeti (L, LUA_REGISTRYINDEX, cbref);
  803. lua_createtable (L, 0, session->mirror_conns->len + 1);
  804. /* Now push master results */
  805. if (session->master_conn && session->master_conn->results) {
  806. lua_pushstring (L, "master");
  807. ucl_object_push_lua (L, session->master_conn->results, true);
  808. lua_settable (L, -3);
  809. }
  810. else {
  811. lua_pushstring (L, "master");
  812. lua_pushstring (L, "no results");
  813. lua_settable (L, -3);
  814. }
  815. for (i = 0; i < session->mirror_conns->len; i ++) {
  816. conn = g_ptr_array_index (session->mirror_conns, i);
  817. if (conn->results) {
  818. lua_pushstring (L, conn->name);
  819. ucl_object_push_lua (L, conn->results, true);
  820. lua_settable (L, -3);
  821. }
  822. else {
  823. lua_pushstring (L, conn->name);
  824. lua_pushstring (L, conn->err ? conn->err : "unknown error");
  825. lua_settable (L, -3);
  826. }
  827. }
  828. if (lua_pcall (L, 1, 0, err_idx) != 0) {
  829. msg_err_session (
  830. "cannot run lua compare script: %s",
  831. lua_tostring (L, -1));
  832. }
  833. lua_settop (L, 0);
  834. }
  835. static void
  836. proxy_session_dtor (struct rspamd_proxy_session *session)
  837. {
  838. guint i;
  839. gint cbref;
  840. struct rspamd_proxy_backend_connection *conn;
  841. if (session->master_conn && session->master_conn->results) {
  842. for (i = 0; i < session->ctx->cmp_refs->len; i++) {
  843. cbref = g_array_index (session->ctx->cmp_refs, gint, i);
  844. proxy_call_cmp_script (session, cbref);
  845. }
  846. }
  847. if (session->master_conn) {
  848. proxy_backend_close_connection (session->master_conn);
  849. }
  850. if (session->client_milter_conn) {
  851. rspamd_milter_session_unref (session->client_milter_conn);
  852. }
  853. else if (session->client_conn) {
  854. rspamd_http_connection_reset (session->client_conn);
  855. rspamd_http_connection_unref (session->client_conn);
  856. }
  857. if (session->map && session->map_len) {
  858. munmap (session->map, session->map_len);
  859. }
  860. for (i = 0; i < session->mirror_conns->len; i ++) {
  861. conn = g_ptr_array_index (session->mirror_conns, i);
  862. if (!(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  863. proxy_backend_close_connection (conn);
  864. }
  865. if (conn->results) {
  866. ucl_object_unref (conn->results);
  867. }
  868. }
  869. if (session->master_conn) {
  870. if (session->master_conn->results) {
  871. ucl_object_unref (session->master_conn->results);
  872. }
  873. if (session->master_conn->task) {
  874. rspamd_session_destroy (session->master_conn->task->s);
  875. }
  876. }
  877. g_ptr_array_free (session->mirror_conns, TRUE);
  878. rspamd_http_message_shmem_unref (session->shmem_ref);
  879. rspamd_http_message_unref (session->client_message);
  880. if (session->client_addr) {
  881. rspamd_inet_address_free (session->client_addr);
  882. }
  883. if (session->client_sock != -1) {
  884. close (session->client_sock);
  885. }
  886. if (session->ctx->sessions_cache) {
  887. rspamd_worker_session_cache_remove (session->ctx->sessions_cache,
  888. session);
  889. }
  890. if (session->pool) {
  891. rspamd_mempool_delete (session->pool);
  892. }
  893. g_free (session);
  894. }
  895. static void
  896. proxy_request_compress (struct rspamd_http_message *msg)
  897. {
  898. guint flags;
  899. ZSTD_CCtx *zctx;
  900. rspamd_fstring_t *body;
  901. const gchar *in;
  902. gsize inlen;
  903. flags = rspamd_http_message_get_flags (msg);
  904. if (!rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  905. if ((flags & RSPAMD_HTTP_FLAG_SHMEM) ||
  906. !(flags & RSPAMD_HTTP_FLAG_HAS_BODY)) {
  907. /* Cannot compress shared or empty message */
  908. return;
  909. }
  910. in = rspamd_http_message_get_body (msg, &inlen);
  911. if (in == NULL || inlen == 0) {
  912. return;
  913. }
  914. body = rspamd_fstring_sized_new (ZSTD_compressBound (inlen));
  915. zctx = ZSTD_createCCtx ();
  916. body->len = ZSTD_compressCCtx (zctx, body->str, body->allocated,
  917. in, inlen, 1);
  918. if (ZSTD_isError (body->len)) {
  919. msg_err ("compression error");
  920. rspamd_fstring_free (body);
  921. ZSTD_freeCCtx (zctx);
  922. return;
  923. }
  924. ZSTD_freeCCtx (zctx);
  925. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  926. rspamd_http_message_add_header (msg, COMPRESSION_HEADER, "zstd");
  927. }
  928. }
  929. static void
  930. proxy_request_decompress (struct rspamd_http_message *msg)
  931. {
  932. rspamd_fstring_t *body;
  933. const gchar *in;
  934. gsize inlen, outlen, r;
  935. ZSTD_DStream *zstream;
  936. ZSTD_inBuffer zin;
  937. ZSTD_outBuffer zout;
  938. if (rspamd_http_message_find_header (msg, COMPRESSION_HEADER)) {
  939. in = rspamd_http_message_get_body (msg, &inlen);
  940. if (in == NULL || inlen == 0) {
  941. return;
  942. }
  943. zstream = ZSTD_createDStream ();
  944. ZSTD_initDStream (zstream);
  945. zin.pos = 0;
  946. zin.src = in;
  947. zin.size = inlen;
  948. if ((outlen = ZSTD_getDecompressedSize (zin.src, zin.size)) == 0) {
  949. outlen = ZSTD_DStreamOutSize ();
  950. }
  951. body = rspamd_fstring_sized_new (outlen);
  952. zout.dst = body->str;
  953. zout.pos = 0;
  954. zout.size = outlen;
  955. while (zin.pos < zin.size) {
  956. r = ZSTD_decompressStream (zstream, &zout, &zin);
  957. if (ZSTD_isError (r)) {
  958. msg_err ("Decompression error: %s", ZSTD_getErrorName (r));
  959. ZSTD_freeDStream (zstream);
  960. rspamd_fstring_free (body);
  961. return;
  962. }
