Index: stable/8/contrib/bind9/lib/dns/tkey.c =================================================================== --- stable/8/contrib/bind9/lib/dns/tkey.c (revision 285976) +++ stable/8/contrib/bind9/lib/dns/tkey.c (revision 285977) @@ -1,1460 +1,1461 @@ /* * Copyright (C) 2004-2014 Internet Systems Consortium, Inc. ("ISC") * Copyright (C) 1999-2001, 2003 Internet Software Consortium. * * Permission to use, copy, modify, and/or distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND ISC DISCLAIMS ALL WARRANTIES WITH * REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY * AND FITNESS. IN NO EVENT SHALL ISC BE LIABLE FOR ANY SPECIAL, DIRECT, * INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM * LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE * OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ /* * $Id$ */ /*! \file */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define TKEY_RANDOM_AMOUNT 16 #define RETERR(x) do { \ result = (x); \ if (result != ISC_R_SUCCESS) \ goto failure; \ } while (0) static void tkey_log(const char *fmt, ...) ISC_FORMAT_PRINTF(1, 2); static void tkey_log(const char *fmt, ...) { va_list ap; va_start(ap, fmt); isc_log_vwrite(dns_lctx, DNS_LOGCATEGORY_GENERAL, DNS_LOGMODULE_REQUEST, ISC_LOG_DEBUG(4), fmt, ap); va_end(ap); } static void _dns_tkey_dumpmessage(dns_message_t *msg) { isc_buffer_t outbuf; unsigned char output[4096]; isc_result_t result; isc_buffer_init(&outbuf, output, sizeof(output)); result = dns_message_totext(msg, &dns_master_style_debug, 0, &outbuf); if (result != ISC_R_SUCCESS) fprintf(stderr, "Warning: dns_message_totext returned: %s\n", dns_result_totext(result)); fprintf(stderr, "%.*s\n", (int)isc_buffer_usedlength(&outbuf), (char *)isc_buffer_base(&outbuf)); } isc_result_t dns_tkeyctx_create(isc_mem_t *mctx, isc_entropy_t *ectx, dns_tkeyctx_t **tctxp) { dns_tkeyctx_t *tctx; REQUIRE(mctx != NULL); REQUIRE(ectx != NULL); REQUIRE(tctxp != NULL && *tctxp == NULL); tctx = isc_mem_get(mctx, sizeof(dns_tkeyctx_t)); if (tctx == NULL) return (ISC_R_NOMEMORY); tctx->mctx = NULL; isc_mem_attach(mctx, &tctx->mctx); tctx->ectx = NULL; isc_entropy_attach(ectx, &tctx->ectx); tctx->dhkey = NULL; tctx->domain = NULL; tctx->gsscred = NULL; tctx->gssapi_keytab = NULL; *tctxp = tctx; return (ISC_R_SUCCESS); } void dns_tkeyctx_destroy(dns_tkeyctx_t **tctxp) { isc_mem_t *mctx; dns_tkeyctx_t *tctx; REQUIRE(tctxp != NULL && *tctxp != NULL); tctx = *tctxp; mctx = tctx->mctx; if (tctx->dhkey != NULL) dst_key_free(&tctx->dhkey); if (tctx->domain != NULL) { if (dns_name_dynamic(tctx->domain)) dns_name_free(tctx->domain, mctx); isc_mem_put(mctx, tctx->domain, sizeof(dns_name_t)); } if (tctx->gssapi_keytab != NULL) { isc_mem_free(mctx, tctx->gssapi_keytab); } if (tctx->gsscred != NULL) dst_gssapi_releasecred(&tctx->gsscred); isc_entropy_detach(&tctx->ectx); isc_mem_put(mctx, tctx, sizeof(dns_tkeyctx_t)); isc_mem_detach(&mctx); *tctxp = NULL; } static isc_result_t add_rdata_to_list(dns_message_t *msg, dns_name_t *name, dns_rdata_t *rdata, isc_uint32_t ttl, dns_namelist_t *namelist) { isc_result_t result; isc_region_t r, newr; dns_rdata_t *newrdata = NULL; dns_name_t *newname = NULL; dns_rdatalist_t *newlist = NULL; dns_rdataset_t *newset = NULL; isc_buffer_t *tmprdatabuf = NULL; RETERR(dns_message_gettemprdata(msg, &newrdata)); dns_rdata_toregion(rdata, &r); RETERR(isc_buffer_allocate(msg->mctx, &tmprdatabuf, r.length)); isc_buffer_availableregion(tmprdatabuf, &newr); memmove(newr.base, r.base, r.length); dns_rdata_fromregion(newrdata, rdata->rdclass, rdata->type, &newr); dns_message_takebuffer(msg, &tmprdatabuf); RETERR(dns_message_gettempname(msg, &newname)); dns_name_init(newname, NULL); RETERR(dns_name_dup(name, msg->mctx, newname)); RETERR(dns_message_gettemprdatalist(msg, &newlist)); newlist->rdclass = newrdata->rdclass; newlist->type = newrdata->type; newlist->covers = 0; newlist->ttl = ttl; ISC_LIST_INIT(newlist->rdata); ISC_LIST_APPEND(newlist->rdata, newrdata, link); RETERR(dns_message_gettemprdataset(msg, &newset)); dns_rdataset_init(newset); RETERR(dns_rdatalist_tordataset(newlist, newset)); ISC_LIST_INIT(newname->list); ISC_LIST_APPEND(newname->list, newset, link); ISC_LIST_APPEND(*namelist, newname, link); return (ISC_R_SUCCESS); failure: if (newrdata != NULL) { if (ISC_LINK_LINKED(newrdata, link)) { INSIST(newlist != NULL); ISC_LIST_UNLINK(newlist->rdata, newrdata, link); } dns_message_puttemprdata(msg, &newrdata); } if (newname != NULL) dns_message_puttempname(msg, &newname); if (newset != NULL) { dns_rdataset_disassociate(newset); dns_message_puttemprdataset(msg, &newset); } if (newlist != NULL) dns_message_puttemprdatalist(msg, &newlist); return (result); } static void free_namelist(dns_message_t *msg, dns_namelist_t *namelist) { dns_name_t *name; dns_rdataset_t *set; while (!ISC_LIST_EMPTY(*namelist)) { name = ISC_LIST_HEAD(*namelist); ISC_LIST_UNLINK(*namelist, name, link); while (!ISC_LIST_EMPTY(name->list)) { set = ISC_LIST_HEAD(name->list); ISC_LIST_UNLINK(name->list, set, link); dns_message_puttemprdataset(msg, &set); } dns_message_puttempname(msg, &name); } } static isc_result_t compute_secret(isc_buffer_t *shared, isc_region_t *queryrandomness, isc_region_t *serverrandomness, isc_buffer_t *secret) { isc_md5_t md5ctx; isc_region_t r, r2; unsigned char digests[32]; unsigned int i; isc_buffer_usedregion(shared, &r); /* * MD5 ( query data | DH value ). */ isc_md5_init(&md5ctx); isc_md5_update(&md5ctx, queryrandomness->base, queryrandomness->length); isc_md5_update(&md5ctx, r.base, r.length); isc_md5_final(&md5ctx, digests); /* * MD5 ( server data | DH value ). */ isc_md5_init(&md5ctx); isc_md5_update(&md5ctx, serverrandomness->base, serverrandomness->length); isc_md5_update(&md5ctx, r.base, r.length); isc_md5_final(&md5ctx, &digests[ISC_MD5_DIGESTLENGTH]); /* * XOR ( DH value, MD5-1 | MD5-2). */ isc_buffer_availableregion(secret, &r); isc_buffer_usedregion(shared, &r2); if (r.length < sizeof(digests) || r.length < r2.length) return (ISC_R_NOSPACE); if (r2.length > sizeof(digests)) { memmove(r.base, r2.base, r2.length); for (i = 0; i < sizeof(digests); i++) r.base[i] ^= digests[i]; isc_buffer_add(secret, r2.length); } else { memmove(r.base, digests, sizeof(digests)); for (i = 0; i < r2.length; i++) r.base[i] ^= r2.base[i]; isc_buffer_add(secret, sizeof(digests)); } return (ISC_R_SUCCESS); } static isc_result_t process_dhtkey(dns_message_t *msg, dns_name_t *signer, dns_name_t *name, dns_rdata_tkey_t *tkeyin, dns_tkeyctx_t *tctx, dns_rdata_tkey_t *tkeyout, dns_tsig_keyring_t *ring, dns_namelist_t *namelist) { isc_result_t result = ISC_R_SUCCESS; dns_name_t *keyname, ourname; dns_rdataset_t *keyset = NULL; dns_rdata_t keyrdata = DNS_RDATA_INIT, ourkeyrdata = DNS_RDATA_INIT; isc_boolean_t found_key = ISC_FALSE, found_incompatible = ISC_FALSE; dst_key_t *pubkey = NULL; isc_buffer_t ourkeybuf, *shared = NULL; isc_region_t r, r2, ourkeyr; unsigned char keydata[DST_KEY_MAXSIZE]; unsigned int sharedsize; isc_buffer_t secret; unsigned char *randomdata = NULL, secretdata[256]; dns_ttl_t ttl = 0; if (tctx->dhkey == NULL) { tkey_log("process_dhtkey: tkey-dhkey not defined"); tkeyout->error = dns_tsigerror_badalg; return (DNS_R_REFUSED); } if (!dns_name_equal(&tkeyin->algorithm, DNS_TSIG_HMACMD5_NAME)) { tkey_log("process_dhtkey: algorithms other than " "hmac-md5 are not supported"); tkeyout->error = dns_tsigerror_badalg; return (ISC_R_SUCCESS); } /* * Look for a DH KEY record that will work with ours. */ for (result = dns_message_firstname(msg, DNS_SECTION_ADDITIONAL); result == ISC_R_SUCCESS && !found_key; result = dns_message_nextname(msg, DNS_SECTION_ADDITIONAL)) { keyname = NULL; dns_message_currentname(msg, DNS_SECTION_ADDITIONAL, &keyname); keyset = NULL; result = dns_message_findtype(keyname, dns_rdatatype_key, 0, &keyset); if (result != ISC_R_SUCCESS) continue; for (result = dns_rdataset_first(keyset); result == ISC_R_SUCCESS && !found_key; result = dns_rdataset_next(keyset)) { dns_rdataset_current(keyset, &keyrdata); pubkey = NULL; result = dns_dnssec_keyfromrdata(keyname, &keyrdata, msg->mctx, &pubkey); if (result != ISC_R_SUCCESS) { dns_rdata_reset(&keyrdata); continue; } if (dst_key_alg(pubkey) == DNS_KEYALG_DH) { if (dst_key_paramcompare(pubkey, tctx->dhkey)) { found_key = ISC_TRUE; ttl = keyset->ttl; break; } else found_incompatible = ISC_TRUE; } dst_key_free(&pubkey); dns_rdata_reset(&keyrdata); } } if (!found_key) { if (found_incompatible) { tkey_log("process_dhtkey: found an incompatible key"); tkeyout->error = dns_tsigerror_badkey; return (ISC_R_SUCCESS); } else { tkey_log("process_dhtkey: failed to find a key"); return (DNS_R_FORMERR); } } RETERR(add_rdata_to_list(msg, keyname, &keyrdata, ttl, namelist)); isc_buffer_init(&ourkeybuf, keydata, sizeof(keydata)); RETERR(dst_key_todns(tctx->dhkey, &ourkeybuf)); isc_buffer_usedregion(&ourkeybuf, &ourkeyr); dns_rdata_fromregion(&ourkeyrdata, dns_rdataclass_any, dns_rdatatype_key, &ourkeyr); dns_name_init(&ourname, NULL); dns_name_clone(dst_key_name(tctx->dhkey), &ourname); /* * XXXBEW The TTL should be obtained from the database, if it exists. */ RETERR(add_rdata_to_list(msg, &ourname, &ourkeyrdata, 0, namelist)); RETERR(dst_key_secretsize(tctx->dhkey, &sharedsize)); RETERR(isc_buffer_allocate(msg->mctx, &shared, sharedsize)); result = dst_key_computesecret(pubkey, tctx->dhkey, shared); if (result != ISC_R_SUCCESS) { tkey_log("process_dhtkey: failed to compute shared secret: %s", isc_result_totext(result)); goto failure; } dst_key_free(&pubkey); isc_buffer_init(&secret, secretdata, sizeof(secretdata)); randomdata = isc_mem_get(tkeyout->mctx, TKEY_RANDOM_AMOUNT); if (randomdata == NULL) goto failure; result = isc_entropy_getdata(tctx->ectx, randomdata, TKEY_RANDOM_AMOUNT, NULL, 0); if (result != ISC_R_SUCCESS) { tkey_log("process_dhtkey: failed to obtain entropy: %s", isc_result_totext(result)); goto failure; } r.base = randomdata; r.length = TKEY_RANDOM_AMOUNT; r2.base = tkeyin->key; r2.length = tkeyin->keylen; RETERR(compute_secret(shared, &r2, &r, &secret)); isc_buffer_free(&shared); RETERR(dns_tsigkey_create(name, &tkeyin->algorithm, isc_buffer_base(&secret), isc_buffer_usedlength(&secret), ISC_TRUE, signer, tkeyin->inception, tkeyin->expire, ring->mctx, ring, NULL)); /* This key is good for a long time */ tkeyout->inception = tkeyin->inception; tkeyout->expire = tkeyin->expire; tkeyout->key = randomdata; tkeyout->keylen = TKEY_RANDOM_AMOUNT; return (ISC_R_SUCCESS); failure: if (!ISC_LIST_EMPTY(*namelist)) free_namelist(msg, namelist); if (shared != NULL) isc_buffer_free(&shared); if (pubkey != NULL) dst_key_free(&pubkey); if (randomdata != NULL) isc_mem_put(tkeyout->mctx, randomdata, TKEY_RANDOM_AMOUNT); return (result); } static isc_result_t process_gsstkey(dns_name_t *name, dns_rdata_tkey_t *tkeyin, dns_tkeyctx_t *tctx, dns_rdata_tkey_t *tkeyout, dns_tsig_keyring_t *ring) { isc_result_t result = ISC_R_SUCCESS; dst_key_t *dstkey = NULL; dns_tsigkey_t *tsigkey = NULL; dns_fixedname_t principal; isc_stdtime_t now; isc_region_t intoken; isc_buffer_t *outtoken = NULL; gss_ctx_id_t gss_ctx = NULL; /* * You have to define either a gss credential (principal) to * accept with tkey-gssapi-credential, or you have to * configure a specific keytab (with tkey-gssapi-keytab) in * order to use gsstkey */ if (tctx->gsscred == NULL && tctx->gssapi_keytab == NULL) { tkey_log("process_gsstkey(): no tkey-gssapi-credential " "or tkey-gssapi-keytab configured"); return (ISC_R_NOPERM); } if (!dns_name_equal(&tkeyin->algorithm, DNS_TSIG_GSSAPI_NAME) && !dns_name_equal(&tkeyin->algorithm, DNS_TSIG_GSSAPIMS_NAME)) { tkeyout->error = dns_tsigerror_badalg; tkey_log("process_gsstkey(): dns_tsigerror_badalg"); /* XXXSRA */ return (ISC_R_SUCCESS); } /* * XXXDCL need to check for key expiry per 4.1.1 * XXXDCL need a way to check fully established, perhaps w/key_flags */ intoken.base = tkeyin->key; intoken.length = tkeyin->keylen; result = dns_tsigkey_find(&tsigkey, name, &tkeyin->algorithm, ring); if (result == ISC_R_SUCCESS) gss_ctx = dst_key_getgssctx(tsigkey->key); dns_fixedname_init(&principal); /* * Note that tctx->gsscred may be NULL if tctx->gssapi_keytab is set */ result = dst_gssapi_acceptctx(tctx->gsscred, tctx->gssapi_keytab, &intoken, &outtoken, &gss_ctx, dns_fixedname_name(&principal), tctx->mctx); if (result == DNS_R_INVALIDTKEY) { if (tsigkey != NULL) dns_tsigkey_detach(&tsigkey); tkeyout->error = dns_tsigerror_badkey; tkey_log("process_gsstkey(): dns_tsigerror_badkey"); /* XXXSRA */ return (ISC_R_SUCCESS); } if (result != DNS_R_CONTINUE && result != ISC_R_SUCCESS) goto failure; /* * XXXDCL Section 4.1.3: Limit GSS_S_CONTINUE_NEEDED to 10 times. */ isc_stdtime_get(&now); if (tsigkey == NULL) { #ifdef GSSAPI OM_uint32 gret, minor, lifetime; #endif isc_uint32_t expire; RETERR(dst_key_fromgssapi(name, gss_ctx, ring->mctx, &dstkey, &intoken)); /* * Limit keys to 1 hour or the context's lifetime whichever * is smaller. */ expire = now + 3600; #ifdef GSSAPI gret = gss_context_time(&minor, gss_ctx, &lifetime); if (gret == GSS_S_COMPLETE && now + lifetime < expire) expire = now + lifetime; #endif RETERR(dns_tsigkey_createfromkey(name, &tkeyin->algorithm, dstkey, ISC_TRUE, dns_fixedname_name(&principal), now, expire, ring->mctx, ring, NULL)); dst_key_free(&dstkey); tkeyout->inception = now; tkeyout->expire = expire; } else { tkeyout->inception = tsigkey->inception; tkeyout->expire = tsigkey->expire; dns_tsigkey_detach(&tsigkey); } if (outtoken) { tkeyout->key = isc_mem_get(tkeyout->mctx, isc_buffer_usedlength(outtoken)); if (tkeyout->key == NULL) { result = ISC_R_NOMEMORY; goto failure; } tkeyout->keylen = isc_buffer_usedlength(outtoken); memmove(tkeyout->key, isc_buffer_base(outtoken), isc_buffer_usedlength(outtoken)); isc_buffer_free(&outtoken); } else { tkeyout->key = isc_mem_get(tkeyout->mctx, tkeyin->keylen); if (tkeyout->key == NULL) { result = ISC_R_NOMEMORY; goto failure; } tkeyout->keylen = tkeyin->keylen; memmove(tkeyout->key, tkeyin->key, tkeyin->keylen); } tkeyout->error = dns_rcode_noerror; tkey_log("process_gsstkey(): dns_tsigerror_noerror"); /* XXXSRA */ return (ISC_R_SUCCESS); failure: if (tsigkey != NULL) dns_tsigkey_detach(&tsigkey); if (dstkey != NULL) dst_key_free(&dstkey); if (outtoken != NULL) isc_buffer_free(&outtoken); tkey_log("process_gsstkey(): %s", isc_result_totext(result)); /* XXXSRA */ return (result); } static isc_result_t process_deletetkey(dns_name_t *signer, dns_name_t *name, dns_rdata_tkey_t *tkeyin, dns_rdata_tkey_t *tkeyout, dns_tsig_keyring_t *ring) { isc_result_t result; dns_tsigkey_t *tsigkey = NULL; dns_name_t *identity; result = dns_tsigkey_find(&tsigkey, name, &tkeyin->algorithm, ring); if (result != ISC_R_SUCCESS) { tkeyout->error = dns_tsigerror_badname; return (ISC_R_SUCCESS); } /* * Only allow a delete if the identity that created the key is the * same as the identity that signed the message. */ identity = dns_tsigkey_identity(tsigkey); if (identity == NULL || !dns_name_equal(identity, signer)) { dns_tsigkey_detach(&tsigkey); return (DNS_R_REFUSED); } /* * Set the key to be deleted when no references are left. If the key * was not generated with TKEY and is in the config file, it may be * reloaded later. */ dns_tsigkey_setdeleted(tsigkey); /* Release the reference */ dns_tsigkey_detach(&tsigkey); return (ISC_R_SUCCESS); } isc_result_t dns_tkey_processquery(dns_message_t *msg, dns_tkeyctx_t *tctx, dns_tsig_keyring_t *ring) { isc_result_t result = ISC_R_SUCCESS; dns_rdata_tkey_t tkeyin, tkeyout; isc_boolean_t freetkeyin = ISC_FALSE; dns_name_t *qname, *name, *keyname, *signer, tsigner; dns_fixedname_t fkeyname; dns_rdataset_t *tkeyset; dns_rdata_t rdata; dns_namelist_t namelist; char tkeyoutdata[512]; isc_buffer_t tkeyoutbuf; REQUIRE(msg != NULL); REQUIRE(tctx != NULL); REQUIRE(ring != NULL); ISC_LIST_INIT(namelist); /* * Interpret the question section. */ result = dns_message_firstname(msg, DNS_SECTION_QUESTION); if (result != ISC_R_SUCCESS) return (DNS_R_FORMERR); qname = NULL; dns_message_currentname(msg, DNS_SECTION_QUESTION, &qname); /* * Look for a TKEY record that matches the question. */ tkeyset = NULL; name = NULL; result = dns_message_findname(msg, DNS_SECTION_ADDITIONAL, qname, dns_rdatatype_tkey, 0, &name, &tkeyset); if (result != ISC_R_SUCCESS) { /* * Try the answer section, since that's where Win2000 * puts it. */ + name = NULL; if (dns_message_findname(msg, DNS_SECTION_ANSWER, qname, dns_rdatatype_tkey, 0, &name, &tkeyset) != ISC_R_SUCCESS) { result = DNS_R_FORMERR; tkey_log("dns_tkey_processquery: couldn't find a TKEY " "matching the question"); goto failure; } } result = dns_rdataset_first(tkeyset); if (result != ISC_R_SUCCESS) { result = DNS_R_FORMERR; goto failure; } dns_rdata_init(&rdata); dns_rdataset_current(tkeyset, &rdata); RETERR(dns_rdata_tostruct(&rdata, &tkeyin, NULL)); freetkeyin = ISC_TRUE; if (tkeyin.error != dns_rcode_noerror) { result = DNS_R_FORMERR; goto failure; } /* * Before we go any farther, verify that the message was signed. * GSSAPI TKEY doesn't require a signature, the rest do. */ dns_name_init(&tsigner, NULL); result = dns_message_signer(msg, &tsigner); if (result != ISC_R_SUCCESS) { if (tkeyin.mode == DNS_TKEYMODE_GSSAPI && result == ISC_R_NOTFOUND) signer = NULL; else { tkey_log("dns_tkey_processquery: query was not " "properly signed - rejecting"); result = DNS_R_FORMERR; goto failure; } } else signer = &tsigner; tkeyout.common.rdclass = tkeyin.common.rdclass; tkeyout.common.rdtype = tkeyin.common.rdtype; ISC_LINK_INIT(&tkeyout.common, link); tkeyout.mctx = msg->mctx; dns_name_init(&tkeyout.algorithm, NULL); dns_name_clone(&tkeyin.algorithm, &tkeyout.algorithm); tkeyout.inception = tkeyout.expire = 0; tkeyout.mode = tkeyin.mode; tkeyout.error = 0; tkeyout.keylen = tkeyout.otherlen = 0; tkeyout.key = tkeyout.other = NULL; /* * A delete operation must have a fully specified key name. If this * is not a delete, we do the following: * if (qname != ".") * keyname = qname + defaultdomain * else * keyname = + defaultdomain */ if (tkeyin.mode != DNS_TKEYMODE_DELETE) { dns_tsigkey_t *tsigkey = NULL; if (tctx->domain == NULL && tkeyin.mode != DNS_TKEYMODE_GSSAPI) { tkey_log("dns_tkey_processquery: tkey-domain not set"); result = DNS_R_REFUSED; goto failure; } dns_fixedname_init(&fkeyname); keyname = dns_fixedname_name(&fkeyname); if (!dns_name_equal(qname, dns_rootname)) { unsigned int n = dns_name_countlabels(qname); RUNTIME_CHECK(dns_name_copy(qname, keyname, NULL) == ISC_R_SUCCESS); dns_name_getlabelsequence(keyname, 0, n - 1, keyname); } else { static char hexdigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; unsigned char randomdata[16]; char randomtext[32]; isc_buffer_t b; unsigned int i, j; result = isc_entropy_getdata(tctx->ectx, randomdata, sizeof(randomdata), NULL, 0); if (result != ISC_R_SUCCESS) goto failure; for (i = 0, j = 0; i < sizeof(randomdata); i++) { unsigned char val = randomdata[i]; randomtext[j++] = hexdigits[val >> 4]; randomtext[j++] = hexdigits[val & 0xF]; } isc_buffer_init(&b, randomtext, sizeof(randomtext)); isc_buffer_add(&b, sizeof(randomtext)); result = dns_name_fromtext(keyname, &b, NULL, 0, NULL); if (result != ISC_R_SUCCESS) goto failure; } if (tkeyin.mode == DNS_TKEYMODE_GSSAPI) { /* Yup. This is a hack */ result = dns_name_concatenate(keyname, dns_rootname, keyname, NULL); if (result != ISC_R_SUCCESS) goto failure; } else { result = dns_name_concatenate(keyname, tctx->domain, keyname, NULL); if (result != ISC_R_SUCCESS) goto failure; } result = dns_tsigkey_find(&tsigkey, keyname, NULL, ring); if (result == ISC_R_SUCCESS) { tkeyout.error = dns_tsigerror_badname; dns_tsigkey_detach(&tsigkey); goto failure_with_tkey; } else if (result != ISC_R_NOTFOUND) goto failure; } else keyname = qname; switch (tkeyin.mode) { case DNS_TKEYMODE_DIFFIEHELLMAN: tkeyout.error = dns_rcode_noerror; RETERR(process_dhtkey(msg, signer, keyname, &tkeyin, tctx, &tkeyout, ring, &namelist)); break; case DNS_TKEYMODE_GSSAPI: tkeyout.error = dns_rcode_noerror; RETERR(process_gsstkey(keyname, &tkeyin, tctx, &tkeyout, ring)); break; case DNS_TKEYMODE_DELETE: tkeyout.error = dns_rcode_noerror; RETERR(process_deletetkey(signer, keyname, &tkeyin, &tkeyout, ring)); break; case DNS_TKEYMODE_SERVERASSIGNED: case DNS_TKEYMODE_RESOLVERASSIGNED: result = DNS_R_NOTIMP; goto failure; default: tkeyout.error = dns_tsigerror_badmode; } failure_with_tkey: dns_rdata_init(&rdata); isc_buffer_init(&tkeyoutbuf, tkeyoutdata, sizeof(tkeyoutdata)); result = dns_rdata_fromstruct(&rdata, tkeyout.common.rdclass, tkeyout.common.rdtype, &tkeyout, &tkeyoutbuf); if (freetkeyin) { dns_rdata_freestruct(&tkeyin); freetkeyin = ISC_FALSE; } if (tkeyout.key != NULL) isc_mem_put(tkeyout.mctx, tkeyout.key, tkeyout.keylen); if (tkeyout.other != NULL) isc_mem_put(tkeyout.mctx, tkeyout.other, tkeyout.otherlen); if (result != ISC_R_SUCCESS) goto failure; RETERR(add_rdata_to_list(msg, keyname, &rdata, 0, &namelist)); RETERR(dns_message_reply(msg, ISC_TRUE)); name = ISC_LIST_HEAD(namelist); while (name != NULL) { dns_name_t *next = ISC_LIST_NEXT(name, link); ISC_LIST_UNLINK(namelist, name, link); dns_message_addname(msg, name, DNS_SECTION_ANSWER); name = next; } return (ISC_R_SUCCESS); failure: if (freetkeyin) dns_rdata_freestruct(&tkeyin); if (!ISC_LIST_EMPTY(namelist)) free_namelist(msg, &namelist); return (result); } static isc_result_t buildquery(dns_message_t *msg, dns_name_t *name, dns_rdata_tkey_t *tkey, isc_boolean_t win2k) { dns_name_t *qname = NULL, *aname = NULL; dns_rdataset_t *question = NULL, *tkeyset = NULL; dns_rdatalist_t *tkeylist = NULL; dns_rdata_t *rdata = NULL; isc_buffer_t *dynbuf = NULL; isc_result_t result; REQUIRE(msg != NULL); REQUIRE(name != NULL); REQUIRE(tkey != NULL); RETERR(dns_message_gettempname(msg, &qname)); RETERR(dns_message_gettempname(msg, &aname)); RETERR(dns_message_gettemprdataset(msg, &question)); dns_rdataset_init(question); dns_rdataset_makequestion(question, dns_rdataclass_any, dns_rdatatype_tkey); RETERR(isc_buffer_allocate(msg->mctx, &dynbuf, 4096)); RETERR(dns_message_gettemprdata(msg, &rdata)); RETERR(dns_rdata_fromstruct(rdata, dns_rdataclass_any, dns_rdatatype_tkey, tkey, dynbuf)); dns_message_takebuffer(msg, &dynbuf); RETERR(dns_message_gettemprdatalist(msg, &tkeylist)); tkeylist->rdclass = dns_rdataclass_any; tkeylist->type = dns_rdatatype_tkey; tkeylist->covers = 0; tkeylist->ttl = 0; ISC_LIST_INIT(tkeylist->rdata); ISC_LIST_APPEND(tkeylist->rdata, rdata, link); RETERR(dns_message_gettemprdataset(msg, &tkeyset)); dns_rdataset_init(tkeyset); RETERR(dns_rdatalist_tordataset(tkeylist, tkeyset)); dns_name_init(qname, NULL); dns_name_clone(name, qname); dns_name_init(aname, NULL); dns_name_clone(name, aname); ISC_LIST_APPEND(qname->list, question, link); ISC_LIST_APPEND(aname->list, tkeyset, link); dns_message_addname(msg, qname, DNS_SECTION_QUESTION); /* * Windows 2000 needs this in the answer section, not the additional * section where the RFC specifies. */ if (win2k) dns_message_addname(msg, aname, DNS_SECTION_ANSWER); else dns_message_addname(msg, aname, DNS_SECTION_ADDITIONAL); return (ISC_R_SUCCESS); failure: if (qname != NULL) dns_message_puttempname(msg, &qname); if (aname != NULL) dns_message_puttempname(msg, &aname); if (question != NULL) { dns_rdataset_disassociate(question); dns_message_puttemprdataset(msg, &question); } if (dynbuf != NULL) isc_buffer_free(&dynbuf); printf("buildquery error\n"); return (result); } isc_result_t dns_tkey_builddhquery(dns_message_t *msg, dst_key_t *key, dns_name_t *name, dns_name_t *algorithm, isc_buffer_t *nonce, isc_uint32_t lifetime) { dns_rdata_tkey_t tkey; dns_rdata_t *rdata = NULL; isc_buffer_t *dynbuf = NULL; isc_region_t r; dns_name_t keyname; dns_namelist_t namelist; isc_result_t result; isc_stdtime_t now; REQUIRE(msg != NULL); REQUIRE(key != NULL); REQUIRE(dst_key_alg(key) == DNS_KEYALG_DH); REQUIRE(dst_key_isprivate(key)); REQUIRE(name != NULL); REQUIRE(algorithm != NULL); tkey.common.rdclass = dns_rdataclass_any; tkey.common.rdtype = dns_rdatatype_tkey; ISC_LINK_INIT(&tkey.common, link); tkey.mctx = msg->mctx; dns_name_init(&tkey.algorithm, NULL); dns_name_clone(algorithm, &tkey.algorithm); isc_stdtime_get(&now); tkey.inception = now; tkey.expire = now + lifetime; tkey.mode = DNS_TKEYMODE_DIFFIEHELLMAN; if (nonce != NULL) isc_buffer_usedregion(nonce, &r); else { r.base = isc_mem_get(msg->mctx, 0); r.length = 0; } tkey.error = 0; tkey.key = r.base; tkey.keylen = r.length; tkey.other = NULL; tkey.otherlen = 0; RETERR(buildquery(msg, name, &tkey, ISC_FALSE)); if (nonce == NULL) isc_mem_put(msg->mctx, r.base, 0); RETERR(dns_message_gettemprdata(msg, &rdata)); RETERR(isc_buffer_allocate(msg->mctx, &dynbuf, 1024)); RETERR(dst_key_todns(key, dynbuf)); isc_buffer_usedregion(dynbuf, &r); dns_rdata_fromregion(rdata, dns_rdataclass_any, dns_rdatatype_key, &r); dns_message_takebuffer(msg, &dynbuf); dns_name_init(&keyname, NULL); dns_name_clone(dst_key_name(key), &keyname); ISC_LIST_INIT(namelist); RETERR(add_rdata_to_list(msg, &keyname, rdata, 0, &namelist)); name = ISC_LIST_HEAD(namelist); while (name != NULL) { dns_name_t *next = ISC_LIST_NEXT(name, link); ISC_LIST_UNLINK(namelist, name, link); dns_message_addname(msg, name, DNS_SECTION_ADDITIONAL); name = next; } return (ISC_R_SUCCESS); failure: if (dynbuf != NULL) isc_buffer_free(&dynbuf); return (result); } isc_result_t dns_tkey_buildgssquery(dns_message_t *msg, dns_name_t *name, dns_name_t *gname, isc_buffer_t *intoken, isc_uint32_t lifetime, gss_ctx_id_t *context, isc_boolean_t win2k, isc_mem_t *mctx, char **err_message) { dns_rdata_tkey_t tkey; isc_result_t result; isc_stdtime_t now; isc_buffer_t token; unsigned char array[4096]; UNUSED(intoken); REQUIRE(msg != NULL); REQUIRE(name != NULL); REQUIRE(gname != NULL); REQUIRE(context != NULL); REQUIRE(mctx != NULL); isc_buffer_init(&token, array, sizeof(array)); result = dst_gssapi_initctx(gname, NULL, &token, context, mctx, err_message); if (result != DNS_R_CONTINUE && result != ISC_R_SUCCESS) return (result); tkey.common.rdclass = dns_rdataclass_any; tkey.common.rdtype = dns_rdatatype_tkey; ISC_LINK_INIT(&tkey.common, link); tkey.mctx = NULL; dns_name_init(&tkey.algorithm, NULL); if (win2k) dns_name_clone(DNS_TSIG_GSSAPIMS_NAME, &tkey.algorithm); else dns_name_clone(DNS_TSIG_GSSAPI_NAME, &tkey.algorithm); isc_stdtime_get(&now); tkey.inception = now; tkey.expire = now + lifetime; tkey.mode = DNS_TKEYMODE_GSSAPI; tkey.error = 0; tkey.key = isc_buffer_base(&token); tkey.keylen = isc_buffer_usedlength(&token); tkey.other = NULL; tkey.otherlen = 0; RETERR(buildquery(msg, name, &tkey, win2k)); return (ISC_R_SUCCESS); failure: return (result); } isc_result_t dns_tkey_builddeletequery(dns_message_t *msg, dns_tsigkey_t *key) { dns_rdata_tkey_t tkey; REQUIRE(msg != NULL); REQUIRE(key != NULL); tkey.common.rdclass = dns_rdataclass_any; tkey.common.rdtype = dns_rdatatype_tkey; ISC_LINK_INIT(&tkey.common, link); tkey.mctx = msg->mctx; dns_name_init(&tkey.algorithm, NULL); dns_name_clone(key->algorithm, &tkey.algorithm); tkey.inception = tkey.expire = 0; tkey.mode = DNS_TKEYMODE_DELETE; tkey.error = 0; tkey.keylen = tkey.otherlen = 0; tkey.key = tkey.other = NULL; return (buildquery(msg, &key->name, &tkey, ISC_FALSE)); } static isc_result_t find_tkey(dns_message_t *msg, dns_name_t **name, dns_rdata_t *rdata, int section) { dns_rdataset_t *tkeyset; isc_result_t result; result = dns_message_firstname(msg, section); while (result == ISC_R_SUCCESS) { *name = NULL; dns_message_currentname(msg, section, name); tkeyset = NULL; result = dns_message_findtype(*name, dns_rdatatype_tkey, 0, &tkeyset); if (result == ISC_R_SUCCESS) { result = dns_rdataset_first(tkeyset); if (result != ISC_R_SUCCESS) return (result); dns_rdataset_current(tkeyset, rdata); return (ISC_R_SUCCESS); } result = dns_message_nextname(msg, section); } if (result == ISC_R_NOMORE) return (ISC_R_NOTFOUND); return (result); } isc_result_t dns_tkey_processdhresponse(dns_message_t *qmsg, dns_message_t *rmsg, dst_key_t *key, isc_buffer_t *nonce, dns_tsigkey_t **outkey, dns_tsig_keyring_t *ring) { dns_rdata_t qtkeyrdata = DNS_RDATA_INIT, rtkeyrdata = DNS_RDATA_INIT; dns_name_t keyname, *tkeyname, *theirkeyname, *ourkeyname, *tempname; dns_rdataset_t *theirkeyset = NULL, *ourkeyset = NULL; dns_rdata_t theirkeyrdata = DNS_RDATA_INIT; dst_key_t *theirkey = NULL; dns_rdata_tkey_t qtkey, rtkey; unsigned char secretdata[256]; unsigned int sharedsize; isc_buffer_t *shared = NULL, secret; isc_region_t r, r2; isc_result_t result; isc_boolean_t freertkey = ISC_FALSE; REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); REQUIRE(key != NULL); REQUIRE(dst_key_alg(key) == DNS_KEYALG_DH); REQUIRE(dst_key_isprivate(key)); if (outkey != NULL) REQUIRE(*outkey == NULL); if (rmsg->rcode != dns_rcode_noerror) return (ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); freertkey = ISC_TRUE; RETERR(find_tkey(qmsg, &tempname, &qtkeyrdata, DNS_SECTION_ADDITIONAL)); RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_DIFFIEHELLMAN || rtkey.mode != qtkey.mode || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm) || rmsg->rcode != dns_rcode_noerror) { tkey_log("dns_tkey_processdhresponse: tkey mode invalid " "or error set(1)"); result = DNS_R_INVALIDTKEY; dns_rdata_freestruct(&qtkey); goto failure; } dns_rdata_freestruct(&qtkey); dns_name_init(&keyname, NULL); dns_name_clone(dst_key_name(key), &keyname); ourkeyname = NULL; ourkeyset = NULL; RETERR(dns_message_findname(rmsg, DNS_SECTION_ANSWER, &keyname, dns_rdatatype_key, 0, &ourkeyname, &ourkeyset)); result = dns_message_firstname(rmsg, DNS_SECTION_ANSWER); while (result == ISC_R_SUCCESS) { theirkeyname = NULL; dns_message_currentname(rmsg, DNS_SECTION_ANSWER, &theirkeyname); if (dns_name_equal(theirkeyname, ourkeyname)) goto next; theirkeyset = NULL; result = dns_message_findtype(theirkeyname, dns_rdatatype_key, 0, &theirkeyset); if (result == ISC_R_SUCCESS) { RETERR(dns_rdataset_first(theirkeyset)); break; } next: result = dns_message_nextname(rmsg, DNS_SECTION_ANSWER); } if (theirkeyset == NULL) { tkey_log("dns_tkey_processdhresponse: failed to find server " "key"); result = ISC_R_NOTFOUND; goto failure; } dns_rdataset_current(theirkeyset, &theirkeyrdata); RETERR(dns_dnssec_keyfromrdata(theirkeyname, &theirkeyrdata, rmsg->mctx, &theirkey)); RETERR(dst_key_secretsize(key, &sharedsize)); RETERR(isc_buffer_allocate(rmsg->mctx, &shared, sharedsize)); RETERR(dst_key_computesecret(theirkey, key, shared)); isc_buffer_init(&secret, secretdata, sizeof(secretdata)); r.base = rtkey.key; r.length = rtkey.keylen; if (nonce != NULL) isc_buffer_usedregion(nonce, &r2); else { r2.base = isc_mem_get(rmsg->mctx, 0); r2.length = 0; } RETERR(compute_secret(shared, &r2, &r, &secret)); if (nonce == NULL) isc_mem_put(rmsg->mctx, r2.base, 0); isc_buffer_usedregion(&secret, &r); result = dns_tsigkey_create(tkeyname, &rtkey.algorithm, r.base, r.length, ISC_TRUE, NULL, rtkey.inception, rtkey.expire, rmsg->mctx, ring, outkey); isc_buffer_free(&shared); dns_rdata_freestruct(&rtkey); dst_key_free(&theirkey); return (result); failure: if (shared != NULL) isc_buffer_free(&shared); if (theirkey != NULL) dst_key_free(&theirkey); if (freertkey) dns_rdata_freestruct(&rtkey); return (result); } isc_result_t dns_tkey_processgssresponse(dns_message_t *qmsg, dns_message_t *rmsg, dns_name_t *gname, gss_ctx_id_t *context, isc_buffer_t *outtoken, dns_tsigkey_t **outkey, dns_tsig_keyring_t *ring, char **err_message) { dns_rdata_t rtkeyrdata = DNS_RDATA_INIT, qtkeyrdata = DNS_RDATA_INIT; dns_name_t *tkeyname; dns_rdata_tkey_t rtkey, qtkey; dst_key_t *dstkey = NULL; isc_buffer_t intoken; isc_result_t result; unsigned char array[1024]; REQUIRE(outtoken != NULL); REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); REQUIRE(gname != NULL); REQUIRE(ring != NULL); if (outkey != NULL) REQUIRE(*outkey == NULL); if (rmsg->rcode != dns_rcode_noerror) return (ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); /* * Win2k puts the item in the ANSWER section, while the RFC * specifies it should be in the ADDITIONAL section. Check first * where it should be, and then where it may be. */ result = find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ADDITIONAL); if (result == ISC_R_NOTFOUND) result = find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ANSWER); if (result != ISC_R_SUCCESS) goto failure; RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_GSSAPI || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm)) { tkey_log("dns_tkey_processgssresponse: tkey mode invalid " "or error set(2) %d", rtkey.error); _dns_tkey_dumpmessage(qmsg); _dns_tkey_dumpmessage(rmsg); result = DNS_R_INVALIDTKEY; goto failure; } isc_buffer_init(outtoken, array, sizeof(array)); isc_buffer_init(&intoken, rtkey.key, rtkey.keylen); RETERR(dst_gssapi_initctx(gname, &intoken, outtoken, context, ring->mctx, err_message)); RETERR(dst_key_fromgssapi(dns_rootname, *context, rmsg->mctx, &dstkey, NULL)); RETERR(dns_tsigkey_createfromkey(tkeyname, DNS_TSIG_GSSAPI_NAME, dstkey, ISC_FALSE, NULL, rtkey.inception, rtkey.expire, ring->mctx, ring, outkey)); dst_key_free(&dstkey); dns_rdata_freestruct(&rtkey); return (result); failure: /* * XXXSRA This probably leaks memory from rtkey and qtkey. */ if (dstkey != NULL) dst_key_free(&dstkey); return (result); } isc_result_t dns_tkey_processdeleteresponse(dns_message_t *qmsg, dns_message_t *rmsg, dns_tsig_keyring_t *ring) { dns_rdata_t qtkeyrdata = DNS_RDATA_INIT, rtkeyrdata = DNS_RDATA_INIT; dns_name_t *tkeyname, *tempname; dns_rdata_tkey_t qtkey, rtkey; dns_tsigkey_t *tsigkey = NULL; isc_result_t result; REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); if (rmsg->rcode != dns_rcode_noerror) return(ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); RETERR(find_tkey(qmsg, &tempname, &qtkeyrdata, DNS_SECTION_ADDITIONAL)); RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_DELETE || rtkey.mode != qtkey.mode || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm) || rmsg->rcode != dns_rcode_noerror) { tkey_log("dns_tkey_processdeleteresponse: tkey mode invalid " "or error set(3)"); result = DNS_R_INVALIDTKEY; dns_rdata_freestruct(&qtkey); dns_rdata_freestruct(&rtkey); goto failure; } dns_rdata_freestruct(&qtkey); RETERR(dns_tsigkey_find(&tsigkey, tkeyname, &rtkey.algorithm, ring)); dns_rdata_freestruct(&rtkey); /* * Mark the key as deleted. */ dns_tsigkey_setdeleted(tsigkey); /* * Release the reference. */ dns_tsigkey_detach(&tsigkey); failure: return (result); } isc_result_t dns_tkey_gssnegotiate(dns_message_t *qmsg, dns_message_t *rmsg, dns_name_t *server, gss_ctx_id_t *context, dns_tsigkey_t **outkey, dns_tsig_keyring_t *ring, isc_boolean_t win2k, char **err_message) { dns_rdata_t rtkeyrdata = DNS_RDATA_INIT, qtkeyrdata = DNS_RDATA_INIT; dns_name_t *tkeyname; dns_rdata_tkey_t rtkey, qtkey; isc_buffer_t intoken, outtoken; dst_key_t *dstkey = NULL; isc_result_t result; unsigned char array[1024]; REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); REQUIRE(server != NULL); if (outkey != NULL) REQUIRE(*outkey == NULL); if (rmsg->rcode != dns_rcode_noerror) return (ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); if (win2k == ISC_TRUE) RETERR(find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ANSWER)); else RETERR(find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ADDITIONAL)); RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_GSSAPI || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm)) { tkey_log("dns_tkey_processdhresponse: tkey mode invalid " "or error set(4)"); result = DNS_R_INVALIDTKEY; goto failure; } isc_buffer_init(&intoken, rtkey.key, rtkey.keylen); isc_buffer_init(&outtoken, array, sizeof(array)); result = dst_gssapi_initctx(server, &intoken, &outtoken, context, ring->mctx, err_message); if (result != DNS_R_CONTINUE && result != ISC_R_SUCCESS) return (result); RETERR(dst_key_fromgssapi(dns_rootname, *context, rmsg->mctx, &dstkey, NULL)); /* * XXXSRA This seems confused. If we got CONTINUE from initctx, * the GSS negotiation hasn't completed yet, so we can't sign * anything yet. */ RETERR(dns_tsigkey_createfromkey(tkeyname, (win2k ? DNS_TSIG_GSSAPIMS_NAME : DNS_TSIG_GSSAPI_NAME), dstkey, ISC_TRUE, NULL, rtkey.inception, rtkey.expire, ring->mctx, ring, outkey)); dst_key_free(&dstkey); dns_rdata_freestruct(&rtkey); return (result); failure: /* * XXXSRA This probably leaks memory from qtkey. */ dns_rdata_freestruct(&rtkey); if (dstkey != NULL) dst_key_free(&dstkey); return (result); } Index: stable/8/crypto/openssh/auth2-chall.c =================================================================== --- stable/8/crypto/openssh/auth2-chall.c (revision 285976) +++ stable/8/crypto/openssh/auth2-chall.c (revision 285977) @@ -1,374 +1,380 @@ /* $OpenBSD: auth2-chall.c,v 1.34 2008/12/09 04:32:22 djm Exp $ */ /* * Copyright (c) 2001 Markus Friedl. All rights reserved. * Copyright (c) 2001 Per Allansson. