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 58KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266
  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 "libutil/map.h"
  19. #include "libutil/upstream.h"
  20. #include "libutil/http.h"
  21. #include "libutil/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. struct timeval io_tv;
  81. gint parser_from_ref;
  82. gint parser_to_ref;
  83. gboolean local;
  84. gboolean self_scan;
  85. gboolean compress;
  86. };
  87. struct rspamd_http_mirror {
  88. gchar *name;
  89. gchar *settings_id;
  90. struct upstream_list *u;
  91. struct rspamd_cryptobox_pubkey *key;
  92. gdouble prob;
  93. gdouble timeout;
  94. struct timeval io_tv;
  95. gint parser_from_ref;
  96. gint parser_to_ref;
  97. gboolean local;
  98. gboolean compress;
  99. };
  100. static const guint64 rspamd_rspamd_proxy_magic = 0xcdeb4fd1fc351980ULL;
  101. struct rspamd_proxy_ctx {
  102. guint64 magic;
  103. /* Events base */
  104. struct event_base *ev_base;
  105. /* DNS resolver */
  106. struct rspamd_dns_resolver *resolver;
  107. /* Config */
  108. struct rspamd_config *cfg;
  109. /* END OF COMMON PART */
  110. gdouble timeout;
  111. struct timeval io_tv;
  112. /* Encryption key for clients */
  113. struct rspamd_cryptobox_keypair *key;
  114. /* Keys cache */
  115. struct rspamd_keypair_cache *keys_cache;
  116. /* Upstreams to use */
  117. GHashTable *upstreams;
  118. /* Mirrors to send traffic to */
  119. GPtrArray *mirrors;
  120. /* Default upstream */
  121. struct rspamd_http_upstream *default_upstream;
  122. /* Local rotating keypair for upstreams */
  123. struct rspamd_cryptobox_keypair *local_key;
  124. struct event rotate_ev;
  125. gdouble rotate_tm;
  126. lua_State *lua_state;
  127. /* Array of callback functions called on end of scan to compare results */
  128. GArray *cmp_refs;
  129. /* Maximum count for retries */
  130. guint max_retries;
  131. /* If we have self_scanning backends, we need to work as a normal worker */
  132. gboolean has_self_scan;
  133. /* It is not HTTP but milter proxy */
  134. gboolean milter;
  135. /* Discard messages instead of rejecting them */
  136. gboolean discard_on_reject;
  137. /* Quarantine messages instead of rejecting them */
  138. gboolean quarantine_on_reject;
  139. /* Milter spam header */
  140. gchar *spam_header;
  141. /* CA name that can be used for client certificates */
  142. gchar *client_ca_name;
  143. /* Milter rejection message */
  144. gchar *reject_message;
  145. /* Sessions cache */
  146. void *sessions_cache;
  147. struct rspamd_milter_context milter_ctx;
  148. /* Language detector */
  149. struct rspamd_lang_detector *lang_det;
  150. };
  151. enum rspamd_backend_flags {
  152. RSPAMD_BACKEND_REPLIED = 1 << 0,
  153. RSPAMD_BACKEND_CLOSED = 1 << 1,
  154. RSPAMD_BACKEND_PARSED = 1 << 2,
  155. };
  156. struct rspamd_proxy_session;
  157. struct rspamd_proxy_backend_connection {
  158. const gchar *name;
  159. struct rspamd_cryptobox_keypair *local_key;
  160. struct rspamd_cryptobox_pubkey *remote_key;
  161. struct upstream *up;
  162. struct rspamd_http_connection *backend_conn;
  163. ucl_object_t *results;
  164. const gchar *err;
  165. struct rspamd_proxy_session *s;
  166. struct timeval *io_tv;
  167. gint backend_sock;
  168. enum rspamd_backend_flags flags;
  169. gint parser_from_ref;
  170. gint parser_to_ref;
  171. struct rspamd_task *task;
  172. };
  173. enum rspamd_proxy_legacy_support {
  174. LEGACY_SUPPORT_NO = 0,
  175. LEGACY_SUPPORT_RSPAMC,
  176. LEGACY_SUPPORT_SPAMC
  177. };
  178. struct rspamd_proxy_session {
  179. struct rspamd_worker *worker;
  180. rspamd_mempool_t *pool;
  181. struct rspamd_proxy_ctx *ctx;
  182. rspamd_inet_addr_t *client_addr;
  183. struct rspamd_http_connection *client_conn;
  184. struct rspamd_milter_session *client_milter_conn;
  185. struct rspamd_http_upstream *backend;
  186. gpointer map;
  187. gchar *fname;
  188. gpointer shmem_ref;
  189. struct rspamd_proxy_backend_connection *master_conn;
  190. struct rspamd_http_message *client_message;
  191. GPtrArray *mirror_conns;
  192. gsize map_len;
  193. gint client_sock;
  194. enum rspamd_proxy_legacy_support legacy_support;
  195. gint retries;
  196. ref_entry_t ref;
  197. };
  198. static gboolean proxy_send_master_message (struct rspamd_proxy_session *session);
  199. static GQuark
  200. rspamd_proxy_quark (void)
  201. {
  202. return g_quark_from_static_string ("rspamd-proxy");
  203. }
  204. static gboolean
  205. rspamd_proxy_parse_lua_parser (lua_State *L, const ucl_object_t *obj,
  206. gint *ref_from, gint *ref_to, GError **err)
  207. {
  208. const gchar *lua_script;
  209. gsize slen;
  210. gint err_idx, ref_idx;
  211. GString *tb = NULL;
  212. gboolean has_ref = FALSE;
  213. g_assert (obj != NULL);
  214. g_assert (ref_from != NULL);
  215. g_assert (ref_to != NULL);
  216. *ref_from = -1;
  217. *ref_to = -1;
  218. lua_script = ucl_object_tolstring (obj, &slen);
  219. lua_pushcfunction (L, &rspamd_lua_traceback);
  220. err_idx = lua_gettop (L);
  221. /* Load data */
  222. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  223. g_set_error (err,
  224. rspamd_proxy_quark (),
  225. EINVAL,
  226. "cannot load lua parser script: %s",
  227. lua_tostring (L, -1));
  228. lua_settop (L, 0); /* Error function */
  229. return FALSE;
  230. }
  231. /* Now do it */
  232. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  233. tb = lua_touserdata (L, -1);
  234. g_set_error (err,
  235. rspamd_proxy_quark (),
  236. EINVAL,
  237. "cannot init lua parser script: %s",
  238. tb->str);
  239. g_string_free (tb, TRUE);
  240. lua_settop (L, 0);
  241. return FALSE;
  242. }
  243. if (lua_istable (L, -1)) {
  244. /*
  245. * We have a table, so we check for two keys:
  246. * 'from' -> function
  247. * 'to' -> function
  248. *
  249. * From converts parent request to a client one
  250. * To converts client request to a parent one
  251. */
  252. lua_pushstring (L, "from");
  253. lua_gettable (L, -2);
  254. if (lua_isfunction (L, -1)) {
  255. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  256. *ref_from = ref_idx;
  257. has_ref = TRUE;
  258. }
  259. lua_pushstring (L, "to");
  260. lua_gettable (L, -2);
  261. if (lua_isfunction (L, -1)) {
  262. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  263. *ref_to = ref_idx;
  264. has_ref = TRUE;
  265. }
  266. }
  267. else if (!lua_isfunction (L, -1)) {
  268. g_set_error (err,
  269. rspamd_proxy_quark (),
  270. EINVAL,
  271. "cannot init lua parser script: "
  272. "must return function");
  273. lua_settop (L, 0);
  274. return FALSE;
  275. }
  276. else {
  277. /* Just parser from protocol */
  278. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  279. *ref_from = ref_idx;
  280. lua_settop (L, 0);
  281. has_ref = TRUE;
  282. }
  283. return has_ref;
  284. }
  285. static gboolean
  286. rspamd_proxy_parse_upstream (rspamd_mempool_t *pool,
  287. const ucl_object_t *obj,
  288. gpointer ud,
  289. struct rspamd_rcl_section *section,
  290. GError **err)
  291. {
  292. const ucl_object_t *elt;
  293. struct rspamd_http_upstream *up = NULL;
  294. struct rspamd_proxy_ctx *ctx;
  295. struct rspamd_rcl_struct_parser *pd = ud;
  296. lua_State *L;
  297. ctx = pd->user_struct;
  298. L = ctx->lua_state;
  299. if (ucl_object_type (obj) != UCL_OBJECT) {
  300. g_set_error (err, rspamd_proxy_quark (), 100,
  301. "upstream option must be an object");
  302. return FALSE;
  303. }
  304. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  305. elt = ucl_object_lookup (obj, "name");
  306. if (elt == NULL) {
  307. if (ucl_object_key (obj)) {
  308. if (strcmp (ucl_object_key (obj), "upstream") == 0) {
  309. /* Iterate over the object and find upstream elements */
  310. ucl_object_iter_t it = NULL;
  311. const ucl_object_t *cur;
  312. gboolean ret = TRUE;
  313. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  314. if (!rspamd_proxy_parse_upstream (pool, cur, ud,
  315. section, err)) {
  316. ret = FALSE;
  317. }
  318. }
  319. return ret;
  320. }
  321. else {
  322. /* Inside upstream */