  963. if (zout.pos == zout.size) {
  964. /* We need to extend output buffer */
  965. zout.size = zout.size * 2 + 1;
  966. body = rspamd_fstring_grow (body, zout.size);
  967. zout.size = body->allocated;
  968. zout.dst = body->str;
  969. }
  970. }
  971. body->len = zout.pos;
  972. ZSTD_freeDStream (zstream);
  973. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  974. rspamd_http_message_remove_header (msg, COMPRESSION_HEADER);
  975. }
  976. }
  977. static struct rspamd_proxy_session *
  978. proxy_session_refresh (struct rspamd_proxy_session *session)
  979. {
  980. struct rspamd_proxy_session *nsession;
  981. nsession = g_malloc0 (sizeof (*nsession));
  982. nsession->client_milter_conn = session->client_milter_conn;
  983. session->client_milter_conn = NULL;
  984. rspamd_milter_update_userdata (nsession->client_milter_conn,
  985. nsession);
  986. nsession->client_addr = session->client_addr;
  987. session->client_addr = NULL;
  988. nsession->ctx = session->ctx;
  989. nsession->worker = session->worker;
  990. nsession->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (), "proxy", 0);
  991. nsession->client_sock = session->client_sock;
  992. session->client_sock = -1;
  993. nsession->mirror_conns = g_ptr_array_sized_new (nsession->ctx->mirrors->len);
  994. REF_INIT_RETAIN (nsession, proxy_session_dtor);
  995. if (nsession->ctx->sessions_cache) {
  996. rspamd_worker_session_cache_add (nsession->ctx->sessions_cache,
  997. nsession->pool->tag.uid, &nsession->ref.refcount, nsession);
  998. }
  999. return nsession;
  1000. }
  1001. static gboolean
  1002. proxy_check_file (struct rspamd_http_message *msg,
  1003. struct rspamd_proxy_session *session)
  1004. {
  1005. const rspamd_ftok_t *tok, *key_tok;
  1006. rspamd_ftok_t srch;
  1007. gchar *file_str;
  1008. GHashTable *query_args;
  1009. GHashTableIter it;
  1010. gpointer k, v;
  1011. struct http_parser_url u;
  1012. rspamd_fstring_t *new_url;
  1013. tok = rspamd_http_message_find_header (msg, "File");
  1014. if (tok) {
  1015. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1016. session->map = rspamd_file_xmap (file_str, PROT_READ, &session->map_len,
  1017. TRUE);
  1018. if (session->map == NULL) {
  1019. if (session->map_len != 0) {
  1020. msg_err_session ("cannot map %s: %s", file_str,
  1021. strerror (errno));
  1022. return FALSE;
  1023. }
  1024. }
  1025. /* Remove header after processing */
  1026. rspamd_http_message_remove_header (msg, "File");
  1027. session->fname = file_str;
  1028. }
  1029. else {
  1030. /* Need to parse query URL */
  1031. if (http_parser_parse_url (RSPAMD_FSTRING_DATA (msg->url),
  1032. RSPAMD_FSTRING_LEN (msg->url), 0, &u) != 0) {
  1033. msg_err_session ("bad request url: %V", msg->url);
  1034. return FALSE;
  1035. }
  1036. if (u.field_set & (1 << UF_QUERY)) {
  1037. /* In case if we have a query, we need to store it somewhere */
  1038. query_args = rspamd_http_message_parse_query (msg);
  1039. srch.begin = "File";
  1040. srch.len = strlen ("File");
  1041. tok = g_hash_table_lookup (query_args, &srch);
  1042. if (tok) {
  1043. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1044. session->map = rspamd_file_xmap (file_str, PROT_READ,
  1045. &session->map_len, TRUE);
  1046. if (session->map == NULL) {
  1047. if (session->map_len != 0) {
  1048. msg_err_session ("cannot map %s: %s", file_str,
  1049. strerror (errno));
  1050. g_hash_table_unref (query_args);
  1051. return FALSE;
  1052. }
  1053. }
  1054. /* We need to create a new URL with file attribute removed */
  1055. new_url = rspamd_fstring_new_init (RSPAMD_FSTRING_DATA (msg->url),
  1056. u.field_data[UF_QUERY].off);
  1057. new_url = rspamd_fstring_append (new_url, "?", 1);
  1058. g_hash_table_iter_init (&it, query_args);
  1059. while (g_hash_table_iter_next (&it, &k, &v)) {
  1060. key_tok = k;
  1061. tok = v;
  1062. if (!rspamd_ftok_icase_equal (key_tok, &srch)) {
  1063. rspamd_printf_fstring (&new_url, "%T=%T&",
  1064. key_tok, tok);
  1065. }
  1066. }
  1067. /* Erase last character (might be either & or ?) */
  1068. rspamd_fstring_erase (new_url, new_url->len - 1, 1);
  1069. rspamd_fstring_free (msg->url);
  1070. msg->url = new_url;
  1071. session->fname = file_str;
  1072. }
  1073. g_hash_table_unref (query_args);
  1074. }
  1075. }
  1076. return TRUE;
  1077. }
  1078. static void
  1079. proxy_backend_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  1080. {
  1081. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1082. struct rspamd_proxy_session *session;
  1083. session = bk_conn->s;
  1084. msg_info_session ("abnormally closing connection from backend: %s:%s, "
  1085. "error: %e",
  1086. bk_conn->name,
  1087. rspamd_inet_address_to_string_pretty (
  1088. rspamd_upstream_addr_cur (bk_conn->up)),
  1089. err);
  1090. if (err) {
  1091. bk_conn->err = rspamd_mempool_strdup (session->pool, err->message);
  1092. }
  1093. rspamd_upstream_fail (bk_conn->up, FALSE, err ? err->message : "unknown");
  1094. proxy_backend_close_connection (bk_conn);
  1095. REF_RELEASE (bk_conn->s);
  1096. }
  1097. static gint
  1098. proxy_backend_mirror_finish_handler (struct rspamd_http_connection *conn,
  1099. struct rspamd_http_message *msg)
  1100. {
  1101. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1102. struct rspamd_proxy_session *session;
  1103. session = bk_conn->s;
  1104. proxy_request_decompress (msg);
  1105. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1106. bk_conn->parser_from_ref, msg, NULL)) {