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "includes.h" #include #include #include #include #include "xmalloc.h" #include "ssh2.h" #include "key.h" #include "hostfile.h" #include "auth.h" #include "buffer.h" #include "packet.h" #include "dispatch.h" #include "log.h" #include "servconf.h" /* import */ extern ServerOptions options; static int auth2_challenge_start(Authctxt *); static int send_userauth_info_request(Authctxt *); static void input_userauth_info_response(int, u_int32_t, void *); #ifdef BSD_AUTH extern KbdintDevice bsdauth_device; #else #ifdef USE_PAM extern KbdintDevice sshpam_device; #endif #ifdef SKEY extern KbdintDevice skey_device; #endif #endif KbdintDevice *devices[] = { #ifdef BSD_AUTH &bsdauth_device, #else #ifdef USE_PAM &sshpam_device, #endif #ifdef SKEY &skey_device, #endif #endif NULL }; typedef struct KbdintAuthctxt KbdintAuthctxt; struct KbdintAuthctxt { char *devices; void *ctxt; KbdintDevice *device; u_int nreq; + u_int devices_done; }; #ifdef USE_PAM void remove_kbdint_device(const char *devname) { int i, j; for (i = 0; devices[i] != NULL; i++) if (strcmp(devices[i]->name, devname) == 0) { for (j = i; devices[j] != NULL; j++) devices[j] = devices[j+1]; i--; } } #endif static KbdintAuthctxt * kbdint_alloc(const char *devs) { KbdintAuthctxt *kbdintctxt; Buffer b; int i; #ifdef USE_PAM if (!options.use_pam) remove_kbdint_device("pam"); #endif kbdintctxt = xmalloc(sizeof(KbdintAuthctxt)); if (strcmp(devs, "") == 0) { buffer_init(&b); for (i = 0; devices[i]; i++) { if (buffer_len(&b) > 0) buffer_append(&b, ",", 1); buffer_append(&b, devices[i]->name, strlen(devices[i]->name)); } buffer_append(&b, "\0", 1); kbdintctxt->devices = xstrdup(buffer_ptr(&b)); buffer_free(&b); } else { kbdintctxt->devices = xstrdup(devs); } debug("kbdint_alloc: devices '%s'", kbdintctxt->devices); kbdintctxt->ctxt = NULL; kbdintctxt->device = NULL; kbdintctxt->nreq = 0; return kbdintctxt; } static void kbdint_reset_device(KbdintAuthctxt *kbdintctxt) { if (kbdintctxt->ctxt) { kbdintctxt->device->free_ctx(kbdintctxt->ctxt); kbdintctxt->ctxt = NULL; } kbdintctxt->device = NULL; } static void kbdint_free(KbdintAuthctxt *kbdintctxt) { if (kbdintctxt->device) kbdint_reset_device(kbdintctxt); if (kbdintctxt->devices) { xfree(kbdintctxt->devices); kbdintctxt->devices = NULL; } xfree(kbdintctxt); } /* get next device */ static int kbdint_next_device(KbdintAuthctxt *kbdintctxt) { size_t len; char *t; int i; if (kbdintctxt->device) kbdint_reset_device(kbdintctxt); do { len = kbdintctxt->devices ? strcspn(kbdintctxt->devices, ",") : 0; if (len == 0) break; - for (i = 0; devices[i]; i++) - if (strncmp(kbdintctxt->devices, devices[i]->name, len) == 0) + for (i = 0; devices[i]; i++) { + if ((kbdintctxt->devices_done & (1 << i)) != 0) + continue; + if (strncmp(kbdintctxt->devices, devices[i]->name, len) == 0) { kbdintctxt->device = devices[i]; + kbdintctxt->devices_done |= 1 << i; + } + } t = kbdintctxt->devices; kbdintctxt->devices = t[len] ? xstrdup(t+len+1) : NULL; xfree(t); debug2("kbdint_next_device: devices %s", kbdintctxt->devices ? kbdintctxt->devices : ""); } while (kbdintctxt->devices && !kbdintctxt->device); return kbdintctxt->device ? 1 : 0; } /* * try challenge-response, set authctxt->postponed if we have to * wait for the response. */ int auth2_challenge(Authctxt *authctxt, char *devs) { debug("auth2_challenge: user=%s devs=%s", authctxt->user ? authctxt->user : "", devs ? devs : ""); if (authctxt->user == NULL || !devs) return 0; if (authctxt->kbdintctxt == NULL) authctxt->kbdintctxt = kbdint_alloc(devs); return auth2_challenge_start(authctxt); } /* unregister kbd-int callbacks and context */ void auth2_challenge_stop(Authctxt *authctxt) { /* unregister callback */ dispatch_set(SSH2_MSG_USERAUTH_INFO_RESPONSE, NULL); if (authctxt->kbdintctxt != NULL) { kbdint_free(authctxt->kbdintctxt); authctxt->kbdintctxt = NULL; } } /* side effect: sets authctxt->postponed if a reply was sent*/ static int auth2_challenge_start(Authctxt *authctxt) { KbdintAuthctxt *kbdintctxt = authctxt->kbdintctxt; debug2("auth2_challenge_start: devices %s", kbdintctxt->devices ? kbdintctxt->devices : ""); if (kbdint_next_device(kbdintctxt) == 0) { auth2_challenge_stop(authctxt); return 0; } debug("auth2_challenge_start: trying authentication method '%s'", kbdintctxt->device->name); if ((kbdintctxt->ctxt = kbdintctxt->device->init_ctx(authctxt)) == NULL) { auth2_challenge_stop(authctxt); return 0; } if (send_userauth_info_request(authctxt) == 0) { auth2_challenge_stop(authctxt); return 0; } dispatch_set(SSH2_MSG_USERAUTH_INFO_RESPONSE, &input_userauth_info_response); authctxt->postponed = 1; return 0; } static int send_userauth_info_request(Authctxt *authctxt) { KbdintAuthctxt *kbdintctxt; char *name, *instr, **prompts; u_int i, *echo_on; kbdintctxt = authctxt->kbdintctxt; if (kbdintctxt->device->query(kbdintctxt->ctxt, &name, &instr, &kbdintctxt->nreq, &prompts, &echo_on)) return 0; packet_start(SSH2_MSG_USERAUTH_INFO_REQUEST); packet_put_cstring(name); packet_put_cstring(instr); packet_put_cstring(""); /* language not used */ packet_put_int(kbdintctxt->nreq); for (i = 0; i < kbdintctxt->nreq; i++) { packet_put_cstring(prompts[i]); packet_put_char(echo_on[i]); } packet_send(); packet_write_wait(); for (i = 0; i < kbdintctxt->nreq; i++) xfree(prompts[i]); xfree(prompts); xfree(echo_on); xfree(name); xfree(instr); return 1; } static void input_userauth_info_response(int type, u_int32_t seq, void *ctxt) { Authctxt *authctxt = ctxt; KbdintAuthctxt *kbdintctxt; int authenticated = 0, res; u_int i, nresp; char **response = NULL, *method; if (authctxt == NULL) fatal("input_userauth_info_response: no authctxt"); kbdintctxt = authctxt->kbdintctxt; if (kbdintctxt == NULL || kbdintctxt->ctxt == NULL) fatal("input_userauth_info_response: no kbdintctxt"); if (kbdintctxt->device == NULL) fatal("input_userauth_info_response: no device"); authctxt->postponed = 0; /* reset */ nresp = packet_get_int(); if (nresp != kbdintctxt->nreq) fatal("input_userauth_info_response: wrong number of replies"); if (nresp > 100) fatal("input_userauth_info_response: too many replies"); if (nresp > 0) { response = xcalloc(nresp, sizeof(char *)); for (i = 0; i < nresp; i++) response[i] = packet_get_string(NULL); } packet_check_eom(); res = kbdintctxt->device->respond(kbdintctxt->ctxt, nresp, response); for (i = 0; i < nresp; i++) { memset(response[i], 'r', strlen(response[i])); xfree(response[i]); } if (response) xfree(response); switch (res) { case 0: /* Success! */ authenticated = authctxt->valid ? 1 : 0; break; case 1: /* Authentication needs further interaction */ if (send_userauth_info_request(authctxt) == 1) authctxt->postponed = 1; break; default: /* Failure! */ break; } xasprintf(&method, "keyboard-interactive/%s", kbdintctxt->device->name); if (!authctxt->postponed) { if (authenticated) { auth2_challenge_stop(authctxt); } else { /* start next device */ /* may set authctxt->postponed */ auth2_challenge_start(authctxt); } } userauth_finish(authctxt, authenticated, method); xfree(method); } void privsep_challenge_enable(void) { #if defined(BSD_AUTH) || defined(USE_PAM) || defined(SKEY) int n = 0; #endif #ifdef BSD_AUTH extern KbdintDevice mm_bsdauth_device; #endif #ifdef USE_PAM extern KbdintDevice mm_sshpam_device; #endif #ifdef SKEY extern KbdintDevice mm_skey_device; #endif #ifdef BSD_AUTH devices[n++] = &mm_bsdauth_device; #else #ifdef USE_PAM devices[n++] = &mm_sshpam_device; #endif #ifdef SKEY devices[n++] = &mm_skey_device; #endif #endif } Index: stable/8/crypto/openssh/sshconnect.c =================================================================== --- stable/8/crypto/openssh/sshconnect.c (revision 285976) +++ stable/8/crypto/openssh/sshconnect.c (revision 285977) @@ -1,1322 +1,1332 @@ /* $OpenBSD: sshconnect.c,v 1.234 2011/05/24 07:15:47 djm Exp $ */ /* $FreeBSD$ */ /* * Author: Tatu Ylonen * Copyright (c) 1995 Tatu Ylonen , Espoo, Finland * All rights reserved * Code to connect to a remote host, and to perform the client side of the * login (authentication) dialog. * * As far as I am concerned, the code I have written for this software * can be used freely for any purpose. Any derived versions of this * software must be clearly marked as such, and if the derived work is * incompatible with the protocol description in the RFC file, it must be * called by a name other than "ssh" or "Secure Shell". */ #include "includes.h" #include #include #include #include #ifdef HAVE_SYS_TIME_H # include #endif #include #include #include #include #include #include #ifdef HAVE_PATHS_H #include #endif #include #include #include #include #include #include #include #include "xmalloc.h" #include "key.h" #include "hostfile.h" #include "ssh.h" #include "rsa.h" #include "buffer.h" #include "packet.h" #include "uidswap.h" #include "compat.h" #include "key.h" #include "sshconnect.h" #include "hostfile.h" #include "log.h" #include "readconf.h" #include "atomicio.h" #include "misc.h" #include "dns.h" #include "roaming.h" #include "ssh2.h" #include "version.h" char *client_version_string = NULL; char *server_version_string = NULL; static int matching_host_key_dns = 0; static pid_t proxy_command_pid = 0; /* import */ extern Options options; extern char *__progname; extern uid_t original_real_uid; extern uid_t original_effective_uid; static int show_other_keys(struct hostkeys *, Key *); static void warn_changed_key(Key *); /* * Connect to the given ssh server using a proxy command. */ static int ssh_proxy_connect(const char *host, u_short port, const char *proxy_command) { char *command_string, *tmp; int pin[2], pout[2]; pid_t pid; char *shell, strport[NI_MAXSERV]; if ((shell = getenv("SHELL")) == NULL || *shell == '\0') shell = _PATH_BSHELL; /* Convert the port number into a string. */ snprintf(strport, sizeof strport, "%hu", port); /* * Build the final command string in the buffer by making the * appropriate substitutions to the given proxy command. * * Use "exec" to avoid "sh -c" processes on some platforms * (e.g. Solaris) */ xasprintf(&tmp, "exec %s", proxy_command); command_string = percent_expand(tmp, "h", host, "p", strport, "r", options.user, (char *)NULL); xfree(tmp); /* Create pipes for communicating with the proxy. */ if (pipe(pin) < 0 || pipe(pout) < 0) fatal("Could not create pipes to communicate with the proxy: %.100s", strerror(errno)); debug("Executing proxy command: %.500s", command_string); /* Fork and execute the proxy command. */ if ((pid = fork()) == 0) { char *argv[10]; /* Child. Permanently give up superuser privileges. */ permanently_drop_suid(original_real_uid); /* Redirect stdin and stdout. */ close(pin[1]); if (pin[0] != 0) { if (dup2(pin[0], 0) < 0) perror("dup2 stdin"); close(pin[0]); } close(pout[0]); if (dup2(pout[1], 1) < 0) perror("dup2 stdout"); /* Cannot be 1 because pin allocated two descriptors. */ close(pout[1]); /* Stderr is left as it is so that error messages get printed on the user's terminal. */ argv[0] = shell; argv[1] = "-c"; argv[2] = command_string; argv[3] = NULL; /* Execute the proxy command. Note that we gave up any extra privileges above. */ signal(SIGPIPE, SIG_DFL); execv(argv[0], argv); perror(argv[0]); exit(1); } /* Parent. */ if (pid < 0) fatal("fork failed: %.100s", strerror(errno)); else proxy_command_pid = pid; /* save pid to clean up later */ /* Close child side of the descriptors. */ close(pin[0]); close(pout[1]); /* Free the command name. */ xfree(command_string); /* Set the connection file descriptors. */ packet_set_connection(pout[0], pin[1]); packet_set_timeout(options.server_alive_interval, options.server_alive_count_max); /* Indicate OK return */ return 0; } void ssh_kill_proxy_command(void) { /* * Send SIGHUP to proxy command if used. We don't wait() in * case it hangs and instead rely on init to reap the child */ if (proxy_command_pid > 1) kill(proxy_command_pid, SIGHUP); } /* * Set TCP receive buffer if requested. * Note: tuning needs to happen after the socket is created but before the * connection happens so winscale is negotiated properly. */ static void ssh_set_socket_recvbuf(int sock) { void *buf = (void *)&options.tcp_rcv_buf; int socksize, sz = sizeof(options.tcp_rcv_buf); socklen_t len = sizeof(int); debug("setsockopt attempting to set SO_RCVBUF to %d", options.tcp_rcv_buf); if (setsockopt(sock, SOL_SOCKET, SO_RCVBUF, buf, sz) >= 0) { getsockopt(sock, SOL_SOCKET, SO_RCVBUF, &socksize, &len); debug("setsockopt SO_RCVBUF: %.100s %d", strerror(errno), socksize); } else error("Couldn't set socket receive buffer to %d: %.100s", options.tcp_rcv_buf, strerror(errno)); } /* * Creates a (possibly privileged) socket for use as the ssh connection. */ static int ssh_create_socket(int privileged, struct addrinfo *ai) { int sock, gaierr; struct addrinfo hints, *res; /* * If we are running as root and want to connect to a privileged * port, bind our own socket to a privileged port. */ if (privileged) { int p = IPPORT_RESERVED - 1; PRIV_START; sock = rresvport_af(&p, ai->ai_family); PRIV_END; if (sock < 0) error("rresvport: af=%d %.100s", ai->ai_family, strerror(errno)); else debug("Allocated local port %d.", p); if (options.tcp_rcv_buf > 0) ssh_set_socket_recvbuf(sock); return sock; } sock = socket(ai->ai_family, ai->ai_socktype, ai->ai_protocol); if (sock < 0) { error("socket: %.100s", strerror(errno)); return -1; } fcntl(sock, F_SETFD, FD_CLOEXEC); if (options.tcp_rcv_buf > 0) ssh_set_socket_recvbuf(sock); /* Bind the socket to an alternative local IP address */ if (options.bind_address == NULL) return sock; memset(&hints, 0, sizeof(hints)); hints.ai_family = ai->ai_family; hints.ai_socktype = ai->ai_socktype; hints.ai_protocol = ai->ai_protocol; hints.ai_flags = AI_PASSIVE; gaierr = getaddrinfo(options.bind_address, NULL, &hints, &res); if (gaierr) { error("getaddrinfo: %s: %s", options.bind_address, ssh_gai_strerror(gaierr)); close(sock); return -1; } if (bind(sock, res->ai_addr, res->ai_addrlen) < 0) { error("bind: %s: %s", options.bind_address, strerror(errno)); close(sock); freeaddrinfo(res); return -1; } freeaddrinfo(res); return sock; } static int timeout_connect(int sockfd, const struct sockaddr *serv_addr, socklen_t addrlen, int *timeoutp) { fd_set *fdset; struct timeval tv, t_start; socklen_t optlen; int optval, rc, result = -1; gettimeofday(&t_start, NULL); if (*timeoutp <= 0) { result = connect(sockfd, serv_addr, addrlen); goto done; } set_nonblock(sockfd); rc = connect(sockfd, serv_addr, addrlen); if (rc == 0) { unset_nonblock(sockfd); result = 0; goto done; } if (errno != EINPROGRESS) { result = -1; goto done; } fdset = (fd_set *)xcalloc(howmany(sockfd + 1, NFDBITS), sizeof(fd_mask)); FD_SET(sockfd, fdset); ms_to_timeval(&tv, *timeoutp); for (;;) { rc = select(sockfd + 1, NULL, fdset, NULL, &tv); if (rc != -1 || errno != EINTR) break; } switch (rc) { case 0: /* Timed out */ errno = ETIMEDOUT; break; case -1: /* Select error */ debug("select: %s", strerror(errno)); break; case 1: /* Completed or failed */ optval = 0; optlen = sizeof(optval); if (getsockopt(sockfd, SOL_SOCKET, SO_ERROR, &optval, &optlen) == -1) { debug("getsockopt: %s", strerror(errno)); break; } if (optval != 0) { errno = optval; break; } result = 0; unset_nonblock(sockfd); break; default: /* Should not occur */ fatal("Bogus return (%d) from select()", rc); } xfree(fdset); done: if (result == 0 && *timeoutp > 0) { ms_subtract_diff(&t_start, timeoutp); if (*timeoutp <= 0) { errno = ETIMEDOUT; result = -1; } } return (result); } /* * Opens a TCP/IP connection to the remote server on the given host. * The address of the remote host will be returned in hostaddr. * If port is 0, the default port will be used. If needpriv is true, * a privileged port will be allocated to make the connection. * This requires super-user privileges if needpriv is true. * Connection_attempts specifies the maximum number of tries (one per * second). If proxy_command is non-NULL, it specifies the command (with %h * and %p substituted for host and port, respectively) to use to contact * the daemon. */ int ssh_connect(const char *host, struct sockaddr_storage * hostaddr, u_short port, int family, int connection_attempts, int *timeout_ms, int want_keepalive, int needpriv, const char *proxy_command) { int gaierr; int on = 1; int sock = -1, attempt; char ntop[NI_MAXHOST], strport[NI_MAXSERV]; struct addrinfo hints, *ai, *aitop; debug2("ssh_connect: needpriv %d", needpriv); /* If a proxy command is given, connect using it. */ if (proxy_command != NULL) return ssh_proxy_connect(host, port, proxy_command); /* No proxy command. */ memset(&hints, 0, sizeof(hints)); hints.ai_family = family; hints.ai_socktype = SOCK_STREAM; snprintf(strport, sizeof strport, "%u", port); if ((gaierr = getaddrinfo(host, strport, &hints, &aitop)) != 0) fatal("%s: Could not resolve hostname %.100s: %s", __progname, host, ssh_gai_strerror(gaierr)); for (attempt = 0; attempt < connection_attempts; attempt++) { if (attempt > 0) { /* Sleep a moment before retrying. */ sleep(1); debug("Trying again..."); } /* * Loop through addresses for this host, and try each one in * sequence until the connection succeeds. */ for (ai = aitop; ai; ai = ai->ai_next) { if (ai->ai_family != AF_INET && ai->ai_family != AF_INET6) continue; if (getnameinfo(ai->ai_addr, ai->ai_addrlen, ntop, sizeof(ntop), strport, sizeof(strport), NI_NUMERICHOST|NI_NUMERICSERV) != 0) { error("ssh_connect: getnameinfo failed"); continue; } debug("Connecting to %.200s [%.100s] port %s.", host, ntop, strport); /* Create a socket for connecting. */ sock = ssh_create_socket(needpriv, ai); if (sock < 0) /* Any error is already output */ continue; if (timeout_connect(sock, ai->ai_addr, ai->ai_addrlen, timeout_ms) >= 0) { /* Successful connection. */ memcpy(hostaddr, ai->ai_addr, ai->ai_addrlen); break; } else { debug("connect to address %s port %s: %s", ntop, strport, strerror(errno)); close(sock); sock = -1; } } if (sock != -1) break; /* Successful connection. */ } freeaddrinfo(aitop); /* Return failure if we didn't get a successful connection. */ if (sock == -1) { error("ssh: connect to host %s port %s: %s", host, strport, strerror(errno)); return (-1); } debug("Connection established."); /* Set SO_KEEPALIVE if requested. */ if (want_keepalive && setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE, (void *)&on, sizeof(on)) < 0) error("setsockopt SO_KEEPALIVE: %.100s", strerror(errno)); /* Set the connection. */ packet_set_connection(sock, sock); packet_set_timeout(options.server_alive_interval, options.server_alive_count_max); return 0; } /* * Waits for the server identification string, and sends our own * identification string. */ void ssh_exchange_identification(int timeout_ms) { char buf[256], remote_version[256]; /* must be same size! */ int remote_major, remote_minor, mismatch; int connection_in = packet_get_connection_in(); int connection_out = packet_get_connection_out(); int minor1 = PROTOCOL_MINOR_1; u_int i, n; size_t len; int fdsetsz, remaining, rc; struct timeval t_start, t_remaining; fd_set *fdset; fdsetsz = howmany(connection_in + 1, NFDBITS) * sizeof(fd_mask); fdset = xcalloc(1, fdsetsz); /* Read other side's version identification. */ remaining = timeout_ms; for (n = 0;;) { for (i = 0; i < sizeof(buf) - 1; i++) { if (timeout_ms > 0) { gettimeofday(&t_start, NULL); ms_to_timeval(&t_remaining, remaining); FD_SET(connection_in, fdset); rc = select(connection_in + 1, fdset, NULL, fdset, &t_remaining); ms_subtract_diff(&t_start, &remaining); if (rc == 0 || remaining <= 0) fatal("Connection timed out during " "banner exchange"); if (rc == -1) { if (errno == EINTR) continue; fatal("ssh_exchange_identification: " "select: %s", strerror(errno)); } } len = roaming_atomicio(read, connection_in, &buf[i], 1); if (len != 1 && errno == EPIPE) fatal("ssh_exchange_identification: " "Connection closed by remote host"); else if (len != 1) fatal("ssh_exchange_identification: " "read: %.100s", strerror(errno)); if (buf[i] == '\r') { buf[i] = '\n'; buf[i + 1] = 0; continue; /**XXX wait for \n */ } if (buf[i] == '\n') { buf[i + 1] = 0; break; } if (++n > 65536) fatal("ssh_exchange_identification: " "No banner received"); } buf[sizeof(buf) - 1] = 0; if (strncmp(buf, "SSH-", 4) == 0) break; debug("ssh_exchange_identification: %s", buf); } server_version_string = xstrdup(buf); xfree(fdset); /* * Check that the versions match. In future this might accept * several versions and set appropriate flags to handle them. */ if (sscanf(server_version_string, "SSH-%d.%d-%[^\n]\n", &remote_major, &remote_minor, remote_version) != 3) fatal("Bad remote protocol version identification: '%.100s'", buf); debug("Remote protocol version %d.%d, remote software version %.100s", remote_major, remote_minor, remote_version); compat_datafellows(remote_version); mismatch = 0; switch (remote_major) { case 1: if (remote_minor == 99 && (options.protocol & SSH_PROTO_2) && !(options.protocol & SSH_PROTO_1_PREFERRED)) { enable_compat20(); break; } if (!(options.protocol & SSH_PROTO_1)) { mismatch = 1; break; } if (remote_minor < 3) { fatal("Remote machine has too old SSH software version."); } else if (remote_minor == 3 || remote_minor == 4) { /* We speak 1.3, too. */ enable_compat13(); minor1 = 3; if (options.forward_agent) { logit("Agent forwarding disabled for protocol 1.3"); options.forward_agent = 0; } } break; case 2: if (options.protocol & SSH_PROTO_2) { enable_compat20(); break; } /* FALLTHROUGH */ default: mismatch = 1; break; } if (mismatch) fatal("Protocol major versions differ: %d vs. %d", (options.protocol & SSH_PROTO_2) ? PROTOCOL_MAJOR_2 : PROTOCOL_MAJOR_1, remote_major); /* Send our own protocol version identification. */ snprintf(buf, sizeof buf, "SSH-%d.%d-%.100s%s%s%s%s", compat20 ? PROTOCOL_MAJOR_2 : PROTOCOL_MAJOR_1, compat20 ? PROTOCOL_MINOR_2 : minor1, SSH_VERSION, options.hpn_disabled ? "" : SSH_VERSION_HPN, *options.version_addendum == '\0' ? "" : " ", options.version_addendum, compat20 ? "\r\n" : "\n"); if (roaming_atomicio(vwrite, connection_out, buf, strlen(buf)) != strlen(buf)) fatal("write: %.100s", strerror(errno)); client_version_string = xstrdup(buf); chop(client_version_string); chop(server_version_string); debug("Local version string %.100s", client_version_string); } /* defaults to 'no' */ static int confirm(const char *prompt) { const char *msg, *again = "Please type 'yes' or 'no': "; char *p; int ret = -1; if (options.batch_mode) return 0; for (msg = prompt;;msg = again) { p = read_passphrase(msg, RP_ECHO); if (p == NULL || (p[0] == '\0') || (p[0] == '\n') || strncasecmp(p, "no", 2) == 0) ret = 0; if (p && strncasecmp(p, "yes", 3) == 0) ret = 1; if (p) xfree(p); if (ret != -1) return ret; } } static int check_host_cert(const char *host, const Key *host_key) { const char *reason; if (key_cert_check_authority(host_key, 1, 0, host, &reason) != 0) { error("%s", reason); return 0; } if (buffer_len(&host_key->cert->critical) != 0) { error("Certificate for %s contains unsupported " "critical options(s)", host); return 0; } return 1; } static int sockaddr_is_local(struct sockaddr *hostaddr) { switch (hostaddr->sa_family) { case AF_INET: return (ntohl(((struct sockaddr_in *)hostaddr)-> sin_addr.s_addr) >> 24) == IN_LOOPBACKNET; case AF_INET6: return IN6_IS_ADDR_LOOPBACK( &(((struct sockaddr_in6 *)hostaddr)->sin6_addr)); default: return 0; } } /* * Prepare the hostname and ip address strings that are used to lookup * host keys in known_hosts files. These may have a port number appended. */ void get_hostfile_hostname_ipaddr(char *hostname, struct sockaddr *hostaddr, u_short port, char **hostfile_hostname, char **hostfile_ipaddr) { char ntop[NI_MAXHOST]; socklen_t addrlen; switch (hostaddr == NULL ? -1 : hostaddr->sa_family) { case -1: addrlen = 0; break; case AF_INET: addrlen = sizeof(struct sockaddr_in); break; case AF_INET6: addrlen = sizeof(struct sockaddr_in6); break; default: addrlen = sizeof(struct sockaddr); break; } /* * We don't have the remote ip-address for connections * using a proxy command */ if (hostfile_ipaddr != NULL) { if (options.proxy_command == NULL) { if (getnameinfo(hostaddr, addrlen, ntop, sizeof(ntop), NULL, 0, NI_NUMERICHOST) != 0) fatal("check_host_key: getnameinfo failed"); *hostfile_ipaddr = put_host_port(ntop, port); } else { *hostfile_ipaddr = xstrdup(""); } } /* * Allow the user to record the key under a different name or * differentiate a non-standard port. This is useful for ssh * tunneling over forwarded connections or if you run multiple * sshd's on different ports on the same machine. */ if (hostfile_hostname != NULL) { if (options.host_key_alias != NULL) { *hostfile_hostname = xstrdup(options.host_key_alias); debug("using hostkeyalias: %s", *hostfile_hostname); } else { *hostfile_hostname = put_host_port(hostname, port); } } } /* * check whether the supplied host key is valid, return -1 if the key * is not valid. user_hostfile[0] will not be updated if 'readonly' is true. */ #define RDRW 0 #define RDONLY 1 #define ROQUIET 2 static int check_host_key(char *hostname, struct sockaddr *hostaddr, u_short port, Key *host_key, int readonly, char **user_hostfiles, u_int num_user_hostfiles, char **system_hostfiles, u_int num_system_hostfiles) { HostStatus host_status; HostStatus ip_status; Key *raw_key = NULL; char *ip = NULL, *host = NULL; char hostline[1000], *hostp, *fp, *ra; char msg[1024]; const char *type; const struct hostkey_entry *host_found, *ip_found; int len, cancelled_forwarding = 0; int local = sockaddr_is_local(hostaddr); int r, want_cert = key_is_cert(host_key), host_ip_differ = 0; struct hostkeys *host_hostkeys, *ip_hostkeys; u_int i; /* * Force accepting of the host key for loopback/localhost. The * problem is that if the home directory is NFS-mounted to multiple * machines, localhost will refer to a different machine in each of * them, and the user will get bogus HOST_CHANGED warnings. This * essentially disables host authentication for localhost; however, * this is probably not a real problem. */ if (options.no_host_authentication_for_localhost == 1 && local && options.host_key_alias == NULL) { debug("Forcing accepting of host key for " "loopback/localhost."); return 0; } /* * Prepare the hostname and address strings used for hostkey lookup. * In some cases, these will have a port number appended. */ get_hostfile_hostname_ipaddr(hostname, hostaddr, port, &host, &ip); /* * Turn off check_host_ip if the connection is to localhost, via proxy * command or if we don't have a hostname to compare with */ if (options.check_host_ip && (local || strcmp(hostname, ip) == 0 || options.proxy_command != NULL)) options.check_host_ip = 0; host_hostkeys = init_hostkeys(); for (i = 0; i < num_user_hostfiles; i++) load_hostkeys(host_hostkeys, host, user_hostfiles[i]); for (i = 0; i < num_system_hostfiles; i++) load_hostkeys(host_hostkeys, host, system_hostfiles[i]); ip_hostkeys = NULL; if (!want_cert && options.check_host_ip) { ip_hostkeys = init_hostkeys(); for (i = 0; i < num_user_hostfiles; i++) load_hostkeys(ip_hostkeys, ip, user_hostfiles[i]); for (i = 0; i < num_system_hostfiles; i++) load_hostkeys(ip_hostkeys, ip, system_hostfiles[i]); } retry: /* Reload these as they may have changed on cert->key downgrade */ want_cert = key_is_cert(host_key); type = key_type(host_key); /* * Check if the host key is present in the user's list of known * hosts or in the systemwide list. */ host_status = check_key_in_hostkeys(host_hostkeys, host_key, &host_found); /* * Also perform check for the ip address, skip the check if we are * localhost, looking for a certificate, or the hostname was an ip * address to begin with. */ if (!want_cert && ip_hostkeys != NULL) { ip_status = check_key_in_hostkeys(ip_hostkeys, host_key, &ip_found); if (host_status == HOST_CHANGED && (ip_status != HOST_CHANGED || (ip_found != NULL && !key_equal(ip_found->key, host_found->key)))) host_ip_differ = 1; } else ip_status = host_status; switch (host_status) { case HOST_OK: /* The host is known and the key matches. */ debug("Host '%.200s' is known and matches the %s host %s.", host, type, want_cert ? "certificate" : "key"); debug("Found %s in %s:%lu", want_cert ? "CA key" : "key", host_found->file, host_found->line); if (want_cert && !check_host_cert(hostname, host_key)) goto fail; if (options.check_host_ip && ip_status == HOST_NEW) { if (readonly || want_cert) logit("%s host key for IP address " "'%.128s' not in list of known hosts.", type, ip); else if (!add_host_to_hostfile(user_hostfiles[0], ip, host_key, options.hash_known_hosts)) logit("Failed to add the %s host key for IP " "address '%.128s' to the list of known " "hosts (%.30s).", type, ip, user_hostfiles[0]); else logit("Warning: Permanently added the %s host " "key for IP address '%.128s' to the list " "of known hosts.", type, ip); } else if (options.visual_host_key) { fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); ra = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_RANDOMART); logit("Host key fingerprint is %s\n%s\n", fp, ra); xfree(ra); xfree(fp); } break; case HOST_NEW: if (options.host_key_alias == NULL && port != 0 && port != SSH_DEFAULT_PORT) { debug("checking without port identifier"); if (check_host_key(hostname, hostaddr, 0, host_key, ROQUIET, user_hostfiles, num_user_hostfiles, system_hostfiles, num_system_hostfiles) == 0) { debug("found matching key w/out port"); break; } } if (readonly || want_cert) goto fail; /* The host is new. */ if (options.strict_host_key_checking == 1) { /* * User has requested strict host key checking. We * will not add the host key automatically. The only * alternative left is to abort. */ error("No %s host key is known for %.200s and you " "have requested strict checking.", type, host); goto fail; } else if (options.strict_host_key_checking == 2) { char msg1[1024], msg2[1024]; if (show_other_keys(host_hostkeys, host_key)) snprintf(msg1, sizeof(msg1), "\nbut keys of different type are already" " known for this host."); else snprintf(msg1, sizeof(msg1), "."); /* The default */ fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); ra = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_RANDOMART); msg2[0] = '\0'; if (options.verify_host_key_dns) { if (matching_host_key_dns) snprintf(msg2, sizeof(msg2), "Matching host key fingerprint" " found in DNS.\n"); else snprintf(msg2, sizeof(msg2), "No matching host key fingerprint" " found in DNS.\n"); } snprintf(msg, sizeof(msg), "The authenticity of host '%.200s (%s)' can't be " "established%s\n" "%s key fingerprint is %s.