  323. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  324. }
  325. }
  326. else {
  327. g_set_error (err, rspamd_proxy_quark (), 100,
  328. "upstream option must have some name definition");
  329. return FALSE;
  330. }
  331. }
  332. else {
  333. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  334. }
  335. up->parser_from_ref = -1;
  336. up->parser_to_ref = -1;
  337. up->timeout = ctx->timeout;
  338. elt = ucl_object_lookup (obj, "key");
  339. if (elt != NULL) {
  340. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  341. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  342. if (up->key == NULL) {
  343. g_set_error (err, rspamd_proxy_quark (), 100,
  344. "cannot read upstream key");
  345. goto err;
  346. }
  347. rspamd_mempool_add_destructor (pool,
  348. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  349. }
  350. elt = ucl_object_lookup (obj, "self_scan");
  351. if (elt && ucl_object_toboolean (elt)) {
  352. up->self_scan = TRUE;
  353. ctx->has_self_scan = TRUE;
  354. }
  355. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  356. if (elt && ucl_object_toboolean (elt)) {
  357. up->compress = TRUE;
  358. }
  359. elt = ucl_object_lookup (obj, "hosts");
  360. if (elt == NULL && !up->self_scan) {
  361. g_set_error (err, rspamd_proxy_quark (), 100,
  362. "upstream option must have some hosts definition");
  363. goto err;
  364. }
  365. if (elt) {
  366. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  367. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  368. g_set_error (err, rspamd_proxy_quark (), 100,
  369. "upstream has bad hosts definition");
  370. goto err;
  371. }
  372. rspamd_mempool_add_destructor (pool,
  373. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  374. }
  375. elt = ucl_object_lookup (obj, "default");
  376. if (elt) {
  377. if (ucl_object_toboolean (elt)) {
  378. ctx->default_upstream = up;
  379. }
  380. }
  381. else if (up->self_scan) {
  382. ctx->default_upstream = up;
  383. }
  384. elt = ucl_object_lookup (obj, "local");
  385. if (elt && ucl_object_toboolean (elt)) {
  386. up->local = TRUE;
  387. }
  388. elt = ucl_object_lookup (obj, "timeout");
  389. if (elt) {
  390. ucl_object_todouble_safe (elt, &up->timeout);
  391. }
  392. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  393. if (elt && ucl_object_type (elt) == UCL_STRING) {
  394. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  395. }
  396. /*
  397. * Accept lua function here in form
  398. * fun :: String -> UCL
  399. */
  400. elt = ucl_object_lookup (obj, "parser");
  401. if (elt) {
  402. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  403. &up->parser_to_ref, err)) {
  404. goto err;
  405. }
  406. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  407. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  408. }
  409. double_to_tv (up->timeout, &up->io_tv);
  410. g_hash_table_insert (ctx->upstreams, up->name, up);
  411. return TRUE;
  412. err:
  413. return FALSE;
  414. }
  415. static gboolean
  416. rspamd_proxy_parse_mirror (rspamd_mempool_t *pool,
  417. const ucl_object_t *obj,
  418. gpointer ud,
  419. struct rspamd_rcl_section *section,
  420. GError **err)
  421. {
  422. const ucl_object_t *elt;
  423. struct rspamd_http_mirror *up = NULL;
  424. struct rspamd_proxy_ctx *ctx;
  425. struct rspamd_rcl_struct_parser *pd = ud;
  426. lua_State *L;
  427. ctx = pd->user_struct;
  428. L = ctx->lua_state;
  429. if (ucl_object_type (obj) != UCL_OBJECT) {
  430. g_set_error (err, rspamd_proxy_quark (), 100,
  431. "mirror option must be an object");
  432. return FALSE;
  433. }
  434. up = rspamd_mempool_alloc0 (pool, sizeof (*up));
  435. elt = ucl_object_lookup (obj, "name");
  436. if (elt == NULL) {
  437. if (ucl_object_key (obj)) {
  438. if (strcmp (ucl_object_key (obj), "mirror") == 0) {
  439. /* Iterate over the object and find upstream elements */
  440. ucl_object_iter_t it = NULL;
  441. const ucl_object_t *cur;
  442. gboolean ret = TRUE;
  443. while ((cur = ucl_object_iterate (obj, &it, true)) != NULL) {
  444. if (!rspamd_proxy_parse_mirror (pool, cur, ud,
  445. section, err)) {
  446. ret = FALSE;
  447. }
  448. }
  449. return ret;
  450. }
  451. else {
  452. /* Inside upstream */
  453. up->name = rspamd_mempool_strdup (pool, ucl_object_key (obj));
  454. }
  455. }
  456. else {
  457. g_set_error (err, rspamd_proxy_quark (), 100,
  458. "mirror option must have some name definition");
  459. return FALSE;
  460. }
  461. }
  462. else {
  463. up->name = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  464. }
  465. up->parser_to_ref = -1;
  466. up->parser_from_ref = -1;
  467. up->timeout = ctx->timeout;
  468. elt = ucl_object_lookup (obj, "key");
  469. if (elt != NULL) {
  470. up->key = rspamd_pubkey_from_base32 (ucl_object_tostring (elt), 0,
  471. RSPAMD_KEYPAIR_KEX, RSPAMD_CRYPTOBOX_MODE_25519);
  472. if (up->key == NULL) {
  473. g_set_error (err, rspamd_proxy_quark (), 100,
  474. "cannot read mirror key");
  475. goto err;
  476. }
  477. rspamd_mempool_add_destructor (pool,
  478. (rspamd_mempool_destruct_t)rspamd_pubkey_unref, up->key);
  479. }
  480. elt = ucl_object_lookup (obj, "hosts");
  481. if (elt == NULL) {
  482. g_set_error (err, rspamd_proxy_quark (), 100,
  483. "mirror option must have some hosts definition");
  484. goto err;
  485. }
  486. up->u = rspamd_upstreams_create (ctx->cfg->ups_ctx);
  487. if (!rspamd_upstreams_from_ucl (up->u, elt, 11333, NULL)) {
  488. g_set_error (err, rspamd_proxy_quark (), 100,
  489. "mirror has bad hosts definition");
  490. goto err;
  491. }
  492. rspamd_mempool_add_destructor (pool,
  493. (rspamd_mempool_destruct_t)rspamd_upstreams_destroy, up->u);
  494. elt = ucl_object_lookup_any (obj, "probability", "prob", NULL);
  495. if (elt) {
  496. up->prob = ucl_object_todouble (elt);
  497. }
  498. else {
  499. up->prob = 1.0;
  500. }
  501. elt = ucl_object_lookup (obj, "local");
  502. if (elt && ucl_object_toboolean (elt)) {
  503. up->local = TRUE;
  504. }
  505. elt = ucl_object_lookup_any (obj, "compress", "compression", NULL);
  506. if (elt && ucl_object_toboolean (elt)) {
  507. up->compress = TRUE;
  508. }
  509. elt = ucl_object_lookup (obj, "timeout");
  510. if (elt) {
  511. ucl_object_todouble_safe (elt, &up->timeout);
  512. }
  513. /*
  514. * Accept lua function here in form
  515. * fun :: String -> UCL
  516. */
  517. elt = ucl_object_lookup (obj, "parser");
  518. if (elt) {
  519. if (!rspamd_proxy_parse_lua_parser (L, elt, &up->parser_from_ref,
  520. &up->parser_to_ref, err)) {
  521. goto err;
  522. }
  523. rspamd_lua_add_ref_dtor (L, pool, up->parser_from_ref);
  524. rspamd_lua_add_ref_dtor (L, pool, up->parser_to_ref);
  525. }
  526. elt = ucl_object_lookup_any (obj, "settings", "settings_id", NULL);
  527. if (elt && ucl_object_type (elt) == UCL_STRING) {
  528. up->settings_id = rspamd_mempool_strdup (pool, ucl_object_tostring (elt));
  529. }
  530. double_to_tv (up->timeout, &up->io_tv);
  531. g_ptr_array_add (ctx->mirrors, up);
  532. return TRUE;
  533. err:
  534. return FALSE;
  535. }
  536. static gboolean
  537. rspamd_proxy_parse_script (rspamd_mempool_t *pool,
  538. const ucl_object_t *obj,
  539. gpointer ud,
  540. struct rspamd_rcl_section *section,
  541. GError **err)
  542. {
  543. struct rspamd_proxy_ctx *ctx;
  544. struct rspamd_rcl_struct_parser *pd = ud;
  545. lua_State *L;
  546. const gchar *lua_script;
  547. gsize slen;
  548. gint err_idx, ref_idx;
  549. GString *tb = NULL;
  550. struct stat st;
  551. ctx = pd->user_struct;
  552. L = ctx->lua_state;
  553. if (ucl_object_type (obj) != UCL_STRING) {
  554. g_set_error (err, rspamd_proxy_quark (), 100,
  555. "script option must be a string with file or lua chunk");
  556. return FALSE;
  557. }
  558. lua_script = ucl_object_tolstring (obj, &slen);
  559. lua_pushcfunction (L, &rspamd_lua_traceback);
  560. err_idx = lua_gettop (L);
  561. if (stat (lua_script, &st) != -1) {
  562. /* Load file */
  563. if (luaL_loadfile (L, lua_script) != 0) {
  564. g_set_error (err,
  565. rspamd_proxy_quark (),
  566. EINVAL,
  567. "cannot load lua parser script: %s",