  1107. msg_warn_session ("cannot parse results from the mirror backend %s:%s",
  1108. bk_conn->name,
  1109. rspamd_inet_address_to_string (
  1110. rspamd_upstream_addr_cur (bk_conn->up)));
  1111. bk_conn->err = "cannot parse ucl";
  1112. }
  1113. msg_info_session ("finished mirror connection to %s", bk_conn->name);
  1114. rspamd_upstream_ok (bk_conn->up);
  1115. proxy_backend_close_connection (bk_conn);
  1116. REF_RELEASE (bk_conn->s);
  1117. return 0;
  1118. }
  1119. static void
  1120. proxy_open_mirror_connections (struct rspamd_proxy_session *session)
  1121. {
  1122. gdouble coin;
  1123. struct rspamd_http_mirror *m;
  1124. guint i;
  1125. struct rspamd_proxy_backend_connection *bk_conn;
  1126. struct rspamd_http_message *msg;
  1127. GError *err = NULL;
  1128. coin = rspamd_random_double ();
  1129. for (i = 0; i < session->ctx->mirrors->len; i ++) {
  1130. m = g_ptr_array_index (session->ctx->mirrors, i);
  1131. if (m->prob < coin) {
  1132. /* No luck */
  1133. continue;
  1134. }
  1135. bk_conn = rspamd_mempool_alloc0 (session->pool,
  1136. sizeof (*bk_conn));
  1137. bk_conn->s = session;
  1138. bk_conn->name = m->name;
  1139. bk_conn->timeout = m->timeout;
  1140. bk_conn->up = rspamd_upstream_get (m->u,
  1141. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1142. bk_conn->parser_from_ref = m->parser_from_ref;
  1143. bk_conn->parser_to_ref = m->parser_to_ref;
  1144. if (bk_conn->up == NULL) {
  1145. msg_err_session ("cannot select upstream for %s", m->name);
  1146. continue;
  1147. }
  1148. bk_conn->backend_sock = rspamd_inet_address_connect (
  1149. rspamd_upstream_addr_next (bk_conn->up),
  1150. SOCK_STREAM, TRUE);
  1151. if (bk_conn->backend_sock == -1) {
  1152. msg_err_session ("cannot connect upstream for %s", m->name);
  1153. rspamd_upstream_fail (bk_conn->up, TRUE, strerror (errno));
  1154. continue;
  1155. }
  1156. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1157. if (msg == NULL) {
  1158. msg_err_session ("cannot copy message to send to a mirror %s: %e",
  1159. m->name, err);
  1160. if (err) {
  1161. g_error_free (err);
  1162. }
  1163. continue;
  1164. }
  1165. if (msg->url->len == 0) {
  1166. msg->url = rspamd_fstring_append (msg->url, "/check", strlen ("/check"));
  1167. }
  1168. if (m->settings_id != NULL) {
  1169. rspamd_http_message_remove_header (msg, "Settings-ID");
  1170. rspamd_http_message_add_header (msg, "Settings-ID", m->settings_id);
  1171. }
  1172. bk_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1173. session->ctx->http_ctx,
  1174. NULL,
  1175. proxy_backend_mirror_error_handler,
  1176. proxy_backend_mirror_finish_handler,
  1177. RSPAMD_HTTP_CLIENT_SIMPLE,
  1178. bk_conn->backend_sock);
  1179. if (m->key) {
  1180. msg->peer_key = rspamd_pubkey_ref (m->key);
  1181. }
  1182. if (m->local ||
  1183. rspamd_inet_address_is_local (rspamd_upstream_addr_cur (bk_conn->up))) {
  1184. if (session->fname) {
  1185. rspamd_http_message_add_header (msg, "File", session->fname);
  1186. }
  1187. msg->method = HTTP_GET;
  1188. rspamd_http_connection_write_message_shared (bk_conn->backend_conn,
  1189. msg, NULL, NULL, bk_conn,
  1190. bk_conn->timeout);
  1191. }
  1192. else {
  1193. if (session->fname) {
  1194. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1195. rspamd_http_message_set_body (msg, session->map, session->map_len);
  1196. }
  1197. msg->method = HTTP_POST;
  1198. if (m->compress) {
  1199. proxy_request_compress (msg);
  1200. if (session->client_milter_conn) {
  1201. rspamd_http_message_add_header (msg, "Content-Type",
  1202. "application/octet-stream");
  1203. }
  1204. }
  1205. else {
  1206. if (session->client_milter_conn) {
  1207. rspamd_http_message_add_header (msg, "Content-Type",
  1208. "text/plain");
  1209. }
  1210. }
  1211. rspamd_http_connection_write_message (bk_conn->backend_conn,
  1212. msg, NULL, NULL, bk_conn,
  1213. bk_conn->timeout);
  1214. }
  1215. g_ptr_array_add (session->mirror_conns, bk_conn);
  1216. REF_RETAIN (session);
  1217. msg_info_session ("send request to %s", m->name);
  1218. }
  1219. }
  1220. static void
  1221. proxy_client_write_error (struct rspamd_proxy_session *session, gint code,
  1222. const gchar *status)
  1223. {
  1224. struct rspamd_http_message *reply;
  1225. if (session->client_milter_conn) {
  1226. rspamd_milter_send_action (session->client_milter_conn,
  1227. RSPAMD_MILTER_TEMPFAIL);
  1228. REF_RELEASE (session);
  1229. }
  1230. else {
  1231. reply = rspamd_http_new_message (HTTP_RESPONSE);
  1232. switch (code) {
  1233. case ETIMEDOUT:
  1234. reply->code = 504;
  1235. reply->status = RSPAMD_FSTRING_LIT ("Gateway timeout");
  1236. break;
  1237. case ECONNRESET:
  1238. case ECONNABORTED:
  1239. reply->code = 502;
  1240. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection reset");
  1241. break;
  1242. case ECONNREFUSED:
  1243. reply->code = 502;
  1244. reply->status = RSPAMD_FSTRING_LIT ("Gateway connection refused");
  1245. break;
  1246. default:
  1247. if (code >= 300) {
  1248. /* Likely HTTP error */
  1249. reply->code = code;
  1250. reply->status = rspamd_fstring_new_init (status, strlen (status));
  1251. }
  1252. else {
  1253. reply->code = 502;
  1254. reply->status = RSPAMD_FSTRING_LIT ("Unknown gateway error: ");
  1255. reply->status = rspamd_fstring_append (reply->status,
  1256. status, strlen (status));
  1257. }
  1258. break;
  1259. }
  1260. rspamd_http_connection_write_message (session->client_conn,