%s%s\n%s" "Are you sure you want to continue connecting " "(yes/no)? ", host, ip, msg1, type, fp, options.visual_host_key ? "\n" : "", options.visual_host_key ? ra : "", msg2); xfree(ra); xfree(fp); if (!confirm(msg)) goto fail; } /* * If not in strict mode, add the key automatically to the * local known_hosts file. */ if (options.check_host_ip && ip_status == HOST_NEW) { snprintf(hostline, sizeof(hostline), "%s,%s", host, ip); hostp = hostline; if (options.hash_known_hosts) { /* Add hash of host and IP separately */ r = add_host_to_hostfile(user_hostfiles[0], host, host_key, options.hash_known_hosts) && add_host_to_hostfile(user_hostfiles[0], ip, host_key, options.hash_known_hosts); } else { /* Add unhashed "host,ip" */ r = add_host_to_hostfile(user_hostfiles[0], hostline, host_key, options.hash_known_hosts); } } else { r = add_host_to_hostfile(user_hostfiles[0], host, host_key, options.hash_known_hosts); hostp = host; } if (!r) logit("Failed to add the host to the list of known " "hosts (%.500s).", user_hostfiles[0]); else logit("Warning: Permanently added '%.200s' (%s) to the " "list of known hosts.", hostp, type); break; case HOST_REVOKED: error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("@ WARNING: REVOKED HOST KEY DETECTED! @"); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("The %s host key for %s is marked as revoked.", type, host); error("This could mean that a stolen key is being used to"); error("impersonate this host."); /* * If strict host key checking is in use, the user will have * to edit the key manually and we can only abort. */ if (options.strict_host_key_checking) { error("%s host key for %.200s was revoked and you have " "requested strict checking.", type, host); goto fail; } goto continue_unsafe; case HOST_CHANGED: if (want_cert) { /* * This is only a debug() since it is valid to have * CAs with wildcard DNS matches that don't match * all hosts that one might visit. */ debug("Host certificate authority does not " "match %s in %s:%lu", CA_MARKER, host_found->file, host_found->line); goto fail; } if (readonly == ROQUIET) goto fail; if (options.check_host_ip && host_ip_differ) { char *key_msg; if (ip_status == HOST_NEW) key_msg = "is unknown"; else if (ip_status == HOST_OK) key_msg = "is unchanged"; else key_msg = "has a different value"; error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("@ WARNING: POSSIBLE DNS SPOOFING DETECTED! @"); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("The %s host key for %s has changed,", type, host); error("and the key for the corresponding IP address %s", ip); error("%s. This could either mean that", key_msg); error("DNS SPOOFING is happening or the IP address for the host"); error("and its host key have changed at the same time."); if (ip_status != HOST_NEW) error("Offending key for IP in %s:%lu", ip_found->file, ip_found->line); } /* The host key has changed. */ warn_changed_key(host_key); error("Add correct host key in %.100s to get rid of this message.", user_hostfiles[0]); error("Offending %s key in %s:%lu", key_type(host_found->key), host_found->file, host_found->line); /* * If strict host key checking is in use, the user will have * to edit the key manually and we can only abort. */ if (options.strict_host_key_checking) { error("%s host key for %.200s has changed and you have " "requested strict checking.", type, host); goto fail; } continue_unsafe: /* * If strict host key checking has not been requested, allow * the connection but without MITM-able authentication or * forwarding. */ if (options.password_authentication) { error("Password authentication is disabled to avoid " "man-in-the-middle attacks."); options.password_authentication = 0; cancelled_forwarding = 1; } if (options.kbd_interactive_authentication) { error("Keyboard-interactive authentication is disabled" " to avoid man-in-the-middle attacks."); options.kbd_interactive_authentication = 0; options.challenge_response_authentication = 0; cancelled_forwarding = 1; } if (options.challenge_response_authentication) { error("Challenge/response authentication is disabled" " to avoid man-in-the-middle attacks."); options.challenge_response_authentication = 0; cancelled_forwarding = 1; } if (options.forward_agent) { error("Agent forwarding is disabled to avoid " "man-in-the-middle attacks."); options.forward_agent = 0; cancelled_forwarding = 1; } if (options.forward_x11) { error("X11 forwarding is disabled to avoid " "man-in-the-middle attacks."); options.forward_x11 = 0; cancelled_forwarding = 1; } if (options.num_local_forwards > 0 || options.num_remote_forwards > 0) { error("Port forwarding is disabled to avoid " "man-in-the-middle attacks."); options.num_local_forwards = options.num_remote_forwards = 0; cancelled_forwarding = 1; } if (options.tun_open != SSH_TUNMODE_NO) { error("Tunnel forwarding is disabled to avoid " "man-in-the-middle attacks."); options.tun_open = SSH_TUNMODE_NO; cancelled_forwarding = 1; } if (options.exit_on_forward_failure && cancelled_forwarding) fatal("Error: forwarding disabled due to host key " "check failure"); /* * XXX Should permit the user to change to use the new id. * This could be done by converting the host key to an * identifying sentence, tell that the host identifies itself * by that sentence, and ask the user if he/she wishes to * accept the authentication. */ break; case HOST_FOUND: fatal("internal error"); break; } if (options.check_host_ip && host_status != HOST_CHANGED && ip_status == HOST_CHANGED) { snprintf(msg, sizeof(msg), "Warning: the %s host key for '%.200s' " "differs from the key for the IP address '%.128s'" "\nOffending key for IP in %s:%lu", type, host, ip, ip_found->file, ip_found->line); if (host_status == HOST_OK) { len = strlen(msg); snprintf(msg + len, sizeof(msg) - len, "\nMatching host key in %s:%lu", host_found->file, host_found->line); } if (options.strict_host_key_checking == 1) { logit("%s", msg); error("Exiting, you have requested strict checking."); goto fail; } else if (options.strict_host_key_checking == 2) { strlcat(msg, "\nAre you sure you want " "to continue connecting (yes/no)? ", sizeof(msg)); if (!confirm(msg)) goto fail; } else { logit("%s", msg); } } xfree(ip); xfree(host); if (host_hostkeys != NULL) free_hostkeys(host_hostkeys); if (ip_hostkeys != NULL) free_hostkeys(ip_hostkeys); return 0; fail: if (want_cert && host_status != HOST_REVOKED) { /* * No matching certificate. Downgrade cert to raw key and * search normally. */ debug("No matching CA found. Retry with plain key"); raw_key = key_from_private(host_key); if (key_drop_cert(raw_key) != 0) fatal("Couldn't drop certificate"); host_key = raw_key; goto retry; } if (raw_key != NULL) key_free(raw_key); xfree(ip); xfree(host); if (host_hostkeys != NULL) free_hostkeys(host_hostkeys); if (ip_hostkeys != NULL) free_hostkeys(ip_hostkeys); return -1; } /* returns 0 if key verifies or -1 if key does NOT verify */ int verify_host_key(char *host, struct sockaddr *hostaddr, Key *host_key) { int flags = 0; char *fp; + Key *plain = NULL; fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); debug("Server host key: %s %s", key_type(host_key), fp); xfree(fp); - /* XXX certs are not yet supported for DNS */ - if (!key_is_cert(host_key) && options.verify_host_key_dns && - verify_host_key_dns(host, hostaddr, host_key, &flags) == 0) { - if (flags & DNS_VERIFY_FOUND) { - - if (options.verify_host_key_dns == 1 && - flags & DNS_VERIFY_MATCH && - flags & DNS_VERIFY_SECURE) - return 0; - - if (flags & DNS_VERIFY_MATCH) { - matching_host_key_dns = 1; - } else { - warn_changed_key(host_key); - error("Update the SSHFP RR in DNS with the new " - "host key to get rid of this message."); + if (options.verify_host_key_dns) { + /* + * XXX certs are not yet supported for DNS, so downgrade + * them and try the plain key. + */ + plain = key_from_private(host_key); + if (key_is_cert(plain)) + key_drop_cert(plain); + if (verify_host_key_dns(host, hostaddr, plain, &flags) == 0) { + if (flags & DNS_VERIFY_FOUND) { + if (options.verify_host_key_dns == 1 && + flags & DNS_VERIFY_MATCH && + flags & DNS_VERIFY_SECURE) { + key_free(plain); + return 0; + } + if (flags & DNS_VERIFY_MATCH) { + matching_host_key_dns = 1; + } else { + warn_changed_key(plain); + error("Update the SSHFP RR in DNS " + "with the new host key to get rid " + "of this message."); + } } } + key_free(plain); } return check_host_key(host, hostaddr, options.port, host_key, RDRW, options.user_hostfiles, options.num_user_hostfiles, options.system_hostfiles, options.num_system_hostfiles); } /* * Starts a dialog with the server, and authenticates the current user on the * server. This does not need any extra privileges. The basic connection * to the server must already have been established before this is called. * If login fails, this function prints an error and never returns. * This function does not require super-user privileges. */ void ssh_login(Sensitive *sensitive, const char *orighost, struct sockaddr *hostaddr, u_short port, struct passwd *pw, int timeout_ms) { char *host, *cp; char *server_user, *local_user; local_user = xstrdup(pw->pw_name); server_user = options.user ? options.user : local_user; /* Convert the user-supplied hostname into all lowercase. */ host = xstrdup(orighost); for (cp = host; *cp; cp++) if (isupper(*cp)) *cp = (char)tolower(*cp); /* Exchange protocol version identification strings with the server. */ ssh_exchange_identification(timeout_ms); /* Put the connection into non-blocking mode. */ packet_set_nonblocking(); /* key exchange */ /* authenticate user */ if (compat20) { ssh_kex2(host, hostaddr, port); ssh_userauth2(local_user, server_user, host, sensitive); } else { ssh_kex(host, hostaddr); ssh_userauth1(local_user, server_user, host, sensitive); } xfree(local_user); } void ssh_put_password(char *password) { int size; char *padded; if (datafellows & SSH_BUG_PASSWORDPAD) { packet_put_cstring(password); return; } size = roundup(strlen(password) + 1, 32); padded = xcalloc(1, size); strlcpy(padded, password, size); packet_put_string(padded, size); memset(padded, 0, size); xfree(padded); } /* print all known host keys for a given host, but skip keys of given type */ static int show_other_keys(struct hostkeys *hostkeys, Key *key) { int type[] = { KEY_RSA1, KEY_RSA, KEY_DSA, KEY_ECDSA, -1}; int i, ret = 0; char *fp, *ra; const struct hostkey_entry *found; for (i = 0; type[i] != -1; i++) { if (type[i] == key->type) continue; if (!lookup_key_in_hostkeys_by_type(hostkeys, type[i], &found)) continue; fp = key_fingerprint(found->key, SSH_FP_MD5, SSH_FP_HEX); ra = key_fingerprint(found->key, SSH_FP_MD5, SSH_FP_RANDOMART); logit("WARNING: %s key found for host %s\n" "in %s:%lu\n" "%s key fingerprint %s.", key_type(found->key), found->host, found->file, found->line, key_type(found->key), fp); if (options.visual_host_key) logit("%s", ra); xfree(ra); xfree(fp); ret = 1; } return ret; } static void warn_changed_key(Key *host_key) { char *fp; fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("@ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @"); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY!"); error("Someone could be eavesdropping on you right now (man-in-the-middle attack)!"); error("It is also possible that a host key has just been changed."); error("The fingerprint for the %s key sent by the remote host is\n%s.", key_type(host_key), fp); error("Please contact your system administrator."); xfree(fp); } /* * Execute a local command */ int ssh_local_cmd(const char *args) { char *shell; pid_t pid; int status; void (*osighand)(int); if (!options.permit_local_command || args == NULL || !*args) return (1); if ((shell = getenv("SHELL")) == NULL || *shell == '\0') shell = _PATH_BSHELL; osighand = signal(SIGCHLD, SIG_DFL); pid = fork(); if (pid == 0) { signal(SIGPIPE, SIG_DFL); debug3("Executing %s -c \"%s\"", shell, args); execl(shell, shell, "-c", args, (char *)NULL); error("Couldn't execute %s -c \"%s\": %s", shell, args, strerror(errno)); _exit(1); } else if (pid == -1) fatal("fork failed: %.100s", strerror(errno)); while (waitpid(pid, &status, 0) == -1) if (errno != EINTR) fatal("Couldn't wait for child: %s", strerror(errno)); signal(SIGCHLD, osighand); if (!WIFEXITED(status)) return (1); return (WEXITSTATUS(status)); } Index: stable/8/sys/netinet/tcp_reass.c =================================================================== --- stable/8/sys/netinet/tcp_reass.c (revision 285976) +++ stable/8/sys/netinet/tcp_reass.c (revision 285977) @@ -1,369 +1,356 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_input.c 8.12 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef TCPDEBUG #include #endif /* TCPDEBUG */ static int tcp_reass_sysctl_maxseg(SYSCTL_HANDLER_ARGS); static int tcp_reass_sysctl_qsize(SYSCTL_HANDLER_ARGS); SYSCTL_NODE(_net_inet_tcp, OID_AUTO, reass, CTLFLAG_RW, 0, "TCP Segment Reassembly Queue"); -static VNET_DEFINE(int, tcp_reass_maxseg) = 0; -#define V_tcp_reass_maxseg VNET(tcp_reass_maxseg) +static int tcp_reass_maxseg = 0; SYSCTL_VNET_PROC(_net_inet_tcp_reass, OID_AUTO, maxsegments, CTLTYPE_INT | CTLFLAG_RDTUN, - &VNET_NAME(tcp_reass_maxseg), 0, &tcp_reass_sysctl_maxseg, "I", + &tcp_reass_maxseg, 0, &tcp_reass_sysctl_maxseg, "I", "Global maximum number of TCP Segments in Reassembly Queue"); -static VNET_DEFINE(int, tcp_reass_qsize) = 0; -#define V_tcp_reass_qsize VNET(tcp_reass_qsize) -SYSCTL_VNET_PROC(_net_inet_tcp_reass, OID_AUTO, cursegments, +static int tcp_reass_qsize = 0; +SYSCTL_PROC(_net_inet_tcp_reass, OID_AUTO, cursegments, CTLTYPE_INT | CTLFLAG_RD, - &VNET_NAME(tcp_reass_qsize), 0, &tcp_reass_sysctl_qsize, "I", + &tcp_reass_qsize, 0, &tcp_reass_sysctl_qsize, "I", "Global number of TCP Segments currently in Reassembly Queue"); -static VNET_DEFINE(int, tcp_reass_overflows) = 0; -#define V_tcp_reass_overflows VNET(tcp_reass_overflows) -SYSCTL_VNET_INT(_net_inet_tcp_reass, OID_AUTO, overflows, +static int tcp_reass_overflows = 0; +SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, overflows, CTLTYPE_INT | CTLFLAG_RD, - &VNET_NAME(tcp_reass_overflows), 0, + &tcp_reass_overflows, 0, "Global number of TCP Segment Reassembly Queue Overflows"); -static VNET_DEFINE(uma_zone_t, tcp_reass_zone); -#define V_tcp_reass_zone VNET(tcp_reass_zone) +static uma_zone_t tcp_reass_zone; /* Initialize TCP reassembly queue */ static void tcp_reass_zone_change(void *tag) { - V_tcp_reass_maxseg = nmbclusters / 16; - uma_zone_set_max(V_tcp_reass_zone, V_tcp_reass_maxseg); + tcp_reass_maxseg = nmbclusters / 16; + uma_zone_set_max(tcp_reass_zone, tcp_reass_maxseg); } void -tcp_reass_init(void) +tcp_reass_global_init(void) { - V_tcp_reass_maxseg = nmbclusters / 16; + tcp_reass_maxseg = nmbclusters / 16; TUNABLE_INT_FETCH("net.inet.tcp.reass.maxsegments", - &V_tcp_reass_maxseg); - V_tcp_reass_zone = uma_zcreate("tcpreass", sizeof (struct tseg_qent), + &tcp_reass_maxseg); + tcp_reass_zone = uma_zcreate("tcpreass", sizeof (struct tseg_qent), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); - uma_zone_set_max(V_tcp_reass_zone, V_tcp_reass_maxseg); + uma_zone_set_max(tcp_reass_zone, tcp_reass_maxseg); EVENTHANDLER_REGISTER(nmbclusters_change, tcp_reass_zone_change, NULL, EVENTHANDLER_PRI_ANY); } -#ifdef VIMAGE void -tcp_reass_destroy(void) -{ - - uma_zdestroy(V_tcp_reass_zone); -} -#endif - -void tcp_reass_flush(struct tcpcb *tp) { struct tseg_qent *qe; INP_WLOCK_ASSERT(tp->t_inpcb); while ((qe = LIST_FIRST(&tp->t_segq)) != NULL) { LIST_REMOVE(qe, tqe_q); m_freem(qe->tqe_m); - uma_zfree(V_tcp_reass_zone, qe); + uma_zfree(tcp_reass_zone, qe); tp->t_segqlen--; } KASSERT((tp->t_segqlen == 0), ("TCP reass queue %p segment count is %d instead of 0 after flush.", tp, tp->t_segqlen)); } static int tcp_reass_sysctl_maxseg(SYSCTL_HANDLER_ARGS) { - V_tcp_reass_maxseg = uma_zone_get_max(V_tcp_reass_zone); + tcp_reass_maxseg = uma_zone_get_max(tcp_reass_zone); return (sysctl_handle_int(oidp, arg1, arg2, req)); } static int tcp_reass_sysctl_qsize(SYSCTL_HANDLER_ARGS) { - V_tcp_reass_qsize = uma_zone_get_cur(V_tcp_reass_zone); + tcp_reass_qsize = uma_zone_get_cur(tcp_reass_zone); return (sysctl_handle_int(oidp, arg1, arg2, req)); } int tcp_reass(struct tcpcb *tp, struct tcphdr *th, int *tlenp, struct mbuf *m) { struct tseg_qent *q; struct tseg_qent *p = NULL; struct tseg_qent *nq; struct tseg_qent *te = NULL; struct socket *so = tp->t_inpcb->inp_socket; char *s = NULL; int flags; struct tseg_qent tqs; INP_WLOCK_ASSERT(tp->t_inpcb); /* * XXX: tcp_reass() is rather inefficient with its data structures * and should be rewritten (see NetBSD for optimizations). */ /* * Call with th==NULL after become established to * force pre-ESTABLISHED data up to user socket. */ if (th == NULL) goto present; /* * Limit the number of segments that can be queued to reduce the * potential for mbuf exhaustion. For best performance, we want to be * able to queue a full window's worth of segments. The size of the * socket receive buffer determines our advertised window and grows * automatically when socket buffer autotuning is enabled. Use it as the * basis for our queue limit. * Always let the missing segment through which caused this queue. * NB: Access to the socket buffer is left intentionally unlocked as we * can tolerate stale information here. * * XXXLAS: Using sbspace(so->so_rcv) instead of so->so_rcv.sb_hiwat * should work but causes packets to be dropped when they shouldn't. * Investigate why and re-evaluate the below limit after the behaviour * is understood. */ if ((th->th_seq != tp->rcv_nxt || !TCPS_HAVEESTABLISHED(tp->t_state)) && tp->t_segqlen >= (so->so_rcv.sb_hiwat / tp->t_maxseg) + 1) { - V_tcp_reass_overflows++; + tcp_reass_overflows++; TCPSTAT_INC(tcps_rcvmemdrop); m_freem(m); *tlenp = 0; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: queue limit reached, " "segment dropped\n", s, __func__); free(s, M_TCPLOG); } return (0); } /* * Allocate a new queue entry. If we can't, or hit the zone limit * just drop the pkt. * * Use a temporary structure on the stack for the missing segment * when the zone is exhausted. Otherwise we may get stuck. */ - te = uma_zalloc(V_tcp_reass_zone, M_NOWAIT); + te = uma_zalloc(tcp_reass_zone, M_NOWAIT); if (te == NULL) { if (th->th_seq != tp->rcv_nxt || !TCPS_HAVEESTABLISHED(tp->t_state)) { TCPSTAT_INC(tcps_rcvmemdrop); m_freem(m); *tlenp = 0; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: global zone limit " "reached, segment dropped\n", s, __func__); free(s, M_TCPLOG); } return (0); } else { bzero(&tqs, sizeof(struct tseg_qent)); te = &tqs; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: global zone limit reached, using " "stack for missing segment\n", s, __func__); free(s, M_TCPLOG); } } } tp->t_segqlen++; /* * Find a segment which begins after this one does. */ LIST_FOREACH(q, &tp->t_segq, tqe_q) { if (SEQ_GT(q->tqe_th->th_seq, th->th_seq)) break; p = q; } /* * If there is a preceding segment, it may provide some of * our data already. If so, drop the data from the incoming * segment. If it provides all of our data, drop us. */ if (p != NULL) { int i; /* conversion to int (in i) handles seq wraparound */ i = p->tqe_th->th_seq + p->tqe_len - th->th_seq; if (i > 0) { if (i >= *tlenp) { TCPSTAT_INC(tcps_rcvduppack); TCPSTAT_ADD(tcps_rcvdupbyte, *tlenp); m_freem(m); if (te != &tqs) - uma_zfree(V_tcp_reass_zone, te); + uma_zfree(tcp_reass_zone, te); tp->t_segqlen--; /* * Try to present any queued data * at the left window edge to the user. * This is needed after the 3-WHS * completes. */ goto present; /* ??? */ } m_adj(m, i); *tlenp -= i; th->th_seq += i; } } tp->t_rcvoopack++; TCPSTAT_INC(tcps_rcvoopack); TCPSTAT_ADD(tcps_rcvoobyte, *tlenp); /* * While we overlap succeeding segments trim them or, * if they are completely covered, dequeue them. */ while (q) { int i = (th->th_seq + *tlenp) - q->tqe_th->th_seq; if (i <= 0) break; if (i < q->tqe_len) { q->tqe_th->th_seq += i; q->tqe_len -= i; m_adj(q->tqe_m, i); break; } nq = LIST_NEXT(q, tqe_q); LIST_REMOVE(q, tqe_q); m_freem(q->tqe_m); - uma_zfree(V_tcp_reass_zone, q); + uma_zfree(tcp_reass_zone, q); tp->t_segqlen--; q = nq; } /* Insert the new segment queue entry into place. */ te->tqe_m = m; te->tqe_th = th; te->tqe_len = *tlenp; if (p == NULL) { LIST_INSERT_HEAD(&tp->t_segq, te, tqe_q); } else { KASSERT(te != &tqs, ("%s: temporary stack based entry not " "first element in queue", __func__)); LIST_INSERT_AFTER(p, te, tqe_q); } present: /* * Present data to user, advancing rcv_nxt through * completed sequence space. */ if (!TCPS_HAVEESTABLISHED(tp->t_state)) return (0); q = LIST_FIRST(&tp->t_segq); if (!q || q->tqe_th->th_seq != tp->rcv_nxt) return (0); SOCKBUF_LOCK(&so->so_rcv); do { tp->rcv_nxt += q->tqe_len; flags = q->tqe_th->th_flags & TH_FIN; nq = LIST_NEXT(q, tqe_q); LIST_REMOVE(q, tqe_q); if (so->so_rcv.sb_state & SBS_CANTRCVMORE) m_freem(q->tqe_m); else sbappendstream_locked(&so->so_rcv, q->tqe_m); if (q != &tqs) - uma_zfree(V_tcp_reass_zone, q); + uma_zfree(tcp_reass_zone, q); tp->t_segqlen--; q = nq; } while (q && q->tqe_th->th_seq == tp->rcv_nxt); ND6_HINT(tp); sorwakeup_locked(so); return (flags); } Index: stable/8/sys/netinet/tcp_subr.c =================================================================== --- stable/8/sys/netinet/tcp_subr.c (revision 285976) +++ stable/8/sys/netinet/tcp_subr.c (revision 285977) @@ -1,2500 +1,2500 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #ifdef INET6 #include #endif #include #include #ifdef INET6 #include #include #include #endif #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #ifdef TCPDEBUG #include #endif #include #ifdef IPSEC #include #include #ifdef INET6 #include #endif #include #include #endif /*IPSEC*/ #include #include #include VNET_DEFINE(int, tcp_mssdflt) = TCP_MSS; #ifdef INET6 VNET_DEFINE(int, tcp_v6mssdflt) = TCP6_MSS; #endif static int sysctl_net_inet_tcp_mss_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_mssdflt = new; } return (error); } SYSCTL_VNET_PROC(_net_inet_tcp, TCPCTL_MSSDFLT, mssdflt, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(tcp_mssdflt), 0, &sysctl_net_inet_tcp_mss_check, "I", "Default TCP Maximum Segment Size"); #ifdef INET6 static int sysctl_net_inet_tcp_mss_v6_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_v6mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_v6mssdflt = new; } return (error); } SYSCTL_VNET_PROC(_net_inet_tcp, TCPCTL_V6MSSDFLT, v6mssdflt, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(tcp_v6mssdflt), 0, &sysctl_net_inet_tcp_mss_v6_check, "I", "Default TCP Maximum Segment Size for IPv6"); #endif static int vnet_sysctl_msec_to_ticks(SYSCTL_HANDLER_ARGS) { VNET_SYSCTL_ARG(req, arg1); return (sysctl_msec_to_ticks(oidp, arg1, arg2, req)); } /* * Minimum MSS we accept and use. This prevents DoS attacks where * we are forced to a ridiculous low MSS like 20 and send hundreds * of packets instead of one. The effect scales with the available * bandwidth and quickly saturates the CPU and network interface * with packet generation and sending. Set to zero to disable MINMSS * checking. This setting prevents us from sending too small packets. */ VNET_DEFINE(int, tcp_minmss) = TCP_MINMSS; SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, minmss, CTLFLAG_RW, &VNET_NAME(tcp_minmss), 0, "Minimum TCP Maximum Segment Size"); VNET_DEFINE(int, tcp_do_rfc1323) = 1; SYSCTL_VNET_INT(_net_inet_tcp, TCPCTL_DO_RFC1323, rfc1323, CTLFLAG_RW, &VNET_NAME(tcp_do_rfc1323), 0, "Enable rfc1323 (high performance TCP) extensions"); static int tcp_log_debug = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_debug, CTLFLAG_RW, &tcp_log_debug, 0, "Log errors caused by incoming TCP segments"); static int tcp_tcbhashsize = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, tcbhashsize, CTLFLAG_RDTUN, &tcp_tcbhashsize, 0, "Size of TCP control-block hashtable"); static int do_tcpdrain = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, do_tcpdrain, CTLFLAG_RW, &do_tcpdrain, 0, "Enable tcp_drain routine for extra help when low on mbufs"); SYSCTL_VNET_UINT(_net_inet_tcp, OID_AUTO, pcbcount, CTLFLAG_RD, &VNET_NAME(tcbinfo.ipi_count), 0, "Number of active PCBs"); static VNET_DEFINE(int, icmp_may_rst) = 1; #define V_icmp_may_rst VNET(icmp_may_rst) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, icmp_may_rst, CTLFLAG_RW, &VNET_NAME(icmp_may_rst), 0, "Certain ICMP unreachable messages may abort connections in SYN_SENT"); static VNET_DEFINE(int, tcp_isn_reseed_interval) = 0; #define V_tcp_isn_reseed_interval VNET(tcp_isn_reseed_interval) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, isn_reseed_interval, CTLFLAG_RW, &VNET_NAME(tcp_isn_reseed_interval), 0, "Seconds between reseeding of ISN secret"); /* * TCP bandwidth limiting sysctls. Note that the default lower bound of * 1024 exists only for debugging. A good production default would be * something like 6100. */ SYSCTL_NODE(_net_inet_tcp, OID_AUTO, inflight, CTLFLAG_RW, 0, "TCP inflight data limiting"); static VNET_DEFINE(int, tcp_inflight_enable) = 0; #define V_tcp_inflight_enable VNET(tcp_inflight_enable) SYSCTL_VNET_INT(_net_inet_tcp_inflight, OID_AUTO, enable, CTLFLAG_RW, &VNET_NAME(tcp_inflight_enable), 0, "Enable automatic TCP inflight data limiting"); static int tcp_inflight_debug = 0; SYSCTL_INT(_net_inet_tcp_inflight, OID_AUTO, debug, CTLFLAG_RW, &tcp_inflight_debug, 0, "Debug TCP inflight calculations"); static VNET_DEFINE(int, tcp_inflight_rttthresh); #define V_tcp_inflight_rttthresh VNET(tcp_inflight_rttthresh) SYSCTL_VNET_PROC(_net_inet_tcp_inflight, OID_AUTO, rttthresh, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(tcp_inflight_rttthresh), 0, vnet_sysctl_msec_to_ticks, "I", "RTT threshold below which inflight will deactivate itself"); static VNET_DEFINE(int, tcp_inflight_min) = 6144; #define V_tcp_inflight_min VNET(tcp_inflight_min) SYSCTL_VNET_INT(_net_inet_tcp_inflight, OID_AUTO, min, CTLFLAG_RW, &VNET_NAME(tcp_inflight_min), 0, "Lower-bound for TCP inflight window"); static VNET_DEFINE(int, tcp_inflight_max) = TCP_MAXWIN << TCP_MAX_WINSHIFT; #define V_tcp_inflight_max VNET(tcp_inflight_max) SYSCTL_VNET_INT(_net_inet_tcp_inflight, OID_AUTO, max, CTLFLAG_RW, &VNET_NAME(tcp_inflight_max), 0, "Upper-bound for TCP inflight window"); static VNET_DEFINE(int, tcp_inflight_stab) = 20; #define V_tcp_inflight_stab VNET(tcp_inflight_stab) SYSCTL_VNET_INT(_net_inet_tcp_inflight, OID_AUTO, stab, CTLFLAG_RW, &VNET_NAME(tcp_inflight_stab), 0, "Inflight Algorithm Stabilization 20 = 2 packets"); #ifdef TCP_SIGNATURE static int tcp_sig_checksigs = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, signature_verify_input, CTLFLAG_RW, &tcp_sig_checksigs, 0, "Verify RFC2385 digests on inbound traffic"); #endif VNET_DEFINE(uma_zone_t, sack_hole_zone); #define V_sack_hole_zone VNET(sack_hole_zone) VNET_DEFINE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST+1]); static struct inpcb *tcp_notify(struct inpcb *, int); static struct inpcb *tcp_mtudisc_notify(struct inpcb *, int); static void tcp_isn_tick(void *); static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr); /* * Target size of TCP PCB hash tables. Must be a power of two. * * Note that this can be overridden by the kernel environment * variable net.inet.tcp.tcbhashsize */ #ifndef TCBHASHSIZE #define TCBHASHSIZE 512 #endif /* * XXX * Callouts should be moved into struct tcp directly. They are currently * separate because the tcpcb structure is exported to userland for sysctl * parsing purposes, which do not know about callouts. */ struct tcpcb_mem { struct tcpcb tcb; struct tcp_timer tt; struct cc_var ccv; struct osd osd; }; static VNET_DEFINE(uma_zone_t, tcpcb_zone); #define V_tcpcb_zone VNET(tcpcb_zone) MALLOC_DEFINE(M_TCPLOG, "tcplog", "TCP address and flags print buffers"); struct callout isn_callout; static struct mtx isn_mtx; #define ISN_LOCK_INIT() mtx_init(&isn_mtx, "isn_mtx", NULL, MTX_DEF) #define ISN_LOCK() mtx_lock(&isn_mtx) #define ISN_UNLOCK() mtx_unlock(&isn_mtx) /* * TCP initialization. */ static void tcp_zone_change(void *tag) { uma_zone_set_max(V_tcbinfo.ipi_zone, maxsockets); uma_zone_set_max(V_tcpcb_zone, maxsockets); tcp_tw_zone_change(); } static int tcp_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp = mem; INP_LOCK_INIT(inp, "inp", "tcpinp"); return (0); } void tcp_init(void) { int hashsize; INP_INFO_LOCK_INIT(&V_tcbinfo, "tcp"); LIST_INIT(&V_tcb); #ifdef VIMAGE V_tcbinfo.ipi_vnet = curvnet; #endif V_tcbinfo.ipi_listhead = &V_tcb; if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_IN, &V_tcp_hhh[HHOOK_TCP_EST_IN], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_OUT, &V_tcp_hhh[HHOOK_TCP_EST_OUT], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); hashsize = TCBHASHSIZE; TUNABLE_INT_FETCH("net.inet.tcp.tcbhashsize", &hashsize); if (!powerof2(hashsize)) { printf("WARNING: TCB hash size not a power of 2\n"); hashsize = 512; /* safe default */ } V_tcbinfo.ipi_hashbase = hashinit(hashsize, M_PCB, &V_tcbinfo.ipi_hashmask); V_tcbinfo.ipi_porthashbase = hashinit(hashsize, M_PCB, &V_tcbinfo.ipi_porthashmask); V_tcbinfo.ipi_zone = uma_zcreate("tcp_inpcb", sizeof(struct inpcb), NULL, NULL, tcp_inpcb_init, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); uma_zone_set_max(V_tcbinfo.ipi_zone, maxsockets); V_tcp_inflight_rttthresh = TCPTV_INFLIGHT_RTTTHRESH; /* * These have to be type stable for the benefit of the timers. */ V_tcpcb_zone = uma_zcreate("tcpcb", sizeof(struct tcpcb_mem), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); uma_zone_set_max(V_tcpcb_zone, maxsockets); tcp_tw_init(); syncache_init(); tcp_hc_init(); - tcp_reass_init(); TUNABLE_INT_FETCH("net.inet.tcp.sack.enable", &V_tcp_do_sack); V_sack_hole_zone = uma_zcreate("sackhole", sizeof(struct sackhole), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); /* Skip initialization of globals for non-default instances. */ if (!IS_DEFAULT_VNET(curvnet)) return; + tcp_reass_global_init(); + /* XXX virtualize those bellow? */ tcp_delacktime = TCPTV_DELACK; tcp_keepinit = TCPTV_KEEP_INIT; tcp_keepidle = TCPTV_KEEP_IDLE; tcp_keepintvl = TCPTV_KEEPINTVL; tcp_maxpersistidle = TCPTV_KEEP_IDLE; tcp_msl = TCPTV_MSL; tcp_rexmit_min = TCPTV_MIN; if (tcp_rexmit_min < 1) tcp_rexmit_min = 1; tcp_rexmit_slop = TCPTV_CPU_VAR; tcp_finwait2_timeout = TCPTV_FINWAIT2_TIMEOUT; tcp_tcbhashsize = hashsize; #ifdef INET6 #define TCP_MINPROTOHDR (sizeof(struct ip6_hdr) + sizeof(struct tcphdr)) #else /* INET6 */ #define TCP_MINPROTOHDR (sizeof(struct tcpiphdr)) #endif /* INET6 */ if (max_protohdr < TCP_MINPROTOHDR) max_protohdr = TCP_MINPROTOHDR; if (max_linkhdr + TCP_MINPROTOHDR > MHLEN) panic("tcp_init"); #undef TCP_MINPROTOHDR ISN_LOCK_INIT(); callout_init(&isn_callout, CALLOUT_MPSAFE); callout_reset(&isn_callout, hz/100, tcp_isn_tick, NULL); EVENTHANDLER_REGISTER(shutdown_pre_sync, tcp_fini, NULL, SHUTDOWN_PRI_DEFAULT); EVENTHANDLER_REGISTER(maxsockets_change, tcp_zone_change, NULL, EVENTHANDLER_PRI_ANY); } #ifdef VIMAGE void tcp_destroy(void) { - tcp_reass_destroy(); tcp_hc_destroy(); syncache_destroy(); tcp_tw_destroy(); /* XXX check that hashes are empty! */ hashdestroy(V_tcbinfo.ipi_hashbase, M_PCB, V_tcbinfo.ipi_hashmask); hashdestroy(V_tcbinfo.ipi_porthashbase, M_PCB, V_tcbinfo.ipi_porthashmask); uma_zdestroy(V_sack_hole_zone); uma_zdestroy(V_tcpcb_zone); uma_zdestroy(V_tcbinfo.ipi_zone); INP_INFO_LOCK_DESTROY(&V_tcbinfo); } #endif void tcp_fini(void *xtp) { callout_stop(&isn_callout); } /* * Fill in the IP and TCP headers for an outgoing packet, given the tcpcb. * tcp_template used to store this data in mbufs, but we now recopy it out * of the tcpcb each time to conserve mbufs. */ void tcpip_fillheaders(struct inpcb *inp, void *ip_ptr, void *tcp_ptr) { struct tcphdr *th = (struct tcphdr *)tcp_ptr; INP_WLOCK_ASSERT(inp); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { struct ip6_hdr *ip6; ip6 = (struct ip6_hdr *)ip_ptr; ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) | (inp->inp_flow & IPV6_FLOWINFO_MASK); ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) | (IPV6_VERSION & IPV6_VERSION_MASK); ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = htons(sizeof(struct tcphdr)); ip6->ip6_src = inp->in6p_laddr; ip6->ip6_dst = inp->in6p_faddr; } else #endif { struct ip *ip; ip = (struct ip *)ip_ptr; ip->ip_v = IPVERSION; ip->ip_hl = 5; ip->ip_tos = inp->inp_ip_tos; ip->ip_len = 0; ip->ip_id = 0; ip->ip_off = 0; ip->ip_ttl = inp->inp_ip_ttl; ip->ip_sum = 0; ip->ip_p = IPPROTO_TCP; ip->ip_src = inp->inp_laddr; ip->ip_dst = inp->inp_faddr; } th->th_sport = inp->inp_lport; th->th_dport = inp->inp_fport; th->th_seq = 0; th->th_ack = 0; th->th_x2 = 0; th->th_off = 5; th->th_flags = 0; th->th_win = 0; th->th_urp = 0; th->th_sum = 0; /* in_pseudo() is called later for ipv4 */ } /* * Create template to be used to send tcp packets on a connection. * Allocates an mbuf and fills in a skeletal tcp/ip header. The only * use for this function is in keepalives, which use tcp_respond. */ struct tcptemp * tcpip_maketemplate(struct inpcb *inp) { struct tcptemp *t; t = malloc(sizeof(*t), M_TEMP, M_NOWAIT); if (t == NULL) return (NULL); tcpip_fillheaders(inp, (void *)&t->tt_ipgen, (void *)&t->tt_t); return (t); } /* * Send a single message to the TCP at address specified by * the given TCP/IP header. If m == NULL, then we make a copy * of the tcpiphdr at ti and send directly to the addressed host. * This is used to force keep alive messages out using the TCP * template for a connection. If flags are given then we send * a message back to the TCP which originated the * segment ti, * and discard the mbuf containing it and any other attached mbufs. * * In any case the ack and sequence number of the transmitted * segment are as specified by the parameters. * * NOTE: If m != NULL, then ti must point to *inside* the mbuf. */ void tcp_respond(struct tcpcb *tp, void *ipgen, struct tcphdr *th, struct mbuf *m, tcp_seq ack, tcp_seq seq, int flags) { int tlen; int win = 0; struct ip *ip; struct tcphdr *nth; #ifdef INET6 struct ip6_hdr *ip6; int isipv6; #endif /* INET6 */ int ipflags = 0; struct inpcb *inp; KASSERT(tp != NULL || m != NULL, ("tcp_respond: tp and m both NULL")); #ifdef INET6 isipv6 = ((struct ip *)ipgen)->ip_v == 6; ip6 = ipgen; #endif /* INET6 */ ip = ipgen; if (tp != NULL) { inp = tp->t_inpcb; KASSERT(inp != NULL, ("tcp control block w/o inpcb")); INP_WLOCK_ASSERT(inp); } else inp = NULL; if (tp != NULL) { if (!(flags & TH_RST)) { win = sbspace(&inp->inp_socket->so_rcv); if (win > (long)TCP_MAXWIN << tp->rcv_scale) win = (long)TCP_MAXWIN << tp->rcv_scale; } } if (m == NULL) { m = m_gethdr(M_DONTWAIT, MT_DATA); if (m == NULL) return; tlen = 0; m->m_data += max_linkhdr; #ifdef INET6 if (isipv6) { bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(struct ip6_hdr)); ip6 = mtod(m, struct ip6_hdr *); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); ip = mtod(m, struct ip *); nth = (struct tcphdr *)(ip + 1); } bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr)); flags = TH_ACK; } else { /* * reuse the mbuf. * XXX MRT We inherrit the FIB, which is lucky. */ m_freem(m->m_next); m->m_next = NULL; m->m_data = (caddr_t)ipgen; /* m_len is set later */ tlen = 0; #define xchg(a,b,type) { type t; t=a; a=b; b=t; } #ifdef INET6 if (isipv6) { xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { xchg(ip->ip_dst.s_addr, ip->ip_src.s_addr, uint32_t); nth = (struct tcphdr *)(ip + 1); } if (th != nth) { /* * this is usually a case when an extension header * exists between the IPv6 header and the * TCP header. */ nth->th_sport = th->th_sport; nth->th_dport = th->th_dport; } xchg(nth->th_dport, nth->th_sport, uint16_t); #undef xchg } #ifdef INET6 if (isipv6) { ip6->ip6_flow = 0; ip6->ip6_vfc = IPV6_VERSION; ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = htons((u_short)(sizeof (struct tcphdr) + tlen)); tlen += sizeof (struct ip6_hdr) + sizeof (struct tcphdr); } else #endif { tlen += sizeof (struct tcpiphdr); ip->ip_len = tlen; ip->ip_ttl = V_ip_defttl; if (V_path_mtu_discovery) ip->ip_off |= IP_DF; } m->m_len = tlen; m->m_pkthdr.len = tlen; m->m_pkthdr.rcvif = NULL; #ifdef MAC if (inp != NULL) { /* * Packet is associated with a socket, so allow the * label of the response to reflect the socket label. */ INP_WLOCK_ASSERT(inp); mac_inpcb_create_mbuf(inp, m); } else { /* * Packet is not associated with a socket, so possibly * update the label in place. */ mac_netinet_tcp_reply(m); } #endif nth->th_seq = htonl(seq); nth->th_ack = htonl(ack); nth->th_x2 = 0; nth->th_off = sizeof (struct tcphdr) >> 2; nth->th_flags = flags; if (tp != NULL) nth->th_win = htons((u_short) (win >> tp->rcv_scale)); else nth->th_win = htons((u_short)win); nth->th_urp = 0; #ifdef INET6 if (isipv6) { nth->th_sum = 0; nth->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr), tlen - sizeof(struct ip6_hdr)); ip6->ip6_hlim = in6_selecthlim(tp != NULL ? tp->t_inpcb : NULL, NULL); } else #endif /* INET6 */ { nth->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htons((u_short)(tlen - sizeof(struct ip) + ip->ip_p))); m->m_pkthdr.csum_flags = CSUM_TCP; m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); } #ifdef TCPDEBUG if (tp == NULL || (inp->inp_socket->so_options & SO_DEBUG)) tcp_trace(TA_OUTPUT, 0, tp, mtod(m, void *), th, 0); #endif #ifdef INET6 if (isipv6) (void) ip6_output(m, NULL, NULL, ipflags, NULL, NULL, inp); else #endif /* INET6 */ (void) ip_output(m, NULL, NULL, ipflags, NULL, inp); } /* * Create a new TCP control block, making an * empty reassembly queue and hooking it to the argument * protocol control block. The `inp' parameter must have * come from the zone allocator set up in tcp_init(). */ struct tcpcb * tcp_newtcpcb(struct inpcb *inp) { struct tcpcb_mem *tm; struct tcpcb *tp; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ tm = uma_zalloc(V_tcpcb_zone, M_NOWAIT | M_ZERO); if (tm == NULL) return (NULL); tp = &tm->tcb; /* Initialise cc_var struct for this tcpcb. */ tp->ccv = &tm->ccv; tp->ccv->type = IPPROTO_TCP; tp->ccv->ccvc.tcp = tp; /* * Use the current system default CC algorithm. */ CC_LIST_RLOCK(); KASSERT(!STAILQ_EMPTY(&cc_list), ("cc_list is empty!")); CC_ALGO(tp) = CC_DEFAULT(); CC_LIST_RUNLOCK(); if (CC_ALGO(tp)->cb_init != NULL) if (CC_ALGO(tp)->cb_init(tp->ccv) > 0) { uma_zfree(V_tcpcb_zone, tm); return (NULL); } tp->osd = &tm->osd; if (khelp_init_osd(HELPER_CLASS_TCP, tp->osd)) { uma_zfree(V_tcpcb_zone, tm); return (NULL); } #ifdef VIMAGE tp->t_vnet = inp->inp_vnet; #endif tp->t_timers = &tm->tt; /* LIST_INIT(&tp->t_segq); */ /* XXX covered by M_ZERO */ tp->t_maxseg = tp->t_maxopd = #ifdef INET6 isipv6 ? V_tcp_v6mssdflt : #endif /* INET6 */ V_tcp_mssdflt; /* Set up our timeouts. */ callout_init(&tp->t_timers->tt_rexmt, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_persist, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_keep, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_2msl, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_delack, CALLOUT_MPSAFE); if (V_tcp_do_rfc1323) tp->t_flags = (TF_REQ_SCALE|TF_REQ_TSTMP); if (V_tcp_do_sack) tp->t_flags |= TF_SACK_PERMIT; TAILQ_INIT(&tp->snd_holes); tp->t_inpcb = inp; /* XXX */ /* * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no * rtt estimate. Set rttvar so that srtt + 4 * rttvar gives * reasonable initial retransmit time. */ tp->t_srtt = TCPTV_SRTTBASE; tp->t_rttvar = ((TCPTV_RTOBASE - TCPTV_SRTTBASE) << TCP_RTTVAR_SHIFT) / 4; tp->t_rttmin = tcp_rexmit_min; tp->t_rxtcur = TCPTV_RTOBASE; tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->snd_bwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->t_rcvtime = ticks; tp->t_bw_rtttime = ticks; /* * IPv4 TTL initialization is necessary for an IPv6 socket as well, * because the socket may be bound to an IPv6 wildcard address, * which may match an IPv4-mapped IPv6 address. */ inp->inp_ip_ttl = V_ip_defttl; inp->inp_ppcb = tp; return (tp); /* XXX */ } /* * Switch the congestion control algorithm back to NewReno for any active * control blocks using an algorithm which is about to go away. * This ensures the CC framework can allow the unload to proceed without leaving * any dangling pointers which would trigger a panic. * Returning non-zero would inform the CC framework that something went wrong * and it would be unsafe to allow the unload to proceed. However, there is no * way for this to occur with this implementation so we always return zero. */ int tcp_ccalgounload(struct cc_algo *unload_algo) { struct cc_algo *tmpalgo; struct inpcb *inp; struct tcpcb *tp; VNET_ITERATOR_DECL(vnet_iter); /* * Check all active control blocks across all network stacks and change * any that are using "unload_algo" back to NewReno. If "unload_algo" * requires cleanup code to be run, call it. */ VNET_LIST_RLOCK(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); INP_INFO_RLOCK(&V_tcbinfo); /* * New connections already part way through being initialised * with the CC algo we're removing will not race with this code * because the INP_INFO_WLOCK is held during initialisation. We * therefore don't enter the loop below until the connection * list has stabilised. */ LIST_FOREACH(inp, &V_tcb, inp_list) { INP_WLOCK(inp); /* Important to skip tcptw structs. */ if (!