  568. lua_tostring (L, -1));
  569. lua_settop (L, 0); /* Error function */
  570. goto err;
  571. }
  572. }
  573. else {
  574. /* Load data directly */
  575. if (luaL_loadbuffer (L, lua_script, slen, "proxy parser") != 0) {
  576. g_set_error (err,
  577. rspamd_proxy_quark (),
  578. EINVAL,
  579. "cannot load lua parser script: %s",
  580. lua_tostring (L, -1));
  581. lua_settop (L, 0); /* Error function */
  582. goto err;
  583. }
  584. }
  585. /* Now do it */
  586. if (lua_pcall (L, 0, 1, err_idx) != 0) {
  587. tb = lua_touserdata (L, -1);
  588. g_set_error (err,
  589. rspamd_proxy_quark (),
  590. EINVAL,
  591. "cannot init lua parser script: %s",
  592. tb->str);
  593. g_string_free (tb, TRUE);
  594. lua_settop (L, 0);
  595. goto err;
  596. }
  597. if (!lua_isfunction (L, -1)) {
  598. g_set_error (err,
  599. rspamd_proxy_quark (),
  600. EINVAL,
  601. "cannot init lua parser script: "
  602. "must return function, %s returned",
  603. lua_typename (L, lua_type (L, -1)));
  604. lua_settop (L, 0);
  605. goto err;
  606. }
  607. ref_idx = luaL_ref (L, LUA_REGISTRYINDEX);
  608. lua_settop (L, 0);
  609. g_array_append_val (ctx->cmp_refs, ref_idx);
  610. return TRUE;
  611. err:
  612. return FALSE;
  613. }
  614. gpointer
  615. init_rspamd_proxy (struct rspamd_config *cfg)
  616. {
  617. struct rspamd_proxy_ctx *ctx;
  618. GQuark type;
  619. type = g_quark_try_string ("rspamd_proxy");
  620. ctx = rspamd_mempool_alloc0 (cfg->cfg_pool,
  621. sizeof (struct rspamd_proxy_ctx));
  622. ctx->magic = rspamd_rspamd_proxy_magic;
  623. ctx->timeout = 120.0;
  624. ctx->upstreams = g_hash_table_new (rspamd_strcase_hash, rspamd_strcase_equal);
  625. rspamd_mempool_add_destructor (cfg->cfg_pool,
  626. (rspamd_mempool_destruct_t)g_hash_table_unref, ctx->upstreams);
  627. ctx->mirrors = g_ptr_array_new ();
  628. rspamd_mempool_add_destructor (cfg->cfg_pool,
  629. (rspamd_mempool_destruct_t)rspamd_ptr_array_free_hard, ctx->mirrors);
  630. ctx->rotate_tm = DEFAULT_ROTATION_TIME;
  631. ctx->cfg = cfg;
  632. ctx->lua_state = cfg->lua_state;
  633. ctx->cmp_refs = g_array_new (FALSE, FALSE, sizeof (gint));
  634. rspamd_mempool_add_destructor (cfg->cfg_pool,
  635. (rspamd_mempool_destruct_t)rspamd_array_free_hard, ctx->cmp_refs);
  636. ctx->max_retries = DEFAULT_RETRIES;
  637. ctx->spam_header = RSPAMD_MILTER_SPAM_HEADER;
  638. rspamd_rcl_register_worker_option (cfg,
  639. type,
  640. "timeout",
  641. rspamd_rcl_parse_struct_time,
  642. ctx,
  643. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, timeout),
  644. RSPAMD_CL_FLAG_TIME_FLOAT,
  645. "IO timeout");
  646. rspamd_rcl_register_worker_option (cfg,
  647. type,
  648. "rotate",
  649. rspamd_rcl_parse_struct_time,
  650. ctx,
  651. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, rotate_tm),
  652. RSPAMD_CL_FLAG_TIME_FLOAT,
  653. "Rotation keys time, default: "
  654. G_STRINGIFY (DEFAULT_ROTATION_TIME) " seconds");
  655. rspamd_rcl_register_worker_option (cfg,
  656. type,
  657. "keypair",
  658. rspamd_rcl_parse_struct_keypair,
  659. ctx,
  660. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, key),
  661. 0,
  662. "Server's keypair");
  663. rspamd_rcl_register_worker_option (cfg,
  664. type,
  665. "upstream",
  666. rspamd_proxy_parse_upstream,
  667. ctx,
  668. 0,
  669. 0,
  670. "List of upstreams");
  671. rspamd_rcl_register_worker_option (cfg,
  672. type,
  673. "mirror",
  674. rspamd_proxy_parse_mirror,
  675. ctx,
  676. 0,
  677. RSPAMD_CL_FLAG_MULTIPLE,
  678. "List of mirrors");
  679. rspamd_rcl_register_worker_option (cfg,
  680. type,
  681. "script",
  682. rspamd_proxy_parse_script,
  683. ctx,
  684. 0,
  685. RSPAMD_CL_FLAG_MULTIPLE,
  686. "Compare script to be executed");
  687. rspamd_rcl_register_worker_option (cfg,
  688. type,
  689. "max_retries",
  690. rspamd_rcl_parse_struct_integer,
  691. ctx,
  692. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, max_retries),
  693. RSPAMD_CL_FLAG_UINT,
  694. "Maximum number of retries for master connection");
  695. rspamd_rcl_register_worker_option (cfg,
  696. type,
  697. "milter",
  698. rspamd_rcl_parse_struct_boolean,
  699. ctx,
  700. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, milter),
  701. 0,
  702. "Accept milter connections, not HTTP");
  703. rspamd_rcl_register_worker_option (cfg,
  704. type,
  705. "discard_on_reject",
  706. rspamd_rcl_parse_struct_boolean,
  707. ctx,
  708. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, discard_on_reject),
  709. 0,
  710. "Tell MTA to discard rejected messages silently");
  711. rspamd_rcl_register_worker_option (cfg,
  712. type,
  713. "quarantine_on_reject",
  714. rspamd_rcl_parse_struct_boolean,
  715. ctx,
  716. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, quarantine_on_reject),
  717. 0,
  718. "Tell MTA to quarantine rejected messages");
  719. rspamd_rcl_register_worker_option (cfg,
  720. type,
  721. "spam_header",
  722. rspamd_rcl_parse_struct_string,
  723. ctx,
  724. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, spam_header),
  725. 0,
  726. "Use the specific spam header (default: X-Spam)");
  727. rspamd_rcl_register_worker_option (cfg,
  728. type,
  729. "client_ca_name",
  730. rspamd_rcl_parse_struct_string,
  731. ctx,
  732. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, client_ca_name),
  733. 0,
  734. "Allow certificates issued by this CA to be treated as client certificates");
  735. rspamd_rcl_register_worker_option (cfg,
  736. type,
  737. "reject_message",
  738. rspamd_rcl_parse_struct_string,
  739. ctx,
  740. G_STRUCT_OFFSET (struct rspamd_proxy_ctx, reject_message),
  741. 0,
  742. "Use custom rejection message");
  743. return ctx;
  744. }
  745. static void
  746. proxy_backend_close_connection (struct rspamd_proxy_backend_connection *conn)
  747. {
  748. if (conn && !(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  749. if (conn->backend_conn) {
  750. rspamd_http_connection_reset (conn->backend_conn);
  751. rspamd_http_connection_unref (conn->backend_conn);
  752. close (conn->backend_sock);
  753. }
  754. conn->flags |= RSPAMD_BACKEND_CLOSED;
  755. }
  756. }
  757. static gboolean
  758. proxy_backend_parse_results (struct rspamd_proxy_session *session,
  759. struct rspamd_proxy_backend_connection *conn,
  760. lua_State *L, gint parser_ref,
  761. const gchar *in, gsize inlen)
  762. {
  763. struct ucl_parser *parser;
  764. GString *tb = NULL;
  765. gint err_idx;
  766. if (inlen == 0 || in == NULL) {
  767. return FALSE;
  768. }
  769. if (parser_ref != -1) {
  770. /* Call parser function */
  771. lua_pushcfunction (L, &rspamd_lua_traceback);
  772. err_idx = lua_gettop (L);
  773. lua_rawgeti (L, LUA_REGISTRYINDEX, parser_ref);
  774. /* XXX: copies all data */
  775. lua_pushlstring (L, in, inlen);
  776. if (lua_pcall (L, 1, 1, err_idx) != 0) {
  777. tb = lua_touserdata (L, -1);
  778. msg_err_session (
  779. "cannot run lua parser script: %s",
  780. tb->str);
  781. g_string_free (tb, TRUE);
  782. lua_settop (L, 0);
  783. return FALSE;
  784. }
  785. conn->results = ucl_object_lua_import (L, -1);
  786. lua_settop (L, 0);
  787. }
  788. else {
  789. parser = ucl_parser_new (0);
  790. if (!ucl_parser_add_chunk (parser, in, inlen)) {
  791. gchar *encoded;
  792. encoded = rspamd_encode_base64 (in, inlen, 0, NULL);
  793. msg_err_session ("cannot parse input: %s", ucl_parser_get_error (
  794. parser));
  795. msg_err_session ("input encoded: %s", encoded);
  796. ucl_parser_free (parser);
  797. g_free (encoded);
  798. return FALSE;
  799. }
  800. conn->results = ucl_parser_get_object (parser);
  801. ucl_parser_free (parser);
  802. }
  803. return TRUE;
  804. }
  805. static void
  806. proxy_call_cmp_script (struct rspamd_proxy_session *session, gint cbref)
  807. {
  808. GString *tb = NULL;
  809. gint err_idx;
  810. guint i;
  811. struct rspamd_proxy_backend_connection *conn;
  812. lua_State *L;
  813. L = session->ctx->lua_state;
  814. lua_pushcfunction (L, &rspamd_lua_traceback);
  815. err_idx = lua_gettop (L);
  816. lua_rawgeti (L, LUA_REGISTRYINDEX, cbref);
  817. lua_createtable (L, 0, session->mirror_conns->len + 1);