  1261. reply, NULL, NULL, session,
  1262. session->ctx->timeout);
  1263. }
  1264. }
  1265. static void
  1266. proxy_backend_master_error_handler (struct rspamd_http_connection *conn, GError *err)
  1267. {
  1268. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1269. struct rspamd_proxy_session *session;
  1270. session = bk_conn->s;
  1271. session->retries ++;
  1272. msg_info_session ("abnormally closing connection from backend: %s, error: %e,"
  1273. " retries left: %d",
  1274. rspamd_inet_address_to_string_pretty (
  1275. rspamd_upstream_addr_cur (session->master_conn->up)),
  1276. err,
  1277. session->ctx->max_retries - session->retries);
  1278. rspamd_upstream_fail (bk_conn->up, FALSE, err ? err->message : "unknown");
  1279. proxy_backend_close_connection (session->master_conn);
  1280. if (session->ctx->max_retries > 0 &&
  1281. session->retries >= session->ctx->max_retries) {
  1282. msg_err_session ("cannot connect to upstream, maximum retries "
  1283. "has been reached: %d", session->retries);
  1284. /* Terminate session immediately */
  1285. proxy_client_write_error (session, err->code, err->message);
  1286. }
  1287. else {
  1288. if (!proxy_send_master_message (session)) {
  1289. proxy_client_write_error (session, err->code, err->message);
  1290. }
  1291. else {
  1292. msg_info_session ("retry connection to: %s"
  1293. " retries left: %d",
  1294. rspamd_inet_address_to_string (
  1295. rspamd_upstream_addr_cur (session->master_conn->up)),
  1296. session->ctx->max_retries - session->retries);
  1297. }
  1298. }
  1299. }
  1300. static gint
  1301. proxy_backend_master_finish_handler (struct rspamd_http_connection *conn,
  1302. struct rspamd_http_message *msg)
  1303. {
  1304. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1305. struct rspamd_proxy_session *session, *nsession;
  1306. rspamd_fstring_t *reply;
  1307. goffset body_offset = -1;
  1308. session = bk_conn->s;
  1309. rspamd_http_connection_steal_msg (session->master_conn->backend_conn);
  1310. proxy_request_decompress (msg);
  1311. rspamd_http_message_remove_header (msg, "Content-Length");
  1312. rspamd_http_message_remove_header (msg, "Key");
  1313. rspamd_http_connection_reset (session->master_conn->backend_conn);
  1314. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1315. bk_conn->parser_from_ref, msg, &body_offset)) {
  1316. msg_warn_session ("cannot parse results from the master backend");
  1317. }
  1318. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1319. /* We need to reformat ucl to fit with legacy spamc protocol */
  1320. if (bk_conn->results) {
  1321. reply = rspamd_fstring_new ();
  1322. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1323. rspamd_ucl_tospamc_output (bk_conn->results, &reply);
  1324. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1325. }
  1326. else {
  1327. rspamd_ucl_torspamc_output (bk_conn->results, &reply);
  1328. }
  1329. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1330. msg->method = HTTP_SYMBOLS;
  1331. }
  1332. else {
  1333. msg_warn_session ("cannot parse results from the master backend, "
  1334. "return them as is");
  1335. }
  1336. }
  1337. rspamd_upstream_ok (bk_conn->up);
  1338. if (session->client_milter_conn) {
  1339. nsession = proxy_session_refresh (session);
  1340. if (body_offset > 0) {
  1341. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1342. session->master_conn->results,
  1343. msg->body_buf.begin + body_offset,
  1344. msg->body_buf.len - body_offset);
  1345. }
  1346. else {
  1347. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1348. session->master_conn->results, NULL, 0);
  1349. }
  1350. REF_RELEASE (session);
  1351. rspamd_http_message_free (msg);
  1352. }
  1353. else {
  1354. rspamd_http_connection_write_message (session->client_conn,
  1355. msg, NULL, NULL, session,
  1356. bk_conn->timeout);
  1357. }
  1358. return 0;
  1359. }
  1360. static void
  1361. rspamd_proxy_scan_self_reply (struct rspamd_task *task)
  1362. {
  1363. struct rspamd_http_message *msg;
  1364. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1365. ucl_object_t *rep = NULL;
  1366. const char *ctype = "application/json";
  1367. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1368. msg->date = time (NULL);
  1369. msg->code = 200;
  1370. switch (task->cmd) {
  1371. case CMD_CHECK:
  1372. case CMD_SKIP:
  1373. case CMD_CHECK_RSPAMC:
  1374. case CMD_CHECK_SPAMC:
  1375. case CMD_CHECK_V2:
  1376. rspamd_task_set_finish_time (task);
  1377. rspamd_protocol_http_reply (msg, task, &rep);
  1378. rspamd_protocol_write_log_pipe (task);
  1379. break;
  1380. case CMD_PING:
  1381. rspamd_http_message_set_body (msg, "pong" CRLF, 6);
  1382. ctype = "text/plain";
  1383. break;
  1384. default:
  1385. msg_err_task ("BROKEN");
  1386. break;
  1387. }
  1388. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1389. if (rep) {
  1390. session->master_conn->results = ucl_object_ref (rep);
  1391. }
  1392. if (session->client_milter_conn) {
  1393. nsession = proxy_session_refresh (session);
  1394. if (task->flags & RSPAMD_TASK_FLAG_MESSAGE_REWRITE) {
  1395. const gchar *start;
  1396. goffset len, hdr_off;
  1397. start = task->msg.begin;
  1398. len = task->msg.len;
  1399. hdr_off = MESSAGE_FIELD (task, raw_headers_content).len;
  1400. if (hdr_off < len) {
  1401. start += hdr_off;
  1402. len -= hdr_off;
  1403. /* The problem here is that we need not end of headers, we need
  1404. * start of body.