(inp->inp_flags & INP_TIMEWAIT) && (tp = intotcpcb(inp)) != NULL) { /* * By holding INP_WLOCK here, we are assured * that the connection is not currently * executing inside the CC module's functions * i.e. it is safe to make the switch back to * NewReno. */ if (CC_ALGO(tp) == unload_algo) { tmpalgo = CC_ALGO(tp); /* NewReno does not require any init. */ CC_ALGO(tp) = &newreno_cc_algo; if (tmpalgo->cb_destroy != NULL) tmpalgo->cb_destroy(tp->ccv); } } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK(); return (0); } /* * Drop a TCP connection, reporting * the specified error. If connection is synchronized, * then send a RST to peer. */ struct tcpcb * tcp_drop(struct tcpcb *tp, int errno) { struct socket *so = tp->t_inpcb->inp_socket; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); if (TCPS_HAVERCVDSYN(tp->t_state)) { tp->t_state = TCPS_CLOSED; (void) tcp_output_reset(tp); TCPSTAT_INC(tcps_drops); } else TCPSTAT_INC(tcps_conndrops); if (errno == ETIMEDOUT && tp->t_softerror) errno = tp->t_softerror; so->so_error = errno; return (tcp_close(tp)); } void tcp_discardcb(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so = inp->inp_socket; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ INP_WLOCK_ASSERT(inp); /* * Make sure that all of our timers are stopped before we * delete the PCB. */ callout_stop(&tp->t_timers->tt_rexmt); callout_stop(&tp->t_timers->tt_persist); callout_stop(&tp->t_timers->tt_keep); callout_stop(&tp->t_timers->tt_2msl); callout_stop(&tp->t_timers->tt_delack); /* * If we got enough samples through the srtt filter, * save the rtt and rttvar in the routing entry. * 'Enough' is arbitrarily defined as 4 rtt samples. * 4 samples is enough for the srtt filter to converge * to within enough % of the correct value; fewer samples * and we could save a bogus rtt. The danger is not high * as tcp quickly recovers from everything. * XXX: Works very well but needs some more statistics! */ if (tp->t_rttupdated >= 4) { struct hc_metrics_lite metrics; u_long ssthresh; bzero(&metrics, sizeof(metrics)); /* * Update the ssthresh always when the conditions below * are satisfied. This gives us better new start value * for the congestion avoidance for new connections. * ssthresh is only set if packet loss occured on a session. * * XXXRW: 'so' may be NULL here, and/or socket buffer may be * being torn down. Ideally this code would not use 'so'. */ ssthresh = tp->snd_ssthresh; if (ssthresh != 0 && ssthresh < so->so_snd.sb_hiwat / 2) { /* * convert the limit from user data bytes to * packets then to packet data bytes. */ ssthresh = (ssthresh + tp->t_maxseg / 2) / tp->t_maxseg; if (ssthresh < 2) ssthresh = 2; ssthresh *= (u_long)(tp->t_maxseg + #ifdef INET6 (isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr) : #endif sizeof (struct tcpiphdr) #ifdef INET6 ) #endif ); } else ssthresh = 0; metrics.rmx_ssthresh = ssthresh; metrics.rmx_rtt = tp->t_srtt; metrics.rmx_rttvar = tp->t_rttvar; /* XXX: This wraps if the pipe is more than 4 Gbit per second */ metrics.rmx_bandwidth = tp->snd_bandwidth; metrics.rmx_cwnd = tp->snd_cwnd; metrics.rmx_sendpipe = 0; metrics.rmx_recvpipe = 0; tcp_hc_update(&inp->inp_inc, &metrics); } /* free the reassembly queue, if any */ tcp_reass_flush(tp); /* Disconnect offload device, if any. */ tcp_offload_detach(tp); tcp_free_sackholes(tp); /* Allow the CC algorithm to clean up after itself. */ if (CC_ALGO(tp)->cb_destroy != NULL) CC_ALGO(tp)->cb_destroy(tp->ccv); khelp_destroy_osd(tp->osd); CC_ALGO(tp) = NULL; inp->inp_ppcb = NULL; tp->t_inpcb = NULL; uma_zfree(V_tcpcb_zone, tp); } /* * Attempt to close a TCP control block, marking it as dropped, and freeing * the socket if we hold the only reference. */ struct tcpcb * tcp_close(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); /* Notify any offload devices of listener close */ if (tp->t_state == TCPS_LISTEN) tcp_offload_listen_close(tp); in_pcbdrop(inp); TCPSTAT_INC(tcps_closed); KASSERT(inp->inp_socket != NULL, ("tcp_close: inp_socket NULL")); so = inp->inp_socket; soisdisconnected(so); if (inp->inp_flags & INP_SOCKREF) { KASSERT(so->so_state & SS_PROTOREF, ("tcp_close: !SS_PROTOREF")); inp->inp_flags &= ~INP_SOCKREF; INP_WUNLOCK(inp); ACCEPT_LOCK(); SOCK_LOCK(so); so->so_state &= ~SS_PROTOREF; sofree(so); return (NULL); } return (tp); } void tcp_drain(void) { VNET_ITERATOR_DECL(vnet_iter); if (!do_tcpdrain) return; VNET_LIST_RLOCK_NOSLEEP(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); struct inpcb *inpb; struct tcpcb *tcpb; /* * Walk the tcpbs, if existing, and flush the reassembly queue, * if there is one... * XXX: The "Net/3" implementation doesn't imply that the TCP * reassembly queue should be flushed, but in a situation * where we're really low on mbufs, this is potentially * usefull. */ INP_INFO_RLOCK(&V_tcbinfo); LIST_FOREACH(inpb, V_tcbinfo.ipi_listhead, inp_list) { if (inpb->inp_flags & INP_TIMEWAIT) continue; INP_WLOCK(inpb); if ((tcpb = intotcpcb(inpb)) != NULL) { tcp_reass_flush(tcpb); tcp_clean_sackreport(tcpb); } INP_WUNLOCK(inpb); } INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK_NOSLEEP(); } /* * Notify a tcp user of an asynchronous error; * store error as soft error, but wake up user * (for now, won't do anything until can select for soft error). * * Do not wake up user since there currently is no mechanism for * reporting soft errors (yet - a kqueue filter may be added). */ static struct inpcb * tcp_notify(struct inpcb *inp, int error) { struct tcpcb *tp; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_notify: tp == NULL")); /* * Ignore some errors if we are hooked up. * If connection hasn't completed, has retransmitted several times, * and receives a second error, give up now. This is better * than waiting a long time to establish a connection that * can never complete. */ if (tp->t_state == TCPS_ESTABLISHED && (error == EHOSTUNREACH || error == ENETUNREACH || error == EHOSTDOWN)) { return (inp); } else if (tp->t_state < TCPS_ESTABLISHED && tp->t_rxtshift > 3 && tp->t_softerror) { tp = tcp_drop(tp, error); if (tp != NULL) return (inp); else return (NULL); } else { tp->t_softerror = error; return (inp); } #if 0 wakeup( &so->so_timeo); sorwakeup(so); sowwakeup(so); #endif } static int tcp_pcblist(SYSCTL_HANDLER_ARGS) { int error, i, m, n, pcb_count; struct inpcb *inp, **inp_list; inp_gen_t gencnt; struct xinpgen xig; /* * The process of preparing the TCB list is too time-consuming and * resource-intensive to repeat twice on every request. */ if (req->oldptr == NULL) { n = V_tcbinfo.ipi_count + syncache_pcbcount(); n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xtcpcb); return (0); } if (req->newptr != NULL) return (EPERM); /* * OK, now we're committed to doing something. */ INP_INFO_RLOCK(&V_tcbinfo); gencnt = V_tcbinfo.ipi_gencnt; n = V_tcbinfo.ipi_count; INP_INFO_RUNLOCK(&V_tcbinfo); m = syncache_pcbcount(); error = sysctl_wire_old_buffer(req, 2 * (sizeof xig) + (n + m) * sizeof(struct xtcpcb)); if (error != 0) return (error); xig.xig_len = sizeof xig; xig.xig_count = n + m; xig.xig_gen = gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return (error); error = syncache_pcblist(req, m, &pcb_count); if (error) return (error); inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); if (inp_list == NULL) return (ENOMEM); INP_INFO_RLOCK(&V_tcbinfo); for (inp = LIST_FIRST(V_tcbinfo.ipi_listhead), i = 0; inp != NULL && i < n; inp = LIST_NEXT(inp, inp_list)) { INP_WLOCK(inp); if (inp->inp_gencnt <= gencnt) { /* * XXX: This use of cr_cansee(), introduced with * TCP state changes, is not quite right, but for * now, better than nothing. */ if (inp->inp_flags & INP_TIMEWAIT) { if (intotw(inp) != NULL) error = cr_cansee(req->td->td_ucred, intotw(inp)->tw_cred); else error = EINVAL; /* Skip this inp. */ } else error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) { in_pcbref(inp); inp_list[i++] = inp; } } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); n = i; error = 0; for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (inp->inp_gencnt <= gencnt) { struct xtcpcb xt; void *inp_ppcb; bzero(&xt, sizeof(xt)); xt.xt_len = sizeof xt; /* XXX should avoid extra copy */ bcopy(inp, &xt.xt_inp, sizeof *inp); inp_ppcb = inp->inp_ppcb; if (inp_ppcb == NULL) bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); else if (inp->inp_flags & INP_TIMEWAIT) { bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); xt.xt_tp.t_state = TCPS_TIME_WAIT; } else bcopy(inp_ppcb, &xt.xt_tp, sizeof xt.xt_tp); if (inp->inp_socket != NULL) sotoxsocket(inp->inp_socket, &xt.xt_socket); else { bzero(&xt.xt_socket, sizeof xt.xt_socket); xt.xt_socket.xso_protocol = IPPROTO_TCP; } xt.xt_inp.inp_gencnt = inp->inp_gencnt; INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xt, sizeof xt); } else INP_RUNLOCK(inp); } INP_INFO_WLOCK(&V_tcbinfo); for (i = 0; i < n; i++) { inp = inp_list[i]; INP_WLOCK(inp); if (!in_pcbrele(inp)) INP_WUNLOCK(inp); } INP_INFO_WUNLOCK(&V_tcbinfo); if (!error) { /* * Give the user an updated idea of our state. * If the generation differs from what we told * her before, she knows that something happened * while we were processing this request, and it * might be necessary to retry. */ INP_INFO_RLOCK(&V_tcbinfo); xig.xig_gen = V_tcbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_tcbinfo.ipi_count + pcb_count; INP_INFO_RUNLOCK(&V_tcbinfo); error = SYSCTL_OUT(req, &xig, sizeof xig); } free(inp_list, M_TEMP); return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_PCBLIST, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0, tcp_pcblist, "S,xtcpcb", "List of active TCP connections"); static int tcp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; struct inpcb *inp; int error; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); INP_INFO_RLOCK(&V_tcbinfo); inp = in_pcblookup_hash(&V_tcbinfo, addrs[1].sin_addr, addrs[1].sin_port, addrs[0].sin_addr, addrs[0].sin_port, 0, NULL); if (inp != NULL) { INP_RLOCK(inp); INP_INFO_RUNLOCK(&V_tcbinfo); if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else { INP_INFO_RUNLOCK(&V_tcbinfo); error = ENOENT; } if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet_tcp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp_getcred, "S,xucred", "Get the xucred of a TCP connection"); #ifdef INET6 static int tcp6_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in6 addrs[2]; struct inpcb *inp; int error, mapped = 0; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); if ((error = sa6_embedscope(&addrs[0], V_ip6_use_defzone)) != 0 || (error = sa6_embedscope(&addrs[1], V_ip6_use_defzone)) != 0) { return (error); } if (IN6_IS_ADDR_V4MAPPED(&addrs[0].sin6_addr)) { if (IN6_IS_ADDR_V4MAPPED(&addrs[1].sin6_addr)) mapped = 1; else return (EINVAL); } INP_INFO_RLOCK(&V_tcbinfo); if (mapped == 1) inp = in_pcblookup_hash(&V_tcbinfo, *(struct in_addr *)&addrs[1].sin6_addr.s6_addr[12], addrs[1].sin6_port, *(struct in_addr *)&addrs[0].sin6_addr.s6_addr[12], addrs[0].sin6_port, 0, NULL); else inp = in6_pcblookup_hash(&V_tcbinfo, &addrs[1].sin6_addr, addrs[1].sin6_port, &addrs[0].sin6_addr, addrs[0].sin6_port, 0, NULL); if (inp != NULL) { INP_RLOCK(inp); INP_INFO_RUNLOCK(&V_tcbinfo); if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else { INP_INFO_RUNLOCK(&V_tcbinfo); error = ENOENT; } if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet6_tcp6, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp6_getcred, "S,xucred", "Get the xucred of a TCP6 connection"); #endif void tcp_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct ip *ip = vip; struct tcphdr *th; struct in_addr faddr; struct inpcb *inp; struct tcpcb *tp; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct icmp *icp; struct in_conninfo inc; tcp_seq icmp_tcp_seq; int mtu; faddr = ((struct sockaddr_in *)sa)->sin_addr; if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc_notify; else if (V_icmp_may_rst && (cmd == PRC_UNREACH_ADMIN_PROHIB || cmd == PRC_UNREACH_PORT || cmd == PRC_TIMXCEED_INTRANS) && ip) notify = tcp_drop_syn_sent; /* * Redirects don't need to be handled up here. */ else if (PRC_IS_REDIRECT(cmd)) return; /* * Source quench is depreciated. */ else if (cmd == PRC_QUENCH) return; /* * Hostdead is ugly because it goes linearly through all PCBs. * XXX: We never get this from ICMP, otherwise it makes an * excellent DoS attack on machines with many connections. */ else if (cmd == PRC_HOSTDEAD) ip = NULL; else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) return; if (ip != NULL) { icp = (struct icmp *)((caddr_t)ip - offsetof(struct icmp, icmp_ip)); th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); INP_INFO_WLOCK(&V_tcbinfo); inp = in_pcblookup_hash(&V_tcbinfo, faddr, th->th_dport, ip->ip_src, th->th_sport, 0, NULL); if (inp != NULL) { INP_WLOCK(inp); if (!(inp->inp_flags & INP_TIMEWAIT) && !(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket == NULL)) { icmp_tcp_seq = htonl(th->th_seq); tp = intotcpcb(inp); if (SEQ_GEQ(icmp_tcp_seq, tp->snd_una) && SEQ_LT(icmp_tcp_seq, tp->snd_max)) { if (cmd == PRC_MSGSIZE) { /* * MTU discovery: * If we got a needfrag set the MTU * in the route to the suggested new * value (if given) and then notify. */ bzero(&inc, sizeof(inc)); inc.inc_faddr = faddr; inc.inc_fibnum = inp->inp_inc.inc_fibnum; mtu = ntohs(icp->icmp_nextmtu); /* * If no alternative MTU was * proposed, try the next smaller * one. ip->ip_len has already * been swapped in icmp_input(). */ if (!mtu) mtu = ip_next_mtu(ip->ip_len, 1); if (mtu < V_tcp_minmss + sizeof(struct tcpiphdr)) mtu = V_tcp_minmss + sizeof(struct tcpiphdr); /* * Only cache the MTU if it * is smaller than the interface * or route MTU. tcp_mtudisc() * will do right thing by itself. */ if (mtu <= tcp_maxmtu(&inc, NULL)) tcp_hc_updatemtu(&inc, mtu); tcp_mtudisc(inp, mtu); } else inp = (*notify)(inp, inetctlerrmap[cmd]); } } if (inp != NULL) INP_WUNLOCK(inp); } else { bzero(&inc, sizeof(inc)); inc.inc_fport = th->th_dport; inc.inc_lport = th->th_sport; inc.inc_faddr = faddr; inc.inc_laddr = ip->ip_src; syncache_unreach(&inc, th); } INP_INFO_WUNLOCK(&V_tcbinfo); } else in_pcbnotifyall(&V_tcbinfo, faddr, inetctlerrmap[cmd], notify); } #ifdef INET6 void tcp6_ctlinput(int cmd, struct sockaddr *sa, void *d) { struct tcphdr th; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct ip6_hdr *ip6; struct mbuf *m; struct ip6ctlparam *ip6cp = NULL; const struct sockaddr_in6 *sa6_src = NULL; int off; struct tcp_portonly { u_int16_t th_sport; u_int16_t th_dport; } *thp; if (sa->sa_family != AF_INET6 || sa->sa_len != sizeof(struct sockaddr_in6)) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc_notify; else if (!PRC_IS_REDIRECT(cmd) && ((unsigned)cmd >= PRC_NCMDS || inet6ctlerrmap[cmd] == 0)) return; /* Source quench is depreciated. */ else if (cmd == PRC_QUENCH) return; /* if the parameter is from icmp6, decode it. */ if (d != NULL) { ip6cp = (struct ip6ctlparam *)d; m = ip6cp->ip6c_m; ip6 = ip6cp->ip6c_ip6; off = ip6cp->ip6c_off; sa6_src = ip6cp->ip6c_src; } else { m = NULL; ip6 = NULL; off = 0; /* fool gcc */ sa6_src = &sa6_any; } if (ip6 != NULL) { struct in_conninfo inc; /* * XXX: We assume that when IPV6 is non NULL, * M and OFF are valid. */ /* check if we can safely examine src and dst ports */ if (m->m_pkthdr.len < off + sizeof(*thp)) return; bzero(&th, sizeof(th)); m_copydata(m, off, sizeof(*thp), (caddr_t)&th); in6_pcbnotify(&V_tcbinfo, sa, th.th_dport, (struct sockaddr *)ip6cp->ip6c_src, th.th_sport, cmd, NULL, notify); bzero(&inc, sizeof(inc)); inc.inc_fport = th.th_dport; inc.inc_lport = th.th_sport; inc.inc6_faddr = ((struct sockaddr_in6 *)sa)->sin6_addr; inc.inc6_laddr = ip6cp->ip6c_src->sin6_addr; inc.inc_flags |= INC_ISIPV6; INP_INFO_WLOCK(&V_tcbinfo); syncache_unreach(&inc, &th); INP_INFO_WUNLOCK(&V_tcbinfo); } else in6_pcbnotify(&V_tcbinfo, sa, 0, (const struct sockaddr *)sa6_src, 0, cmd, NULL, notify); } #endif /* INET6 */ /* * Following is where TCP initial sequence number generation occurs. * * There are two places where we must use initial sequence numbers: * 1. In SYN-ACK packets. * 2. In SYN packets. * * All ISNs for SYN-ACK packets are generated by the syncache. See * tcp_syncache.c for details. * * The ISNs in SYN packets must be monotonic; TIME_WAIT recycling * depends on this property. In addition, these ISNs should be * unguessable so as to prevent connection hijacking. To satisfy * the requirements of this situation, the algorithm outlined in * RFC 1948 is used, with only small modifications. * * Implementation details: * * Time is based off the system timer, and is corrected so that it * increases by one megabyte per second. This allows for proper * recycling on high speed LANs while still leaving over an hour * before rollover. * * As reading the *exact* system time is too expensive to be done * whenever setting up a TCP connection, we increment the time * offset in two ways. First, a small random positive increment * is added to isn_offset for each connection that is set up. * Second, the function tcp_isn_tick fires once per clock tick * and increments isn_offset as necessary so that sequence numbers * are incremented at approximately ISN_BYTES_PER_SECOND. The * random positive increments serve only to ensure that the same * exact sequence number is never sent out twice (as could otherwise * happen when a port is recycled in less than the system tick * interval.) * * net.inet.tcp.isn_reseed_interval controls the number of seconds * between seeding of isn_secret. This is normally set to zero, * as reseeding should not be necessary. * * Locking of the global variables isn_secret, isn_last_reseed, isn_offset, * isn_offset_old, and isn_ctx is performed using the TCP pcbinfo lock. In * general, this means holding an exclusive (write) lock. */ #define ISN_BYTES_PER_SECOND 1048576 #define ISN_STATIC_INCREMENT 4096 #define ISN_RANDOM_INCREMENT (4096 - 1) static VNET_DEFINE(u_char, isn_secret[32]); static VNET_DEFINE(int, isn_last_reseed); static VNET_DEFINE(u_int32_t, isn_offset); static VNET_DEFINE(u_int32_t, isn_offset_old); #define V_isn_secret VNET(isn_secret) #define V_isn_last_reseed VNET(isn_last_reseed) #define V_isn_offset VNET(isn_offset) #define V_isn_offset_old VNET(isn_offset_old) tcp_seq tcp_new_isn(struct tcpcb *tp) { MD5_CTX isn_ctx; u_int32_t md5_buffer[4]; tcp_seq new_isn; INP_WLOCK_ASSERT(tp->t_inpcb); ISN_LOCK(); /* Seed if this is the first use, reseed if requested. */ if ((V_isn_last_reseed == 0) || ((V_tcp_isn_reseed_interval > 0) && (((u_int)V_isn_last_reseed + (u_int)V_tcp_isn_reseed_interval*hz) < (u_int)ticks))) { read_random(&V_isn_secret, sizeof(V_isn_secret)); V_isn_last_reseed = ticks; } /* Compute the md5 hash and return the ISN. */ MD5Init(&isn_ctx); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_fport, sizeof(u_short)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_lport, sizeof(u_short)); #ifdef INET6 if ((tp->t_inpcb->inp_vflag & INP_IPV6) != 0) { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_faddr, sizeof(struct in6_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_laddr, sizeof(struct in6_addr)); } else #endif { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_faddr, sizeof(struct in_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_laddr, sizeof(struct in_addr)); } MD5Update(&isn_ctx, (u_char *) &V_isn_secret, sizeof(V_isn_secret)); MD5Final((u_char *) &md5_buffer, &isn_ctx); new_isn = (tcp_seq) md5_buffer[0]; V_isn_offset += ISN_STATIC_INCREMENT + (arc4random() & ISN_RANDOM_INCREMENT); new_isn += V_isn_offset; ISN_UNLOCK(); return (new_isn); } /* * Increment the offset to the next ISN_BYTES_PER_SECOND / 100 boundary * to keep time flowing at a relatively constant rate. If the random * increments have already pushed us past the projected offset, do nothing. */ static void tcp_isn_tick(void *xtp) { VNET_ITERATOR_DECL(vnet_iter); u_int32_t projected_offset; VNET_LIST_RLOCK_NOSLEEP(); ISN_LOCK(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); /* XXX appease INVARIANTS */ projected_offset = V_isn_offset_old + ISN_BYTES_PER_SECOND / 100; if (SEQ_GT(projected_offset, V_isn_offset)) V_isn_offset = projected_offset; V_isn_offset_old = V_isn_offset; CURVNET_RESTORE(); } ISN_UNLOCK(); VNET_LIST_RUNLOCK_NOSLEEP(); callout_reset(&isn_callout, hz/100, tcp_isn_tick, NULL); } /* * When a specific ICMP unreachable message is received and the * connection state is SYN-SENT, drop the connection. This behavior * is controlled by the icmp_may_rst sysctl. */ struct inpcb * tcp_drop_syn_sent(struct inpcb *inp, int errno) { struct tcpcb *tp; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); if (tp->t_state != TCPS_SYN_SENT) return (inp); tp = tcp_drop(tp, errno); if (tp != NULL) return (inp); else return (NULL); } /* * When `need fragmentation' ICMP is received, update our idea of the MSS * based on the new value. Also nudge TCP to send something, since we * know the packet we just sent was dropped. * This duplicates some code in the tcp_mss() function in tcp_input.c. */ static struct inpcb * tcp_mtudisc_notify(struct inpcb *inp, int error) { return (tcp_mtudisc(inp, -1)); } struct inpcb * tcp_mtudisc(struct inpcb *inp, int mtuoffer) { struct tcpcb *tp; struct socket *so; INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_mtudisc: tp == NULL")); tcp_mss_update(tp, -1, mtuoffer, NULL, NULL); so = inp->inp_socket; SOCKBUF_LOCK(&so->so_snd); /* If the mss is larger than the socket buffer, decrease the mss. */ if (so->so_snd.sb_hiwat < tp->t_maxseg) tp->t_maxseg = so->so_snd.sb_hiwat; SOCKBUF_UNLOCK(&so->so_snd); TCPSTAT_INC(tcps_mturesent); tp->t_rtttime = 0; tp->snd_nxt = tp->snd_una; tcp_free_sackholes(tp); tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_SACK_PERMIT) EXIT_FASTRECOVERY(tp->t_flags); tcp_output_send(tp); return (inp); } /* * Look-up the routing entry to the peer of this inpcb. If no route * is found and it cannot be allocated, then return 0. This routine * is called by TCP routines that access the rmx structure and by * tcp_mss_update to get the peer/interface MTU. */ u_long tcp_maxmtu(struct in_conninfo *inc, int *flags) { struct route sro; struct sockaddr_in *dst; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu with NULL in_conninfo pointer")); bzero(&sro, sizeof(sro)); if (inc->inc_faddr.s_addr != INADDR_ANY) { dst = (struct sockaddr_in *)&sro.ro_dst; dst->sin_family = AF_INET; dst->sin_len = sizeof(*dst); dst->sin_addr = inc->inc_faddr; in_rtalloc_ign(&sro, 0, inc->inc_fibnum); } if (sro.ro_rt != NULL) { ifp = sro.ro_rt->rt_ifp; if (sro.ro_rt->rt_rmx.rmx_mtu == 0) maxmtu = ifp->if_mtu; else maxmtu = min(sro.ro_rt->rt_rmx.rmx_mtu, ifp->if_mtu); /* Report additional interface capabilities. */ if (flags != NULL) { if (ifp->if_capenable & IFCAP_TSO4 && ifp->if_hwassist & CSUM_TSO) *flags |= CSUM_TSO; } RTFREE(sro.ro_rt); } return (maxmtu); } #ifdef INET6 u_long tcp_maxmtu6(struct in_conninfo *inc, int *flags) { struct route_in6 sro6; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu6 with NULL in_conninfo pointer")); bzero(&sro6, sizeof(sro6)); if (!IN6_IS_ADDR_UNSPECIFIED(&inc->inc6_faddr)) { sro6.ro_dst.sin6_family = AF_INET6; sro6.ro_dst.sin6_len = sizeof(struct sockaddr_in6); sro6.ro_dst.sin6_addr = inc->inc6_faddr; in6_rtalloc_ign(&sro6, 0, inc->inc_fibnum); } if (sro6.ro_rt != NULL) { ifp = sro6.ro_rt->rt_ifp; if (sro6.ro_rt->rt_rmx.rmx_mtu == 0) maxmtu = IN6_LINKMTU(sro6.ro_rt->rt_ifp); else maxmtu = min(sro6.ro_rt->rt_rmx.rmx_mtu, IN6_LINKMTU(sro6.ro_rt->rt_ifp)); /* Report additional interface capabilities. */ if (flags != NULL) { if (ifp->if_capenable & IFCAP_TSO6 && ifp->if_hwassist & CSUM_TSO) *flags |= CSUM_TSO; } RTFREE(sro6.ro_rt); } return (maxmtu); } #endif /* INET6 */ #ifdef IPSEC /* compute ESP/AH header size for TCP, including outer IP header. */ size_t ipsec_hdrsiz_tcp(struct tcpcb *tp) { struct inpcb *inp; struct mbuf *m; size_t hdrsiz; struct ip *ip; #ifdef INET6 struct ip6_hdr *ip6; #endif struct tcphdr *th; if ((tp == NULL) || ((inp = tp->t_inpcb) == NULL)) return (0); MGETHDR(m, M_DONTWAIT, MT_DATA); if (!m) return (0); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)(ip6 + 1); m->m_pkthdr.len = m->m_len = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); tcpip_fillheaders(inp, ip6, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } else #endif /* INET6 */ { ip = mtod(m, struct ip *); th = (struct tcphdr *)(ip + 1); m->m_pkthdr.len = m->m_len = sizeof(struct tcpiphdr); tcpip_fillheaders(inp, ip, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } m_free(m); return (hdrsiz); } #endif /* IPSEC */ /* * TCP BANDWIDTH DELAY PRODUCT WINDOW LIMITING * * This code attempts to calculate the bandwidth-delay product as a * means of determining the optimal window size to maximize bandwidth, * minimize RTT, and avoid the over-allocation of buffers on interfaces and * routers. This code also does a fairly good job keeping RTTs in check * across slow links like modems. We implement an algorithm which is very * similar (but not meant to be) TCP/Vegas. The code operates on the * transmitter side of a TCP connection and so only effects the transmit * side of the connection. * * BACKGROUND: TCP makes no provision for the management of buffer space * at the end points or at the intermediate routers and switches. A TCP * stream, whether using NewReno or not, will eventually buffer as * many packets as it is able and the only reason this typically works is * due to the fairly small default buffers made available for a connection * (typicaly 16K or 32K). As machines use larger windows and/or window * scaling it is now fairly easy for even a single TCP connection to blow-out * all available buffer space not only on the local interface, but on * intermediate routers and switches as well. NewReno makes a misguided * attempt to 'solve' this problem by waiting for an actual failure to occur, * then backing off, then steadily increasing the window again until another * failure occurs, ad-infinitum. This results in terrible oscillation that * is only made worse as network loads increase and the idea of intentionally * blowing out network buffers is, frankly, a terrible way to manage network * resources. * * It is far better to limit the transmit window prior to the failure * condition being achieved. There are two general ways to do this: First * you can 'scan' through different transmit window sizes and locate the * point where the RTT stops increasing, indicating that you have filled the * pipe, then scan backwards until you note that RTT stops decreasing, then * repeat ad-infinitum. This method works in principle but has severe * implementation issues due to RTT variances, timer granularity, and * instability in the algorithm which can lead to many false positives and * create oscillations as well as interact badly with other TCP streams * implementing the same algorithm. * * The second method is to limit the window to the bandwidth delay product * of the link. This is the method we implement. RTT variances and our * own manipulation of the congestion window, bwnd, can potentially * destabilize the algorithm. For this reason we have to stabilize the * elements used to calculate the window. We do this by using the minimum * observed RTT, the long term average of the observed bandwidth, and * by adding two segments worth of slop. It isn't perfect but it is able * to react to changing conditions and gives us a very stable basis on * which to extend the algorithm. */ void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq) { u_long bw; u_long bwnd; int save_ticks; INP_WLOCK_ASSERT(tp->t_inpcb); /* * If inflight_enable is disabled in the middle of a tcp connection, * make sure snd_bwnd is effectively disabled. */ if (V_tcp_inflight_enable == 0 || tp->t_rttlow < V_tcp_inflight_rttthresh) { tp->snd_bwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->snd_bandwidth = 0; return; } /* * Figure out the bandwidth. Due to the tick granularity this * is a very rough number and it MUST be averaged over a fairly * long period of time. XXX we need to take into account a link * that is not using all available bandwidth, but for now our * slop will ramp us up if this case occurs and the bandwidth later * increases. * * Note: if ticks rollover 'bw' may wind up negative. We must * effectively reset t_bw_rtttime for this case. */ save_ticks = ticks; if ((u_int)(save_ticks - tp->t_bw_rtttime) < 1) return; bw = (int64_t)(ack_seq - tp->t_bw_rtseq) * hz / (save_ticks - tp->t_bw_rtttime); tp->t_bw_rtttime = save_ticks; tp->t_bw_rtseq = ack_seq; if (tp->t_bw_rtttime == 0 || (int)bw < 0) return; bw = ((int64_t)tp->snd_bandwidth * 15 + bw) >> 4; tp->snd_bandwidth = bw; /* * Calculate the semi-static bandwidth delay product, plus two maximal * segments. The additional slop puts us squarely in the sweet * spot and also handles the bandwidth run-up case and stabilization. * Without the slop we could be locking ourselves into a lower * bandwidth. * * Situations Handled: * (1) Prevents over-queueing of packets on LANs, especially on * high speed LANs, allowing larger TCP buffers to be * specified, and also does a good job preventing * over-queueing of packets over choke points like modems * (at least for the transmit side). * * (2) Is able to handle changing network loads (bandwidth * drops so bwnd drops, bandwidth increases so bwnd * increases). * * (3) Theoretically should stabilize in the face of multiple * connections implementing the same algorithm (this may need * a little work). * * (4) Stability value (defaults to 20 = 2 maximal packets) can * be adjusted with a sysctl but typically only needs to be * on very slow connections. A value no smaller then 5 * should be used, but only reduce this default if you have * no other choice. */ #define USERTT ((tp->t_srtt + tp->t_rttbest) / 2) bwnd = (int64_t)bw * USERTT / (hz << TCP_RTT_SHIFT) + V_tcp_inflight_stab * tp->t_maxseg / 10; #undef USERTT if (tcp_inflight_debug > 0) { static int ltime; if ((u_int)(ticks - ltime) >= hz / tcp_inflight_debug) { ltime = ticks; printf("%p bw %ld rttbest %d srtt %d bwnd %ld\n", tp, bw, tp->t_rttbest, tp->t_srtt, bwnd ); } } if ((long)bwnd < V_tcp_inflight_min) bwnd = V_tcp_inflight_min; if (bwnd > V_tcp_inflight_max) bwnd = V_tcp_inflight_max; if ((long)bwnd < tp->t_maxseg * 2) bwnd = tp->t_maxseg * 2; tp->snd_bwnd = bwnd; } #ifdef TCP_SIGNATURE /* * Callback function invoked by m_apply() to digest TCP segment data * contained within an mbuf chain. */ static int tcp_signature_apply(void *fstate, void *data, u_int len) { MD5Update(fstate, (u_char *)data, len); return (0); } /* * Compute TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * _unused * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * direction direction of flow (IPSEC_DIR_INBOUND or OUTBOUND) * * We do this over ip, tcphdr, segment data, and the key in the SADB. * When called from tcp_input(), we can be sure that th_sum has been * zeroed out and verified already. * * Return 0 if successful, otherwise return -1. * * XXX The key is retrieved from the system's PF_KEY SADB, by keying a * search with the destination IP address, and a 'magic SPI' to be * determined by the application. This is hardcoded elsewhere to 1179 * right now. Another branch of this code exists which uses the SPD to * specify per-application flows but it is unstable. */ int tcp_signature_compute(struct mbuf *m, int _unused, int len, int optlen, u_char *buf, u_int direction) { union sockaddr_union dst; struct ippseudo ippseudo; MD5_CTX ctx; int doff; struct ip *ip; struct ipovly *ipovly; struct secasvar *sav; struct tcphdr *th; #ifdef INET6 struct ip6_hdr *ip6; struct in6_addr in6; char ip6buf[INET6_ADDRSTRLEN]; uint32_t plen; uint16_t nhdr; #endif u_short savecsum; KASSERT(m != NULL, ("NULL mbuf chain")); KASSERT(buf != NULL, ("NULL signature pointer")); /* Extract the destination from the IP header in the mbuf. */ bzero(&dst, sizeof(union sockaddr_union)); ip = mtod(m, struct ip *); #ifdef INET6 ip6 = NULL; /* Make the compiler happy. */ #endif switch (ip->ip_v) { case IPVERSION: dst.sa.sa_len = sizeof(struct sockaddr_in); dst.sa.sa_family = AF_INET; dst.sin.sin_addr = (direction == IPSEC_DIR_INBOUND) ? ip->ip_src : ip->ip_dst; break; #ifdef INET6 case (IPV6_VERSION >> 4): ip6 = mtod(m, struct ip6_hdr *); dst.sa.sa_len = sizeof(struct sockaddr_in6); dst.sa.sa_family = AF_INET6; dst.sin6.sin6_addr = (direction == IPSEC_DIR_INBOUND) ? ip6->ip6_src : ip6->ip6_dst; break; #endif default: return (EINVAL); /* NOTREACHED */ break; } /* Look up an SADB entry which matches the address of the peer. */ sav = KEY_ALLOCSA(&dst, IPPROTO_TCP, htonl(TCP_SIG_SPI)); if (sav == NULL) { ipseclog((LOG_ERR, "%s: SADB lookup failed for %s\n", __func__, (ip->ip_v == IPVERSION) ? inet_ntoa(dst.sin.sin_addr) : #ifdef INET6 (ip->ip_v == (IPV6_VERSION >> 4)) ? ip6_sprintf(ip6buf, &dst.sin6.sin6_addr) : #endif "(unsupported)")); return (EINVAL); } MD5Init(&ctx); /* * Step 1: Update MD5 hash with IP(v6) pseudo-header. * * XXX The ippseudo header MUST be digested in network byte order, * or else we'll fail the regression test. Assume all fields we've * been doing arithmetic on have been in host byte order. * XXX One cannot depend on ipovly->ih_len here. When called from * tcp_output(), the underlying ip_len member has not yet been set. */ switch (ip->ip_v) { case IPVERSION: ipovly = (struct ipovly *)ip; ippseudo.ippseudo_src = ipovly->ih_src; ippseudo.ippseudo_dst = ipovly->ih_dst; ippseudo.ippseudo_pad = 0; ippseudo.ippseudo_p = IPPROTO_TCP; ippseudo.ippseudo_len = htons(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&ippseudo, sizeof(struct ippseudo)); th = (struct tcphdr *)((u_char *)ip + sizeof(struct ip)); doff = sizeof(struct ip) + sizeof(struct tcphdr) + optlen; break; #ifdef INET6 /* * RFC 2385, 2.0 Proposal * For IPv6, the pseudo-header is as described in RFC 2460, namely the * 128-bit source IPv6 address, 128-bit destination IPv6 address, zero- * extended next header value (to form 32 bits), and 32-bit segment * length. * Note: Upper-Layer Packet Length comes before Next Header. */ case (IPV6_VERSION >> 4): in6 = ip6->ip6_src; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); in6 = ip6->ip6_dst; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); plen = htonl(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&plen, sizeof(uint32_t)); nhdr = 0; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); nhdr = IPPROTO_TCP; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); th = (struct tcphdr *)((u_char *)ip6 + sizeof(struct ip6_hdr)); doff = sizeof(struct ip6_hdr) + sizeof(struct tcphdr) + optlen; break; #endif default: return (EINVAL); /* NOTREACHED */ break; } /* * Step 2: Update MD5 hash with TCP header, excluding options. * The TCP checksum must be set to zero. */ savecsum = th->th_sum; th->th_sum = 0; MD5Update(&ctx, (char *)th, sizeof(struct tcphdr)); th->th_sum = savecsum; /* * Step 3: Update MD5 hash with TCP segment data. * Use m_apply() to avoid an early m_pullup(). */ if (len > 0) m_apply(m, doff, len, tcp_signature_apply, &ctx); /* * Step 4: Update MD5 hash with shared secret. */ MD5Update(&ctx, sav->key_auth->key_data, _KEYLEN(sav->key_auth)); MD5Final(buf, &ctx); key_sa_recordxfer(sav, m); KEY_FREESAV(&sav); return (0); } /* * Verify the TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * direction direction of flow (IPSEC_DIR_INBOUND or OUTBOUND) * * Return 1 if successful, otherwise return 0. */ int tcp_signature_verify(struct mbuf *m, int off0, int tlen, int optlen, struct tcpopt *to, struct tcphdr *th, u_int tcpbflag) { char tmpdigest[TCP_SIGLEN]; if (tcp_sig_checksigs == 0) return (1); if ((tcpbflag & TF_SIGNATURE) == 0) { if ((to->to_flags & TOF_SIGNATURE) != 0) { /* * If this socket is not expecting signature but * the segment contains signature just fail. */ TCPSTAT_INC(tcps_sig_err_sigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } /* Signature is not expected, and not present in segment. */ return (1); } /* * If this socket is expecting signature but the segment does not * contain any just fail. */ if ((to->to_flags & TOF_SIGNATURE) == 0) { TCPSTAT_INC(tcps_sig_err_nosigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (tcp_signature_compute(m, off0, tlen, optlen, &tmpdigest[0], IPSEC_DIR_INBOUND) == -1) { TCPSTAT_INC(tcps_sig_err_buildsig); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (bcmp(to->to_signature, &tmpdigest[0], TCP_SIGLEN) != 0) { TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } TCPSTAT_INC(tcps_sig_rcvgoodsig); return (1); } #endif /* TCP_SIGNATURE */ static int sysctl_drop(SYSCTL_HANDLER_ARGS) { /* addrs[0] is a foreign socket, addrs[1] is a local one. */ struct sockaddr_storage addrs[2]; struct inpcb *inp; struct tcpcb *tp; struct tcptw *tw; struct sockaddr_in *fin, *lin; #ifdef INET6 struct sockaddr_in6 *fin6, *lin6; #endif int error; inp = NULL; fin = lin = NULL; #ifdef INET6 fin6 = lin6 = NULL; #endif error = 0; if (req->oldptr != NULL || req->oldlen != 0) return (EINVAL); if (req->newptr == NULL) return (EPERM); if (req->newlen < sizeof(addrs)) return (ENOMEM); error = SYSCTL_IN(req, &addrs, sizeof(addrs)); if (error) return (error); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: fin6 = (struct sockaddr_in6 *)&addrs[0]; lin6 = (struct sockaddr_in6 *)&addrs[1]; if (fin6->sin6_len != sizeof(struct sockaddr_in6) || lin6->sin6_len != sizeof(struct sockaddr_in6)) return (EINVAL); if (IN6_IS_ADDR_V4MAPPED(&fin6->sin6_addr)) { if (!IN6_IS_ADDR_V4MAPPED(&lin6->sin6_addr)) return (EINVAL); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[0]); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[1]); fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; break; } error = sa6_embedscope(fin6, V_ip6_use_defzone); if (error) return (error); error = sa6_embedscope(lin6, V_ip6_use_defzone); if (error) return (error); break; #endif case AF_INET: fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; if (fin->sin_len != sizeof(struct sockaddr_in) || lin->sin_len != sizeof(struct sockaddr_in)) return (EINVAL); break; default: return (EINVAL); } INP_INFO_WLOCK(&V_tcbinfo); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: inp = in6_pcblookup_hash(&V_tcbinfo, &fin6->sin6_addr, fin6->sin6_port, &lin6->sin6_addr, lin6->sin6_port, 0, NULL); break; #endif case AF_INET: inp = in_pcblookup_hash(&V_tcbinfo, fin->sin_addr, fin->sin_port, lin->sin_addr, lin->sin_port, 0, NULL); break; } if (inp != NULL) { INP_WLOCK(inp); if (inp->inp_flags & INP_TIMEWAIT) { /* * XXXRW: There currently exists a state where an * inpcb is present, but its timewait state has been * discarded. For now, don't allow dropping of this * type of inpcb. */ tw = intotw(inp); if (tw != NULL) tcp_twclose(tw, 0); else INP_WUNLOCK(inp); } else if (!