  818. /* Now push master results */
  819. if (session->master_conn && session->master_conn->results) {
  820. lua_pushstring (L, "master");
  821. ucl_object_push_lua (L, session->master_conn->results, true);
  822. lua_settable (L, -3);
  823. }
  824. else {
  825. lua_pushstring (L, "master");
  826. lua_pushstring (L, "no results");
  827. lua_settable (L, -3);
  828. }
  829. for (i = 0; i < session->mirror_conns->len; i ++) {
  830. conn = g_ptr_array_index (session->mirror_conns, i);
  831. if (conn->results) {
  832. lua_pushstring (L, conn->name);
  833. ucl_object_push_lua (L, conn->results, true);
  834. lua_settable (L, -3);
  835. }
  836. else {
  837. lua_pushstring (L, conn->name);
  838. lua_pushstring (L, conn->err ? conn->err : "unknown error");
  839. lua_settable (L, -3);
  840. }
  841. }
  842. if (lua_pcall (L, 1, 0, err_idx) != 0) {
  843. tb = lua_touserdata (L, -1);
  844. msg_err_session (
  845. "cannot run lua compare script: %s",
  846. tb->str);
  847. g_string_free (tb, TRUE);
  848. }
  849. lua_settop (L, 0);
  850. }
  851. static void
  852. proxy_session_dtor (struct rspamd_proxy_session *session)
  853. {
  854. guint i;
  855. gint cbref;
  856. struct rspamd_proxy_backend_connection *conn;
  857. if (session->master_conn && session->master_conn->results) {
  858. for (i = 0; i < session->ctx->cmp_refs->len; i++) {
  859. cbref = g_array_index (session->ctx->cmp_refs, gint, i);
  860. proxy_call_cmp_script (session, cbref);
  861. }
  862. }
  863. if (session->master_conn) {
  864. proxy_backend_close_connection (session->master_conn);
  865. }
  866. if (session->client_milter_conn) {
  867. rspamd_milter_session_unref (session->client_milter_conn);
  868. }
  869. else if (session->client_conn) {
  870. rspamd_http_connection_reset (session->client_conn);
  871. rspamd_http_connection_unref (session->client_conn);
  872. }
  873. if (session->map && session->map_len) {
  874. munmap (session->map, session->map_len);
  875. }
  876. for (i = 0; i < session->mirror_conns->len; i ++) {
  877. conn = g_ptr_array_index (session->mirror_conns, i);
  878. if (!(conn->flags & RSPAMD_BACKEND_CLOSED)) {
  879. proxy_backend_close_connection (conn);
  880. }
  881. if (conn->results) {
  882. ucl_object_unref (conn->results);
  883. }
  884. }
  885. if (session->master_conn) {
  886. if (session->master_conn->results) {
  887. ucl_object_unref (session->master_conn->results);
  888. }
  889. if (session->master_conn->task) {
  890. rspamd_session_destroy (session->master_conn->task->s);
  891. }
  892. }
  893. g_ptr_array_free (session->mirror_conns, TRUE);
  894. rspamd_http_message_shmem_unref (session->shmem_ref);
  895. rspamd_http_message_unref (session->client_message);
  896. if (session->client_addr) {
  897. rspamd_inet_address_free (session->client_addr);
  898. }
  899. if (session->client_sock != -1) {
  900. close (session->client_sock);
  901. }
  902. if (session->ctx->sessions_cache) {
  903. rspamd_worker_session_cache_remove (session->ctx->sessions_cache,
  904. session);
  905. }
  906. if (session->pool) {
  907. rspamd_mempool_delete (session->pool);
  908. }
  909. g_free (session);
  910. }
  911. static void
  912. proxy_request_compress (struct rspamd_http_message *msg)
  913. {
  914. guint flags;
  915. ZSTD_CCtx *zctx;
  916. rspamd_fstring_t *body;
  917. const gchar *in;
  918. gsize inlen;
  919. flags = rspamd_http_message_get_flags (msg);
  920. if (!rspamd_http_message_find_header (msg, "Compression")) {
  921. if ((flags & RSPAMD_HTTP_FLAG_SHMEM) ||
  922. !(flags & RSPAMD_HTTP_FLAG_HAS_BODY)) {
  923. /* Cannot compress shared or empty message */
  924. return;
  925. }
  926. in = rspamd_http_message_get_body (msg, &inlen);
  927. if (in == NULL || inlen == 0) {
  928. return;
  929. }
  930. body = rspamd_fstring_sized_new (ZSTD_compressBound (inlen));
  931. zctx = ZSTD_createCCtx ();
  932. body->len = ZSTD_compressCCtx (zctx, body->str, body->allocated,
  933. in, inlen, 1);
  934. if (ZSTD_isError (body->len)) {
  935. msg_err ("compression error");
  936. rspamd_fstring_free (body);
  937. ZSTD_freeCCtx (zctx);
  938. return;
  939. }
  940. ZSTD_freeCCtx (zctx);
  941. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  942. rspamd_http_message_add_header (msg, "Compression", "zstd");
  943. }
  944. }
  945. static void
  946. proxy_request_decompress (struct rspamd_http_message *msg)
  947. {
  948. rspamd_fstring_t *body;
  949. const gchar *in;
  950. gsize inlen, outlen, r;
  951. ZSTD_DStream *zstream;
  952. ZSTD_inBuffer zin;
  953. ZSTD_outBuffer zout;
  954. if (rspamd_http_message_find_header (msg, "Compression")) {
  955. in = rspamd_http_message_get_body (msg, &inlen);
  956. if (in == NULL || inlen == 0) {
  957. return;
  958. }
  959. zstream = ZSTD_createDStream ();
  960. ZSTD_initDStream (zstream);
  961. zin.pos = 0;
  962. zin.src = in;
  963. zin.size = inlen;
  964. if ((outlen = ZSTD_getDecompressedSize (zin.src, zin.size)) == 0) {
  965. outlen = ZSTD_DStreamOutSize ();
  966. }
  967. body = rspamd_fstring_sized_new (outlen);
  968. zout.dst = body->str;
  969. zout.pos = 0;
  970. zout.size = outlen;
  971. while (zin.pos < zin.size) {
  972. r = ZSTD_decompressStream (zstream, &zout, &zin);
  973. if (ZSTD_isError (r)) {
  974. msg_err ("Decompression error: %s", ZSTD_getErrorName (r));
  975. ZSTD_freeDStream (zstream);
  976. rspamd_fstring_free (body);
  977. return;
  978. }
  979. if (zout.pos == zout.size) {
  980. /* We need to extend output buffer */
  981. zout.size = zout.size * 1.5 + 1.0;
  982. body = rspamd_fstring_grow (body, zout.size);
  983. zout.size = body->allocated;
  984. zout.dst = body->str;
  985. }
  986. }
  987. body->len = zout.pos;
  988. ZSTD_freeDStream (zstream);
  989. rspamd_http_message_set_body_from_fstring_steal (msg, body);
  990. rspamd_http_message_remove_header (msg, "Compression");
  991. }
  992. return;
  993. }
  994. static struct rspamd_proxy_session *
  995. proxy_session_refresh (struct rspamd_proxy_session *session)
  996. {
  997. struct rspamd_proxy_session *nsession;
  998. nsession = g_malloc0 (sizeof (*nsession));
  999. nsession->client_milter_conn = session->client_milter_conn;
  1000. session->client_milter_conn = NULL;
  1001. rspamd_milter_update_userdata (nsession->client_milter_conn,
  1002. nsession);
  1003. nsession->client_addr = session->client_addr;
  1004. session->client_addr = NULL;
  1005. nsession->ctx = session->ctx;
  1006. nsession->worker = session->worker;
  1007. nsession->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (), "proxy");
  1008. nsession->client_sock = session->client_sock;
  1009. session->client_sock = -1;
  1010. nsession->mirror_conns = g_ptr_array_sized_new (nsession->ctx->mirrors->len);
  1011. REF_INIT_RETAIN (nsession, proxy_session_dtor);
  1012. if (nsession->ctx->sessions_cache) {
  1013. rspamd_worker_session_cache_add (nsession->ctx->sessions_cache,
  1014. nsession->pool->tag.uid, &nsession->ref.refcount, nsession);
  1015. }
  1016. return nsession;
  1017. }
  1018. static gboolean
  1019. proxy_check_file (struct rspamd_http_message *msg,
  1020. struct rspamd_proxy_session *session)
  1021. {
  1022. const rspamd_ftok_t *tok, *key_tok;
  1023. rspamd_ftok_t srch;
  1024. gchar *file_str;
  1025. GHashTable *query_args;
  1026. GHashTableIter it;
  1027. gpointer k, v;
  1028. struct http_parser_url u;
  1029. rspamd_fstring_t *new_url;
  1030. tok = rspamd_http_message_find_header (msg, "File");
  1031. if (tok) {
  1032. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1033. session->map = rspamd_file_xmap (file_str, PROT_READ, &session->map_len,
  1034. TRUE);
  1035. if (session->map == NULL) {
  1036. if (session->map_len != 0) {
  1037. msg_err_session ("cannot map %s: %s", file_str,
  1038. strerror (errno));
  1039. return FALSE;
  1040. }
  1041. }
  1042. /* Remove header after processing */
  1043. rspamd_http_message_remove_header (msg, "File");
  1044. session->fname = file_str;
  1045. }
  1046. else {
  1047. /* Need to parse query URL */
  1048. if (http_parser_parse_url (RSPAMD_FSTRING_DATA (msg->url),
  1049. RSPAMD_FSTRING_LEN (msg->url), 0, &u) != 0) {