  1405. *
  1406. * Hence, we need to skip one \r\n till there is anything else in
  1407. * a line.
  1408. */
  1409. if (*start == '\r' && len > 0) {
  1410. start++;
  1411. len--;
  1412. }
  1413. if (*start == '\n' && len > 0) {
  1414. start++;
  1415. len--;
  1416. }
  1417. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1418. session->master_conn->results, start, len);
  1419. }
  1420. else {
  1421. /* XXX: should never happen! */
  1422. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1423. session->master_conn->results, NULL, 0);
  1424. }
  1425. }
  1426. else {
  1427. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1428. session->master_conn->results, NULL, 0);
  1429. }
  1430. rspamd_http_message_free (msg);
  1431. REF_RELEASE (session);
  1432. }
  1433. else {
  1434. rspamd_http_connection_reset (session->client_conn);
  1435. rspamd_http_connection_write_message (session->client_conn,
  1436. msg,
  1437. NULL,
  1438. ctype,
  1439. session,
  1440. 0);
  1441. }
  1442. }
  1443. static gboolean
  1444. rspamd_proxy_task_fin (void *ud)
  1445. {
  1446. struct rspamd_task *task = ud;
  1447. msg_debug_task ("finish task");
  1448. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1449. rspamd_proxy_scan_self_reply (task);
  1450. return TRUE;
  1451. }
  1452. if (!rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL)) {
  1453. rspamd_proxy_scan_self_reply (task);
  1454. return TRUE;
  1455. }
  1456. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1457. rspamd_proxy_scan_self_reply (task);
  1458. return TRUE;
  1459. }
  1460. /* One more iteration */
  1461. return FALSE;
  1462. }
  1463. static gboolean
  1464. rspamd_proxy_self_scan (struct rspamd_proxy_session *session)
  1465. {
  1466. struct rspamd_task *task;
  1467. struct rspamd_http_message *msg;
  1468. const gchar *data;
  1469. gsize len;
  1470. msg = session->client_message;
  1471. task = rspamd_task_new (session->worker, session->ctx->cfg,
  1472. session->pool, session->ctx->lang_det,
  1473. session->ctx->event_loop, FALSE);
  1474. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1475. if (session->ctx->milter) {
  1476. task->protocol_flags |= RSPAMD_TASK_PROTOCOL_FLAG_MILTER|
  1477. RSPAMD_TASK_PROTOCOL_FLAG_BODY_BLOCK;
  1478. }
  1479. task->sock = -1;
  1480. if (session->client_milter_conn) {
  1481. task->client_addr = rspamd_inet_address_copy (
  1482. session->client_milter_conn->addr);
  1483. }
  1484. else {
  1485. task->client_addr = rspamd_inet_address_copy (session->client_addr);
  1486. }
  1487. task->fin_arg = session;
  1488. task->resolver = session->ctx->resolver;
  1489. /* TODO: allow to disable autolearn in protocol */
  1490. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1491. task->s = rspamd_session_create (task->task_pool, rspamd_proxy_task_fin,
  1492. NULL, (event_finalizer_t )rspamd_task_free, task);
  1493. data = rspamd_http_message_get_body (msg, &len);
  1494. if (session->backend->settings_id) {
  1495. rspamd_http_message_remove_header (msg, "Settings-ID");
  1496. rspamd_http_message_add_header (msg, "Settings-ID",
  1497. session->backend->settings_id);
  1498. }
  1499. /* Process message */
  1500. if (!rspamd_protocol_handle_request (task, msg)) {
  1501. msg_err_task ("cannot handle request: %e", task->err);
  1502. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1503. }
  1504. else {
  1505. if (task->cmd == CMD_PING) {
  1506. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1507. }
  1508. else {
  1509. if (!rspamd_task_load_message (task, msg, data, len)) {
  1510. msg_err_task ("cannot load message: %e", task->err);
  1511. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1512. }
  1513. }
  1514. }
  1515. /* Set global timeout for the task */
  1516. if (session->ctx->default_upstream->timeout > 0.0) {
  1517. task->timeout_ev.data = task;
  1518. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1519. session->ctx->default_upstream->timeout,
  1520. session->ctx->default_upstream->timeout);
  1521. ev_timer_start (task->event_loop, &task->timeout_ev);
  1522. }
  1523. else if (session->ctx->has_self_scan) {
  1524. if (session->ctx->cfg->task_timeout > 0) {
  1525. task->timeout_ev.data = task;
  1526. ev_timer_init (&task->timeout_ev, rspamd_task_timeout,
  1527. session->ctx->cfg->task_timeout,
  1528. session->ctx->default_upstream->timeout);
  1529. ev_timer_start (task->event_loop, &task->timeout_ev);
  1530. }
  1531. }
  1532. session->master_conn->task = task;
  1533. rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL);
  1534. rspamd_session_pending (task->s);
  1535. return TRUE;
  1536. }
  1537. static gboolean
  1538. proxy_send_master_message (struct rspamd_proxy_session *session)
  1539. {
  1540. struct rspamd_http_message *msg;
  1541. struct rspamd_http_upstream *backend = NULL;
  1542. const rspamd_ftok_t *host;
  1543. GError *err = NULL;
  1544. gchar hostbuf[512];
  1545. host = rspamd_http_message_find_header (session->client_message, "Host");
  1546. if (host == NULL) {
  1547. backend = session->ctx->default_upstream;
  1548. }
  1549. else {
  1550. rspamd_strlcpy (hostbuf, host->begin, MIN(host->len + 1, sizeof (hostbuf)));
  1551. backend = g_hash_table_lookup (session->ctx->upstreams, hostbuf);
  1552. if (backend == NULL) {
  1553. backend = session->ctx->default_upstream;
  1554. }
  1555. }
  1556. if (backend == NULL) {