(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket->so_options & SO_ACCEPTCONN)) { tp = intotcpcb(inp); tp = tcp_drop(tp, ECONNABORTED); if (tp != NULL) INP_WUNLOCK(inp); } else INP_WUNLOCK(inp); } else error = ESRCH; INP_INFO_WUNLOCK(&V_tcbinfo); return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_DROP, drop, CTLTYPE_STRUCT|CTLFLAG_WR|CTLFLAG_SKIP, NULL, 0, sysctl_drop, "", "Drop TCP connection"); /* * Generate a standardized TCP log line for use throughout the * tcp subsystem. Memory allocation is done with M_NOWAIT to * allow use in the interrupt context. * * NB: The caller MUST free(s, M_TCPLOG) the returned string. * NB: The function may return NULL if memory allocation failed. * * Due to header inclusion and ordering limitations the struct ip * and ip6_hdr pointers have to be passed as void pointers. */ char * tcp_log_vain(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_in_vain == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } char * tcp_log_addrs(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_debug == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { char *s, *sp; size_t size; struct ip *ip; #ifdef INET6 const struct ip6_hdr *ip6; ip6 = (const struct ip6_hdr *)ip6hdr; #endif /* INET6 */ ip = (struct ip *)ip4hdr; /* * The log line looks like this: * "TCP: [1.2.3.4]:50332 to [1.2.3.4]:80 tcpflags 0x2" */ size = sizeof("TCP: []:12345 to []:12345 tcpflags 0x2<>") + sizeof(PRINT_TH_FLAGS) + 1 + #ifdef INET6 2 * INET6_ADDRSTRLEN; #else 2 * INET_ADDRSTRLEN; #endif /* INET6 */ s = malloc(size, M_TCPLOG, M_ZERO|M_NOWAIT); if (s == NULL) return (NULL); strcat(s, "TCP: ["); sp = s + strlen(s); if (inc && ((inc->inc_flags & INC_ISIPV6) == 0)) { inet_ntoa_r(inc->inc_faddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); inet_ntoa_r(inc->inc_laddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); #ifdef INET6 } else if (inc) { ip6_sprintf(sp, &inc->inc6_faddr); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); ip6_sprintf(sp, &inc->inc6_laddr); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); } else if (ip6 && th) { ip6_sprintf(sp, &ip6->ip6_src); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); ip6_sprintf(sp, &ip6->ip6_dst); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); #endif /* INET6 */ } else if (ip && th) { inet_ntoa_r(ip->ip_src, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); inet_ntoa_r(ip->ip_dst, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); } else { free(s, M_TCPLOG); return (NULL); } sp = s + strlen(s); if (th) sprintf(sp, " tcpflags 0x%b", th->th_flags, PRINT_TH_FLAGS); if (*(s + size - 1) != '\0') panic("%s: string too long", __func__); return (s); } Index: stable/8/sys/netinet/tcp_var.h =================================================================== --- stable/8/sys/netinet/tcp_var.h (revision 285976) +++ stable/8/sys/netinet/tcp_var.h (revision 285977) @@ -1,730 +1,727 @@ /*- * Copyright (c) 1982, 1986, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 * $FreeBSD$ */ #ifndef _NETINET_TCP_VAR_H_ #define _NETINET_TCP_VAR_H_ #include #ifdef _KERNEL #include /* * Kernel variables for tcp. */ VNET_DECLARE(int, tcp_do_rfc1323); #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) #endif /* _KERNEL */ /* TCP segment queue entry */ struct tseg_qent { LIST_ENTRY(tseg_qent) tqe_q; int tqe_len; /* TCP segment data length */ struct tcphdr *tqe_th; /* a pointer to tcp header */ struct mbuf *tqe_m; /* mbuf contains packet */ }; LIST_HEAD(tsegqe_head, tseg_qent); struct sackblk { tcp_seq start; /* start seq no. of sack block */ tcp_seq end; /* end seq no. */ }; struct sackhole { tcp_seq start; /* start seq no. of hole */ tcp_seq end; /* end seq no. */ tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ }; struct sackhint { struct sackhole *nexthole; int sack_bytes_rexmit; tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */ }; struct tcptemp { u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ struct tcphdr tt_t; }; #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ /* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ #ifdef INET6 #define ND6_HINT(tp) \ do { \ if ((tp) && (tp)->t_inpcb && \ ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ nd6_nud_hint(NULL, NULL, 0); \ } while (0) #else #define ND6_HINT(tp) #endif /* * Tcp control block, one per tcp; fields: * Organized for 16 byte cacheline efficiency. */ struct tcpcb { struct tsegqe_head t_segq; /* segment reassembly queue */ void *t_pspare[2]; /* new reassembly queue */ int t_segqlen; /* segment reassembly queue length */ int t_dupacks; /* consecutive dup acks recd */ struct tcp_timer *t_timers; /* All the TCP timers in one struct */ struct inpcb *t_inpcb; /* back pointer to internet pcb */ int t_state; /* state of this connection */ u_int t_flags; struct vnet *t_vnet; /* back pointer to parent vnet */ tcp_seq snd_una; /* send unacknowledged */ tcp_seq snd_max; /* highest sequence number sent; * used to recognize retransmits */ tcp_seq snd_nxt; /* send next */ tcp_seq snd_up; /* send urgent pointer */ tcp_seq snd_wl1; /* window update seg seq number */ tcp_seq snd_wl2; /* window update seg ack number */ tcp_seq iss; /* initial send sequence number */ tcp_seq irs; /* initial receive sequence number */ tcp_seq rcv_nxt; /* receive next */ tcp_seq rcv_adv; /* advertised window */ u_long rcv_wnd; /* receive window */ tcp_seq rcv_up; /* receive urgent pointer */ u_long snd_wnd; /* send window */ u_long snd_cwnd; /* congestion-controlled window */ u_long snd_bwnd; /* bandwidth-controlled window */ u_long snd_ssthresh; /* snd_cwnd size threshold for * for slow start exponential to * linear switch */ u_long snd_bandwidth; /* calculated bandwidth or 0 */ tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ u_int t_maxopd; /* mss plus options */ u_int t_rcvtime; /* inactivity time */ u_int t_starttime; /* time connection was established */ u_int t_rtttime; /* RTT measurement start time */ tcp_seq t_rtseq; /* sequence number being timed */ u_int t_bw_rtttime; /* used for bandwidth calculation */ tcp_seq t_bw_rtseq; /* used for bandwidth calculation */ int t_rxtcur; /* current retransmit value (ticks) */ u_int t_maxseg; /* maximum segment size */ int t_srtt; /* smoothed round-trip time */ int t_rttvar; /* variance in round-trip time */ int t_rxtshift; /* log(2) of rexmt exp. backoff */ u_int t_rttmin; /* minimum rtt allowed */ u_int t_rttbest; /* best rtt we've seen */ u_long t_rttupdated; /* number of times rtt sampled */ u_long max_sndwnd; /* largest window peer has offered */ int t_softerror; /* possible error not yet reported */ /* out-of-band data */ char t_oobflags; /* have some */ char t_iobc; /* input character */ /* RFC 1323 variables */ u_char snd_scale; /* window scaling for send window */ u_char rcv_scale; /* window scaling for recv window */ u_char request_r_scale; /* pending window scaling */ u_int32_t ts_recent; /* timestamp echo data */ u_int ts_recent_age; /* when last updated */ u_int32_t ts_offset; /* our timestamp offset */ tcp_seq last_ack_sent; /* experimental */ u_long snd_cwnd_prev; /* cwnd prior to retransmit */ u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ u_int t_badrxtwin; /* window for retransmit recovery */ u_char snd_limited; /* segments limited transmitted */ /* SACK related state */ int snd_numholes; /* number of holes seen by sender */ TAILQ_HEAD(sackhole_head, sackhole) snd_holes; /* SACK scoreboard (sorted) */ tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ int rcv_numsacks; /* # distinct sack blks present */ struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ tcp_seq sack_newdata; /* New data xmitted in this recovery episode starts at this seq number */ struct sackhint sackhint; /* SACK scoreboard hint */ int t_rttlow; /* smallest observerved RTT */ u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ int rfbuf_cnt; /* recv buffer autoscaling byte count */ struct toe_usrreqs *t_tu; /* offload operations vector */ void *t_toe; /* TOE pcb pointer */ int t_bytes_acked; /* # bytes acked during current RTT */ int t_sndzerowin; /* zero-window updates sent */ struct cc_algo *cc_algo; /* congestion control algorithm */ struct cc_var *ccv; /* congestion control specific vars */ struct osd *osd; /* storage for Khelp module data */ void *t_pspare2[3]; /* 3 TBD */ uint64_t _pad[10]; /* 7 UTO, 3 TBD (1-2 CC/RTT?) */ uint64_t t_sndrexmitpack;/* retransmit packets sent */ uint64_t t_rcvoopack; /* out-of-order packets received */ }; /* * Flags and utility macros for the t_flags field. */ #define TF_ACKNOW 0x000001 /* ack peer immediately */ #define TF_DELACK 0x000002 /* ack, but try to delay it */ #define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ #define TF_NOOPT 0x000008 /* don't use tcp options */ #define TF_SENTFIN 0x000010 /* have sent FIN */ #define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ #define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ #define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ #define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ #define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ #define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ #define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ #define TF_NOPUSH 0x001000 /* don't push */ #define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ #define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ #define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ #define TF_LASTIDLE 0x040000 /* connection was previously idle */ #define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ #define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ #define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ #define TF_FORCEDATA 0x800000 /* force out a byte */ #define TF_TSO 0x1000000 /* TSO enabled on this connection */ #define TF_TOE 0x2000000 /* this connection is offloaded */ #define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ #define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ #define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) /* * Flags for the t_oobflags field. */ #define TCPOOB_HAVEDATA 0x01 #define TCPOOB_HADDATA 0x02 #ifdef TCP_SIGNATURE /* * Defines which are needed by the xform_tcp module and tcp_[in|out]put * for SADB verification and lookup. */ #define TCP_SIGLEN 16 /* length of computed digest in bytes */ #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ /* * Only a single SA per host may be specified at this time. An SPI is * needed in order for the KEY_ALLOCSA() lookup to work. */ #define TCP_SIG_SPI 0x1000 #endif /* TCP_SIGNATURE */ /* * Structure to hold TCP options that are only used during segment * processing (in tcp_input), but not held in the tcpcb. * It's basically used to reduce the number of parameters * to tcp_dooptions and tcp_addoptions. * The binary order of the to_flags is relevant for packing of the * options in tcp_addoptions. */ struct tcpopt { u_int64_t to_flags; /* which options are present */ #define TOF_MSS 0x0001 /* maximum segment size */ #define TOF_SCALE 0x0002 /* window scaling */ #define TOF_SACKPERM 0x0004 /* SACK permitted */ #define TOF_TS 0x0010 /* timestamp */ #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ #define TOF_SACK 0x0080 /* Peer sent SACK option */ #define TOF_MAXOPT 0x0100 u_int32_t to_tsval; /* new timestamp */ u_int32_t to_tsecr; /* reflected timestamp */ u_char *to_sacks; /* pointer to the first SACK blocks */ u_char *to_signature; /* pointer to the TCP-MD5 signature */ u_int16_t to_mss; /* maximum segment size */ u_int8_t to_wscale; /* window scaling */ u_int8_t to_nsacks; /* number of SACK blocks */ }; /* * Flags for tcp_dooptions. */ #define TO_SYN 0x01 /* parse SYN-only options */ struct hc_metrics_lite { /* must stay in sync with hc_metrics */ u_long rmx_mtu; /* MTU for this path */ u_long rmx_ssthresh; /* outbound gateway buffer limit */ u_long rmx_rtt; /* estimated round trip time */ u_long rmx_rttvar; /* estimated rtt variance */ u_long rmx_bandwidth; /* estimated bandwidth */ u_long rmx_cwnd; /* congestion window */ u_long rmx_sendpipe; /* outbound delay-bandwidth product */ u_long rmx_recvpipe; /* inbound delay-bandwidth product */ }; #ifndef _NETINET_IN_PCB_H_ struct in_conninfo; #endif /* _NETINET_IN_PCB_H_ */ struct tcptw { struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ tcp_seq snd_nxt; tcp_seq rcv_nxt; tcp_seq iss; tcp_seq irs; u_short last_win; /* cached window value */ u_short tw_so_options; /* copy of so_options */ struct ucred *tw_cred; /* user credentials */ u_int32_t t_recent; u_int32_t ts_offset; /* our timestamp offset */ u_int t_starttime; int tw_time; TAILQ_ENTRY(tcptw) tw_2msl; }; #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) /* * The smoothed round-trip time and estimated variance * are stored as fixed point numbers scaled by the values below. * For convenience, these scales are also used in smoothing the average * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). * With these scales, srtt has 3 bits to the right of the binary point, * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the * binary point, and is smoothed with an ALPHA of 0.75. */ #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ /* * The initial retransmission should happen at rtt + 4 * rttvar. * Because of the way we do the smoothing, srtt and rttvar * will each average +1/2 tick of bias. When we compute * the retransmit timer, we want 1/2 tick of rounding and * 1 extra tick because of +-1/2 tick uncertainty in the * firing of the timer. The bias will give us exactly the * 1.5 tick we need. But, because the bias is * statistical, we have to test that we don't drop below * the minimum feasible timer (which is 2 ticks). * This version of the macro adapted from a paper by Lawrence * Brakmo and Larry Peterson which outlines a problem caused * by insufficient precision in the original implementation, * which results in inappropriately large RTO values for very * fast networks. */ #define TCP_REXMTVAL(tp) \ max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) /* * TCP statistics. * Many of these should be kept per connection, * but that's inconvenient at the moment. */ struct tcpstat { u_long tcps_connattempt; /* connections initiated */ u_long tcps_accepts; /* connections accepted */ u_long tcps_connects; /* connections established */ u_long tcps_drops; /* connections dropped */ u_long tcps_conndrops; /* embryonic connections dropped */ u_long tcps_minmssdrops; /* average minmss too low drops */ u_long tcps_closed; /* conn. closed (includes drops) */ u_long tcps_segstimed; /* segs where we tried to get rtt */ u_long tcps_rttupdated; /* times we succeeded */ u_long tcps_delack; /* delayed acks sent */ u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ u_long tcps_rexmttimeo; /* retransmit timeouts */ u_long tcps_persisttimeo; /* persist timeouts */ u_long tcps_keeptimeo; /* keepalive timeouts */ u_long tcps_keepprobe; /* keepalive probes sent */ u_long tcps_keepdrops; /* connections dropped in keepalive */ u_long tcps_sndtotal; /* total packets sent */ u_long tcps_sndpack; /* data packets sent */ u_long tcps_sndbyte; /* data bytes sent */ u_long tcps_sndrexmitpack; /* data packets retransmitted */ u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ u_long tcps_sndrexmitbad; /* unnecessary packet retransmissions */ u_long tcps_sndacks; /* ack-only packets sent */ u_long tcps_sndprobe; /* window probes sent */ u_long tcps_sndurg; /* packets sent with URG only */ u_long tcps_sndwinup; /* window update-only packets sent */ u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ u_long tcps_rcvtotal; /* total packets received */ u_long tcps_rcvpack; /* packets received in sequence */ u_long tcps_rcvbyte; /* bytes received in sequence */ u_long tcps_rcvbadsum; /* packets received with ccksum errs */ u_long tcps_rcvbadoff; /* packets received with bad offset */ u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ u_long tcps_rcvshort; /* packets received too short */ u_long tcps_rcvduppack; /* duplicate-only packets received */ u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ u_long tcps_rcvpartduppack; /* packets with some duplicate data */ u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ u_long tcps_rcvoopack; /* out-of-order packets received */ u_long tcps_rcvoobyte; /* out-of-order bytes received */ u_long tcps_rcvpackafterwin; /* packets with data after window */ u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ u_long tcps_rcvafterclose; /* packets rcvd after "close" */ u_long tcps_rcvwinprobe; /* rcvd window probe packets */ u_long tcps_rcvdupack; /* rcvd duplicate acks */ u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ u_long tcps_rcvackpack; /* rcvd ack packets */ u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ u_long tcps_rcvwinupd; /* rcvd window update packets */ u_long tcps_pawsdrop; /* segments dropped due to PAWS */ u_long tcps_predack; /* times hdr predict ok for acks */ u_long tcps_preddat; /* times hdr predict ok for data pkts */ u_long tcps_pcbcachemiss; u_long tcps_cachedrtt; /* times cached RTT in route updated */ u_long tcps_cachedrttvar; /* times cached rttvar updated */ u_long tcps_cachedssthresh; /* times cached ssthresh updated */ u_long tcps_usedrtt; /* times RTT initialized from route */ u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */ u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/ u_long tcps_persistdrop; /* timeout in persist state */ u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */ u_long tcps_mturesent; /* resends due to MTU discovery */ u_long tcps_listendrop; /* listen queue overflows */ u_long tcps_badrst; /* ignored RSTs in the window */ u_long tcps_sc_added; /* entry added to syncache */ u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */ u_long tcps_sc_dupsyn; /* duplicate SYN packet */ u_long tcps_sc_dropped; /* could not reply to packet */ u_long tcps_sc_completed; /* successful extraction of entry */ u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */ u_long tcps_sc_reset; /* RST removed entry from syncache */ u_long tcps_sc_stale; /* timed out or listen socket gone */ u_long tcps_sc_aborted; /* syncache entry aborted */ u_long tcps_sc_badack; /* removed due to bad ACK */ u_long tcps_sc_unreach; /* ICMP unreachable received */ u_long tcps_sc_zonefail; /* zalloc() failed */ u_long tcps_sc_sendcookie; /* SYN cookie sent */ u_long tcps_sc_recvcookie; /* SYN cookie received */ u_long tcps_hc_added; /* entry added to hostcache */ u_long tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */ u_long tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ /* SACK related stats */ u_long tcps_sack_recovery_episode; /* SACK recovery episodes */ u_long tcps_sack_rexmits; /* SACK rexmit segments */ u_long tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ u_long tcps_sack_rcv_blocks; /* SACK blocks (options) received */ u_long tcps_sack_send_blocks; /* SACK blocks (options) sent */ u_long tcps_sack_sboverflow; /* times scoreboard overflowed */ /* ECN related stats */ u_long tcps_ecn_ce; /* ECN Congestion Experienced */ u_long tcps_ecn_ect0; /* ECN Capable Transport */ u_long tcps_ecn_ect1; /* ECN Capable Transport */ u_long tcps_ecn_shs; /* ECN successful handshakes */ u_long tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ /* TCP_SIGNATURE related stats */ u_long tcps_sig_rcvgoodsig; /* Total matching signature received */ u_long tcps_sig_rcvbadsig; /* Total bad signature received */ u_long tcps_sig_err_buildsig; /* Mismatching signature received */ u_long tcps_sig_err_sigopt; /* No signature expected by socket */ u_long tcps_sig_err_nosigopt; /* No signature provided by segment */ u_long _pad[7]; /* 6 UTO, 1 TBD */ }; #ifdef _KERNEL /* * In-kernel consumers can use these accessor macros directly to update * stats. */ #define TCPSTAT_ADD(name, val) V_tcpstat.name += (val) #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) /* * Kernel module consumers must use this accessor macro. */ void kmod_tcpstat_inc(int statnum); #define KMOD_TCPSTAT_INC(name) \ kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(u_long)) /* * TCP specific helper hook point identifiers. */ #define HHOOK_TCP_EST_IN 0 #define HHOOK_TCP_EST_OUT 1 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT struct tcp_hhook_data { struct tcpcb *tp; struct tcphdr *th; struct tcpopt *to; long len; int tso; tcp_seq curack; }; #endif /* * TCB structure exported to user-land via sysctl(3). * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been * included. Not all of our clients do. */ #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) struct xtcpcb { size_t xt_len; struct inpcb xt_inp; struct tcpcb xt_tp; struct xsocket xt_socket; u_quad_t xt_alignment_hack; }; #endif /* * Names for TCP sysctl objects */ #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ #define TCPCTL_MSSDFLT 3 /* MSS default */ #define TCPCTL_STATS 4 /* statistics (read-only) */ #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ #define TCPCTL_SENDSPACE 8 /* send buffer space */ #define TCPCTL_RECVSPACE 9 /* receive buffer space */ #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ #define TCPCTL_DROP 15 /* drop tcp connection */ #define TCPCTL_MAXID 16 #define TCPCTL_FINWAIT2_TIMEOUT 17 #define TCPCTL_NAMES { \ { 0, 0 }, \ { "rfc1323", CTLTYPE_INT }, \ { "mssdflt", CTLTYPE_INT }, \ { "stats", CTLTYPE_STRUCT }, \ { "rttdflt", CTLTYPE_INT }, \ { "keepidle", CTLTYPE_INT }, \ { "keepintvl", CTLTYPE_INT }, \ { "sendspace", CTLTYPE_INT }, \ { "recvspace", CTLTYPE_INT }, \ { "keepinit", CTLTYPE_INT }, \ { "pcblist", CTLTYPE_STRUCT }, \ { "delacktime", CTLTYPE_INT }, \ { "v6mssdflt", CTLTYPE_INT }, \ { "maxid", CTLTYPE_INT }, \ } #ifdef _KERNEL #ifdef SYSCTL_DECL SYSCTL_DECL(_net_inet_tcp); SYSCTL_DECL(_net_inet_tcp_sack); MALLOC_DECLARE(M_TCPLOG); #endif VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ VNET_DECLARE(struct inpcbinfo, tcbinfo); VNET_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ extern int tcp_log_in_vain; VNET_DECLARE(int, tcp_mssdflt); /* XXX */ VNET_DECLARE(int, tcp_minmss); VNET_DECLARE(int, tcp_delack_enabled); VNET_DECLARE(int, tcp_do_rfc3390); VNET_DECLARE(int, path_mtu_discovery); VNET_DECLARE(int, ss_fltsz); VNET_DECLARE(int, ss_fltsz_local); VNET_DECLARE(int, tcp_do_rfc3465); VNET_DECLARE(int, tcp_abc_l_var); #define V_tcb VNET(tcb) #define V_tcbinfo VNET(tcbinfo) #define V_tcpstat VNET(tcpstat) #define V_tcp_mssdflt VNET(tcp_mssdflt) #define V_tcp_minmss VNET(tcp_minmss) #define V_tcp_delack_enabled VNET(tcp_delack_enabled) #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) #define V_path_mtu_discovery VNET(path_mtu_discovery) #define V_ss_fltsz VNET(ss_fltsz) #define V_ss_fltsz_local VNET(ss_fltsz_local) #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) #define V_tcp_abc_l_var VNET(tcp_abc_l_var) VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ #define V_tcp_do_sack VNET(tcp_do_sack) #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ VNET_DECLARE(int, tcp_ecn_maxretries); #define V_tcp_do_ecn VNET(tcp_do_ecn) #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); #define V_tcp_hhh VNET(tcp_hhh) int tcp_addoptions(struct tcpopt *, u_char *); int tcp_ccalgounload(struct cc_algo *unload_algo); struct tcpcb * tcp_close(struct tcpcb *); void tcp_discardcb(struct tcpcb *); void tcp_twstart(struct tcpcb *); #if 0 int tcp_twrecycleable(struct tcptw *tw); #endif void tcp_twclose(struct tcptw *_tw, int _reuse); void tcp_ctlinput(int, struct sockaddr *, void *); int tcp_ctloutput(struct socket *, struct sockopt *); struct tcpcb * tcp_drop(struct tcpcb *, int); void tcp_drain(void); void tcp_init(void); #ifdef VIMAGE void tcp_destroy(void); #endif void tcp_fini(void *); char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, const void *); char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, const void *); int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); -void tcp_reass_init(void); +void tcp_reass_global_init(void); void tcp_reass_flush(struct tcpcb *); -#ifdef VIMAGE -void tcp_reass_destroy(void); -#endif void tcp_input(struct mbuf *, int); u_long tcp_maxmtu(struct in_conninfo *, int *); u_long tcp_maxmtu6(struct in_conninfo *, int *); void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, int *); void tcp_mss(struct tcpcb *, int); int tcp_mssopt(struct in_conninfo *); struct inpcb * tcp_drop_syn_sent(struct inpcb *, int); struct inpcb * tcp_mtudisc(struct inpcb *, int); struct tcpcb * tcp_newtcpcb(struct inpcb *); int tcp_output(struct tcpcb *); void tcp_respond(struct tcpcb *, void *, struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); void tcp_tw_init(void); #ifdef VIMAGE void tcp_tw_destroy(void); #endif void tcp_tw_zone_change(void); int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, struct mbuf *, int); int tcp_twrespond(struct tcptw *, int); void tcp_setpersist(struct tcpcb *); #ifdef TCP_SIGNATURE int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int); int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *, struct tcphdr *, u_int); #endif void tcp_slowtimo(void); struct tcptemp * tcpip_maketemplate(struct inpcb *); void tcpip_fillheaders(struct inpcb *, void *, void *); void tcp_timer_activate(struct tcpcb *, int, u_int); int tcp_timer_active(struct tcpcb *, int); void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq); /* * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) */ void tcp_hc_init(void); #ifdef VIMAGE void tcp_hc_destroy(void); #endif void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); u_long tcp_hc_getmtu(struct in_conninfo *); void tcp_hc_updatemtu(struct in_conninfo *, u_long); void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); extern struct pr_usrreqs tcp_usrreqs; extern u_long tcp_sendspace; extern u_long tcp_recvspace; tcp_seq tcp_new_isn(struct tcpcb *); void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); void tcp_clean_sackreport(struct tcpcb *tp); void tcp_sack_adjust(struct tcpcb *tp); struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); void tcp_free_sackholes(struct tcpcb *tp); int tcp_newreno(struct tcpcb *, struct tcphdr *); u_long tcp_seq_subtract(u_long, u_long ); void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); #endif /* _KERNEL */ #endif /* _NETINET_TCP_VAR_H_ */ Index: stable/9/contrib/bind9/lib/dns/tkey.c =================================================================== --- stable/9/contrib/bind9/lib/dns/tkey.c (revision 285976) +++ stable/9/contrib/bind9/lib/dns/tkey.c (revision 285977) @@ -1,1471 +1,1472 @@ /* * Copyright (C) 2004-2015 Internet Systems Consortium, Inc. ("ISC") * Copyright (C) 1999-2001, 2003 Internet Software Consortium. * * Permission to use, copy, modify, and/or distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND ISC DISCLAIMS ALL WARRANTIES WITH * REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY * AND FITNESS. IN NO EVENT SHALL ISC BE LIABLE FOR ANY SPECIAL, DIRECT, * INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM * LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE * OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ /* * $Id$ */ /*! \file */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define TKEY_RANDOM_AMOUNT 16 #define RETERR(x) do { \ result = (x); \ if (result != ISC_R_SUCCESS) \ goto failure; \ } while (0) static void tkey_log(const char *fmt, ...) ISC_FORMAT_PRINTF(1, 2); static void tkey_log(const char *fmt, ...) { va_list ap; va_start(ap, fmt); isc_log_vwrite(dns_lctx, DNS_LOGCATEGORY_GENERAL, DNS_LOGMODULE_REQUEST, ISC_LOG_DEBUG(4), fmt, ap); va_end(ap); } static void _dns_tkey_dumpmessage(dns_message_t *msg) { isc_buffer_t outbuf; unsigned char output[4096]; isc_result_t result; isc_buffer_init(&outbuf, output, sizeof(output)); result = dns_message_totext(msg, &dns_master_style_debug, 0, &outbuf); if (result != ISC_R_SUCCESS) fprintf(stderr, "Warning: dns_message_totext returned: %s\n", dns_result_totext(result)); fprintf(stderr, "%.