  1050. msg_err_session ("bad request url: %V", msg->url);
  1051. return FALSE;
  1052. }
  1053. if (u.field_set & (1 << UF_QUERY)) {
  1054. /* In case if we have a query, we need to store it somewhere */
  1055. query_args = rspamd_http_message_parse_query (msg);
  1056. srch.begin = "File";
  1057. srch.len = strlen ("File");
  1058. tok = g_hash_table_lookup (query_args, &srch);
  1059. if (tok) {
  1060. file_str = rspamd_mempool_ftokdup (session->pool, tok);
  1061. session->map = rspamd_file_xmap (file_str, PROT_READ,
  1062. &session->map_len, TRUE);
  1063. if (session->map == NULL) {
  1064. if (session->map_len != 0) {
  1065. msg_err_session ("cannot map %s: %s", file_str,
  1066. strerror (errno));
  1067. g_hash_table_unref (query_args);
  1068. return FALSE;
  1069. }
  1070. }
  1071. /* We need to create a new URL with file attribute removed */
  1072. new_url = rspamd_fstring_new_init (RSPAMD_FSTRING_DATA (msg->url),
  1073. u.field_data[UF_QUERY].off);
  1074. new_url = rspamd_fstring_append (new_url, "?", 1);
  1075. g_hash_table_iter_init (&it, query_args);
  1076. while (g_hash_table_iter_next (&it, &k, &v)) {
  1077. key_tok = k;
  1078. tok = v;
  1079. if (!rspamd_ftok_icase_equal (key_tok, &srch)) {
  1080. rspamd_printf_fstring (&new_url, "%T=%T&",
  1081. key_tok, tok);
  1082. }
  1083. }
  1084. /* Erase last character (might be either & or ?) */
  1085. rspamd_fstring_erase (new_url, new_url->len - 1, 1);
  1086. rspamd_fstring_free (msg->url);
  1087. msg->url = new_url;
  1088. session->fname = file_str;
  1089. }
  1090. g_hash_table_unref (query_args);
  1091. }
  1092. }
  1093. return TRUE;
  1094. }
  1095. static void
  1096. proxy_backend_mirror_error_handler (struct rspamd_http_connection *conn, GError *err)
  1097. {
  1098. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1099. struct rspamd_proxy_session *session;
  1100. session = bk_conn->s;
  1101. msg_info_session ("abnormally closing connection from backend: %s:%s, "
  1102. "error: %e",
  1103. bk_conn->name,
  1104. rspamd_inet_address_to_string (rspamd_upstream_addr (bk_conn->up)),
  1105. err);
  1106. if (err) {
  1107. bk_conn->err = rspamd_mempool_strdup (session->pool, err->message);
  1108. }
  1109. rspamd_upstream_fail (bk_conn->up, FALSE);
  1110. proxy_backend_close_connection (bk_conn);
  1111. REF_RELEASE (bk_conn->s);
  1112. }
  1113. static gint
  1114. proxy_backend_mirror_finish_handler (struct rspamd_http_connection *conn,
  1115. struct rspamd_http_message *msg)
  1116. {
  1117. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1118. struct rspamd_proxy_session *session;
  1119. session = bk_conn->s;
  1120. proxy_request_decompress (msg);
  1121. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1122. bk_conn->parser_from_ref, msg->body_buf.begin, msg->body_buf.len)) {
  1123. msg_warn_session ("cannot parse results from the mirror backend %s:%s",
  1124. bk_conn->name,
  1125. rspamd_inet_address_to_string (rspamd_upstream_addr (bk_conn->up)));
  1126. bk_conn->err = "cannot parse ucl";
  1127. }
  1128. msg_info_session ("finished mirror connection to %s", bk_conn->name);
  1129. rspamd_upstream_ok (bk_conn->up);
  1130. proxy_backend_close_connection (bk_conn);
  1131. REF_RELEASE (bk_conn->s);
  1132. return 0;
  1133. }
  1134. static void
  1135. proxy_open_mirror_connections (struct rspamd_proxy_session *session)
  1136. {
  1137. gdouble coin;
  1138. struct rspamd_http_mirror *m;
  1139. guint i;
  1140. struct rspamd_proxy_backend_connection *bk_conn;
  1141. struct rspamd_http_message *msg;
  1142. GError *err = NULL;
  1143. coin = rspamd_random_double ();
  1144. for (i = 0; i < session->ctx->mirrors->len; i ++) {
  1145. m = g_ptr_array_index (session->ctx->mirrors, i);
  1146. if (m->prob < coin) {
  1147. /* No luck */
  1148. continue;
  1149. }
  1150. bk_conn = rspamd_mempool_alloc0 (session->pool,
  1151. sizeof (*bk_conn));
  1152. bk_conn->s = session;
  1153. bk_conn->name = m->name;
  1154. bk_conn->io_tv = &m->io_tv;
  1155. bk_conn->up = rspamd_upstream_get (m->u,
  1156. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1157. bk_conn->parser_from_ref = m->parser_from_ref;
  1158. bk_conn->parser_to_ref = m->parser_to_ref;
  1159. if (bk_conn->up == NULL) {
  1160. msg_err_session ("cannot select upstream for %s", m->name);
  1161. continue;
  1162. }
  1163. bk_conn->backend_sock = rspamd_inet_address_connect (
  1164. rspamd_upstream_addr (bk_conn->up),
  1165. SOCK_STREAM, TRUE);
  1166. if (bk_conn->backend_sock == -1) {
  1167. msg_err_session ("cannot connect upstream for %s", m->name);
  1168. rspamd_upstream_fail (bk_conn->up, TRUE);
  1169. continue;
  1170. }
  1171. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1172. if (msg == NULL) {
  1173. msg_err_session ("cannot copy message to send to a mirror %s: %e",
  1174. m->name, err);
  1175. if (err) {
  1176. g_error_free (err);
  1177. }
  1178. continue;
  1179. }
  1180. if (msg->url->len == 0) {
  1181. msg->url = rspamd_fstring_append (msg->url, "/check", strlen ("/check"));
  1182. }
  1183. if (m->settings_id != NULL) {
  1184. rspamd_http_message_remove_header (msg, "Settings-ID");
  1185. rspamd_http_message_add_header (msg, "Settings-ID", m->settings_id);
  1186. }
  1187. bk_conn->backend_conn = rspamd_http_connection_new (NULL,
  1188. proxy_backend_mirror_error_handler,
  1189. proxy_backend_mirror_finish_handler,
  1190. RSPAMD_HTTP_CLIENT_SIMPLE,
  1191. RSPAMD_HTTP_CLIENT,
  1192. session->ctx->keys_cache,
  1193. NULL);
  1194. if (m->key) {
  1195. rspamd_http_connection_set_key (bk_conn->backend_conn,
  1196. session->ctx->local_key);
  1197. msg->peer_key = rspamd_pubkey_ref (m->key);
  1198. }
  1199. if (m->local ||
  1200. rspamd_inet_address_is_local (
  1201. rspamd_upstream_addr (bk_conn->up), FALSE)) {
  1202. if (session->fname) {
  1203. rspamd_http_message_add_header (msg, "File", session->fname);
  1204. }
  1205. msg->method = HTTP_GET;
  1206. rspamd_http_connection_write_message_shared (bk_conn->backend_conn,
  1207. msg, NULL, NULL, bk_conn,
  1208. bk_conn->backend_sock,
  1209. bk_conn->io_tv, session->ctx->ev_base);
  1210. }
  1211. else {
  1212. if (session->fname) {
  1213. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1214. rspamd_http_message_set_body (msg, session->map, session->map_len);
  1215. }
  1216. msg->method = HTTP_POST;
  1217. if (m->compress) {
  1218. proxy_request_compress (msg);
  1219. if (session->client_milter_conn) {
  1220. rspamd_http_message_add_header (msg, "Content-Type",
  1221. "application/octet-stream");
  1222. }
  1223. }
  1224. else {
  1225. if (session->client_milter_conn) {
  1226. rspamd_http_message_add_header (msg, "Content-Type",
  1227. "text/plain");
  1228. }
  1229. }
  1230. rspamd_http_connection_write_message (bk_conn->backend_conn,
  1231. msg, NULL, NULL, bk_conn,
  1232. bk_conn->backend_sock,
  1233. bk_conn->io_tv, session->ctx->ev_base);
  1234. }
  1235. g_ptr_array_add (session->mirror_conns, bk_conn);
  1236. REF_RETAIN (session);
  1237. msg_info_session ("send request to %s", m->name);
  1238. }
  1239. }
  1240. static void
  1241. proxy_client_write_error (struct rspamd_proxy_session *session, gint code,
  1242. const gchar *status)
  1243. {
  1244. struct rspamd_http_message *reply;
  1245. if (session->client_milter_conn) {
  1246. rspamd_milter_send_action (session->client_milter_conn,
  1247. RSPAMD_MILTER_TEMPFAIL);
  1248. REF_RELEASE (session);
  1249. }
  1250. else {
  1251. reply = rspamd_http_new_message (HTTP_RESPONSE);
  1252. reply->code = code;
  1253. reply->status = rspamd_fstring_new_init (status, strlen (status));
  1254. rspamd_http_connection_write_message (session->client_conn,
  1255. reply, NULL, NULL, session, session->client_sock,
  1256. &session->ctx->io_tv, session->ctx->ev_base);
  1257. }
  1258. }
  1259. static void
  1260. proxy_backend_master_error_handler (struct rspamd_http_connection *conn, GError *err)
  1261. {
  1262. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1263. struct rspamd_proxy_session *session;
  1264. session = bk_conn->s;
  1265. msg_info_session ("abnormally closing connection from backend: %s, error: %e,"