  1557. /* No backend */
  1558. msg_err_session ("cannot find upstream for %s", host ? hostbuf : "default");
  1559. goto err;
  1560. }
  1561. else {
  1562. session->backend = backend;
  1563. if (backend->self_scan) {
  1564. return rspamd_proxy_self_scan (session);
  1565. }
  1566. retry:
  1567. if (session->ctx->max_retries &&
  1568. session->retries > session->ctx->max_retries) {
  1569. msg_err_session ("cannot connect to upstream, maximum retries "
  1570. "has been reached: %d", session->retries);
  1571. goto err;
  1572. }
  1573. /* Provide hash key if hashing based on source address is desired */
  1574. guint hash_len;
  1575. gpointer hash_key = rspamd_inet_address_get_hash_key (session->client_addr,
  1576. &hash_len);
  1577. if (session->ctx->max_retries > 1 &&
  1578. session->retries == session->ctx->max_retries) {
  1579. session->master_conn->up = rspamd_upstream_get_except (backend->u,
  1580. session->master_conn->up,
  1581. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1582. hash_key, hash_len);
  1583. }
  1584. else {
  1585. session->master_conn->up = rspamd_upstream_get (backend->u,
  1586. RSPAMD_UPSTREAM_ROUND_ROBIN,
  1587. hash_key, hash_len);
  1588. }
  1589. session->master_conn->timeout = backend->timeout;
  1590. if (session->master_conn->up == NULL) {
  1591. msg_err_session ("cannot select upstream for %s",
  1592. host ? hostbuf : "default");
  1593. goto err;
  1594. }
  1595. session->master_conn->backend_sock = rspamd_inet_address_connect (
  1596. rspamd_upstream_addr_next (session->master_conn->up),
  1597. SOCK_STREAM, TRUE);
  1598. if (session->master_conn->backend_sock == -1) {
  1599. msg_err_session ("cannot connect upstream: %s(%s)",
  1600. host ? hostbuf : "default",
  1601. rspamd_inet_address_to_string_pretty (
  1602. rspamd_upstream_addr_cur (
  1603. session->master_conn->up)));
  1604. rspamd_upstream_fail (session->master_conn->up, TRUE,
  1605. strerror (errno));
  1606. session->retries ++;
  1607. goto retry;
  1608. }
  1609. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1610. if (msg == NULL) {
  1611. msg_err_session ("cannot copy message to send it to the upstream: %e",
  1612. err);
  1613. if (err) {
  1614. g_error_free (err);
  1615. }
  1616. goto err; /* No fallback here */
  1617. }
  1618. session->master_conn->backend_conn = rspamd_http_connection_new_client_socket (
  1619. session->ctx->http_ctx,
  1620. NULL,
  1621. proxy_backend_master_error_handler,
  1622. proxy_backend_master_finish_handler,
  1623. RSPAMD_HTTP_CLIENT_SIMPLE,
  1624. session->master_conn->backend_sock);
  1625. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1626. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1627. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1628. if (backend->key) {
  1629. msg->peer_key = rspamd_pubkey_ref (backend->key);
  1630. }
  1631. if (backend->settings_id != NULL) {
  1632. rspamd_http_message_remove_header (msg, "Settings-ID");
  1633. rspamd_http_message_add_header (msg, "Settings-ID",
  1634. backend->settings_id);
  1635. }
  1636. if (backend->local ||
  1637. rspamd_inet_address_is_local (
  1638. rspamd_upstream_addr_cur (
  1639. session->master_conn->up))) {
  1640. if (session->fname) {
  1641. rspamd_http_message_add_header (msg, "File", session->fname);
  1642. }
  1643. msg->method = HTTP_GET;
  1644. rspamd_http_connection_write_message_shared (
  1645. session->master_conn->backend_conn,
  1646. msg, NULL, NULL, session->master_conn,
  1647. session->master_conn->timeout);
  1648. }
  1649. else {
  1650. if (session->fname) {
  1651. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1652. rspamd_http_message_set_body (msg,
  1653. session->map, session->map_len);
  1654. }
  1655. msg->method = HTTP_POST;
  1656. if (backend->compress) {
  1657. proxy_request_compress (msg);
  1658. if (session->client_milter_conn) {
  1659. rspamd_http_message_add_header (msg, "Content-Type",
  1660. "application/octet-stream");
  1661. }
  1662. }
  1663. else {
  1664. if (session->client_milter_conn) {
  1665. rspamd_http_message_add_header (msg, "Content-Type",
  1666. "text/plain");
  1667. }
  1668. }
  1669. rspamd_http_connection_write_message (
  1670. session->master_conn->backend_conn,
  1671. msg, NULL, NULL, session->master_conn,
  1672. session->master_conn->timeout);
  1673. }
  1674. }
  1675. return TRUE;
  1676. err:
  1677. if (session->client_milter_conn) {
  1678. rspamd_milter_send_action (session->client_milter_conn,
  1679. RSPAMD_MILTER_TEMPFAIL);
  1680. REF_RELEASE (session);
  1681. }
  1682. else {
  1683. rspamd_http_connection_steal_msg (session->client_conn);
  1684. rspamd_http_connection_reset (session->client_conn);
  1685. proxy_client_write_error (session, 404, "Backend not found");
  1686. }
  1687. return FALSE;
  1688. }
  1689. static void
  1690. proxy_client_error_handler (struct rspamd_http_connection *conn, GError *err)
  1691. {
  1692. struct rspamd_proxy_session *session = conn->ud;
  1693. msg_info_session ("abnormally closing connection from: %s, error: %s",
  1694. rspamd_inet_address_to_string (session->client_addr), err->message);
  1695. /* Terminate session immediately */
  1696. proxy_backend_close_connection (session->master_conn);
  1697. REF_RELEASE (session);
  1698. }
  1699. static gint
  1700. proxy_client_finish_handler (struct rspamd_http_connection *conn,
  1701. struct rspamd_http_message *msg)