*s\n", (int)isc_buffer_usedlength(&outbuf), (char *)isc_buffer_base(&outbuf)); } isc_result_t dns_tkeyctx_create(isc_mem_t *mctx, isc_entropy_t *ectx, dns_tkeyctx_t **tctxp) { dns_tkeyctx_t *tctx; REQUIRE(mctx != NULL); REQUIRE(ectx != NULL); REQUIRE(tctxp != NULL && *tctxp == NULL); tctx = isc_mem_get(mctx, sizeof(dns_tkeyctx_t)); if (tctx == NULL) return (ISC_R_NOMEMORY); tctx->mctx = NULL; isc_mem_attach(mctx, &tctx->mctx); tctx->ectx = NULL; isc_entropy_attach(ectx, &tctx->ectx); tctx->dhkey = NULL; tctx->domain = NULL; tctx->gsscred = NULL; tctx->gssapi_keytab = NULL; *tctxp = tctx; return (ISC_R_SUCCESS); } void dns_tkeyctx_destroy(dns_tkeyctx_t **tctxp) { isc_mem_t *mctx; dns_tkeyctx_t *tctx; REQUIRE(tctxp != NULL && *tctxp != NULL); tctx = *tctxp; mctx = tctx->mctx; if (tctx->dhkey != NULL) dst_key_free(&tctx->dhkey); if (tctx->domain != NULL) { if (dns_name_dynamic(tctx->domain)) dns_name_free(tctx->domain, mctx); isc_mem_put(mctx, tctx->domain, sizeof(dns_name_t)); } if (tctx->gssapi_keytab != NULL) { isc_mem_free(mctx, tctx->gssapi_keytab); } if (tctx->gsscred != NULL) dst_gssapi_releasecred(&tctx->gsscred); isc_entropy_detach(&tctx->ectx); isc_mem_put(mctx, tctx, sizeof(dns_tkeyctx_t)); isc_mem_detach(&mctx); *tctxp = NULL; } static isc_result_t add_rdata_to_list(dns_message_t *msg, dns_name_t *name, dns_rdata_t *rdata, isc_uint32_t ttl, dns_namelist_t *namelist) { isc_result_t result; isc_region_t r, newr; dns_rdata_t *newrdata = NULL; dns_name_t *newname = NULL; dns_rdatalist_t *newlist = NULL; dns_rdataset_t *newset = NULL; isc_buffer_t *tmprdatabuf = NULL; RETERR(dns_message_gettemprdata(msg, &newrdata)); dns_rdata_toregion(rdata, &r); RETERR(isc_buffer_allocate(msg->mctx, &tmprdatabuf, r.length)); isc_buffer_availableregion(tmprdatabuf, &newr); memmove(newr.base, r.base, r.length); dns_rdata_fromregion(newrdata, rdata->rdclass, rdata->type, &newr); dns_message_takebuffer(msg, &tmprdatabuf); RETERR(dns_message_gettempname(msg, &newname)); dns_name_init(newname, NULL); RETERR(dns_name_dup(name, msg->mctx, newname)); RETERR(dns_message_gettemprdatalist(msg, &newlist)); newlist->rdclass = newrdata->rdclass; newlist->type = newrdata->type; newlist->covers = 0; newlist->ttl = ttl; ISC_LIST_INIT(newlist->rdata); ISC_LIST_APPEND(newlist->rdata, newrdata, link); RETERR(dns_message_gettemprdataset(msg, &newset)); dns_rdataset_init(newset); RETERR(dns_rdatalist_tordataset(newlist, newset)); ISC_LIST_INIT(newname->list); ISC_LIST_APPEND(newname->list, newset, link); ISC_LIST_APPEND(*namelist, newname, link); return (ISC_R_SUCCESS); failure: if (newrdata != NULL) { if (ISC_LINK_LINKED(newrdata, link)) { INSIST(newlist != NULL); ISC_LIST_UNLINK(newlist->rdata, newrdata, link); } dns_message_puttemprdata(msg, &newrdata); } if (newname != NULL) dns_message_puttempname(msg, &newname); if (newset != NULL) { dns_rdataset_disassociate(newset); dns_message_puttemprdataset(msg, &newset); } if (newlist != NULL) dns_message_puttemprdatalist(msg, &newlist); return (result); } static void free_namelist(dns_message_t *msg, dns_namelist_t *namelist) { dns_name_t *name; dns_rdataset_t *set; while (!ISC_LIST_EMPTY(*namelist)) { name = ISC_LIST_HEAD(*namelist); ISC_LIST_UNLINK(*namelist, name, link); while (!ISC_LIST_EMPTY(name->list)) { set = ISC_LIST_HEAD(name->list); ISC_LIST_UNLINK(name->list, set, link); dns_message_puttemprdataset(msg, &set); } dns_message_puttempname(msg, &name); } } static isc_result_t compute_secret(isc_buffer_t *shared, isc_region_t *queryrandomness, isc_region_t *serverrandomness, isc_buffer_t *secret) { isc_md5_t md5ctx; isc_region_t r, r2; unsigned char digests[32]; unsigned int i; isc_buffer_usedregion(shared, &r); /* * MD5 ( query data | DH value ). */ isc_md5_init(&md5ctx); isc_md5_update(&md5ctx, queryrandomness->base, queryrandomness->length); isc_md5_update(&md5ctx, r.base, r.length); isc_md5_final(&md5ctx, digests); /* * MD5 ( server data | DH value ). */ isc_md5_init(&md5ctx); isc_md5_update(&md5ctx, serverrandomness->base, serverrandomness->length); isc_md5_update(&md5ctx, r.base, r.length); isc_md5_final(&md5ctx, &digests[ISC_MD5_DIGESTLENGTH]); /* * XOR ( DH value, MD5-1 | MD5-2). */ isc_buffer_availableregion(secret, &r); isc_buffer_usedregion(shared, &r2); if (r.length < sizeof(digests) || r.length < r2.length) return (ISC_R_NOSPACE); if (r2.length > sizeof(digests)) { memmove(r.base, r2.base, r2.length); for (i = 0; i < sizeof(digests); i++) r.base[i] ^= digests[i]; isc_buffer_add(secret, r2.length); } else { memmove(r.base, digests, sizeof(digests)); for (i = 0; i < r2.length; i++) r.base[i] ^= r2.base[i]; isc_buffer_add(secret, sizeof(digests)); } return (ISC_R_SUCCESS); } static isc_result_t process_dhtkey(dns_message_t *msg, dns_name_t *signer, dns_name_t *name, dns_rdata_tkey_t *tkeyin, dns_tkeyctx_t *tctx, dns_rdata_tkey_t *tkeyout, dns_tsig_keyring_t *ring, dns_namelist_t *namelist) { isc_result_t result = ISC_R_SUCCESS; dns_name_t *keyname, ourname; dns_rdataset_t *keyset = NULL; dns_rdata_t keyrdata = DNS_RDATA_INIT, ourkeyrdata = DNS_RDATA_INIT; isc_boolean_t found_key = ISC_FALSE, found_incompatible = ISC_FALSE; dst_key_t *pubkey = NULL; isc_buffer_t ourkeybuf, *shared = NULL; isc_region_t r, r2, ourkeyr; unsigned char keydata[DST_KEY_MAXSIZE]; unsigned int sharedsize; isc_buffer_t secret; unsigned char *randomdata = NULL, secretdata[256]; dns_ttl_t ttl = 0; if (tctx->dhkey == NULL) { tkey_log("process_dhtkey: tkey-dhkey not defined"); tkeyout->error = dns_tsigerror_badalg; return (DNS_R_REFUSED); } if (!dns_name_equal(&tkeyin->algorithm, DNS_TSIG_HMACMD5_NAME)) { tkey_log("process_dhtkey: algorithms other than " "hmac-md5 are not supported"); tkeyout->error = dns_tsigerror_badalg; return (ISC_R_SUCCESS); } /* * Look for a DH KEY record that will work with ours. */ for (result = dns_message_firstname(msg, DNS_SECTION_ADDITIONAL); result == ISC_R_SUCCESS && !found_key; result = dns_message_nextname(msg, DNS_SECTION_ADDITIONAL)) { keyname = NULL; dns_message_currentname(msg, DNS_SECTION_ADDITIONAL, &keyname); keyset = NULL; result = dns_message_findtype(keyname, dns_rdatatype_key, 0, &keyset); if (result != ISC_R_SUCCESS) continue; for (result = dns_rdataset_first(keyset); result == ISC_R_SUCCESS && !found_key; result = dns_rdataset_next(keyset)) { dns_rdataset_current(keyset, &keyrdata); pubkey = NULL; result = dns_dnssec_keyfromrdata(keyname, &keyrdata, msg->mctx, &pubkey); if (result != ISC_R_SUCCESS) { dns_rdata_reset(&keyrdata); continue; } if (dst_key_alg(pubkey) == DNS_KEYALG_DH) { if (dst_key_paramcompare(pubkey, tctx->dhkey)) { found_key = ISC_TRUE; ttl = keyset->ttl; break; } else found_incompatible = ISC_TRUE; } dst_key_free(&pubkey); dns_rdata_reset(&keyrdata); } } if (!found_key) { if (found_incompatible) { tkey_log("process_dhtkey: found an incompatible key"); tkeyout->error = dns_tsigerror_badkey; return (ISC_R_SUCCESS); } else { tkey_log("process_dhtkey: failed to find a key"); return (DNS_R_FORMERR); } } RETERR(add_rdata_to_list(msg, keyname, &keyrdata, ttl, namelist)); isc_buffer_init(&ourkeybuf, keydata, sizeof(keydata)); RETERR(dst_key_todns(tctx->dhkey, &ourkeybuf)); isc_buffer_usedregion(&ourkeybuf, &ourkeyr); dns_rdata_fromregion(&ourkeyrdata, dns_rdataclass_any, dns_rdatatype_key, &ourkeyr); dns_name_init(&ourname, NULL); dns_name_clone(dst_key_name(tctx->dhkey), &ourname); /* * XXXBEW The TTL should be obtained from the database, if it exists. */ RETERR(add_rdata_to_list(msg, &ourname, &ourkeyrdata, 0, namelist)); RETERR(dst_key_secretsize(tctx->dhkey, &sharedsize)); RETERR(isc_buffer_allocate(msg->mctx, &shared, sharedsize)); result = dst_key_computesecret(pubkey, tctx->dhkey, shared); if (result != ISC_R_SUCCESS) { tkey_log("process_dhtkey: failed to compute shared secret: %s", isc_result_totext(result)); goto failure; } dst_key_free(&pubkey); isc_buffer_init(&secret, secretdata, sizeof(secretdata)); randomdata = isc_mem_get(tkeyout->mctx, TKEY_RANDOM_AMOUNT); if (randomdata == NULL) goto failure; result = isc_entropy_getdata(tctx->ectx, randomdata, TKEY_RANDOM_AMOUNT, NULL, 0); if (result != ISC_R_SUCCESS) { tkey_log("process_dhtkey: failed to obtain entropy: %s", isc_result_totext(result)); goto failure; } r.base = randomdata; r.length = TKEY_RANDOM_AMOUNT; r2.base = tkeyin->key; r2.length = tkeyin->keylen; RETERR(compute_secret(shared, &r2, &r, &secret)); isc_buffer_free(&shared); RETERR(dns_tsigkey_create(name, &tkeyin->algorithm, isc_buffer_base(&secret), isc_buffer_usedlength(&secret), ISC_TRUE, signer, tkeyin->inception, tkeyin->expire, ring->mctx, ring, NULL)); /* This key is good for a long time */ tkeyout->inception = tkeyin->inception; tkeyout->expire = tkeyin->expire; tkeyout->key = randomdata; tkeyout->keylen = TKEY_RANDOM_AMOUNT; return (ISC_R_SUCCESS); failure: if (!ISC_LIST_EMPTY(*namelist)) free_namelist(msg, namelist); if (shared != NULL) isc_buffer_free(&shared); if (pubkey != NULL) dst_key_free(&pubkey); if (randomdata != NULL) isc_mem_put(tkeyout->mctx, randomdata, TKEY_RANDOM_AMOUNT); return (result); } static isc_result_t process_gsstkey(dns_name_t *name, dns_rdata_tkey_t *tkeyin, dns_tkeyctx_t *tctx, dns_rdata_tkey_t *tkeyout, dns_tsig_keyring_t *ring) { isc_result_t result = ISC_R_SUCCESS; dst_key_t *dstkey = NULL; dns_tsigkey_t *tsigkey = NULL; dns_fixedname_t principal; isc_stdtime_t now; isc_region_t intoken; isc_buffer_t *outtoken = NULL; gss_ctx_id_t gss_ctx = NULL; /* * You have to define either a gss credential (principal) to * accept with tkey-gssapi-credential, or you have to * configure a specific keytab (with tkey-gssapi-keytab) in * order to use gsstkey */ if (tctx->gsscred == NULL && tctx->gssapi_keytab == NULL) { tkey_log("process_gsstkey(): no tkey-gssapi-credential " "or tkey-gssapi-keytab configured"); return (ISC_R_NOPERM); } if (!dns_name_equal(&tkeyin->algorithm, DNS_TSIG_GSSAPI_NAME) && !dns_name_equal(&tkeyin->algorithm, DNS_TSIG_GSSAPIMS_NAME)) { tkeyout->error = dns_tsigerror_badalg; tkey_log("process_gsstkey(): dns_tsigerror_badalg"); /* XXXSRA */ return (ISC_R_SUCCESS); } /* * XXXDCL need to check for key expiry per 4.1.1 * XXXDCL need a way to check fully established, perhaps w/key_flags */ intoken.base = tkeyin->key; intoken.length = tkeyin->keylen; result = dns_tsigkey_find(&tsigkey, name, &tkeyin->algorithm, ring); if (result == ISC_R_SUCCESS) gss_ctx = dst_key_getgssctx(tsigkey->key); dns_fixedname_init(&principal); /* * Note that tctx->gsscred may be NULL if tctx->gssapi_keytab is set */ result = dst_gssapi_acceptctx(tctx->gsscred, tctx->gssapi_keytab, &intoken, &outtoken, &gss_ctx, dns_fixedname_name(&principal), tctx->mctx); if (result == DNS_R_INVALIDTKEY) { if (tsigkey != NULL) dns_tsigkey_detach(&tsigkey); tkeyout->error = dns_tsigerror_badkey; tkey_log("process_gsstkey(): dns_tsigerror_badkey"); /* XXXSRA */ return (ISC_R_SUCCESS); } if (result != DNS_R_CONTINUE && result != ISC_R_SUCCESS) goto failure; /* * XXXDCL Section 4.1.3: Limit GSS_S_CONTINUE_NEEDED to 10 times. */ isc_stdtime_get(&now); if (tsigkey == NULL) { #ifdef GSSAPI OM_uint32 gret, minor, lifetime; #endif isc_uint32_t expire; RETERR(dst_key_fromgssapi(name, gss_ctx, ring->mctx, &dstkey, &intoken)); /* * Limit keys to 1 hour or the context's lifetime whichever * is smaller. */ expire = now + 3600; #ifdef GSSAPI gret = gss_context_time(&minor, gss_ctx, &lifetime); if (gret == GSS_S_COMPLETE && now + lifetime < expire) expire = now + lifetime; #endif RETERR(dns_tsigkey_createfromkey(name, &tkeyin->algorithm, dstkey, ISC_TRUE, dns_fixedname_name(&principal), now, expire, ring->mctx, ring, NULL)); dst_key_free(&dstkey); tkeyout->inception = now; tkeyout->expire = expire; } else { tkeyout->inception = tsigkey->inception; tkeyout->expire = tsigkey->expire; dns_tsigkey_detach(&tsigkey); } if (outtoken) { tkeyout->key = isc_mem_get(tkeyout->mctx, isc_buffer_usedlength(outtoken)); if (tkeyout->key == NULL) { result = ISC_R_NOMEMORY; goto failure; } tkeyout->keylen = isc_buffer_usedlength(outtoken); memmove(tkeyout->key, isc_buffer_base(outtoken), isc_buffer_usedlength(outtoken)); isc_buffer_free(&outtoken); } else { tkeyout->key = isc_mem_get(tkeyout->mctx, tkeyin->keylen); if (tkeyout->key == NULL) { result = ISC_R_NOMEMORY; goto failure; } tkeyout->keylen = tkeyin->keylen; memmove(tkeyout->key, tkeyin->key, tkeyin->keylen); } tkeyout->error = dns_rcode_noerror; tkey_log("process_gsstkey(): dns_tsigerror_noerror"); /* XXXSRA */ return (ISC_R_SUCCESS); failure: if (tsigkey != NULL) dns_tsigkey_detach(&tsigkey); if (dstkey != NULL) dst_key_free(&dstkey); if (outtoken != NULL) isc_buffer_free(&outtoken); tkey_log("process_gsstkey(): %s", isc_result_totext(result)); /* XXXSRA */ return (result); } static isc_result_t process_deletetkey(dns_name_t *signer, dns_name_t *name, dns_rdata_tkey_t *tkeyin, dns_rdata_tkey_t *tkeyout, dns_tsig_keyring_t *ring) { isc_result_t result; dns_tsigkey_t *tsigkey = NULL; dns_name_t *identity; result = dns_tsigkey_find(&tsigkey, name, &tkeyin->algorithm, ring); if (result != ISC_R_SUCCESS) { tkeyout->error = dns_tsigerror_badname; return (ISC_R_SUCCESS); } /* * Only allow a delete if the identity that created the key is the * same as the identity that signed the message. */ identity = dns_tsigkey_identity(tsigkey); if (identity == NULL || !dns_name_equal(identity, signer)) { dns_tsigkey_detach(&tsigkey); return (DNS_R_REFUSED); } /* * Set the key to be deleted when no references are left. If the key * was not generated with TKEY and is in the config file, it may be * reloaded later. */ dns_tsigkey_setdeleted(tsigkey); /* Release the reference */ dns_tsigkey_detach(&tsigkey); return (ISC_R_SUCCESS); } isc_result_t dns_tkey_processquery(dns_message_t *msg, dns_tkeyctx_t *tctx, dns_tsig_keyring_t *ring) { isc_result_t result = ISC_R_SUCCESS; dns_rdata_tkey_t tkeyin, tkeyout; isc_boolean_t freetkeyin = ISC_FALSE; dns_name_t *qname, *name, *keyname, *signer, tsigner; dns_fixedname_t fkeyname; dns_rdataset_t *tkeyset; dns_rdata_t rdata; dns_namelist_t namelist; char tkeyoutdata[512]; isc_buffer_t tkeyoutbuf; REQUIRE(msg != NULL); REQUIRE(tctx != NULL); REQUIRE(ring != NULL); ISC_LIST_INIT(namelist); /* * Interpret the question section. */ result = dns_message_firstname(msg, DNS_SECTION_QUESTION); if (result != ISC_R_SUCCESS) return (DNS_R_FORMERR); qname = NULL; dns_message_currentname(msg, DNS_SECTION_QUESTION, &qname); /* * Look for a TKEY record that matches the question. */ tkeyset = NULL; name = NULL; result = dns_message_findname(msg, DNS_SECTION_ADDITIONAL, qname, dns_rdatatype_tkey, 0, &name, &tkeyset); if (result != ISC_R_SUCCESS) { /* * Try the answer section, since that's where Win2000 * puts it. */ + name = NULL; if (dns_message_findname(msg, DNS_SECTION_ANSWER, qname, dns_rdatatype_tkey, 0, &name, &tkeyset) != ISC_R_SUCCESS) { result = DNS_R_FORMERR; tkey_log("dns_tkey_processquery: couldn't find a TKEY " "matching the question"); goto failure; } } result = dns_rdataset_first(tkeyset); if (result != ISC_R_SUCCESS) { result = DNS_R_FORMERR; goto failure; } dns_rdata_init(&rdata); dns_rdataset_current(tkeyset, &rdata); RETERR(dns_rdata_tostruct(&rdata, &tkeyin, NULL)); freetkeyin = ISC_TRUE; if (tkeyin.error != dns_rcode_noerror) { result = DNS_R_FORMERR; goto failure; } /* * Before we go any farther, verify that the message was signed. * GSSAPI TKEY doesn't require a signature, the rest do. */ dns_name_init(&tsigner, NULL); result = dns_message_signer(msg, &tsigner); if (result != ISC_R_SUCCESS) { if (tkeyin.mode == DNS_TKEYMODE_GSSAPI && result == ISC_R_NOTFOUND) signer = NULL; else { tkey_log("dns_tkey_processquery: query was not " "properly signed - rejecting"); result = DNS_R_FORMERR; goto failure; } } else signer = &tsigner; tkeyout.common.rdclass = tkeyin.common.rdclass; tkeyout.common.rdtype = tkeyin.common.rdtype; ISC_LINK_INIT(&tkeyout.common, link); tkeyout.mctx = msg->mctx; dns_name_init(&tkeyout.algorithm, NULL); dns_name_clone(&tkeyin.algorithm, &tkeyout.algorithm); tkeyout.inception = tkeyout.expire = 0; tkeyout.mode = tkeyin.mode; tkeyout.error = 0; tkeyout.keylen = tkeyout.otherlen = 0; tkeyout.key = tkeyout.other = NULL; /* * A delete operation must have a fully specified key name. If this * is not a delete, we do the following: * if (qname != ".") * keyname = qname + defaultdomain * else * keyname = + defaultdomain */ if (tkeyin.mode != DNS_TKEYMODE_DELETE) { dns_tsigkey_t *tsigkey = NULL; if (tctx->domain == NULL && tkeyin.mode != DNS_TKEYMODE_GSSAPI) { tkey_log("dns_tkey_processquery: tkey-domain not set"); result = DNS_R_REFUSED; goto failure; } dns_fixedname_init(&fkeyname); keyname = dns_fixedname_name(&fkeyname); if (!dns_name_equal(qname, dns_rootname)) { unsigned int n = dns_name_countlabels(qname); RUNTIME_CHECK(dns_name_copy(qname, keyname, NULL) == ISC_R_SUCCESS); dns_name_getlabelsequence(keyname, 0, n - 1, keyname); } else { static char hexdigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; unsigned char randomdata[16]; char randomtext[32]; isc_buffer_t b; unsigned int i, j; result = isc_entropy_getdata(tctx->ectx, randomdata, sizeof(randomdata), NULL, 0); if (result != ISC_R_SUCCESS) goto failure; for (i = 0, j = 0; i < sizeof(randomdata); i++) { unsigned char val = randomdata[i]; randomtext[j++] = hexdigits[val >> 4]; randomtext[j++] = hexdigits[val & 0xF]; } isc_buffer_init(&b, randomtext, sizeof(randomtext)); isc_buffer_add(&b, sizeof(randomtext)); result = dns_name_fromtext(keyname, &b, NULL, 0, NULL); if (result != ISC_R_SUCCESS) goto failure; } if (tkeyin.mode == DNS_TKEYMODE_GSSAPI) { /* Yup. This is a hack */ result = dns_name_concatenate(keyname, dns_rootname, keyname, NULL); if (result != ISC_R_SUCCESS) goto failure; } else { result = dns_name_concatenate(keyname, tctx->domain, keyname, NULL); if (result != ISC_R_SUCCESS) goto failure; } result = dns_tsigkey_find(&tsigkey, keyname, NULL, ring); if (result == ISC_R_SUCCESS) { tkeyout.error = dns_tsigerror_badname; dns_tsigkey_detach(&tsigkey); goto failure_with_tkey; } else if (result != ISC_R_NOTFOUND) goto failure; } else keyname = qname; switch (tkeyin.mode) { case DNS_TKEYMODE_DIFFIEHELLMAN: tkeyout.error = dns_rcode_noerror; RETERR(process_dhtkey(msg, signer, keyname, &tkeyin, tctx, &tkeyout, ring, &namelist)); break; case DNS_TKEYMODE_GSSAPI: tkeyout.error = dns_rcode_noerror; RETERR(process_gsstkey(keyname, &tkeyin, tctx, &tkeyout, ring)); break; case DNS_TKEYMODE_DELETE: tkeyout.error = dns_rcode_noerror; RETERR(process_deletetkey(signer, keyname, &tkeyin, &tkeyout, ring)); break; case DNS_TKEYMODE_SERVERASSIGNED: case DNS_TKEYMODE_RESOLVERASSIGNED: result = DNS_R_NOTIMP; goto failure; default: tkeyout.error = dns_tsigerror_badmode; } failure_with_tkey: dns_rdata_init(&rdata); isc_buffer_init(&tkeyoutbuf, tkeyoutdata, sizeof(tkeyoutdata)); result = dns_rdata_fromstruct(&rdata, tkeyout.common.rdclass, tkeyout.common.rdtype, &tkeyout, &tkeyoutbuf); if (freetkeyin) { dns_rdata_freestruct(&tkeyin); freetkeyin = ISC_FALSE; } if (tkeyout.key != NULL) isc_mem_put(tkeyout.mctx, tkeyout.key, tkeyout.keylen); if (tkeyout.other != NULL) isc_mem_put(tkeyout.mctx, tkeyout.other, tkeyout.otherlen); if (result != ISC_R_SUCCESS) goto failure; RETERR(add_rdata_to_list(msg, keyname, &rdata, 0, &namelist)); RETERR(dns_message_reply(msg, ISC_TRUE)); name = ISC_LIST_HEAD(namelist); while (name != NULL) { dns_name_t *next = ISC_LIST_NEXT(name, link); ISC_LIST_UNLINK(namelist, name, link); dns_message_addname(msg, name, DNS_SECTION_ANSWER); name = next; } return (ISC_R_SUCCESS); failure: if (freetkeyin) dns_rdata_freestruct(&tkeyin); if (!ISC_LIST_EMPTY(namelist)) free_namelist(msg, &namelist); return (result); } static isc_result_t buildquery(dns_message_t *msg, dns_name_t *name, dns_rdata_tkey_t *tkey, isc_boolean_t win2k) { dns_name_t *qname = NULL, *aname = NULL; dns_rdataset_t *question = NULL, *tkeyset = NULL; dns_rdatalist_t *tkeylist = NULL; dns_rdata_t *rdata = NULL; isc_buffer_t *dynbuf = NULL, *anamebuf = NULL, *qnamebuf = NULL; isc_result_t result; REQUIRE(msg != NULL); REQUIRE(name != NULL); REQUIRE(tkey != NULL); RETERR(dns_message_gettempname(msg, &qname)); RETERR(dns_message_gettempname(msg, &aname)); RETERR(dns_message_gettemprdataset(msg, &question)); dns_rdataset_init(question); dns_rdataset_makequestion(question, dns_rdataclass_any, dns_rdatatype_tkey); RETERR(isc_buffer_allocate(msg->mctx, &dynbuf, 4096)); RETERR(isc_buffer_allocate(msg->mctx, &anamebuf, DNS_NAME_MAXWIRE)); RETERR(isc_buffer_allocate(msg->mctx, &qnamebuf, DNS_NAME_MAXWIRE)); RETERR(dns_message_gettemprdata(msg, &rdata)); RETERR(dns_rdata_fromstruct(rdata, dns_rdataclass_any, dns_rdatatype_tkey, tkey, dynbuf)); dns_message_takebuffer(msg, &dynbuf); RETERR(dns_message_gettemprdatalist(msg, &tkeylist)); tkeylist->rdclass = dns_rdataclass_any; tkeylist->type = dns_rdatatype_tkey; tkeylist->covers = 0; tkeylist->ttl = 0; ISC_LIST_INIT(tkeylist->rdata); ISC_LIST_APPEND(tkeylist->rdata, rdata, link); RETERR(dns_message_gettemprdataset(msg, &tkeyset)); dns_rdataset_init(tkeyset); RETERR(dns_rdatalist_tordataset(tkeylist, tkeyset)); dns_name_init(qname, NULL); dns_name_copy(name, qname, qnamebuf); dns_name_init(aname, NULL); dns_name_copy(name, aname, anamebuf); ISC_LIST_APPEND(qname->list, question, link); ISC_LIST_APPEND(aname->list, tkeyset, link); dns_message_addname(msg, qname, DNS_SECTION_QUESTION); dns_message_takebuffer(msg, &qnamebuf); /* * Windows 2000 needs this in the answer section, not the additional * section where the RFC specifies. */ if (win2k) dns_message_addname(msg, aname, DNS_SECTION_ANSWER); else dns_message_addname(msg, aname, DNS_SECTION_ADDITIONAL); dns_message_takebuffer(msg, &anamebuf); return (ISC_R_SUCCESS); failure: if (qname != NULL) dns_message_puttempname(msg, &qname); if (aname != NULL) dns_message_puttempname(msg, &aname); if (question != NULL) { dns_rdataset_disassociate(question); dns_message_puttemprdataset(msg, &question); } if (dynbuf != NULL) isc_buffer_free(&dynbuf); if (qnamebuf != NULL) isc_buffer_free(&qnamebuf); if (anamebuf != NULL) isc_buffer_free(&anamebuf); printf("buildquery error\n"); return (result); } isc_result_t dns_tkey_builddhquery(dns_message_t *msg, dst_key_t *key, dns_name_t *name, dns_name_t *algorithm, isc_buffer_t *nonce, isc_uint32_t lifetime) { dns_rdata_tkey_t tkey; dns_rdata_t *rdata = NULL; isc_buffer_t *dynbuf = NULL; isc_region_t r; dns_name_t keyname; dns_namelist_t namelist; isc_result_t result; isc_stdtime_t now; REQUIRE(msg != NULL); REQUIRE(key != NULL); REQUIRE(dst_key_alg(key) == DNS_KEYALG_DH); REQUIRE(dst_key_isprivate(key)); REQUIRE(name != NULL); REQUIRE(algorithm != NULL); tkey.common.rdclass = dns_rdataclass_any; tkey.common.rdtype = dns_rdatatype_tkey; ISC_LINK_INIT(&tkey.common, link); tkey.mctx = msg->mctx; dns_name_init(&tkey.algorithm, NULL); dns_name_clone(algorithm, &tkey.algorithm); isc_stdtime_get(&now); tkey.inception = now; tkey.expire = now + lifetime; tkey.mode = DNS_TKEYMODE_DIFFIEHELLMAN; if (nonce != NULL) isc_buffer_usedregion(nonce, &r); else { r.base = isc_mem_get(msg->mctx, 0); r.length = 0; } tkey.error = 0; tkey.key = r.base; tkey.keylen = r.length; tkey.other = NULL; tkey.otherlen = 0; RETERR(buildquery(msg, name, &tkey, ISC_FALSE)); if (nonce == NULL) isc_mem_put(msg->mctx, r.base, 0); RETERR(dns_message_gettemprdata(msg, &rdata)); RETERR(isc_buffer_allocate(msg->mctx, &dynbuf, 1024)); RETERR(dst_key_todns(key, dynbuf)); isc_buffer_usedregion(dynbuf, &r); dns_rdata_fromregion(rdata, dns_rdataclass_any, dns_rdatatype_key, &r); dns_message_takebuffer(msg, &dynbuf); dns_name_init(&keyname, NULL); dns_name_clone(dst_key_name(key), &keyname); ISC_LIST_INIT(namelist); RETERR(add_rdata_to_list(msg, &keyname, rdata, 0, &namelist)); name = ISC_LIST_HEAD(namelist); while (name != NULL) { dns_name_t *next = ISC_LIST_NEXT(name, link); ISC_LIST_UNLINK(namelist, name, link); dns_message_addname(msg, name, DNS_SECTION_ADDITIONAL); name = next; } return (ISC_R_SUCCESS); failure: if (dynbuf != NULL) isc_buffer_free(&dynbuf); return (result); } isc_result_t dns_tkey_buildgssquery(dns_message_t *msg, dns_name_t *name, dns_name_t *gname, isc_buffer_t *intoken, isc_uint32_t lifetime, gss_ctx_id_t *context, isc_boolean_t win2k, isc_mem_t *mctx, char **err_message) { dns_rdata_tkey_t tkey; isc_result_t result; isc_stdtime_t now; isc_buffer_t token; unsigned char array[4096]; UNUSED(intoken); REQUIRE(msg != NULL); REQUIRE(name != NULL); REQUIRE(gname != NULL); REQUIRE(context != NULL); REQUIRE(mctx != NULL); isc_buffer_init(&token, array, sizeof(array)); result = dst_gssapi_initctx(gname, NULL, &token, context, mctx, err_message); if (result != DNS_R_CONTINUE && result != ISC_R_SUCCESS) return (result); tkey.common.rdclass = dns_rdataclass_any; tkey.common.rdtype = dns_rdatatype_tkey; ISC_LINK_INIT(&tkey.common, link); tkey.mctx = NULL; dns_name_init(&tkey.algorithm, NULL); if (win2k) dns_name_clone(DNS_TSIG_GSSAPIMS_NAME, &tkey.algorithm); else dns_name_clone(DNS_TSIG_GSSAPI_NAME, &tkey.algorithm); isc_stdtime_get(&now); tkey.inception = now; tkey.expire = now + lifetime; tkey.mode = DNS_TKEYMODE_GSSAPI; tkey.error = 0; tkey.key = isc_buffer_base(&token); tkey.keylen = isc_buffer_usedlength(&token); tkey.other = NULL; tkey.otherlen = 0; RETERR(buildquery(msg, name, &tkey, win2k)); return (ISC_R_SUCCESS); failure: return (result); } isc_result_t dns_tkey_builddeletequery(dns_message_t *msg, dns_tsigkey_t *key) { dns_rdata_tkey_t tkey; REQUIRE(msg != NULL); REQUIRE(key != NULL); tkey.common.rdclass = dns_rdataclass_any; tkey.common.rdtype = dns_rdatatype_tkey; ISC_LINK_INIT(&tkey.common, link); tkey.mctx = msg->mctx; dns_name_init(&tkey.algorithm, NULL); dns_name_clone(key->algorithm, &tkey.algorithm); tkey.inception = tkey.expire = 0; tkey.mode = DNS_TKEYMODE_DELETE; tkey.error = 0; tkey.keylen = tkey.otherlen = 0; tkey.key = tkey.other = NULL; return (buildquery(msg, &key->name, &tkey, ISC_FALSE)); } static isc_result_t find_tkey(dns_message_t *msg, dns_name_t **name, dns_rdata_t *rdata, int section) { dns_rdataset_t *tkeyset; isc_result_t result; result = dns_message_firstname(msg, section); while (result == ISC_R_SUCCESS) { *name = NULL; dns_message_currentname(msg, section, name); tkeyset = NULL; result = dns_message_findtype(*name, dns_rdatatype_tkey, 0, &tkeyset); if (result == ISC_R_SUCCESS) { result = dns_rdataset_first(tkeyset); if (result != ISC_R_SUCCESS) return (result); dns_rdataset_current(tkeyset, rdata); return (ISC_R_SUCCESS); } result = dns_message_nextname(msg, section); } if (result == ISC_R_NOMORE) return (ISC_R_NOTFOUND); return (result); } isc_result_t dns_tkey_processdhresponse(dns_message_t *qmsg, dns_message_t *rmsg, dst_key_t *key, isc_buffer_t *nonce, dns_tsigkey_t **outkey, dns_tsig_keyring_t *ring) { dns_rdata_t qtkeyrdata = DNS_RDATA_INIT, rtkeyrdata = DNS_RDATA_INIT; dns_name_t keyname, *tkeyname, *theirkeyname, *ourkeyname, *tempname; dns_rdataset_t *theirkeyset = NULL, *ourkeyset = NULL; dns_rdata_t theirkeyrdata = DNS_RDATA_INIT; dst_key_t *theirkey = NULL; dns_rdata_tkey_t qtkey, rtkey; unsigned char secretdata[256]; unsigned int sharedsize; isc_buffer_t *shared = NULL, secret; isc_region_t r, r2; isc_result_t result; isc_boolean_t freertkey = ISC_FALSE; REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); REQUIRE(key != NULL); REQUIRE(dst_key_alg(key) == DNS_KEYALG_DH); REQUIRE(dst_key_isprivate(key)); if (outkey != NULL) REQUIRE(*outkey == NULL); if (rmsg->rcode != dns_rcode_noerror) return (ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); freertkey = ISC_TRUE; RETERR(find_tkey(qmsg, &tempname, &qtkeyrdata, DNS_SECTION_ADDITIONAL)); RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_DIFFIEHELLMAN || rtkey.mode != qtkey.mode || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm) || rmsg->rcode != dns_rcode_noerror) { tkey_log("dns_tkey_processdhresponse: tkey mode invalid " "or error set(1)"); result = DNS_R_INVALIDTKEY; dns_rdata_freestruct(&qtkey); goto failure; } dns_rdata_freestruct(&qtkey); dns_name_init(&keyname, NULL); dns_name_clone(dst_key_name(key), &keyname); ourkeyname = NULL; ourkeyset = NULL; RETERR(dns_message_findname(rmsg, DNS_SECTION_ANSWER, &keyname, dns_rdatatype_key, 0, &ourkeyname, &ourkeyset)); result = dns_message_firstname(rmsg, DNS_SECTION_ANSWER); while (result == ISC_R_SUCCESS) { theirkeyname = NULL; dns_message_currentname(rmsg, DNS_SECTION_ANSWER, &theirkeyname); if (dns_name_equal(theirkeyname, ourkeyname)) goto next; theirkeyset = NULL; result = dns_message_findtype(theirkeyname, dns_rdatatype_key, 0, &theirkeyset); if (result == ISC_R_SUCCESS) { RETERR(dns_rdataset_first(theirkeyset)); break; } next: result = dns_message_nextname(rmsg, DNS_SECTION_ANSWER); } if (theirkeyset == NULL) { tkey_log("dns_tkey_processdhresponse: failed to find server " "key"); result = ISC_R_NOTFOUND; goto failure; } dns_rdataset_current(theirkeyset, &theirkeyrdata); RETERR(dns_dnssec_keyfromrdata(theirkeyname, &theirkeyrdata, rmsg->mctx, &theirkey)); RETERR(dst_key_secretsize(key, &sharedsize)); RETERR(isc_buffer_allocate(rmsg->mctx, &shared, sharedsize)); RETERR(dst_key_computesecret(theirkey, key, shared)); isc_buffer_init(&secret, secretdata, sizeof(secretdata)); r.base = rtkey.key; r.length = rtkey.keylen; if (nonce != NULL) isc_buffer_usedregion(nonce, &r2); else { r2.base = isc_mem_get(rmsg->mctx, 0); r2.length = 0; } RETERR(compute_secret(shared, &r2, &r, &secret)); if (nonce == NULL) isc_mem_put(rmsg->mctx, r2.base, 0); isc_buffer_usedregion(&secret, &r); result = dns_tsigkey_create(tkeyname, &rtkey.algorithm, r.base, r.length, ISC_TRUE, NULL, rtkey.inception, rtkey.expire, rmsg->mctx, ring, outkey); isc_buffer_free(&shared); dns_rdata_freestruct(&rtkey); dst_key_free(&theirkey); return (result); failure: if (shared != NULL) isc_buffer_free(&shared); if (theirkey != NULL) dst_key_free(&theirkey); if (freertkey) dns_rdata_freestruct(&rtkey); return (result); } isc_result_t dns_tkey_processgssresponse(dns_message_t *qmsg, dns_message_t *rmsg, dns_name_t *gname, gss_ctx_id_t *context, isc_buffer_t *outtoken, dns_tsigkey_t **outkey, dns_tsig_keyring_t *ring, char **err_message) { dns_rdata_t rtkeyrdata = DNS_RDATA_INIT, qtkeyrdata = DNS_RDATA_INIT; dns_name_t *tkeyname; dns_rdata_tkey_t rtkey, qtkey; dst_key_t *dstkey = NULL; isc_buffer_t intoken; isc_result_t result; unsigned char array[1024]; REQUIRE(outtoken != NULL); REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); REQUIRE(gname != NULL); REQUIRE(ring != NULL); if (outkey != NULL) REQUIRE(*outkey == NULL); if (rmsg->rcode != dns_rcode_noerror) return (ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); /* * Win2k puts the item in the ANSWER section, while the RFC * specifies it should be in the ADDITIONAL section. Check first * where it should be, and then where it may be. */ result = find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ADDITIONAL); if (result == ISC_R_NOTFOUND) result = find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ANSWER); if (result != ISC_R_SUCCESS) goto failure; RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_GSSAPI || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm)) { tkey_log("dns_tkey_processgssresponse: tkey mode invalid " "or error set(2) %d", rtkey.error); _dns_tkey_dumpmessage(qmsg); _dns_tkey_dumpmessage(rmsg); result = DNS_R_INVALIDTKEY; goto failure; } isc_buffer_init(outtoken, array, sizeof(array)); isc_buffer_init(&intoken, rtkey.key, rtkey.keylen); RETERR(dst_gssapi_initctx(gname, &intoken, outtoken, context, ring->mctx, err_message)); RETERR(dst_key_fromgssapi(dns_rootname, *context, rmsg->mctx, &dstkey, NULL)); RETERR(dns_tsigkey_createfromkey(tkeyname, DNS_TSIG_GSSAPI_NAME, dstkey, ISC_FALSE, NULL, rtkey.inception, rtkey.expire, ring->mctx, ring, outkey)); dst_key_free(&dstkey); dns_rdata_freestruct(&rtkey); return (result); failure: /* * XXXSRA This probably leaks memory from rtkey and qtkey. */ if (dstkey != NULL) dst_key_free(&dstkey); return (result); } isc_result_t dns_tkey_processdeleteresponse(dns_message_t *qmsg, dns_message_t *rmsg, dns_tsig_keyring_t *ring) { dns_rdata_t qtkeyrdata = DNS_RDATA_INIT, rtkeyrdata = DNS_RDATA_INIT; dns_name_t *tkeyname, *tempname; dns_rdata_tkey_t qtkey, rtkey; dns_tsigkey_t *tsigkey = NULL; isc_result_t result; REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); if (rmsg->rcode != dns_rcode_noerror) return(ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); RETERR(find_tkey(qmsg, &tempname, &qtkeyrdata, DNS_SECTION_ADDITIONAL)); RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_DELETE || rtkey.mode != qtkey.mode || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm) || rmsg->rcode != dns_rcode_noerror) { tkey_log("dns_tkey_processdeleteresponse: tkey mode invalid " "or error set(3)"); result = DNS_R_INVALIDTKEY; dns_rdata_freestruct(&qtkey); dns_rdata_freestruct(&rtkey); goto failure; } dns_rdata_freestruct(&qtkey); RETERR(dns_tsigkey_find(&tsigkey, tkeyname, &rtkey.algorithm, ring)); dns_rdata_freestruct(&rtkey); /* * Mark the key as deleted. */ dns_tsigkey_setdeleted(tsigkey); /* * Release the reference. */ dns_tsigkey_detach(&tsigkey); failure: return (result); } isc_result_t dns_tkey_gssnegotiate(dns_message_t *qmsg, dns_message_t *rmsg, dns_name_t *server, gss_ctx_id_t *context, dns_tsigkey_t **outkey, dns_tsig_keyring_t *ring, isc_boolean_t win2k, char **err_message) { dns_rdata_t rtkeyrdata = DNS_RDATA_INIT, qtkeyrdata = DNS_RDATA_INIT; dns_name_t *tkeyname; dns_rdata_tkey_t rtkey, qtkey; isc_buffer_t intoken, outtoken; dst_key_t *dstkey = NULL; isc_result_t result; unsigned char array[1024]; isc_boolean_t freertkey = ISC_FALSE; REQUIRE(qmsg != NULL); REQUIRE(rmsg != NULL); REQUIRE(server != NULL); if (outkey != NULL) REQUIRE(*outkey == NULL); if (rmsg->rcode != dns_rcode_noerror) return (ISC_RESULTCLASS_DNSRCODE + rmsg->rcode); RETERR(find_tkey(rmsg, &tkeyname, &rtkeyrdata, DNS_SECTION_ANSWER)); RETERR(dns_rdata_tostruct(&rtkeyrdata, &rtkey, NULL)); freertkey = ISC_TRUE; if (win2k == ISC_TRUE) RETERR(find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ANSWER)); else RETERR(find_tkey(qmsg, &tkeyname, &qtkeyrdata, DNS_SECTION_ADDITIONAL)); RETERR(dns_rdata_tostruct(&qtkeyrdata, &qtkey, NULL)); if (rtkey.error != dns_rcode_noerror || rtkey.mode != DNS_TKEYMODE_GSSAPI || !dns_name_equal(&rtkey.algorithm, &qtkey.algorithm)) { tkey_log("dns_tkey_processdhresponse: tkey mode invalid " "or error set(4)"); result = DNS_R_INVALIDTKEY; goto failure; } isc_buffer_init(&intoken, rtkey.key, rtkey.keylen); isc_buffer_init(&outtoken, array, sizeof(array)); result = dst_gssapi_initctx(server, &intoken, &outtoken, context, ring->mctx, err_message); if (result != DNS_R_CONTINUE && result != ISC_R_SUCCESS) return (result); RETERR(dst_key_fromgssapi(dns_rootname, *context, rmsg->mctx, &dstkey, NULL)); /* * XXXSRA This seems confused. If we got CONTINUE from initctx, * the GSS negotiation hasn't completed yet, so we can't sign * anything yet. */ RETERR(dns_tsigkey_createfromkey(tkeyname, (win2k ? DNS_TSIG_GSSAPIMS_NAME : DNS_TSIG_GSSAPI_NAME), dstkey, ISC_TRUE, NULL, rtkey.inception, rtkey.expire, ring->mctx, ring, outkey)); dst_key_free(&dstkey); dns_rdata_freestruct(&rtkey); return (result); failure: /* * XXXSRA This probably leaks memory from qtkey. */ if (freertkey) dns_rdata_freestruct(&rtkey); if (dstkey != NULL) dst_key_free(&dstkey); return (result); } Index: stable/9/crypto/openssh/auth2-chall.c =================================================================== --- stable/9/crypto/openssh/auth2-chall.c (revision 285976) +++ stable/9/crypto/openssh/auth2-chall.c (revision 285977) @@ -1,374 +1,379 @@ /* $OpenBSD: auth2-chall.c,v 1.41 2014/02/02 03:44:31 djm Exp $ */ /* * Copyright (c) 2001 Markus Friedl. All rights reserved. * Copyright (c) 2001 Per Allansson. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "includes.h" #include #include #include #include #include "xmalloc.h" #include "ssh2.h" #include "key.h" #include "hostfile.h" #include "auth.h" #include "buffer.h" #include "packet.h" #include "dispatch.h" #include "log.h" #include "servconf.h" /* import */ extern ServerOptions options; static int auth2_challenge_start(Authctxt *); static int send_userauth_info_request(Authctxt *); static void input_userauth_info_response(int, u_int32_t, void *); #ifdef BSD_AUTH extern KbdintDevice bsdauth_device; #else #ifdef USE_PAM extern KbdintDevice sshpam_device; #endif #ifdef SKEY extern KbdintDevice skey_device; #endif #endif KbdintDevice *devices[] = { #ifdef BSD_AUTH &bsdauth_device, #else #ifdef USE_PAM &sshpam_device, #endif #ifdef SKEY &skey_device, #endif #endif NULL }; typedef struct KbdintAuthctxt KbdintAuthctxt; struct KbdintAuthctxt { char *devices; void *ctxt; KbdintDevice *device; u_int nreq; + u_int devices_done; }; #ifdef USE_PAM void remove_kbdint_device(const char *devname) { int i, j; for (i = 0; devices[i] != NULL; i++) if (strcmp(devices[i]->name, devname) == 0) { for (j = i; devices[j] != NULL; j++) devices[j] = devices[j+1]; i--; } } #endif static KbdintAuthctxt * kbdint_alloc(const char *devs) { KbdintAuthctxt *kbdintctxt; Buffer b; int i; #ifdef USE_PAM if (!options.use_pam) remove_kbdint_device("pam"); #endif kbdintctxt = xcalloc(1, sizeof(KbdintAuthctxt)); if (strcmp(devs, "") == 0) { buffer_init(&b); for (i = 0; devices[i]; i++) { if (buffer_len(&b) > 0) buffer_append(&b, ",", 1); buffer_append(&b, devices[i]->name, strlen(devices[i]->name)); } buffer_append(&b, "\0", 1); kbdintctxt->devices = xstrdup(buffer_ptr(&b)); buffer_free(&b); } else { kbdintctxt->devices = xstrdup(devs); } debug("kbdint_alloc: devices '%s'", kbdintctxt->devices); kbdintctxt->ctxt = NULL; kbdintctxt->device = NULL; kbdintctxt->nreq = 0; return kbdintctxt; } static void kbdint_reset_device(KbdintAuthctxt *kbdintctxt) { if (kbdintctxt->ctxt) { kbdintctxt->device->free_ctx(kbdintctxt->ctxt); kbdintctxt->ctxt = NULL; } kbdintctxt->device = NULL; } static void kbdint_free(KbdintAuthctxt *kbdintctxt) { if (kbdintctxt->device) kbdint_reset_device(kbdintctxt); free(kbdintctxt->devices); explicit_bzero(kbdintctxt, sizeof(*kbdintctxt)); free(kbdintctxt); } /* get next device */ static int kbdint_next_device(Authctxt *authctxt, KbdintAuthctxt *kbdintctxt) { size_t len; char *t; int i; if (kbdintctxt->device) kbdint_reset_device(kbdintctxt); do { len = kbdintctxt->devices ? strcspn(kbdintctxt->devices, ",") : 0; if (len == 0) break; for (i = 0; devices[i]; i++) { - if (!auth2_method_allowed(authctxt, + if ((kbdintctxt->devices_done & (1 << i)) != 0 || + !auth2_method_allowed(authctxt, "keyboard-interactive", devices[i]->name)) continue; - if (strncmp(kbdintctxt->devices, devices[i]->name, len) == 0) + if (strncmp(kbdintctxt->devices, devices[i]->name, + len) == 0) { kbdintctxt->device = devices[i]; + kbdintctxt->devices_done |= 1 << i; + } } t = kbdintctxt->devices; kbdintctxt->devices = t[len] ? xstrdup(t+len+1) : NULL; free(t); debug2("kbdint_next_device: devices %s", kbdintctxt->devices ? kbdintctxt->devices : ""); } while (kbdintctxt->devices && !kbdintctxt->device); return kbdintctxt->device ? 