  1266. " retries left: %d",
  1267. rspamd_inet_address_to_string (rspamd_upstream_addr (session->master_conn->up)),
  1268. err,
  1269. session->ctx->max_retries - session->retries);
  1270. session->retries ++;
  1271. rspamd_upstream_fail (bk_conn->up, FALSE);
  1272. proxy_backend_close_connection (session->master_conn);
  1273. if (session->ctx->max_retries &&
  1274. session->retries > session->ctx->max_retries) {
  1275. msg_err_session ("cannot connect to upstream, maximum retries "
  1276. "has been reached: %d", session->retries);
  1277. /* Terminate session immediately */
  1278. proxy_client_write_error (session, err->code, err->message);
  1279. }
  1280. else {
  1281. if (!proxy_send_master_message (session)) {
  1282. proxy_client_write_error (session, err->code, err->message);
  1283. }
  1284. else {
  1285. msg_info_session ("retry connection to: %s"
  1286. " retries left: %d",
  1287. rspamd_inet_address_to_string (
  1288. rspamd_upstream_addr (session->master_conn->up)),
  1289. session->ctx->max_retries - session->retries);
  1290. }
  1291. }
  1292. }
  1293. static gint
  1294. proxy_backend_master_finish_handler (struct rspamd_http_connection *conn,
  1295. struct rspamd_http_message *msg)
  1296. {
  1297. struct rspamd_proxy_backend_connection *bk_conn = conn->ud;
  1298. struct rspamd_proxy_session *session, *nsession;
  1299. rspamd_fstring_t *reply;
  1300. session = bk_conn->s;
  1301. rspamd_http_connection_steal_msg (session->master_conn->backend_conn);
  1302. proxy_request_decompress (msg);
  1303. rspamd_http_message_remove_header (msg, "Content-Length");
  1304. rspamd_http_message_remove_header (msg, "Key");
  1305. rspamd_http_connection_reset (session->master_conn->backend_conn);
  1306. if (!proxy_backend_parse_results (session, bk_conn, session->ctx->lua_state,
  1307. bk_conn->parser_from_ref, msg->body_buf.begin, msg->body_buf.len)) {
  1308. msg_warn_session ("cannot parse results from the master backend");
  1309. }
  1310. if (session->legacy_support > LEGACY_SUPPORT_NO) {
  1311. /* We need to reformat ucl to fit with legacy spamc protocol */
  1312. if (bk_conn->results) {
  1313. reply = rspamd_fstring_new ();
  1314. if (session->legacy_support == LEGACY_SUPPORT_SPAMC) {
  1315. rspamd_ucl_tospamc_output (bk_conn->results, &reply);
  1316. msg->flags |= RSPAMD_HTTP_FLAG_SPAMC;
  1317. }
  1318. else {
  1319. rspamd_ucl_torspamc_output (bk_conn->results, &reply);
  1320. }
  1321. rspamd_http_message_set_body_from_fstring_steal (msg, reply);
  1322. msg->method = HTTP_SYMBOLS;
  1323. }
  1324. else {
  1325. msg_warn_session ("cannot parse results from the master backend, "
  1326. "return them as is");
  1327. }
  1328. }
  1329. rspamd_upstream_ok (bk_conn->up);
  1330. if (session->client_milter_conn) {
  1331. nsession = proxy_session_refresh (session);
  1332. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1333. session->master_conn->results);
  1334. REF_RELEASE (session);
  1335. rspamd_http_message_free (msg);
  1336. }
  1337. else {
  1338. rspamd_http_connection_write_message (session->client_conn,
  1339. msg, NULL, NULL, session, session->client_sock,
  1340. bk_conn->io_tv, session->ctx->ev_base);
  1341. }
  1342. return 0;
  1343. }
  1344. static void
  1345. rspamd_proxy_scan_self_reply (struct rspamd_task *task)
  1346. {
  1347. struct rspamd_http_message *msg;
  1348. struct rspamd_proxy_session *session = task->fin_arg, *nsession;
  1349. ucl_object_t *rep = NULL;
  1350. const char *ctype = "application/json";
  1351. msg = rspamd_http_new_message (HTTP_RESPONSE);
  1352. msg->date = time (NULL);
  1353. msg->code = 200;
  1354. switch (task->cmd) {
  1355. case CMD_REPORT_IFSPAM:
  1356. case CMD_REPORT:
  1357. case CMD_CHECK:
  1358. case CMD_SYMBOLS:
  1359. case CMD_PROCESS:
  1360. case CMD_SKIP:
  1361. case CMD_CHECK_V2:
  1362. rspamd_task_set_finish_time (task);
  1363. rspamd_protocol_http_reply (msg, task, &rep);
  1364. rspamd_protocol_write_log_pipe (task);
  1365. break;
  1366. case CMD_PING:
  1367. rspamd_http_message_set_body (msg, "pong" CRLF, 6);
  1368. ctype = "text/plain";
  1369. break;
  1370. case CMD_OTHER:
  1371. msg_err_task ("BROKEN");
  1372. break;
  1373. }
  1374. session->master_conn->flags |= RSPAMD_BACKEND_CLOSED;
  1375. if (rep) {
  1376. session->master_conn->results = ucl_object_ref (rep);
  1377. }
  1378. if (session->client_milter_conn) {
  1379. nsession = proxy_session_refresh (session);
  1380. rspamd_milter_send_task_results (nsession->client_milter_conn,
  1381. session->master_conn->results);
  1382. rspamd_http_message_free (msg);
  1383. REF_RELEASE (session);
  1384. }
  1385. else {
  1386. rspamd_http_connection_reset (session->client_conn);
  1387. rspamd_http_connection_write_message (session->client_conn,
  1388. msg,
  1389. NULL,
  1390. ctype,
  1391. session,
  1392. session->client_sock,
  1393. NULL,
  1394. session->ctx->ev_base);
  1395. }
  1396. }
  1397. static gboolean
  1398. rspamd_proxy_task_fin (void *ud)
  1399. {
  1400. struct rspamd_task *task = ud;
  1401. msg_debug_task ("finish task");
  1402. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1403. rspamd_proxy_scan_self_reply (task);
  1404. return TRUE;
  1405. }
  1406. if (!rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL)) {
  1407. rspamd_proxy_scan_self_reply (task);
  1408. return TRUE;
  1409. }
  1410. if (RSPAMD_TASK_IS_PROCESSED (task)) {
  1411. rspamd_proxy_scan_self_reply (task);
  1412. return TRUE;
  1413. }
  1414. /* One more iteration */
  1415. return FALSE;
  1416. }
  1417. static gboolean
  1418. rspamd_proxy_self_scan (struct rspamd_proxy_session *session)
  1419. {
  1420. struct rspamd_task *task;
  1421. struct rspamd_http_message *msg;
  1422. const gchar *data;
  1423. gsize len;
  1424. msg = session->client_message;
  1425. task = rspamd_task_new (session->worker, session->ctx->cfg,
  1426. session->pool, session->ctx->lang_det,
  1427. session->ctx->ev_base);
  1428. task->flags |= RSPAMD_TASK_FLAG_MIME;
  1429. task->sock = -1;
  1430. if (session->client_milter_conn) {
  1431. task->client_addr = rspamd_inet_address_copy (
  1432. session->client_milter_conn->addr);
  1433. }
  1434. else {
  1435. task->client_addr = rspamd_inet_address_copy (session->client_addr);
  1436. }
  1437. task->fin_arg = session;
  1438. task->resolver = session->ctx->resolver;
  1439. /* TODO: allow to disable autolearn in protocol */
  1440. task->flags |= RSPAMD_TASK_FLAG_LEARN_AUTO;
  1441. task->s = rspamd_session_create (task->task_pool, rspamd_proxy_task_fin,
  1442. NULL, (event_finalizer_t )rspamd_task_free, task);
  1443. data = rspamd_http_message_get_body (msg, &len);
  1444. if (session->backend->settings_id) {
  1445. rspamd_http_message_remove_header (msg, "Settings-ID");
  1446. rspamd_http_message_add_header (msg, "Settings-ID",
  1447. session->backend->settings_id);
  1448. }
  1449. /* Process message */
  1450. if (!rspamd_protocol_handle_request (task, msg)) {
  1451. msg_err_task ("cannot handle request: %e", task->err);
  1452. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1453. }
  1454. else {
  1455. if (task->cmd == CMD_PING) {
  1456. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1457. }
  1458. else {
  1459. if (!rspamd_task_load_message (task, msg, data, len)) {
  1460. msg_err_task ("cannot load message: %e", task->err);
  1461. task->flags |= RSPAMD_TASK_FLAG_SKIP;
  1462. }
  1463. }
  1464. }
  1465. /* Set global timeout for the task */
  1466. if (session->ctx->default_upstream->timeout > 0.0) {
  1467. struct timeval task_tv;
  1468. event_set (&task->timeout_ev, -1, EV_TIMEOUT, rspamd_task_timeout,
  1469. task);
  1470. event_base_set (session->ctx->ev_base, &task->timeout_ev);
  1471. double_to_tv (session->ctx->default_upstream->timeout, &task_tv);
  1472. event_add (&task->timeout_ev, &task_tv);
  1473. }
  1474. else if (session->ctx->has_self_scan) {
  1475. if (session->ctx->cfg->task_timeout > 0) {
  1476. struct timeval task_tv;
  1477. event_set (&task->timeout_ev, -1, EV_TIMEOUT, rspamd_task_timeout,
  1478. task);
  1479. event_base_set (session->ctx->ev_base, &task->timeout_ev);