  1702. {
  1703. struct rspamd_proxy_session *session = conn->ud;
  1704. if (!session->master_conn) {
  1705. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1706. sizeof (*session->master_conn));
  1707. session->master_conn->s = session;
  1708. session->master_conn->name = "master";
  1709. /* Reset spamc legacy */
  1710. if (msg->method >= HTTP_SYMBOLS) {
  1711. msg->method = HTTP_POST;
  1712. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1713. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1714. msg_info_session ("enabling legacy spamc mode for session");
  1715. }
  1716. else {
  1717. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1718. msg_info_session ("enabling legacy rspamc mode for session");
  1719. }
  1720. }
  1721. if (msg->url->len == 0) {
  1722. msg->url = rspamd_fstring_append (msg->url,
  1723. "/" MSG_CMD_CHECK_V2, strlen ("/" MSG_CMD_CHECK_V2));
  1724. }
  1725. if (!proxy_check_file (msg, session)) {
  1726. goto err;
  1727. }
  1728. session->client_message = rspamd_http_connection_steal_msg (
  1729. session->client_conn);
  1730. session->shmem_ref = rspamd_http_message_shmem_ref (session->client_message);
  1731. rspamd_http_message_remove_header (msg, "Content-Length");
  1732. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1733. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1734. rspamd_http_message_remove_header (msg, "Connection");
  1735. rspamd_http_message_remove_header (msg, "Key");
  1736. proxy_open_mirror_connections (session);
  1737. rspamd_http_connection_reset (session->client_conn);
  1738. proxy_send_master_message (session);
  1739. }
  1740. else {
  1741. msg_info_session ("finished master connection");
  1742. proxy_backend_close_connection (session->master_conn);
  1743. REF_RELEASE (session);
  1744. }
  1745. return 0;
  1746. err:
  1747. rspamd_http_connection_steal_msg (session->client_conn);
  1748. rspamd_http_message_remove_header (msg, "Content-Length");
  1749. rspamd_http_message_remove_header (msg, "Key");
  1750. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1751. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1752. rspamd_http_message_remove_header (msg, "Connection");
  1753. rspamd_http_connection_reset (session->client_conn);
  1754. proxy_client_write_error (session, 404, "Backend not found");
  1755. return 0;
  1756. }
  1757. static void
  1758. proxy_milter_finish_handler (gint fd,
  1759. struct rspamd_milter_session *rms,
  1760. void *ud)
  1761. {
  1762. struct rspamd_proxy_session *session = ud;
  1763. struct rspamd_http_message *msg;
  1764. session->client_milter_conn = rms;
  1765. if (rms->message == NULL || rms->message->len == 0) {
  1766. msg_info_session ("finished milter connection");
  1767. proxy_backend_close_connection (session->master_conn);
  1768. REF_RELEASE (session);
  1769. }
  1770. else {
  1771. if (!session->master_conn) {
  1772. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1773. sizeof (*session->master_conn));
  1774. }
  1775. msg = rspamd_milter_to_http (rms);
  1776. session->master_conn->s = session;
  1777. session->master_conn->name = "master";
  1778. session->client_message = msg;
  1779. proxy_open_mirror_connections (session);
  1780. proxy_send_master_message (session);
  1781. }
  1782. }
  1783. static void
  1784. proxy_milter_error_handler (gint fd,
  1785. struct rspamd_milter_session *rms,
  1786. void *ud, GError *err)
  1787. {
  1788. struct rspamd_proxy_session *session = ud;
  1789. msg_info_session ("abnormally closing milter connection from: %s, "
  1790. "error: %e",
  1791. rspamd_inet_address_to_string_pretty (session->client_addr),
  1792. err);
  1793. /* Terminate session immediately */
  1794. proxy_backend_close_connection (session->master_conn);
  1795. REF_RELEASE (session);
  1796. }
  1797. static void
  1798. proxy_accept_socket (EV_P_ ev_io *w, int revents)
  1799. {
  1800. struct rspamd_worker *worker = (struct rspamd_worker *)w->data;
  1801. struct rspamd_proxy_ctx *ctx;
  1802. rspamd_inet_addr_t *addr;
  1803. struct rspamd_proxy_session *session;
  1804. gint nfd;
  1805. ctx = worker->ctx;
  1806. if ((nfd =
  1807. rspamd_accept_from_socket (w->fd, &addr,
  1808. rspamd_worker_throttle_accept_events, worker->accept_events)) == -1) {
  1809. msg_warn ("accept failed: %s", strerror (errno));
  1810. return;
  1811. }
  1812. /* Check for EAGAIN */
  1813. if (nfd == 0) {
  1814. return;
  1815. }
  1816. session = g_malloc0 (sizeof (*session));
  1817. REF_INIT_RETAIN (session, proxy_session_dtor);
  1818. session->client_sock = nfd;
  1819. session->client_addr = addr;
  1820. session->mirror_conns = g_ptr_array_sized_new (ctx->mirrors->len);
  1821. session->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (),
  1822. "proxy", 0);
  1823. session->ctx = ctx;
  1824. session->worker = worker;
  1825. if (ctx->sessions_cache) {
  1826. rspamd_worker_session_cache_add (ctx->sessions_cache,
  1827. session->pool->tag.uid, &session->ref.refcount, session);
  1828. }
  1829. if (!ctx->milter) {
  1830. session->client_conn = rspamd_http_connection_new_server (
  1831. ctx->http_ctx,
  1832. nfd,
  1833. NULL,
  1834. proxy_client_error_handler,
  1835. proxy_client_finish_handler,
  1836. 0);
  1837. if (ctx->key) {
  1838. rspamd_http_connection_set_key (session->client_conn, ctx->key);
  1839. }
  1840. msg_info_session ("accepted http connection from %s port %d",
  1841. rspamd_inet_address_to_string (addr),
  1842. rspamd_inet_address_get_port (addr));
  1843. rspamd_http_connection_read_message_shared (session->client_conn,
  1844. session,
  1845. session->ctx->timeout);
  1846. }
  1847. else {
  1848. msg_info_session ("accepted milter connection from %s port %d",
  1849. rspamd_inet_address_to_string (addr),
  1850. rspamd_inet_address_get_port (addr));
  1851. #ifdef TCP_NODELAY
  1852. #ifndef SOL_TCP
  1853. #define SOL_TCP IPPROTO_TCP
  1854. #endif
  1855. if (rspamd_inet_address_get_af (addr) != AF_UNIX) {
  1856. gint sopt = 1;
  1857. if (setsockopt (nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof (sopt)) ==
  1858. -1) {
  1859. msg_warn_session ("cannot set TCP_NODELAY: %s",
  1860. strerror (errno));
  1861. }
  1862. }
  1863. #endif
  1864. rspamd_milter_handle_socket (nfd, 0.0,
  1865. session->pool,
  1866. ctx->event_loop,
  1867. proxy_milter_finish_handler,
  1868. proxy_milter_error_handler,
  1869. session);
  1870. }
  1871. }
  1872. static void
  1873. adjust_upstreams_limits (struct rspamd_proxy_ctx *ctx)
  1874. {
  1875. struct rspamd_http_upstream *backend;
  1876. gpointer k, v;
  1877. GHashTableIter it;
  1878. /*
  1879. * We set error time equal to max_retries * backend_timeout and max_errors
  1880. * to max_retries - 1
  1881. *
  1882. * So if we failed to scan a message on a backend for some reasons, we
  1883. * will try to re-resolve it faster
  1884. */
  1885. g_hash_table_iter_init (&it, ctx->upstreams);
  1886. while (g_hash_table_iter_next (&it, &k, &v)) {
  1887. backend = (struct rspamd_http_upstream *)v;
  1888. if (!backend->self_scan && backend->u) {
  1889. rspamd_upstreams_set_limits (backend->u,
  1890. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1891. ctx->max_retries - 1, 0);
  1892. }
  1893. }
  1894. }
  1895. void
  1896. start_rspamd_proxy (struct rspamd_worker *worker)
  1897. {
  1898. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1899. gboolean is_controller = FALSE;
  1900. g_assert (rspamd_worker_check_context (worker->ctx, rspamd_rspamd_proxy_magic));
  1901. ctx->cfg = worker->srv->cfg;
  1902. ctx->event_loop = rspamd_prepare_worker (worker, "rspamd_proxy",
  1903. proxy_accept_socket);
  1904. ctx->resolver = rspamd_dns_resolver_init (worker->srv->logger,
  1905. ctx->event_loop,
  1906. worker->srv->cfg);
  1907. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  1908. ctx->event_loop, ctx->resolver->r);
  1909. ctx->http_ctx = rspamd_http_context_create (ctx->cfg, ctx->event_loop,
  1910. ctx->cfg->ups_ctx);
  1911. rspamd_mempool_add_destructor (ctx->cfg->cfg_pool,
  1912. (rspamd_mempool_destruct_t)rspamd_http_context_free,
  1913. ctx->http_ctx);
  1914. if (ctx->has_self_scan) {
  1915. /* Additional initialisation needed */
  1916. rspamd_worker_init_scanner (worker, ctx->event_loop, ctx->resolver,
  1917. &ctx->lang_det);
  1918. if (worker->index == 0) {
  1919. /*
  1920. * If there are no controllers and no normal workers,
  1921. * then pretend that we are a controller
  1922. */
  1923. gboolean controller_seen = FALSE;
  1924. GList *cur;
  1925. cur = worker->srv->cfg->workers;
  1926. while (cur) {
  1927. struct rspamd_worker_conf *cf;
  1928. cf = (struct rspamd_worker_conf *)cur->data;
  1929. if ((cf->type == g_quark_from_static_string ("controller")) ||
  1930. (cf->type == g_quark_from_static_string ("normal"))) {
  1931. if (cf->enabled && cf->count >= 0) {
  1932. controller_seen = TRUE;
  1933. break;
  1934. }
  1935. }
  1936. cur = g_list_next (cur);
  1937. }
  1938. if (!controller_seen) {
  1939. msg_info ("no controller or normal workers defined, execute "
  1940. "controller periodics in this worker");
  1941. worker->flags |= RSPAMD_WORKER_CONTROLLER;
  1942. is_controller = TRUE;
  1943. }
  1944. }
  1945. }
  1946. else {
  1947. worker->flags &= ~RSPAMD_WORKER_SCANNER;
  1948. }
  1949. if (worker->srv->cfg->enable_sessions_cache) {
  1950. ctx->sessions_cache = rspamd_worker_session_cache_new (worker,
  1951. ctx->event_loop);
  1952. }
  1953. ctx->milter_ctx.spam_header = ctx->spam_header;
  1954. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  1955. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  1956. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  1957. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  1958. ctx->milter_ctx.reject_message = ctx->reject_message;
  1959. ctx->milter_ctx.cfg = ctx->cfg;
  1960. rspamd_milter_init_library (&ctx->milter_ctx);
  1961. if (is_controller) {
  1962. rspamd_worker_init_controller (worker, NULL);
  1963. }
  1964. else {
  1965. if (ctx->has_self_scan) {
  1966. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  1967. worker, RSPAMD_MAP_WATCH_SCANNER);
  1968. }
  1969. else {
  1970. rspamd_map_watch (worker->srv->cfg, ctx->event_loop, ctx->resolver,
  1971. worker, RSPAMD_MAP_WATCH_WORKER);
  1972. }
  1973. }
  1974. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->event_loop,
  1975. worker);
  1976. adjust_upstreams_limits (ctx);
  1977. ev_loop (ctx->event_loop, 0);
  1978. rspamd_worker_block_signals ();
  1979. if (ctx->has_self_scan) {
  1980. rspamd_stat_close ();
  1981. }
  1982. if (is_controller) {
  1983. rspamd_controller_on_terminate (worker, NULL);
  1984. }
  1985. REF_RELEASE (ctx->cfg);
  1986. rspamd_log_close (worker->srv->logger);
  1987. exit (EXIT_SUCCESS);
  1988. }