  1480. double_to_tv (session->ctx->cfg->task_timeout, &task_tv);
  1481. event_add (&task->timeout_ev, &task_tv);
  1482. }
  1483. }
  1484. session->master_conn->task = task;
  1485. rspamd_task_process (task, RSPAMD_TASK_PROCESS_ALL);
  1486. rspamd_session_pending (task->s);
  1487. return TRUE;
  1488. }
  1489. static gboolean
  1490. proxy_send_master_message (struct rspamd_proxy_session *session)
  1491. {
  1492. struct rspamd_http_message *msg;
  1493. struct rspamd_http_upstream *backend = NULL;
  1494. const rspamd_ftok_t *host;
  1495. GError *err = NULL;
  1496. gchar hostbuf[512];
  1497. host = rspamd_http_message_find_header (session->client_message, "Host");
  1498. if (host == NULL) {
  1499. backend = session->ctx->default_upstream;
  1500. }
  1501. else {
  1502. rspamd_strlcpy (hostbuf, host->begin, MIN(host->len + 1, sizeof (hostbuf)));
  1503. backend = g_hash_table_lookup (session->ctx->upstreams, hostbuf);
  1504. if (backend == NULL) {
  1505. backend = session->ctx->default_upstream;
  1506. }
  1507. }
  1508. if (backend == NULL) {
  1509. /* No backend */
  1510. msg_err_session ("cannot find upstream for %s", host ? hostbuf : "default");
  1511. goto err;
  1512. }
  1513. else {
  1514. session->backend = backend;
  1515. if (backend->self_scan) {
  1516. return rspamd_proxy_self_scan (session);
  1517. }
  1518. retry:
  1519. if (session->ctx->max_retries &&
  1520. session->retries > session->ctx->max_retries) {
  1521. msg_err_session ("cannot connect to upstream, maximum retries "
  1522. "has been reached: %d", session->retries);
  1523. goto err;
  1524. }
  1525. session->master_conn->up = rspamd_upstream_get (backend->u,
  1526. RSPAMD_UPSTREAM_ROUND_ROBIN, NULL, 0);
  1527. session->master_conn->io_tv = &backend->io_tv;
  1528. if (session->master_conn->up == NULL) {
  1529. msg_err_session ("cannot select upstream for %s",
  1530. host ? hostbuf : "default");
  1531. goto err;
  1532. }
  1533. session->master_conn->backend_sock = rspamd_inet_address_connect (
  1534. rspamd_upstream_addr (session->master_conn->up),
  1535. SOCK_STREAM, TRUE);
  1536. if (session->master_conn->backend_sock == -1) {
  1537. msg_err_session ("cannot connect upstream: %s(%s)",
  1538. host ? hostbuf : "default",
  1539. rspamd_inet_address_to_string (rspamd_upstream_addr (
  1540. session->master_conn->up)));
  1541. rspamd_upstream_fail (session->master_conn->up, TRUE);
  1542. session->retries ++;
  1543. goto retry;
  1544. }
  1545. msg = rspamd_http_connection_copy_msg (session->client_message, &err);
  1546. if (msg == NULL) {
  1547. msg_err_session ("cannot copy message to send it to the upstream: %e",
  1548. err);
  1549. if (err) {
  1550. g_error_free (err);
  1551. }
  1552. goto err; /* No fallback here */
  1553. }
  1554. session->master_conn->backend_conn = rspamd_http_connection_new (
  1555. NULL,
  1556. proxy_backend_master_error_handler,
  1557. proxy_backend_master_finish_handler,
  1558. RSPAMD_HTTP_CLIENT_SIMPLE,
  1559. RSPAMD_HTTP_CLIENT,
  1560. session->ctx->keys_cache,
  1561. NULL);
  1562. session->master_conn->flags &= ~RSPAMD_BACKEND_CLOSED;
  1563. session->master_conn->parser_from_ref = backend->parser_from_ref;
  1564. session->master_conn->parser_to_ref = backend->parser_to_ref;
  1565. if (backend->key) {
  1566. msg->peer_key = rspamd_pubkey_ref (backend->key);
  1567. rspamd_http_connection_set_key (session->master_conn->backend_conn,
  1568. session->ctx->local_key);
  1569. }
  1570. if (backend->settings_id != NULL) {
  1571. rspamd_http_message_remove_header (msg, "Settings-ID");
  1572. rspamd_http_message_add_header (msg, "Settings-ID",
  1573. backend->settings_id);
  1574. }
  1575. if (backend->local ||
  1576. rspamd_inet_address_is_local (
  1577. rspamd_upstream_addr (session->master_conn->up), FALSE)) {
  1578. if (session->fname) {
  1579. rspamd_http_message_add_header (msg, "File", session->fname);
  1580. }
  1581. msg->method = HTTP_GET;
  1582. rspamd_http_connection_write_message_shared (
  1583. session->master_conn->backend_conn,
  1584. msg, NULL, NULL, session->master_conn,
  1585. session->master_conn->backend_sock,
  1586. session->master_conn->io_tv, session->ctx->ev_base);
  1587. }
  1588. else {
  1589. if (session->fname) {
  1590. msg->flags &= ~RSPAMD_HTTP_FLAG_SHMEM;
  1591. rspamd_http_message_set_body (msg,
  1592. session->map, session->map_len);
  1593. }
  1594. msg->method = HTTP_POST;
  1595. if (backend->compress) {
  1596. proxy_request_compress (msg);
  1597. if (session->client_milter_conn) {
  1598. rspamd_http_message_add_header (msg, "Content-Type",
  1599. "application/octet-stream");
  1600. }
  1601. }
  1602. else {
  1603. if (session->client_milter_conn) {
  1604. rspamd_http_message_add_header (msg, "Content-Type",
  1605. "text/plain");
  1606. }
  1607. }
  1608. rspamd_http_connection_write_message (
  1609. session->master_conn->backend_conn,
  1610. msg, NULL, NULL, session->master_conn,
  1611. session->master_conn->backend_sock,
  1612. session->master_conn->io_tv, session->ctx->ev_base);
  1613. }
  1614. }
  1615. return TRUE;
  1616. err:
  1617. if (session->client_milter_conn) {
  1618. rspamd_milter_send_action (session->client_milter_conn,
  1619. RSPAMD_MILTER_TEMPFAIL);
  1620. REF_RELEASE (session);
  1621. }
  1622. else {
  1623. rspamd_http_connection_steal_msg (session->client_conn);
  1624. rspamd_http_connection_reset (session->client_conn);
  1625. proxy_client_write_error (session, 404, "Backend not found");
  1626. }
  1627. return FALSE;
  1628. }
  1629. static void
  1630. proxy_client_error_handler (struct rspamd_http_connection *conn, GError *err)
  1631. {
  1632. struct rspamd_proxy_session *session = conn->ud;
  1633. msg_info_session ("abnormally closing connection from: %s, error: %s",
  1634. rspamd_inet_address_to_string (session->client_addr), err->message);
  1635. /* Terminate session immediately */
  1636. proxy_backend_close_connection (session->master_conn);
  1637. REF_RELEASE (session);
  1638. }
  1639. static gint
  1640. proxy_client_finish_handler (struct rspamd_http_connection *conn,
  1641. struct rspamd_http_message *msg)
  1642. {
  1643. struct rspamd_proxy_session *session = conn->ud;
  1644. if (!session->master_conn) {
  1645. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1646. sizeof (*session->master_conn));
  1647. session->master_conn->s = session;
  1648. session->master_conn->name = "master";
  1649. /* Reset spamc legacy */
  1650. if (msg->method >= HTTP_SYMBOLS) {
  1651. msg->method = HTTP_POST;
  1652. if (msg->flags & RSPAMD_HTTP_FLAG_SPAMC) {
  1653. session->legacy_support = LEGACY_SUPPORT_SPAMC;
  1654. msg_info_session ("enabling legacy spamc mode for session");
  1655. }
  1656. else {
  1657. session->legacy_support = LEGACY_SUPPORT_RSPAMC;
  1658. msg_info_session ("enabling legacy rspamc mode for session");
  1659. }
  1660. }
  1661. if (msg->url->len == 0) {
  1662. msg->url = rspamd_fstring_append (msg->url,
  1663. "/" MSG_CMD_CHECK_V2, strlen ("/" MSG_CMD_CHECK_V2));
  1664. }
  1665. if (!proxy_check_file (msg, session)) {
  1666. goto err;
  1667. }
  1668. session->client_message = rspamd_http_connection_steal_msg (
  1669. session->client_conn);
  1670. session->shmem_ref = rspamd_http_message_shmem_ref (session->client_message);
  1671. rspamd_http_message_remove_header (msg, "Content-Length");
  1672. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1673. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1674. rspamd_http_message_remove_header (msg, "Connection");
  1675. rspamd_http_message_remove_header (msg, "Key");
  1676. proxy_open_mirror_connections (session);
  1677. rspamd_http_connection_reset (session->client_conn);
  1678. proxy_send_master_message (session);
  1679. }
  1680. else {
  1681. msg_info_session ("finished master connection");
  1682. proxy_backend_close_connection (session->master_conn);
  1683. REF_RELEASE (session);
  1684. }
  1685. return 0;
  1686. err:
  1687. rspamd_http_connection_steal_msg (session->client_conn);
  1688. rspamd_http_message_remove_header (msg, "Content-Length");
  1689. rspamd_http_message_remove_header (msg, "Key");
  1690. rspamd_http_message_remove_header (msg, "Transfer-Encoding");
  1691. rspamd_http_message_remove_header (msg, "Keep-Alive");
  1692. rspamd_http_message_remove_header (msg, "Connection");
  1693. rspamd_http_connection_reset (session->client_conn);
  1694. proxy_client_write_error (session, 404, "Backend not found");
  1695. return 0;
  1696. }
  1697. static void
  1698. proxy_milter_finish_handler (gint fd,
  1699. struct rspamd_milter_session *rms,
  1700. void *ud)
  1701. {
  1702. struct rspamd_proxy_session *session = ud;
  1703. struct rspamd_http_message *msg;
  1704. session->client_milter_conn = rms;
  1705. if (rms->message == NULL || rms->message->len == 0) {
  1706. msg_info_session ("finished milter connection");
  1707. proxy_backend_close_connection (session->master_conn);
  1708. REF_RELEASE (session);
  1709. }
  1710. else {
  1711. if (!session->master_conn) {
  1712. session->master_conn = rspamd_mempool_alloc0 (session->pool,
  1713. sizeof (*session->master_conn));
  1714. }
  1715. msg = rspamd_milter_to_http (rms);
  1716. session->master_conn->s = session;
  1717. session->master_conn->name = "master";
  1718. session->client_message = msg;
  1719. proxy_open_mirror_connections (session);
  1720. proxy_send_master_message (session);
  1721. }
  1722. }
  1723. static void
  1724. proxy_milter_error_handler (gint fd,
  1725. struct rspamd_milter_session *rms,
  1726. void *ud, GError *err)
  1727. {
  1728. struct rspamd_proxy_session *session = ud;
  1729. msg_info_session ("abnormally closing milter connection from: %s, "
  1730. "error: %s", rspamd_inet_address_to_string (session->client_addr),
  1731. err->message);
  1732. /* Terminate session immediately */
  1733. proxy_backend_close_connection (session->master_conn);
  1734. REF_RELEASE (session);
  1735. }
  1736. static void
  1737. proxy_accept_socket (gint fd, short what, void *arg)
  1738. {
  1739. struct rspamd_worker *worker = (struct rspamd_worker *) arg;
  1740. struct rspamd_proxy_ctx *ctx;
  1741. rspamd_inet_addr_t *addr;
  1742. struct rspamd_proxy_session *session;
  1743. gint nfd;
  1744. ctx = worker->ctx;
  1745. if ((nfd =
  1746. rspamd_accept_from_socket (fd, &addr, worker->accept_events)) == -1) {
  1747. msg_warn ("accept failed: %s", strerror (errno));
  1748. return;
  1749. }
  1750. /* Check for EAGAIN */
  1751. if (nfd == 0) {
  1752. return;
  1753. }
  1754. session = g_malloc0 (sizeof (*session));
  1755. REF_INIT_RETAIN (session, proxy_session_dtor);
  1756. session->client_sock = nfd;
  1757. session->client_addr = addr;
  1758. session->mirror_conns = g_ptr_array_sized_new (ctx->mirrors->len);
  1759. session->pool = rspamd_mempool_new (rspamd_mempool_suggest_size (),
  1760. "proxy");
  1761. session->ctx = ctx;
  1762. session->worker = worker;
  1763. if (ctx->sessions_cache) {
  1764. rspamd_worker_session_cache_add (ctx->sessions_cache,
  1765. session->pool->tag.uid, &session->ref.refcount, session);
  1766. }
  1767. if (!ctx->milter) {
  1768. session->client_conn = rspamd_http_connection_new (NULL,
  1769. proxy_client_error_handler,
  1770. proxy_client_finish_handler,
  1771. 0,
  1772. RSPAMD_HTTP_SERVER,
  1773. ctx->keys_cache,
  1774. NULL);
  1775. if (ctx->key) {
  1776. rspamd_http_connection_set_key (session->client_conn, ctx->key);
  1777. }
  1778. msg_info_session ("accepted http connection from %s port %d",
  1779. rspamd_inet_address_to_string (addr),
  1780. rspamd_inet_address_get_port (addr));
  1781. rspamd_http_connection_read_message_shared (session->client_conn,
  1782. session,
  1783. nfd,
  1784. &ctx->io_tv,
  1785. ctx->ev_base);
  1786. }
  1787. else {
  1788. msg_info_session ("accepted milter connection from %s port %d",
  1789. rspamd_inet_address_to_string (addr),
  1790. rspamd_inet_address_get_port (addr));
  1791. #ifdef TCP_NODELAY
  1792. #ifndef SOL_TCP
  1793. #define SOL_TCP IPPROTO_TCP
  1794. #endif
  1795. if (rspamd_inet_address_get_af (addr) != AF_UNIX) {
  1796. gint sopt = 1;
  1797. if (setsockopt (nfd, SOL_TCP, TCP_NODELAY, &sopt, sizeof (sopt)) ==
  1798. -1) {
  1799. msg_warn_session ("cannot set TCP_NODELAY: %s",
  1800. strerror (errno));
  1801. }
  1802. }
  1803. #endif
  1804. rspamd_milter_handle_socket (nfd, NULL,
  1805. session->pool,
  1806. ctx->ev_base,
  1807. proxy_milter_finish_handler,
  1808. proxy_milter_error_handler,
  1809. session);
  1810. }
  1811. }
  1812. static void
  1813. proxy_rotate_key (gint fd, short what, void *arg)
  1814. {
  1815. struct timeval rot_tv;
  1816. struct rspamd_proxy_ctx *ctx = arg;
  1817. gpointer kp;
  1818. double_to_tv (ctx->rotate_tm, &rot_tv);
  1819. rot_tv.tv_sec += ottery_rand_range (rot_tv.tv_sec);
  1820. event_del (&ctx->rotate_ev);
  1821. event_add (&ctx->rotate_ev, &rot_tv);
  1822. kp = ctx->local_key;
  1823. ctx->local_key = rspamd_keypair_new (RSPAMD_KEYPAIR_KEX,
  1824. RSPAMD_CRYPTOBOX_MODE_25519);
  1825. rspamd_keypair_unref (kp);
  1826. }
  1827. static void
  1828. adjust_upstreams_limits (struct rspamd_proxy_ctx *ctx)
  1829. {
  1830. struct rspamd_http_upstream *backend;
  1831. gpointer k, v;
  1832. GHashTableIter it;
  1833. /*
  1834. * We set error time equal to max_retries * backend_timeout and max_errors
  1835. * to max_retries - 1
  1836. *
  1837. * So if we failed to scan a message on a backend for some reasons, we
  1838. * will try to re-resolve it faster
  1839. */
  1840. g_hash_table_iter_init (&it, ctx->upstreams);
  1841. while (g_hash_table_iter_next (&it, &k, &v)) {
  1842. backend = (struct rspamd_http_upstream *)v;
  1843. if (!backend->self_scan && backend->u) {
  1844. rspamd_upstreams_set_limits (backend->u,
  1845. NAN, NAN, ctx->max_retries * backend->timeout, NAN,
  1846. ctx->max_retries - 1, 0);
  1847. }
  1848. }
  1849. }
  1850. void
  1851. start_rspamd_proxy (struct rspamd_worker *worker) {
  1852. struct rspamd_proxy_ctx *ctx = worker->ctx;
  1853. struct timeval rot_tv;
  1854. ctx->cfg = worker->srv->cfg;
  1855. ctx->ev_base = rspamd_prepare_worker (worker, "rspamd_proxy",
  1856. proxy_accept_socket);
  1857. ctx->resolver = dns_resolver_init (worker->srv->logger,
  1858. ctx->ev_base,
  1859. worker->srv->cfg);
  1860. double_to_tv (ctx->timeout, &ctx->io_tv);
  1861. rspamd_map_watch (worker->srv->cfg, ctx->ev_base, ctx->resolver, worker, 0);
  1862. rspamd_upstreams_library_config (worker->srv->cfg, ctx->cfg->ups_ctx,
  1863. ctx->ev_base, ctx->resolver->r);
  1864. /* XXX: stupid default */
  1865. ctx->keys_cache = rspamd_keypair_cache_new (256);
  1866. ctx->local_key = rspamd_keypair_new (RSPAMD_KEYPAIR_KEX,
  1867. RSPAMD_CRYPTOBOX_MODE_25519);
  1868. double_to_tv (ctx->rotate_tm, &rot_tv);
  1869. rot_tv.tv_sec += ottery_rand_range (rot_tv.tv_sec);
  1870. event_set (&ctx->rotate_ev, -1, EV_TIMEOUT, proxy_rotate_key, ctx);
  1871. event_base_set (ctx->ev_base, &ctx->rotate_ev);
  1872. event_add (&ctx->rotate_ev, &rot_tv);
  1873. if (ctx->has_self_scan) {
  1874. /* Additional initialisation needed */
  1875. rspamd_worker_init_scanner (worker, ctx->ev_base, ctx->resolver,
  1876. &ctx->lang_det);
  1877. }
  1878. if (worker->srv->cfg->enable_sessions_cache) {
  1879. ctx->sessions_cache = rspamd_worker_session_cache_new (worker,
  1880. ctx->ev_base);
  1881. }
  1882. ctx->milter_ctx.spam_header = ctx->spam_header;
  1883. ctx->milter_ctx.discard_on_reject = ctx->discard_on_reject;
  1884. ctx->milter_ctx.quarantine_on_reject = ctx->quarantine_on_reject;
  1885. ctx->milter_ctx.sessions_cache = ctx->sessions_cache;
  1886. ctx->milter_ctx.client_ca_name = ctx->client_ca_name;
  1887. ctx->milter_ctx.reject_message = ctx->reject_message;
  1888. rspamd_milter_init_library (&ctx->milter_ctx);
  1889. rspamd_lua_run_postloads (ctx->cfg->lua_state, ctx->cfg, ctx->ev_base,
  1890. worker);
  1891. adjust_upstreams_limits (ctx);
  1892. event_base_loop (ctx->ev_base, 0);
  1893. rspamd_worker_block_signals ();
  1894. if (ctx->has_self_scan) {
  1895. rspamd_stat_close ();
  1896. }
  1897. rspamd_keypair_cache_destroy (ctx->keys_cache);
  1898. REF_RELEASE (ctx->cfg);
  1899. rspamd_log_close (worker->srv->logger, TRUE);
  1900. exit (EXIT_SUCCESS);
  1901. }