1 : 0; } /* * try challenge-response, set authctxt->postponed if we have to * wait for the response. */ int auth2_challenge(Authctxt *authctxt, char *devs) { debug("auth2_challenge: user=%s devs=%s", authctxt->user ? authctxt->user : "", devs ? devs : ""); if (authctxt->user == NULL || !devs) return 0; if (authctxt->kbdintctxt == NULL) authctxt->kbdintctxt = kbdint_alloc(devs); return auth2_challenge_start(authctxt); } /* unregister kbd-int callbacks and context */ void auth2_challenge_stop(Authctxt *authctxt) { /* unregister callback */ dispatch_set(SSH2_MSG_USERAUTH_INFO_RESPONSE, NULL); if (authctxt->kbdintctxt != NULL) { kbdint_free(authctxt->kbdintctxt); authctxt->kbdintctxt = NULL; } } /* side effect: sets authctxt->postponed if a reply was sent*/ static int auth2_challenge_start(Authctxt *authctxt) { KbdintAuthctxt *kbdintctxt = authctxt->kbdintctxt; debug2("auth2_challenge_start: devices %s", kbdintctxt->devices ? kbdintctxt->devices : ""); if (kbdint_next_device(authctxt, kbdintctxt) == 0) { auth2_challenge_stop(authctxt); return 0; } debug("auth2_challenge_start: trying authentication method '%s'", kbdintctxt->device->name); if ((kbdintctxt->ctxt = kbdintctxt->device->init_ctx(authctxt)) == NULL) { auth2_challenge_stop(authctxt); return 0; } if (send_userauth_info_request(authctxt) == 0) { auth2_challenge_stop(authctxt); return 0; } dispatch_set(SSH2_MSG_USERAUTH_INFO_RESPONSE, &input_userauth_info_response); authctxt->postponed = 1; return 0; } static int send_userauth_info_request(Authctxt *authctxt) { KbdintAuthctxt *kbdintctxt; char *name, *instr, **prompts; u_int i, *echo_on; kbdintctxt = authctxt->kbdintctxt; if (kbdintctxt->device->query(kbdintctxt->ctxt, &name, &instr, &kbdintctxt->nreq, &prompts, &echo_on)) return 0; packet_start(SSH2_MSG_USERAUTH_INFO_REQUEST); packet_put_cstring(name); packet_put_cstring(instr); packet_put_cstring(""); /* language not used */ packet_put_int(kbdintctxt->nreq); for (i = 0; i < kbdintctxt->nreq; i++) { packet_put_cstring(prompts[i]); packet_put_char(echo_on[i]); } packet_send(); packet_write_wait(); for (i = 0; i < kbdintctxt->nreq; i++) free(prompts[i]); free(prompts); free(echo_on); free(name); free(instr); return 1; } static void input_userauth_info_response(int type, u_int32_t seq, void *ctxt) { Authctxt *authctxt = ctxt; KbdintAuthctxt *kbdintctxt; int authenticated = 0, res; u_int i, nresp; const char *devicename = NULL; char **response = NULL; if (authctxt == NULL) fatal("input_userauth_info_response: no authctxt"); kbdintctxt = authctxt->kbdintctxt; if (kbdintctxt == NULL || kbdintctxt->ctxt == NULL) fatal("input_userauth_info_response: no kbdintctxt"); if (kbdintctxt->device == NULL) fatal("input_userauth_info_response: no device"); authctxt->postponed = 0; /* reset */ nresp = packet_get_int(); if (nresp != kbdintctxt->nreq) fatal("input_userauth_info_response: wrong number of replies"); if (nresp > 100) fatal("input_userauth_info_response: too many replies"); if (nresp > 0) { response = xcalloc(nresp, sizeof(char *)); for (i = 0; i < nresp; i++) response[i] = packet_get_string(NULL); } packet_check_eom(); res = kbdintctxt->device->respond(kbdintctxt->ctxt, nresp, response); for (i = 0; i < nresp; i++) { explicit_bzero(response[i], strlen(response[i])); free(response[i]); } free(response); switch (res) { case 0: /* Success! */ authenticated = authctxt->valid ? 1 : 0; break; case 1: /* Authentication needs further interaction */ if (send_userauth_info_request(authctxt) == 1) authctxt->postponed = 1; break; default: /* Failure! */ break; } devicename = kbdintctxt->device->name; if (!authctxt->postponed) { if (authenticated) { auth2_challenge_stop(authctxt); } else { /* start next device */ /* may set authctxt->postponed */ auth2_challenge_start(authctxt); } } userauth_finish(authctxt, authenticated, "keyboard-interactive", devicename); } void privsep_challenge_enable(void) { #if defined(BSD_AUTH) || defined(USE_PAM) || defined(SKEY) int n = 0; #endif #ifdef BSD_AUTH extern KbdintDevice mm_bsdauth_device; #endif #ifdef USE_PAM extern KbdintDevice mm_sshpam_device; #endif #ifdef SKEY extern KbdintDevice mm_skey_device; #endif #ifdef BSD_AUTH devices[n++] = &mm_bsdauth_device; #else #ifdef USE_PAM devices[n++] = &mm_sshpam_device; #endif #ifdef SKEY devices[n++] = &mm_skey_device; #endif #endif } Index: stable/9/crypto/openssh/sshconnect.c =================================================================== --- stable/9/crypto/openssh/sshconnect.c (revision 285976) +++ stable/9/crypto/openssh/sshconnect.c (revision 285977) @@ -1,1433 +1,1443 @@ /* $OpenBSD: sshconnect.c,v 1.246 2014/02/06 22:21:01 djm Exp $ */ /* $FreeBSD$ */ /* * Author: Tatu Ylonen * Copyright (c) 1995 Tatu Ylonen , Espoo, Finland * All rights reserved * Code to connect to a remote host, and to perform the client side of the * login (authentication) dialog. * * As far as I am concerned, the code I have written for this software * can be used freely for any purpose. Any derived versions of this * software must be clearly marked as such, and if the derived work is * incompatible with the protocol description in the RFC file, it must be * called by a name other than "ssh" or "Secure Shell". */ #include "includes.h" __RCSID("$FreeBSD$"); #include #include #include #include #ifdef HAVE_SYS_TIME_H # include #endif #include #include #include #include #include #include #include #ifdef HAVE_PATHS_H #include #endif #include #include #include #include #include #include #include #include "xmalloc.h" #include "key.h" #include "hostfile.h" #include "ssh.h" #include "rsa.h" #include "buffer.h" #include "packet.h" #include "uidswap.h" #include "compat.h" #include "key.h" #include "sshconnect.h" #include "hostfile.h" #include "log.h" #include "readconf.h" #include "atomicio.h" #include "misc.h" #include "dns.h" #include "roaming.h" #include "monitor_fdpass.h" #include "ssh2.h" #include "version.h" char *client_version_string = NULL; char *server_version_string = NULL; static int matching_host_key_dns = 0; static pid_t proxy_command_pid = 0; /* import */ extern Options options; extern char *__progname; extern uid_t original_real_uid; extern uid_t original_effective_uid; static int show_other_keys(struct hostkeys *, Key *); static void warn_changed_key(Key *); /* Expand a proxy command */ static char * expand_proxy_command(const char *proxy_command, const char *user, const char *host, int port) { char *tmp, *ret, strport[NI_MAXSERV]; snprintf(strport, sizeof strport, "%d", port); xasprintf(&tmp, "exec %s", proxy_command); ret = percent_expand(tmp, "h", host, "p", strport, "r", options.user, (char *)NULL); free(tmp); return ret; } /* * Connect to the given ssh server using a proxy command that passes a * a connected fd back to us. */ static int ssh_proxy_fdpass_connect(const char *host, u_short port, const char *proxy_command) { char *command_string; int sp[2], sock; pid_t pid; char *shell; if ((shell = getenv("SHELL")) == NULL) shell = _PATH_BSHELL; if (socketpair(AF_UNIX, SOCK_STREAM, 0, sp) < 0) fatal("Could not create socketpair to communicate with " "proxy dialer: %.100s", strerror(errno)); command_string = expand_proxy_command(proxy_command, options.user, host, port); debug("Executing proxy dialer command: %.500s", command_string); /* Fork and execute the proxy command. */ if ((pid = fork()) == 0) { char *argv[10]; /* Child. Permanently give up superuser privileges. */ permanently_drop_suid(original_real_uid); close(sp[1]); /* Redirect stdin and stdout. */ if (sp[0] != 0) { if (dup2(sp[0], 0) < 0) perror("dup2 stdin"); } if (sp[0] != 1) { if (dup2(sp[0], 1) < 0) perror("dup2 stdout"); } if (sp[0] >= 2) close(sp[0]); /* * Stderr is left as it is so that error messages get * printed on the user's terminal. */ argv[0] = shell; argv[1] = "-c"; argv[2] = command_string; argv[3] = NULL; /* * Execute the proxy command. * Note that we gave up any extra privileges above. */ execv(argv[0], argv); perror(argv[0]); exit(1); } /* Parent. */ if (pid < 0) fatal("fork failed: %.100s", strerror(errno)); close(sp[0]); free(command_string); if ((sock = mm_receive_fd(sp[1])) == -1) fatal("proxy dialer did not pass back a connection"); while (waitpid(pid, NULL, 0) == -1) if (errno != EINTR) fatal("Couldn't wait for child: %s", strerror(errno)); /* Set the connection file descriptors. */ packet_set_connection(sock, sock); return 0; } /* * Connect to the given ssh server using a proxy command. */ static int ssh_proxy_connect(const char *host, u_short port, const char *proxy_command) { char *command_string; int pin[2], pout[2]; pid_t pid; char *shell; if ((shell = getenv("SHELL")) == NULL || *shell == '\0') shell = _PATH_BSHELL; /* Create pipes for communicating with the proxy. */ if (pipe(pin) < 0 || pipe(pout) < 0) fatal("Could not create pipes to communicate with the proxy: %.100s", strerror(errno)); command_string = expand_proxy_command(proxy_command, options.user, host, port); debug("Executing proxy command: %.500s", command_string); /* Fork and execute the proxy command. */ if ((pid = fork()) == 0) { char *argv[10]; /* Child. Permanently give up superuser privileges. */ permanently_drop_suid(original_real_uid); /* Redirect stdin and stdout. */ close(pin[1]); if (pin[0] != 0) { if (dup2(pin[0], 0) < 0) perror("dup2 stdin"); close(pin[0]); } close(pout[0]); if (dup2(pout[1], 1) < 0) perror("dup2 stdout"); /* Cannot be 1 because pin allocated two descriptors. */ close(pout[1]); /* Stderr is left as it is so that error messages get printed on the user's terminal. */ argv[0] = shell; argv[1] = "-c"; argv[2] = command_string; argv[3] = NULL; /* Execute the proxy command. Note that we gave up any extra privileges above. */ signal(SIGPIPE, SIG_DFL); execv(argv[0], argv); perror(argv[0]); exit(1); } /* Parent. */ if (pid < 0) fatal("fork failed: %.100s", strerror(errno)); else proxy_command_pid = pid; /* save pid to clean up later */ /* Close child side of the descriptors. */ close(pin[0]); close(pout[1]); /* Free the command name. */ free(command_string); /* Set the connection file descriptors. */ packet_set_connection(pout[0], pin[1]); /* Indicate OK return */ return 0; } void ssh_kill_proxy_command(void) { /* * Send SIGHUP to proxy command if used. We don't wait() in * case it hangs and instead rely on init to reap the child */ if (proxy_command_pid > 1) kill(proxy_command_pid, SIGHUP); } /* * Set TCP receive buffer if requested. * Note: tuning needs to happen after the socket is created but before the * connection happens so winscale is negotiated properly. */ static void ssh_set_socket_recvbuf(int sock) { void *buf = (void *)&options.tcp_rcv_buf; int socksize, sz = sizeof(options.tcp_rcv_buf); socklen_t len = sizeof(int); debug("setsockopt attempting to set SO_RCVBUF to %d", options.tcp_rcv_buf); if (setsockopt(sock, SOL_SOCKET, SO_RCVBUF, buf, sz) >= 0) { getsockopt(sock, SOL_SOCKET, SO_RCVBUF, &socksize, &len); debug("setsockopt SO_RCVBUF: %.100s %d", strerror(errno), socksize); } else error("Couldn't set socket receive buffer to %d: %.100s", options.tcp_rcv_buf, strerror(errno)); } /* * Creates a (possibly privileged) socket for use as the ssh connection. */ static int ssh_create_socket(int privileged, struct addrinfo *ai) { int sock, r, gaierr; struct addrinfo hints, *res = NULL; sock = socket(ai->ai_family, ai->ai_socktype, ai->ai_protocol); if (sock < 0) { error("socket: %s", strerror(errno)); return -1; } fcntl(sock, F_SETFD, FD_CLOEXEC); if (options.tcp_rcv_buf > 0) ssh_set_socket_recvbuf(sock); /* Bind the socket to an alternative local IP address */ if (options.bind_address == NULL && !privileged) return sock; if (options.bind_address) { memset(&hints, 0, sizeof(hints)); hints.ai_family = ai->ai_family; hints.ai_socktype = ai->ai_socktype; hints.ai_protocol = ai->ai_protocol; hints.ai_flags = AI_PASSIVE; gaierr = getaddrinfo(options.bind_address, NULL, &hints, &res); if (gaierr) { error("getaddrinfo: %s: %s", options.bind_address, ssh_gai_strerror(gaierr)); close(sock); return -1; } } /* * If we are running as root and want to connect to a privileged * port, bind our own socket to a privileged port. */ if (privileged) { PRIV_START; r = bindresvport_sa(sock, res ? res->ai_addr : NULL); PRIV_END; if (r < 0) { error("bindresvport_sa: af=%d %s", ai->ai_family, strerror(errno)); goto fail; } } else { if (bind(sock, res->ai_addr, res->ai_addrlen) < 0) { error("bind: %s: %s", options.bind_address, strerror(errno)); fail: close(sock); freeaddrinfo(res); return -1; } } if (res != NULL) freeaddrinfo(res); return sock; } static int timeout_connect(int sockfd, const struct sockaddr *serv_addr, socklen_t addrlen, int *timeoutp) { fd_set *fdset; struct timeval tv, t_start; socklen_t optlen; int optval, rc, result = -1; gettimeofday(&t_start, NULL); if (*timeoutp <= 0) { result = connect(sockfd, serv_addr, addrlen); goto done; } set_nonblock(sockfd); rc = connect(sockfd, serv_addr, addrlen); if (rc == 0) { unset_nonblock(sockfd); result = 0; goto done; } if (errno != EINPROGRESS) { result = -1; goto done; } fdset = (fd_set *)xcalloc(howmany(sockfd + 1, NFDBITS), sizeof(fd_mask)); FD_SET(sockfd, fdset); ms_to_timeval(&tv, *timeoutp); for (;;) { rc = select(sockfd + 1, NULL, fdset, NULL, &tv); if (rc != -1 || errno != EINTR) break; } switch (rc) { case 0: /* Timed out */ errno = ETIMEDOUT; break; case -1: /* Select error */ debug("select: %s", strerror(errno)); break; case 1: /* Completed or failed */ optval = 0; optlen = sizeof(optval); if (getsockopt(sockfd, SOL_SOCKET, SO_ERROR, &optval, &optlen) == -1) { debug("getsockopt: %s", strerror(errno)); break; } if (optval != 0) { errno = optval; break; } result = 0; unset_nonblock(sockfd); break; default: /* Should not occur */ fatal("Bogus return (%d) from select()", rc); } free(fdset); done: if (result == 0 && *timeoutp > 0) { ms_subtract_diff(&t_start, timeoutp); if (*timeoutp <= 0) { errno = ETIMEDOUT; result = -1; } } return (result); } /* * Opens a TCP/IP connection to the remote server on the given host. * The address of the remote host will be returned in hostaddr. * If port is 0, the default port will be used. If needpriv is true, * a privileged port will be allocated to make the connection. * This requires super-user privileges if needpriv is true. * Connection_attempts specifies the maximum number of tries (one per * second). If proxy_command is non-NULL, it specifies the command (with %h * and %p substituted for host and port, respectively) to use to contact * the daemon. */ static int ssh_connect_direct(const char *host, struct addrinfo *aitop, struct sockaddr_storage *hostaddr, u_short port, int family, int connection_attempts, int *timeout_ms, int want_keepalive, int needpriv) { int on = 1; int sock = -1, attempt; char ntop[NI_MAXHOST], strport[NI_MAXSERV]; struct addrinfo *ai; debug2("ssh_connect: needpriv %d", needpriv); for (attempt = 0; attempt < connection_attempts; attempt++) { if (attempt > 0) { /* Sleep a moment before retrying. */ sleep(1); debug("Trying again..."); } /* * Loop through addresses for this host, and try each one in * sequence until the connection succeeds. */ for (ai = aitop; ai; ai = ai->ai_next) { if (ai->ai_family != AF_INET && ai->ai_family != AF_INET6) continue; if (getnameinfo(ai->ai_addr, ai->ai_addrlen, ntop, sizeof(ntop), strport, sizeof(strport), NI_NUMERICHOST|NI_NUMERICSERV) != 0) { error("ssh_connect: getnameinfo failed"); continue; } debug("Connecting to %.200s [%.100s] port %s.", host, ntop, strport); /* Create a socket for connecting. */ sock = ssh_create_socket(needpriv, ai); if (sock < 0) /* Any error is already output */ continue; if (timeout_connect(sock, ai->ai_addr, ai->ai_addrlen, timeout_ms) >= 0) { /* Successful connection. */ memcpy(hostaddr, ai->ai_addr, ai->ai_addrlen); break; } else { debug("connect to address %s port %s: %s", ntop, strport, strerror(errno)); close(sock); sock = -1; } } if (sock != -1) break; /* Successful connection. */ } /* Return failure if we didn't get a successful connection. */ if (sock == -1) { error("ssh: connect to host %s port %s: %s", host, strport, strerror(errno)); return (-1); } debug("Connection established."); /* Set SO_KEEPALIVE if requested. */ if (want_keepalive && setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE, (void *)&on, sizeof(on)) < 0) error("setsockopt SO_KEEPALIVE: %.100s", strerror(errno)); /* Set the connection. */ packet_set_connection(sock, sock); return 0; } int ssh_connect(const char *host, struct addrinfo *addrs, struct sockaddr_storage *hostaddr, u_short port, int family, int connection_attempts, int *timeout_ms, int want_keepalive, int needpriv) { if (options.proxy_command == NULL) { return ssh_connect_direct(host, addrs, hostaddr, port, family, connection_attempts, timeout_ms, want_keepalive, needpriv); } else if (strcmp(options.proxy_command, "-") == 0) { packet_set_connection(STDIN_FILENO, STDOUT_FILENO); return 0; /* Always succeeds */ } else if (options.proxy_use_fdpass) { return ssh_proxy_fdpass_connect(host, port, options.proxy_command); } return ssh_proxy_connect(host, port, options.proxy_command); } static void send_client_banner(int connection_out, int minor1) { /* Send our own protocol version identification. */ xasprintf(&client_version_string, "SSH-%d.%d-%.100s%s%s%s%s", compat20 ? PROTOCOL_MAJOR_2 : PROTOCOL_MAJOR_1, compat20 ? PROTOCOL_MINOR_2 : minor1, SSH_VERSION, options.hpn_disabled ? "" : SSH_VERSION_HPN, *options.version_addendum == '\0' ? "" : " ", options.version_addendum, compat20 ? "\r\n" : "\n"); if (roaming_atomicio(vwrite, connection_out, client_version_string, strlen(client_version_string)) != strlen(client_version_string)) fatal("write: %.100s", strerror(errno)); chop(client_version_string); debug("Local version string %.100s", client_version_string); } /* * Waits for the server identification string, and sends our own * identification string. */ void ssh_exchange_identification(int timeout_ms) { char buf[256], remote_version[256]; /* must be same size! */ int remote_major, remote_minor, mismatch; int connection_in = packet_get_connection_in(); int connection_out = packet_get_connection_out(); int minor1 = PROTOCOL_MINOR_1, client_banner_sent = 0; u_int i, n; size_t len; int fdsetsz, remaining, rc; struct timeval t_start, t_remaining; fd_set *fdset; fdsetsz = howmany(connection_in + 1, NFDBITS) * sizeof(fd_mask); fdset = xcalloc(1, fdsetsz); /* * If we are SSH2-only then we can send the banner immediately and * save a round-trip. */ if (options.protocol == SSH_PROTO_2) { enable_compat20(); send_client_banner(connection_out, 0); client_banner_sent = 1; } /* Read other side's version identification. */ remaining = timeout_ms; for (n = 0;;) { for (i = 0; i < sizeof(buf) - 1; i++) { if (timeout_ms > 0) { gettimeofday(&t_start, NULL); ms_to_timeval(&t_remaining, remaining); FD_SET(connection_in, fdset); rc = select(connection_in + 1, fdset, NULL, fdset, &t_remaining); ms_subtract_diff(&t_start, &remaining); if (rc == 0 || remaining <= 0) fatal("Connection timed out during " "banner exchange"); if (rc == -1) { if (errno == EINTR) continue; fatal("ssh_exchange_identification: " "select: %s", strerror(errno)); } } len = roaming_atomicio(read, connection_in, &buf[i], 1); if (len != 1 && errno == EPIPE) fatal("ssh_exchange_identification: " "Connection closed by remote host"); else if (len != 1) fatal("ssh_exchange_identification: " "read: %.100s", strerror(errno)); if (buf[i] == '\r') { buf[i] = '\n'; buf[i + 1] = 0; continue; /**XXX wait for \n */ } if (buf[i] == '\n') { buf[i + 1] = 0; break; } if (++n > 65536) fatal("ssh_exchange_identification: " "No banner received"); } buf[sizeof(buf) - 1] = 0; if (strncmp(buf, "SSH-", 4) == 0) break; debug("ssh_exchange_identification: %s", buf); } server_version_string = xstrdup(buf); free(fdset); /* * Check that the versions match. In future this might accept * several versions and set appropriate flags to handle them. */ if (sscanf(server_version_string, "SSH-%d.%d-%[^\n]\n", &remote_major, &remote_minor, remote_version) != 3) fatal("Bad remote protocol version identification: '%.100s'", buf); debug("Remote protocol version %d.%d, remote software version %.100s", remote_major, remote_minor, remote_version); compat_datafellows(remote_version); mismatch = 0; switch (remote_major) { case 1: if (remote_minor == 99 && (options.protocol & SSH_PROTO_2) && !(options.protocol & SSH_PROTO_1_PREFERRED)) { enable_compat20(); break; } if (!(options.protocol & SSH_PROTO_1)) { mismatch = 1; break; } if (remote_minor < 3) { fatal("Remote machine has too old SSH software version."); } else if (remote_minor == 3 || remote_minor == 4) { /* We speak 1.3, too. */ enable_compat13(); minor1 = 3; if (options.forward_agent) { logit("Agent forwarding disabled for protocol 1.3"); options.forward_agent = 0; } } break; case 2: if (options.protocol & SSH_PROTO_2) { enable_compat20(); break; } /* FALLTHROUGH */ default: mismatch = 1; break; } if (mismatch) fatal("Protocol major versions differ: %d vs. %d", (options.protocol & SSH_PROTO_2) ? PROTOCOL_MAJOR_2 : PROTOCOL_MAJOR_1, remote_major); if ((datafellows & SSH_BUG_DERIVEKEY) != 0) fatal("Server version \"%.100s\" uses unsafe key agreement; " "refusing connection", remote_version); if ((datafellows & SSH_BUG_RSASIGMD5) != 0) logit("Server version \"%.100s\" uses unsafe RSA signature " "scheme; disabling use of RSA keys", remote_version); if (!client_banner_sent) send_client_banner(connection_out, minor1); chop(server_version_string); } /* defaults to 'no' */ static int confirm(const char *prompt) { const char *msg, *again = "Please type 'yes' or 'no': "; char *p; int ret = -1; if (options.batch_mode) return 0; for (msg = prompt;;msg = again) { p = read_passphrase(msg, RP_ECHO); if (p == NULL || (p[0] == '\0') || (p[0] == '\n') || strncasecmp(p, "no", 2) == 0) ret = 0; if (p && strncasecmp(p, "yes", 3) == 0) ret = 1; free(p); if (ret != -1) return ret; } } static int check_host_cert(const char *host, const Key *host_key) { const char *reason; if (key_cert_check_authority(host_key, 1, 0, host, &reason) != 0) { error("%s", reason); return 0; } if (buffer_len(&host_key->cert->critical) != 0) { error("Certificate for %s contains unsupported " "critical options(s)", host); return 0; } return 1; } static int sockaddr_is_local(struct sockaddr *hostaddr) { switch (hostaddr->sa_family) { case AF_INET: return (ntohl(((struct sockaddr_in *)hostaddr)-> sin_addr.s_addr) >> 24) == IN_LOOPBACKNET; case AF_INET6: return IN6_IS_ADDR_LOOPBACK( &(((struct sockaddr_in6 *)hostaddr)->sin6_addr)); default: return 0; } } /* * Prepare the hostname and ip address strings that are used to lookup * host keys in known_hosts files. These may have a port number appended. */ void get_hostfile_hostname_ipaddr(char *hostname, struct sockaddr *hostaddr, u_short port, char **hostfile_hostname, char **hostfile_ipaddr) { char ntop[NI_MAXHOST]; socklen_t addrlen; switch (hostaddr == NULL ? -1 : hostaddr->sa_family) { case -1: addrlen = 0; break; case AF_INET: addrlen = sizeof(struct sockaddr_in); break; case AF_INET6: addrlen = sizeof(struct sockaddr_in6); break; default: addrlen = sizeof(struct sockaddr); break; } /* * We don't have the remote ip-address for connections * using a proxy command */ if (hostfile_ipaddr != NULL) { if (options.proxy_command == NULL) { if (getnameinfo(hostaddr, addrlen, ntop, sizeof(ntop), NULL, 0, NI_NUMERICHOST) != 0) fatal("check_host_key: getnameinfo failed"); *hostfile_ipaddr = put_host_port(ntop, port); } else { *hostfile_ipaddr = xstrdup(""); } } /* * Allow the user to record the key under a different name or * differentiate a non-standard port. This is useful for ssh * tunneling over forwarded connections or if you run multiple * sshd's on different ports on the same machine. */ if (hostfile_hostname != NULL) { if (options.host_key_alias != NULL) { *hostfile_hostname = xstrdup(options.host_key_alias); debug("using hostkeyalias: %s", *hostfile_hostname); } else { *hostfile_hostname = put_host_port(hostname, port); } } } /* * check whether the supplied host key is valid, return -1 if the key * is not valid. user_hostfile[0] will not be updated if 'readonly' is true. */ #define RDRW 0 #define RDONLY 1 #define ROQUIET 2 static int check_host_key(char *hostname, struct sockaddr *hostaddr, u_short port, Key *host_key, int readonly, char **user_hostfiles, u_int num_user_hostfiles, char **system_hostfiles, u_int num_system_hostfiles) { HostStatus host_status; HostStatus ip_status; Key *raw_key = NULL; char *ip = NULL, *host = NULL; char hostline[1000], *hostp, *fp, *ra; char msg[1024]; const char *type; const struct hostkey_entry *host_found, *ip_found; int len, cancelled_forwarding = 0; int local = sockaddr_is_local(hostaddr); int r, want_cert = key_is_cert(host_key), host_ip_differ = 0; struct hostkeys *host_hostkeys, *ip_hostkeys; u_int i; /* * Force accepting of the host key for loopback/localhost. The * problem is that if the home directory is NFS-mounted to multiple * machines, localhost will refer to a different machine in each of * them, and the user will get bogus HOST_CHANGED warnings. This * essentially disables host authentication for localhost; however, * this is probably not a real problem. */ if (options.no_host_authentication_for_localhost == 1 && local && options.host_key_alias == NULL) { debug("Forcing accepting of host key for " "loopback/localhost."); return 0; } /* * Prepare the hostname and address strings used for hostkey lookup. * In some cases, these will have a port number appended. */ get_hostfile_hostname_ipaddr(hostname, hostaddr, port, &host, &ip); /* * Turn off check_host_ip if the connection is to localhost, via proxy * command or if we don't have a hostname to compare with */ if (options.check_host_ip && (local || strcmp(hostname, ip) == 0 || options.proxy_command != NULL)) options.check_host_ip = 0; host_hostkeys = init_hostkeys(); for (i = 0; i < num_user_hostfiles; i++) load_hostkeys(host_hostkeys, host, user_hostfiles[i]); for (i = 0; i < num_system_hostfiles; i++) load_hostkeys(host_hostkeys, host, system_hostfiles[i]); ip_hostkeys = NULL; if (!want_cert && options.check_host_ip) { ip_hostkeys = init_hostkeys(); for (i = 0; i < num_user_hostfiles; i++) load_hostkeys(ip_hostkeys, ip, user_hostfiles[i]); for (i = 0; i < num_system_hostfiles; i++) load_hostkeys(ip_hostkeys, ip, system_hostfiles[i]); } retry: /* Reload these as they may have changed on cert->key downgrade */ want_cert = key_is_cert(host_key); type = key_type(host_key); /* * Check if the host key is present in the user's list of known * hosts or in the systemwide list. */ host_status = check_key_in_hostkeys(host_hostkeys, host_key, &host_found); /* * Also perform check for the ip address, skip the check if we are * localhost, looking for a certificate, or the hostname was an ip * address to begin with. */ if (!want_cert && ip_hostkeys != NULL) { ip_status = check_key_in_hostkeys(ip_hostkeys, host_key, &ip_found); if (host_status == HOST_CHANGED && (ip_status != HOST_CHANGED || (ip_found != NULL && !key_equal(ip_found->key, host_found->key)))) host_ip_differ = 1; } else ip_status = host_status; switch (host_status) { case HOST_OK: /* The host is known and the key matches. */ debug("Host '%.200s' is known and matches the %s host %s.", host, type, want_cert ? "certificate" : "key"); debug("Found %s in %s:%lu", want_cert ? "CA key" : "key", host_found->file, host_found->line); if (want_cert && !check_host_cert(hostname, host_key)) goto fail; if (options.check_host_ip && ip_status == HOST_NEW) { if (readonly || want_cert) logit("%s host key for IP address " "'%.128s' not in list of known hosts.", type, ip); else if (!add_host_to_hostfile(user_hostfiles[0], ip, host_key, options.hash_known_hosts)) logit("Failed to add the %s host key for IP " "address '%.128s' to the list of known " "hosts (%.30s).", type, ip, user_hostfiles[0]); else logit("Warning: Permanently added the %s host " "key for IP address '%.128s' to the list " "of known hosts.", type, ip); } else if (options.visual_host_key) { fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); ra = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_RANDOMART); logit("Host key fingerprint is %s\n%s\n", fp, ra); free(ra); free(fp); } break; case HOST_NEW: if (options.host_key_alias == NULL && port != 0 && port != SSH_DEFAULT_PORT) { debug("checking without port identifier"); if (check_host_key(hostname, hostaddr, 0, host_key, ROQUIET, user_hostfiles, num_user_hostfiles, system_hostfiles, num_system_hostfiles) == 0) { debug("found matching key w/out port"); break; } } if (readonly || want_cert) goto fail; /* The host is new. */ if (options.strict_host_key_checking == 1) { /* * User has requested strict host key checking. We * will not add the host key automatically. The only * alternative left is to abort. */ error("No %s host key is known for %.200s and you " "have requested strict checking.", type, host); goto fail; } else if (options.strict_host_key_checking == 2) { char msg1[1024], msg2[1024]; if (show_other_keys(host_hostkeys, host_key)) snprintf(msg1, sizeof(msg1), "\nbut keys of different type are already" " known for this host."); else snprintf(msg1, sizeof(msg1), "."); /* The default */ fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); ra = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_RANDOMART); msg2[0] = '\0'; if (options.verify_host_key_dns) { if (matching_host_key_dns) snprintf(msg2, sizeof(msg2), "Matching host key fingerprint" " found in DNS.\n"); else snprintf(msg2, sizeof(msg2), "No matching host key fingerprint" " found in DNS.\n"); } snprintf(msg, sizeof(msg), "The authenticity of host '%.200s (%s)' can't be " "established%s\n" "%s key fingerprint is %s.%s%s\n%s" "Are you sure you want to continue connecting " "(yes/no)? ", host, ip, msg1, type, fp, options.visual_host_key ? "\n" : "", options.visual_host_key ? ra : "", msg2); free(ra); free(fp); if (!confirm(msg)) goto fail; } /* * If not in strict mode, add the key automatically to the * local known_hosts file. */ if (options.check_host_ip && ip_status == HOST_NEW) { snprintf(hostline, sizeof(hostline), "%s,%s", host, ip); hostp = hostline; if (options.hash_known_hosts) { /* Add hash of host and IP separately */ r = add_host_to_hostfile(user_hostfiles[0], host, host_key, options.hash_known_hosts) && add_host_to_hostfile(user_hostfiles[0], ip, host_key, options.hash_known_hosts); } else { /* Add unhashed "host,ip" */ r = add_host_to_hostfile(user_hostfiles[0], hostline, host_key, options.hash_known_hosts); } } else { r = add_host_to_hostfile(user_hostfiles[0], host, host_key, options.hash_known_hosts); hostp = host; } if (!r) logit("Failed to add the host to the list of known " "hosts (%.500s).", user_hostfiles[0]); else logit("Warning: Permanently added '%.200s' (%s) to the " "list of known hosts.", hostp, type); break; case HOST_REVOKED: error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("@ WARNING: REVOKED HOST KEY DETECTED! @"); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("The %s host key for %s is marked as revoked.", type, host); error("This could mean that a stolen key is being used to"); error("impersonate this host."); /* * If strict host key checking is in use, the user will have * to edit the key manually and we can only abort. */ if (options.strict_host_key_checking) { error("%s host key for %.200s was revoked and you have " "requested strict checking.", type, host); goto fail; } goto continue_unsafe; case HOST_CHANGED: if (want_cert) { /* * This is only a debug() since it is valid to have * CAs with wildcard DNS matches that don't match * all hosts that one might visit. */ debug("Host certificate authority does not " "match %s in %s:%lu", CA_MARKER, host_found->file, host_found->line); goto fail; } if (readonly == ROQUIET) goto fail; if (options.check_host_ip && host_ip_differ) { char *key_msg; if (ip_status == HOST_NEW) key_msg = "is unknown"; else if (ip_status == HOST_OK) key_msg = "is unchanged"; else key_msg = "has a different value"; error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("@ WARNING: POSSIBLE DNS SPOOFING DETECTED! @"); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("The %s host key for %s has changed,", type, host); error("and the key for the corresponding IP address %s", ip); error("%s. This could either mean that", key_msg); error("DNS SPOOFING is happening or the IP address for the host"); error("and its host key have changed at the same time."); if (ip_status != HOST_NEW) error("Offending key for IP in %s:%lu", ip_found->file, ip_found->line); } /* The host key has changed. */ warn_changed_key(host_key); error("Add correct host key in %.100s to get rid of this message.", user_hostfiles[0]); error("Offending %s key in %s:%lu", key_type(host_found->key), host_found->file, host_found->line); /* * If strict host key checking is in use, the user will have * to edit the key manually and we can only abort. */ if (options.strict_host_key_checking) { error("%s host key for %.200s has changed and you have " "requested strict checking.", type, host); goto fail; } continue_unsafe: /* * If strict host key checking has not been requested, allow * the connection but without MITM-able authentication or * forwarding. */ if (options.password_authentication) { error("Password authentication is disabled to avoid " "man-in-the-middle attacks."); options.password_authentication = 0; cancelled_forwarding = 1; } if (options.kbd_interactive_authentication) { error("Keyboard-interactive authentication is disabled" " to avoid man-in-the-middle attacks."); options.kbd_interactive_authentication = 0; options.challenge_response_authentication = 0; cancelled_forwarding = 1; } if (options.challenge_response_authentication) { error("Challenge/response authentication is disabled" " to avoid man-in-the-middle attacks."); options.challenge_response_authentication = 0; cancelled_forwarding = 1; } if (options.forward_agent) { error("Agent forwarding is disabled to avoid " "man-in-the-middle attacks."); options.forward_agent = 0; cancelled_forwarding = 1; } if (options.forward_x11) { error("X11 forwarding is disabled to avoid " "man-in-the-middle attacks."); options.forward_x11 = 0; cancelled_forwarding = 1; } if (options.num_local_forwards > 0 || options.num_remote_forwards > 0) { error("Port forwarding is disabled to avoid " "man-in-the-middle attacks."); options.num_local_forwards = options.num_remote_forwards = 0; cancelled_forwarding = 1; } if (options.tun_open != SSH_TUNMODE_NO) { error("Tunnel forwarding is disabled to avoid " "man-in-the-middle attacks."); options.tun_open = SSH_TUNMODE_NO; cancelled_forwarding = 1; } if (options.exit_on_forward_failure && cancelled_forwarding) fatal("Error: forwarding disabled due to host key " "check failure"); /* * XXX Should permit the user to change to use the new id. * This could be done by converting the host key to an * identifying sentence, tell that the host identifies itself * by that sentence, and ask the user if he/she wishes to * accept the authentication. */ break; case HOST_FOUND: fatal("internal error"); break; } if (options.check_host_ip && host_status != HOST_CHANGED && ip_status == HOST_CHANGED) { snprintf(msg, sizeof(msg), "Warning: the %s host key for '%.200s' " "differs from the key for the IP address '%.128s'" "\nOffending key for IP in %s:%lu", type, host, ip, ip_found->file, ip_found->line); if (host_status == HOST_OK) { len = strlen(msg); snprintf(msg + len, sizeof(msg) - len, "\nMatching host key in %s:%lu", host_found->file, host_found->line); } if (options.strict_host_key_checking == 1) { logit("%s", msg); error("Exiting, you have requested strict checking."); goto fail; } else if (options.strict_host_key_checking == 2) { strlcat(msg, "\nAre you sure you want " "to continue connecting (yes/no)? ", sizeof(msg)); if (!confirm(msg)) goto fail; } else { logit("%s", msg); } } free(ip); free(host); if (host_hostkeys != NULL) free_hostkeys(host_hostkeys); if (ip_hostkeys != NULL) free_hostkeys(ip_hostkeys); return 0; fail: if (want_cert && host_status != HOST_REVOKED) { /* * No matching certificate. Downgrade cert to raw key and * search normally. */ debug("No matching CA found. Retry with plain key"); raw_key = key_from_private(host_key); if (key_drop_cert(raw_key) != 0) fatal("Couldn't drop certificate"); host_key = raw_key; goto retry; } if (raw_key != NULL) key_free(raw_key); free(ip); free(host); if (host_hostkeys != NULL) free_hostkeys(host_hostkeys); if (ip_hostkeys != NULL) free_hostkeys(ip_hostkeys); return -1; } /* returns 0 if key verifies or -1 if key does NOT verify */ int verify_host_key(char *host, struct sockaddr *hostaddr, Key *host_key) { int flags = 0; char *fp; + Key *plain = NULL; fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); debug("Server host key: %s %s", key_type(host_key), fp); free(fp); - /* XXX certs are not yet supported for DNS */ - if (!key_is_cert(host_key) && options.verify_host_key_dns && - verify_host_key_dns(host, hostaddr, host_key, &flags) == 0) { - if (flags & DNS_VERIFY_FOUND) { - - if (options.verify_host_key_dns == 1 && - flags & DNS_VERIFY_MATCH && - flags & DNS_VERIFY_SECURE) - return 0; - - if (flags & DNS_VERIFY_MATCH) { - matching_host_key_dns = 1; - } else { - warn_changed_key(host_key); - error("Update the SSHFP RR in DNS with the new " - "host key to get rid of this message."); + if (options.verify_host_key_dns) { + /* + * XXX certs are not yet supported for DNS, so downgrade + * them and try the plain key. + */ + plain = key_from_private(host_key); + if (key_is_cert(plain)) + key_drop_cert(plain); + if (verify_host_key_dns(host, hostaddr, plain, &flags) == 0) { + if (flags & DNS_VERIFY_FOUND) { + if (options.verify_host_key_dns == 1 && + flags & DNS_VERIFY_MATCH && + flags & DNS_VERIFY_SECURE) { + key_free(plain); + return 0; + } + if (flags & DNS_VERIFY_MATCH) { + matching_host_key_dns = 1; + } else { + warn_changed_key(plain); + error("Update the SSHFP RR in DNS " + "with the new host key to get rid " + "of this message."); + } } } + key_free(plain); } return check_host_key(host, hostaddr, options.port, host_key, RDRW, options.user_hostfiles, options.num_user_hostfiles, options.system_hostfiles, options.num_system_hostfiles); } /* * Starts a dialog with the server, and authenticates the current user on the * server. This does not need any extra privileges. The basic connection * to the server must already have been established before this is called. * If login fails, this function prints an error and never returns. * This function does not require super-user privileges. */ void ssh_login(Sensitive *sensitive, const char *orighost, struct sockaddr *hostaddr, u_short port, struct passwd *pw, int timeout_ms) { char *host; char *server_user, *local_user; local_user = xstrdup(pw->pw_name); server_user = options.user ? options.user : local_user; /* Convert the user-supplied hostname into all lowercase. */ host = xstrdup(orighost); lowercase(host); /* Exchange protocol version identification strings with the server. */ ssh_exchange_identification(timeout_ms); /* Put the connection into non-blocking mode. */ packet_set_nonblocking(); /* key exchange */ /* authenticate user */ if (compat20) { ssh_kex2(host, hostaddr, port); ssh_userauth2(local_user, server_user, host, sensitive); } else { ssh_kex(host, hostaddr); ssh_userauth1(local_user, server_user, host, sensitive); } free(local_user); } void ssh_put_password(char *password) { int size; char *padded; if (datafellows & SSH_BUG_PASSWORDPAD) { packet_put_cstring(password); return; } size = roundup(strlen(password) + 1, 32); padded = xcalloc(1, size); strlcpy(padded, password, size); packet_put_string(padded, size); explicit_bzero(padded, size); free(padded); } /* print all known host keys for a given host, but skip keys of given type */ static int show_other_keys(struct hostkeys *hostkeys, Key *key) { int type[] = { KEY_RSA1, KEY_RSA, KEY_DSA, KEY_ECDSA, KEY_ED25519, -1 }; int i, ret = 0; char *fp, *ra; const struct hostkey_entry *found; for (i = 0; type[i] != -1; i++) { if (type[i] == key->type) continue; if (!lookup_key_in_hostkeys_by_type(hostkeys, type[i], &found)) continue; fp = key_fingerprint(found->key, SSH_FP_MD5, SSH_FP_HEX); ra = key_fingerprint(found->key, SSH_FP_MD5, SSH_FP_RANDOMART); logit("WARNING: %s key found for host %s\n" "in %s:%lu\n" "%s key fingerprint %s.", key_type(found->key), found->host, found->file, found->line, key_type(found->key), fp); if (options.visual_host_key) logit("%s", ra); free(ra); free(fp); ret = 1; } return ret; } static void warn_changed_key(Key *host_key) { char *fp; fp = key_fingerprint(host_key, SSH_FP_MD5, SSH_FP_HEX); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("@ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @"); error("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@"); error("IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY!"); error("Someone could be eavesdropping on you right now (man-in-the-middle attack)!"); error("It is also possible that a host key has just been changed."); error("The fingerprint for the %s key sent by the remote host is\n%s.", key_type(host_key), fp); error("Please contact your system administrator."); free(fp); } /* * Execute a local command */ int ssh_local_cmd(const char *args) { char *shell; pid_t pid; int status; void (*osighand)(int); if (!options.permit_local_command || args == NULL || !*args) return (1); if ((shell = getenv("SHELL")) == NULL || *shell == '\0') shell = _PATH_BSHELL; osighand = signal(SIGCHLD, SIG_DFL); pid = fork(); if (pid == 0) { signal(SIGPIPE, SIG_DFL); debug3("Executing %s -c \"%s\"", shell, args); execl(shell, shell, "-c", args, (char *)NULL); error("Couldn't execute %s -c \"%s\": %s", shell, args, strerror(errno)); _exit(1); } else if (pid == -1) fatal("fork failed: %.100s", strerror(errno)); while (waitpid(pid, &status, 0) == -1) if (errno != EINTR) fatal("Couldn't wait for child: %s", strerror(errno)); signal(SIGCHLD, osighand); if (!WIFEXITED(status)) return (1); return (WEXITSTATUS(status)); } Index: stable/9/sys/netinet/tcp_reass.c =================================================================== --- stable/9/sys/netinet/tcp_reass.c (revision 285976) +++ stable/9/sys/netinet/tcp_reass.c (revision 285977) @@ -1,363 +1,351 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_input.c 8.12 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef TCPDEBUG #include #endif /* TCPDEBUG */ static int tcp_reass_sysctl_qsize(SYSCTL_HANDLER_ARGS); static SYSCTL_NODE(_net_inet_tcp, OID_AUTO, reass, CTLFLAG_RW, 0, "TCP Segment Reassembly Queue"); -static VNET_DEFINE(int, tcp_reass_maxseg) = 0; -#define V_tcp_reass_maxseg VNET(tcp_reass_maxseg) -SYSCTL_VNET_INT(_net_inet_tcp_reass, OID_AUTO, maxsegments, CTLFLAG_RDTUN, - &VNET_NAME(tcp_reass_maxseg), 0, +static int tcp_reass_maxseg = 0; +SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, maxsegments, CTLFLAG_RDTUN, + &tcp_reass_maxseg, 0, "Global maximum number of TCP Segments in Reassembly Queue"); -SYSCTL_VNET_PROC(_net_inet_tcp_reass, OID_AUTO, cursegments, +SYSCTL_PROC(_net_inet_tcp_reass, OID_AUTO, cursegments, (CTLTYPE_INT | CTLFLAG_RD), NULL, 0, &tcp_reass_sysctl_qsize, "I", "Global number of TCP Segments currently in Reassembly Queue"); -static VNET_DEFINE(int, tcp_reass_overflows) = 0; -#define V_tcp_reass_overflows VNET(tcp_reass_overflows) -SYSCTL_VNET_INT(_net_inet_tcp_reass, OID_AUTO, overflows, +static int tcp_reass_overflows = 0; +SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, overflows, CTLFLAG_RD, - &VNET_NAME(tcp_reass_overflows), 0, + &tcp_reass_overflows, 0, "Global number of TCP Segment Reassembly Queue Overflows"); -static VNET_DEFINE(uma_zone_t, tcp_reass_zone); -#define V_tcp_reass_zone VNET(tcp_reass_zone) +static uma_zone_t tcp_reass_zone; /* Initialize TCP reassembly queue */ static void tcp_reass_zone_change(void *tag) { /* Set the zone limit and read back the effective value. */ - V_tcp_reass_maxseg = nmbclusters / 16; - V_tcp_reass_maxseg = uma_zone_set_max(V_tcp_reass_zone, - V_tcp_reass_maxseg); + tcp_reass_maxseg = nmbclusters / 16; + tcp_reass_maxseg = uma_zone_set_max(tcp_reass_zone, + tcp_reass_maxseg); } void -tcp_reass_init(void) +tcp_reass_global_init(void) { - V_tcp_reass_maxseg = nmbclusters / 16; + tcp_reass_maxseg = nmbclusters / 16; TUNABLE_INT_FETCH("net.inet.tcp.reass.maxsegments", - &V_tcp_reass_maxseg); - V_tcp_reass_zone = uma_zcreate("tcpreass", sizeof (struct tseg_qent), + &tcp_reass_maxseg); + tcp_reass_zone = uma_zcreate("tcpreass", sizeof (struct tseg_qent), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); /* Set the zone limit and read back the effective value. */ - V_tcp_reass_maxseg = uma_zone_set_max(V_tcp_reass_zone, - V_tcp_reass_maxseg); + tcp_reass_maxseg = uma_zone_set_max(tcp_reass_zone, + tcp_reass_maxseg); EVENTHANDLER_REGISTER(nmbclusters_change, tcp_reass_zone_change, NULL, EVENTHANDLER_PRI_ANY); } -#ifdef VIMAGE void -tcp_reass_destroy(void) -{ - - uma_zdestroy(V_tcp_reass_zone); -} -#endif - -void tcp_reass_flush(struct tcpcb *tp) { struct tseg_qent *qe; INP_WLOCK_ASSERT(tp->t_inpcb); while ((qe = LIST_FIRST(&tp->t_segq)) != NULL) { LIST_REMOVE(qe, tqe_q); m_freem(qe->tqe_m); - uma_zfree(V_tcp_reass_zone, qe); + uma_zfree(tcp_reass_zone, qe); tp->t_segqlen--; } KASSERT((tp->t_segqlen == 0), ("TCP reass queue %p segment count is %d instead of 0 after flush.", tp, tp->t_segqlen)); } static int tcp_reass_sysctl_qsize(SYSCTL_HANDLER_ARGS) { int qsize; - qsize = uma_zone_get_cur(V_tcp_reass_zone); + qsize = uma_zone_get_cur(tcp_reass_zone); return (sysctl_handle_int(oidp, &qsize, 0, req)); } int tcp_reass(struct tcpcb *tp, struct tcphdr *th, int *tlenp, struct mbuf *m) { struct tseg_qent *q; struct tseg_qent *p = NULL; struct tseg_qent *nq; struct tseg_qent *te = NULL; struct socket *so = tp->t_inpcb->inp_socket; char *s = NULL; int flags; struct tseg_qent tqs; INP_WLOCK_ASSERT(tp->t_inpcb); /* * XXX: tcp_reass() is rather inefficient with its data structures * and should be rewritten (see NetBSD for optimizations). */ /* * Call with th==NULL after become established to * force pre-ESTABLISHED data up to user socket. */ if (th == NULL) goto present; /* * Limit the number of segments that can be queued to reduce the * potential for mbuf exhaustion. For best performance, we want to be * able to queue a full window's worth of segments. The size of the * socket receive buffer determines our advertised window and grows * automatically when socket buffer autotuning is enabled. Use it as the * basis for our queue limit. * Always let the missing segment through which caused this queue. * NB: Access to the socket buffer is left intentionally unlocked as we * can tolerate stale information here. * * XXXLAS: Using sbspace(so->so_rcv) instead of so->so_rcv.sb_hiwat * should work but causes packets to be dropped when they shouldn't. * Investigate why and re-evaluate the below limit after the behaviour * is understood. */ if ((th->th_seq != tp->rcv_nxt || !TCPS_HAVEESTABLISHED(tp->t_state)) && tp->t_segqlen >= (so->so_rcv.sb_hiwat / tp->t_maxseg) + 1) { - V_tcp_reass_overflows++; + tcp_reass_overflows++; TCPSTAT_INC(tcps_rcvmemdrop); m_freem(m); *tlenp = 0; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: queue limit reached, " "segment dropped\n", s, __func__); free(s, M_TCPLOG); } return (0); } /* * Allocate a new queue entry. If we can't, or hit the zone limit * just drop the pkt. * * Use a temporary structure on the stack for the missing segment * when the zone is exhausted. Otherwise we may get stuck. */ - te = uma_zalloc(V_tcp_reass_zone, M_NOWAIT); + te = uma_zalloc(tcp_reass_zone, M_NOWAIT); if (te == NULL) { if (th->th_seq != tp->rcv_nxt || !TCPS_HAVEESTABLISHED(tp->t_state)) { TCPSTAT_INC(tcps_rcvmemdrop); m_freem(m); *tlenp = 0; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: global zone limit " "reached, segment dropped\n", s, __func__); free(s, M_TCPLOG); } return (0); } else { bzero(&tqs, sizeof(struct tseg_qent)); te = &tqs; if ((s = tcp_log_addrs(&tp->t_inpcb->inp_inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: global zone limit reached, using " "stack for missing segment\n", s, __func__); free(s, M_TCPLOG); } } } tp->t_segqlen++; /* * Find a segment which begins after this one does. */ LIST_FOREACH(q, &tp->t_segq, tqe_q) { if (SEQ_GT(q->tqe_th->th_seq, th->th_seq)) break; p = q; } /* * If there is a preceding segment, it may provide some of * our data already. If so, drop the data from the incoming * segment. If it provides all of our data, drop us. */ if (p != NULL) { int i; /* conversion to int (in i) handles seq wraparound */ i = p->tqe_th->th_seq + p->tqe_len - th->th_seq; if (i > 0) { if (i >= *tlenp) { TCPSTAT_INC(tcps_rcvduppack); TCPSTAT_ADD(tcps_rcvdupbyte, *tlenp); m_freem(m); if (te != &tqs) - uma_zfree(V_tcp_reass_zone, te); + uma_zfree(tcp_reass_zone, te); tp->t_segqlen--; /* * Try to present any queued data * at the left window edge to the user. * This is needed after the 3-WHS * completes. */ goto present; /* ??? */ } m_adj(m, i); *tlenp -= i; th->th_seq += i; } } tp->t_rcvoopack++; TCPSTAT_INC(tcps_rcvoopack); TCPSTAT_ADD(tcps_rcvoobyte, *tlenp); /* * While we overlap succeeding segments trim them or, * if they are completely covered, dequeue them. */ while (q) { int i = (th->th_seq + *tlenp) - q->tqe_th->th_seq; if (i <= 0) break; if (i < q->tqe_len) { q->tqe_th->th_seq += i; q->tqe_len -= i; m_adj(q->tqe_m, i); break; } nq = LIST_NEXT(q, tqe_q); LIST_REMOVE(q, tqe_q); m_freem(q->tqe_m); - uma_zfree(V_tcp_reass_zone, q); + uma_zfree(tcp_reass_zone, q); tp->t_segqlen--; q = nq; } /* Insert the new segment queue entry into place. */ te->tqe_m = m; te->tqe_th = th; te->tqe_len = *tlenp; if (p == NULL) { LIST_INSERT_HEAD(&tp->t_segq, te, tqe_q); } else { KASSERT(te != &tqs, ("%s: temporary stack based entry not " "first element in queue", __func__)); LIST_INSERT_AFTER(p, te, tqe_q); } present: /* * Present data to user, advancing rcv_nxt through * completed sequence space. */ if (!TCPS_HAVEESTABLISHED(tp->t_state)) return (0); q = LIST_FIRST(&tp->t_segq); if (!q || q->tqe_th->th_seq != tp->rcv_nxt) return (0); SOCKBUF_LOCK(&so->so_rcv); do { tp->rcv_nxt += q->tqe_len; flags = q->tqe_th->th_flags & TH_FIN; nq = LIST_NEXT(q, tqe_q); LIST_REMOVE(q, tqe_q); if (so->so_rcv.sb_state & SBS_CANTRCVMORE) m_freem(q->tqe_m); else sbappendstream_locked(&so->so_rcv, q->tqe_m); if (q != &tqs) - uma_zfree(V_tcp_reass_zone, q); + uma_zfree(tcp_reass_zone, q); tp->t_segqlen--; q = nq; } while (q && q->tqe_th->th_seq == tp->rcv_nxt); ND6_HINT(tp); sorwakeup_locked(so); return (flags); } Index: stable/9/sys/netinet/tcp_subr.c =================================================================== --- stable/9/sys/netinet/tcp_subr.c (revision 285976) +++ stable/9/sys/netinet/tcp_subr.c (revision 285977) @@ -1,2328 +1,2328 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #include #include #include #include #endif #include #include #include #include #include #ifdef INET6 #include #endif #include #ifdef TCPDEBUG #include #endif #ifdef INET6 #include #endif #ifdef TCP_OFFLOAD #include #endif #ifdef IPSEC #include #include #ifdef INET6 #include #endif #include #include #endif /*IPSEC*/ #include #include #include VNET_DEFINE(int, tcp_mssdflt) = TCP_MSS; #ifdef INET6 VNET_DEFINE(int, tcp_v6mssdflt) = TCP6_MSS; #endif static int sysctl_net_inet_tcp_mss_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_mssdflt = new; } return (error); } SYSCTL_VNET_PROC(_net_inet_tcp, TCPCTL_MSSDFLT, mssdflt, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(tcp_mssdflt), 0, &sysctl_net_inet_tcp_mss_check, "I", "Default TCP Maximum Segment Size"); #ifdef INET6 static int sysctl_net_inet_tcp_mss_v6_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_v6mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_v6mssdflt = new; } return (error); } SYSCTL_VNET_PROC(_net_inet_tcp, TCPCTL_V6MSSDFLT, v6mssdflt, CTLTYPE_INT|CTLFLAG_RW, &VNET_NAME(tcp_v6mssdflt), 0, &sysctl_net_inet_tcp_mss_v6_check, "I", "Default TCP Maximum Segment Size for IPv6"); #endif /* INET6 */ /* * Minimum MSS we accept and use. This prevents DoS attacks where * we are forced to a ridiculous low MSS like 20 and send hundreds * of packets instead of one. The effect scales with the available * bandwidth and quickly saturates the CPU and network interface * with packet generation and sending. Set to zero to disable MINMSS * checking. This setting prevents us from sending too small packets. */ VNET_DEFINE(int, tcp_minmss) = TCP_MINMSS; SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, minmss, CTLFLAG_RW, &VNET_NAME(tcp_minmss), 0, "Minimum TCP Maximum Segment Size"); VNET_DEFINE(int, tcp_do_rfc1323) = 1; SYSCTL_VNET_INT(_net_inet_tcp, TCPCTL_DO_RFC1323, rfc1323, CTLFLAG_RW, &VNET_NAME(tcp_do_rfc1323), 0, "Enable rfc1323 (high performance TCP) extensions"); static int tcp_log_debug = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_debug, CTLFLAG_RW, &tcp_log_debug, 0, "Log errors caused by incoming TCP segments"); static int tcp_tcbhashsize = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, tcbhashsize, CTLFLAG_RDTUN, &tcp_tcbhashsize, 0, "Size of TCP control-block hashtable"); static int do_tcpdrain = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, do_tcpdrain, CTLFLAG_RW, &do_tcpdrain, 0, "Enable tcp_drain routine for extra help when low on mbufs"); SYSCTL_VNET_UINT(_net_inet_tcp, OID_AUTO, pcbcount, CTLFLAG_RD, &VNET_NAME(tcbinfo.ipi_count), 0, "Number of active PCBs"); static VNET_DEFINE(int, icmp_may_rst) = 1; #define V_icmp_may_rst VNET(icmp_may_rst) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, icmp_may_rst, CTLFLAG_RW, &VNET_NAME(icmp_may_rst), 0, "Certain ICMP unreachable messages may abort connections in SYN_SENT"); static VNET_DEFINE(int, tcp_isn_reseed_interval) = 0; #define V_tcp_isn_reseed_interval VNET(tcp_isn_reseed_interval) SYSCTL_VNET_INT(_net_inet_tcp, OID_AUTO, isn_reseed_interval, CTLFLAG_RW, &VNET_NAME(tcp_isn_reseed_interval), 0, "Seconds between reseeding of ISN secret"); static int tcp_soreceive_stream = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, soreceive_stream, CTLFLAG_RDTUN, &tcp_soreceive_stream, 0, "Using soreceive_stream for TCP sockets"); #ifdef TCP_SIGNATURE static int tcp_sig_checksigs = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, signature_verify_input, CTLFLAG_RW, &tcp_sig_checksigs, 0, "Verify RFC2385 digests on inbound traffic"); #endif VNET_DEFINE(uma_zone_t, sack_hole_zone); #define V_sack_hole_zone VNET(sack_hole_zone) VNET_DEFINE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST+1]); static struct inpcb *tcp_notify(struct inpcb *, int); static struct inpcb *tcp_mtudisc_notify(struct inpcb *, int); static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr); /* * Target size of TCP PCB hash tables. Must be a power of two. * * Note that this can be overridden by the kernel environment * variable net.inet.tcp.tcbhashsize */ #ifndef TCBHASHSIZE #define TCBHASHSIZE 512 #endif /* * XXX * Callouts should be moved into struct tcp directly. They are currently * separate because the tcpcb structure is exported to userland for sysctl * parsing purposes, which do not know about callouts. */ struct tcpcb_mem { struct tcpcb tcb; struct tcp_timer tt; struct cc_var ccv; struct osd osd; }; static VNET_DEFINE(uma_zone_t, tcpcb_zone); #define V_tcpcb_zone VNET(tcpcb_zone) MALLOC_DEFINE(M_TCPLOG, "tcplog", "TCP address and flags print buffers"); static struct mtx isn_mtx; #define ISN_LOCK_INIT() mtx_init(&isn_mtx, "isn_mtx", NULL, MTX_DEF) #define ISN_LOCK() mtx_lock(&isn_mtx) #define ISN_UNLOCK() mtx_unlock(&isn_mtx) /* * TCP initialization. */ static void tcp_zone_change(void *tag) { uma_zone_set_max(V_tcbinfo.ipi_zone, maxsockets); uma_zone_set_max(V_tcpcb_zone, maxsockets); tcp_tw_zone_change(); } static int tcp_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp = mem; INP_LOCK_INIT(inp, "inp", "tcpinp"); return (0); } void tcp_init(void) { int hashsize; if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_IN, &V_tcp_hhh[HHOOK_TCP_EST_IN], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_OUT, &V_tcp_hhh[HHOOK_TCP_EST_OUT], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); hashsize = TCBHASHSIZE; TUNABLE_INT_FETCH("net.inet.tcp.tcbhashsize", &hashsize); if (!powerof2(hashsize)) { printf("WARNING: TCB hash size not a power of 2\n"); hashsize = 512; /* safe default */ } in_pcbinfo_init(&V_tcbinfo, "tcp", &V_tcb, hashsize, hashsize, "tcp_inpcb", tcp_inpcb_init, NULL, UMA_ZONE_NOFREE, IPI_HASHFIELDS_4TUPLE); /* * These have to be type stable for the benefit of the timers. */ V_tcpcb_zone = uma_zcreate("tcpcb", sizeof(struct tcpcb_mem), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); uma_zone_set_max(V_tcpcb_zone, maxsockets); tcp_tw_init(); syncache_init(); tcp_hc_init(); - tcp_reass_init(); TUNABLE_INT_FETCH("net.inet.tcp.sack.enable", &V_tcp_do_sack); V_sack_hole_zone = uma_zcreate("sackhole", sizeof(struct sackhole), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); /* Skip initialization of globals for non-default instances. */ if (!IS_DEFAULT_VNET(curvnet)) return; + tcp_reass_global_init(); + /* XXX virtualize those bellow? */ tcp_delacktime = TCPTV_DELACK; tcp_keepinit = TCPTV_KEEP_INIT; tcp_keepidle = TCPTV_KEEP_IDLE; tcp_keepintvl = TCPTV_KEEPINTVL; tcp_maxpersistidle = TCPTV_KEEP_IDLE; tcp_msl = TCPTV_MSL; tcp_rexmit_min = TCPTV_MIN; if (tcp_rexmit_min < 1) tcp_rexmit_min = 1; tcp_rexmit_slop = TCPTV_CPU_VAR; tcp_finwait2_timeout = TCPTV_FINWAIT2_TIMEOUT; tcp_tcbhashsize = hashsize; TUNABLE_INT_FETCH("net.inet.tcp.soreceive_stream", &tcp_soreceive_stream); if (tcp_soreceive_stream) { #ifdef INET tcp_usrreqs.pru_soreceive = soreceive_stream; #endif #ifdef INET6 tcp6_usrreqs.pru_soreceive = soreceive_stream; #endif /* INET6 */ } #ifdef INET6 #define TCP_MINPROTOHDR (sizeof(struct ip6_hdr) + sizeof(struct tcphdr)) #else /* INET6 */ #define TCP_MINPROTOHDR (sizeof(struct tcpiphdr)) #endif /* INET6 */ if (max_protohdr < TCP_MINPROTOHDR) max_protohdr = TCP_MINPROTOHDR; if (max_linkhdr + TCP_MINPROTOHDR > MHLEN) panic("tcp_init"); #undef TCP_MINPROTOHDR ISN_LOCK_INIT(); EVENTHANDLER_REGISTER(shutdown_pre_sync, tcp_fini, NULL, SHUTDOWN_PRI_DEFAULT); EVENTHANDLER_REGISTER(maxsockets_change, tcp_zone_change, NULL, EVENTHANDLER_PRI_ANY); } #ifdef VIMAGE void tcp_destroy(void) { - tcp_reass_destroy(); tcp_hc_destroy(); syncache_destroy(); tcp_tw_destroy(); in_pcbinfo_destroy(&V_tcbinfo); uma_zdestroy(V_sack_hole_zone); uma_zdestroy(V_tcpcb_zone); } #endif void tcp_fini(void *xtp) { } /* * Fill in the IP and TCP headers for an outgoing packet, given the tcpcb. * tcp_template used to store this data in mbufs, but we now recopy it out * of the tcpcb each time to conserve mbufs. */ void tcpip_fillheaders(struct inpcb *inp, void *ip_ptr, void *tcp_ptr) { struct tcphdr *th = (struct tcphdr *)tcp_ptr; INP_WLOCK_ASSERT(inp); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { struct ip6_hdr *ip6; ip6 = (struct ip6_hdr *)ip_ptr; ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) | (inp->inp_flow & IPV6_FLOWINFO_MASK); ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) | (IPV6_VERSION & IPV6_VERSION_MASK); ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = htons(sizeof(struct tcphdr)); ip6->ip6_src = inp->in6p_laddr; ip6->ip6_dst = inp->in6p_faddr; } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET { struct ip *ip; ip = (struct ip *)ip_ptr; ip->ip_v = IPVERSION; ip->ip_hl = 5; ip->ip_tos = inp->inp_ip_tos; ip->ip_len = 0; ip->ip_id = 0; ip->ip_off = 0; ip->ip_ttl = inp->inp_ip_ttl; ip->ip_sum = 0; ip->ip_p = IPPROTO_TCP; ip->ip_src = inp->inp_laddr; ip->ip_dst = inp->inp_faddr; } #endif /* INET */ th->th_sport = inp->inp_lport; th->th_dport = inp->inp_fport; th->th_seq = 0; th->th_ack = 0; th->th_x2 = 0; th->th_off = 5; th->th_flags = 0; th->th_win = 0; th->th_urp = 0; th->th_sum = 0; /* in_pseudo() is called later for ipv4 */ } /* * Create template to be used to send tcp packets on a connection. * Allocates an mbuf and fills in a skeletal tcp/ip header. The only * use for this function is in keepalives, which use tcp_respond. */ struct tcptemp * tcpip_maketemplate(struct inpcb *inp) { struct tcptemp *t; t = malloc(sizeof(*t), M_TEMP, M_NOWAIT); if (t == NULL) return (NULL); tcpip_fillheaders(inp, (void *)&t->tt_ipgen, (void *)&t->tt_t); return (t); } /* * Send a single message to the TCP at address specified by * the given TCP/IP header. If m == NULL, then we make a copy * of the tcpiphdr at ti and send directly to the addressed host. * This is used to force keep alive messages out using the TCP * template for a connection. If flags are given then we send * a message back to the TCP which originated the * segment ti, * and discard the mbuf containing it and any other attached mbufs. * * In any case the ack and sequence number of the transmitted * segment are as specified by the parameters. * * NOTE: If m != NULL, then ti must point to *inside* the mbuf. */ void tcp_respond(struct tcpcb *tp, void *ipgen, struct tcphdr *th, struct mbuf *m, tcp_seq ack, tcp_seq seq, int flags) { int tlen; int win = 0; struct ip *ip; struct tcphdr *nth; #ifdef INET6 struct ip6_hdr *ip6; int isipv6; #endif /* INET6 */ int ipflags = 0; struct inpcb *inp; KASSERT(tp != NULL || m != NULL, ("tcp_respond: tp and m both NULL")); #ifdef INET6 isipv6 = ((struct ip *)ipgen)->ip_v == (IPV6_VERSION >> 4); ip6 = ipgen; #endif /* INET6 */ ip = ipgen; if (tp != NULL) { inp = tp->t_inpcb; KASSERT(inp != NULL, ("tcp control block w/o inpcb")); INP_WLOCK_ASSERT(inp); } else inp = NULL; if (tp != NULL) { if (!(flags & TH_RST)) { win = sbspace(&inp->inp_socket->so_rcv); if (win > (long)TCP_MAXWIN << tp->rcv_scale) win = (long)TCP_MAXWIN << tp->rcv_scale; } } if (m == NULL) { m = m_gethdr(M_DONTWAIT, MT_DATA); if (m == NULL) return; tlen = 0; m->m_data += max_linkhdr; #ifdef INET6 if (isipv6) { bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(struct ip6_hdr)); ip6 = mtod(m, struct ip6_hdr *); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); ip = mtod(m, struct ip *); nth = (struct tcphdr *)(ip + 1); } bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr)); flags = TH_ACK; } else { /* * reuse the mbuf. * XXX MRT We inherrit the FIB, which is lucky. */ m_freem(m->m_next); m->m_next = NULL; m->m_data = (caddr_t)ipgen; m_addr_changed(m); /* m_len is set later */ tlen = 0; #define xchg(a,b,type) { type t; t=a; a=b; b=t; } #ifdef INET6 if (isipv6) { xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { xchg(ip->ip_dst.s_addr, ip->ip_src.s_addr, uint32_t); nth = (struct tcphdr *)(ip + 1); } if (th != nth) { /* * this is usually a case when an extension header * exists between the IPv6 header and the * TCP header. */ nth->th_sport = th->th_sport; nth->th_dport = th->th_dport; } xchg(nth->th_dport, nth->th_sport, uint16_t); #undef xchg } #ifdef INET6 if (isipv6) { ip6->ip6_flow = 0; ip6->ip6_vfc = IPV6_VERSION; ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = 0; /* Set in ip6_output(). */ tlen += sizeof (struct ip6_hdr) + sizeof (struct tcphdr); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { tlen += sizeof (struct tcpiphdr); ip->ip_len = tlen; ip->ip_ttl = V_ip_defttl; if (V_path_mtu_discovery) ip->ip_off |= IP_DF; } #endif m->m_len = tlen; m->m_pkthdr.len = tlen; m->m_pkthdr.rcvif = NULL; #ifdef MAC if (inp != NULL) { /* * Packet is associated with a socket, so allow the * label of the response to reflect the socket label. */ INP_WLOCK_ASSERT(inp); mac_inpcb_create_mbuf(inp, m); } else { /* * Packet is not associated with a socket, so possibly * update the label in place. */ mac_netinet_tcp_reply(m); } #endif nth->th_seq = htonl(seq); nth->th_ack = htonl(ack); nth->th_x2 = 0; nth->th_off = sizeof (struct tcphdr) >> 2; nth->th_flags = flags; if (tp != NULL) nth->th_win = htons((u_short) (win >> tp->rcv_scale)); else nth->th_win = htons((u_short)win); nth->th_urp = 0; m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); #ifdef INET6 if (isipv6) { m->m_pkthdr.csum_flags = CSUM_TCP_IPV6; nth->th_sum = in6_cksum_pseudo(ip6, tlen - sizeof(struct ip6_hdr), IPPROTO_TCP, 0); ip6->ip6_hlim = in6_selecthlim(tp != NULL ? tp->t_inpcb : NULL, NULL); } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET { m->m_pkthdr.csum_flags = CSUM_TCP; nth->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htons((u_short)(tlen - sizeof(struct ip) + ip->ip_p))); } #endif /* INET */ #ifdef TCPDEBUG if (tp == NULL || (inp->inp_socket->so_options & SO_DEBUG)) tcp_trace(TA_OUTPUT, 0, tp, mtod(m, void *), th, 0); #endif #ifdef INET6 if (isipv6) (void) ip6_output(m, NULL, NULL, ipflags, NULL, NULL, inp); #endif /* INET6 */ #if defined(INET) && defined(INET6) else #endif #ifdef INET (void) ip_output(m, NULL, NULL, ipflags, NULL, inp); #endif } /* * Create a new TCP control block, making an * empty reassembly queue and hooking it to the argument * protocol control block. The `inp' parameter must have * come from the zone allocator set up in tcp_init(). */ struct tcpcb * tcp_newtcpcb(struct inpcb *inp) { struct tcpcb_mem *tm; struct tcpcb *tp; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ tm = uma_zalloc(V_tcpcb_zone, M_NOWAIT | M_ZERO); if (tm == NULL) return (NULL); tp = &tm->tcb; /* Initialise cc_var struct for this tcpcb. */ tp->ccv = &tm->ccv; tp->ccv->type = IPPROTO_TCP; tp->ccv->ccvc.tcp = tp; /* * Use the current system default CC algorithm. */ CC_LIST_RLOCK(); KASSERT(!STAILQ_EMPTY(&cc_list), ("cc_list is empty!")); CC_ALGO(tp) = CC_DEFAULT(); CC_LIST_RUNLOCK(); if (CC_ALGO(tp)->cb_init != NULL) if (CC_ALGO(tp)->cb_init(tp->ccv) > 0) { uma_zfree(V_tcpcb_zone, tm); return (NULL); } tp->osd = &tm->osd; if (khelp_init_osd(HELPER_CLASS_TCP, tp->osd)) { uma_zfree(V_tcpcb_zone, tm); return (NULL); } #ifdef VIMAGE tp->t_vnet = inp->inp_vnet; #endif tp->t_timers = &tm->tt; /* LIST_INIT(&tp->t_segq); */ /* XXX covered by M_ZERO */ tp->t_maxseg = tp->t_maxopd = #ifdef INET6 isipv6 ? V_tcp_v6mssdflt : #endif /* INET6 */ V_tcp_mssdflt; /* Set up our timeouts. */ callout_init(&tp->t_timers->tt_rexmt, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_persist, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_keep, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_2msl, CALLOUT_MPSAFE); callout_init(&tp->t_timers->tt_delack, CALLOUT_MPSAFE); if (V_tcp_do_rfc1323) tp->t_flags = (TF_REQ_SCALE|TF_REQ_TSTMP); if (V_tcp_do_sack) tp->t_flags |= TF_SACK_PERMIT; TAILQ_INIT(&tp->snd_holes); tp->t_inpcb = inp; /* XXX */ /* * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no * rtt estimate. Set rttvar so that srtt + 4 * rttvar gives * reasonable initial retransmit time. */ tp->t_srtt = TCPTV_SRTTBASE; tp->t_rttvar = ((TCPTV_RTOBASE - TCPTV_SRTTBASE) << TCP_RTTVAR_SHIFT) / 4; tp->t_rttmin = tcp_rexmit_min; tp->t_rxtcur = TCPTV_RTOBASE; tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->t_rcvtime = ticks; /* * IPv4 TTL initialization is necessary for an IPv6 socket as well, * because the socket may be bound to an IPv6 wildcard address, * which may match an IPv4-mapped IPv6 address. */ inp->inp_ip_ttl = V_ip_defttl; inp->inp_ppcb = tp; return (tp); /* XXX */ } /* * Switch the congestion control algorithm back to NewReno for any active * control blocks using an algorithm which is about to go away. * This ensures the CC framework can allow the unload to proceed without leaving * any dangling pointers which would trigger a panic. * Returning non-zero would inform the CC framework that something went wrong * and it would be unsafe to allow the unload to proceed. However, there is no * way for this to occur with this implementation so we always return zero. */ int tcp_ccalgounload(struct cc_algo *unload_algo) { struct cc_algo *tmpalgo; struct inpcb *inp; struct tcpcb *tp; VNET_ITERATOR_DECL(vnet_iter); /* * Check all active control blocks across all network stacks and change * any that are using "unload_algo" back to NewReno. If "unload_algo" * requires cleanup code to be run, call it. */ VNET_LIST_RLOCK(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); INP_INFO_RLOCK(&V_tcbinfo); /* * New connections already part way through being initialised * with the CC algo we're removing will not race with this code * because the INP_INFO_WLOCK is held during initialisation. We * therefore don't enter the loop below until the connection * list has stabilised. */ LIST_FOREACH(inp, &V_tcb, inp_list) { INP_WLOCK(inp); /* Important to skip tcptw structs. */ if (!(inp->inp_flags & INP_TIMEWAIT) && (tp = intotcpcb(inp)) != NULL) { /* * By holding INP_WLOCK here, we are assured * that the connection is not currently * executing inside the CC module's functions * i.e. it is safe to make the switch back to * NewReno. */ if (CC_ALGO(tp) == unload_algo) { tmpalgo = CC_ALGO(tp); /* NewReno does not require any init. */ CC_ALGO(tp) = &newreno_cc_algo; if (tmpalgo->cb_destroy != NULL) tmpalgo->cb_destroy(tp->ccv); } } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK(); return (0); } /* * Drop a TCP connection, reporting * the specified error. If connection is synchronized, * then send a RST to peer. */ struct tcpcb * tcp_drop(struct tcpcb *tp, int errno) { struct socket *so = tp->t_inpcb->inp_socket; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); if (TCPS_HAVERCVDSYN(tp->t_state)) { tp->t_state = TCPS_CLOSED; (void) tcp_output(tp); TCPSTAT_INC(tcps_drops); } else TCPSTAT_INC(tcps_conndrops); if (errno == ETIMEDOUT && tp->t_softerror) errno = tp->t_softerror; so->so_error = errno; return (tcp_close(tp)); } void tcp_discardcb(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so = inp->inp_socket; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ INP_WLOCK_ASSERT(inp); /* * Make sure that all of our timers are stopped before we delete the * PCB. * * XXXRW: Really, we would like to use callout_drain() here in order * to avoid races experienced in tcp_timer.c where a timer is already * executing at this point. However, we can't, both because we're * running in a context where we can't sleep, and also because we * hold locks required by the timers. What we instead need to do is * test to see if callout_drain() is required, and if so, defer some * portion of the remainder of tcp_discardcb() to an asynchronous * context that can callout_drain() and then continue. Some care * will be required to ensure that no further processing takes place * on the tcpcb, even though it hasn't been freed (a flag?). */ callout_stop(&tp->t_timers->tt_rexmt); callout_stop(&tp->t_timers->tt_persist); callout_stop(&tp->t_timers->tt_keep); callout_stop(&tp->t_timers->tt_2msl); callout_stop(&tp->t_timers->tt_delack); /* * If we got enough samples through the srtt filter, * save the rtt and rttvar in the routing entry. * 'Enough' is arbitrarily defined as 4 rtt samples. * 4 samples is enough for the srtt filter to converge * to within enough % of the correct value; fewer samples * and we could save a bogus rtt. The danger is not high * as tcp quickly recovers from everything. * XXX: Works very well but needs some more statistics! */ if (tp->t_rttupdated >= 4) { struct hc_metrics_lite metrics; u_long ssthresh; bzero(&metrics, sizeof(metrics)); /* * Update the ssthresh always when the conditions below * are satisfied. This gives us better new start value * for the congestion avoidance for new connections. * ssthresh is only set if packet loss occured on a session. * * XXXRW: 'so' may be NULL here, and/or socket buffer may be * being torn down. Ideally this code would not use 'so'. */ ssthresh = tp->snd_ssthresh; if (ssthresh != 0 && ssthresh < so->so_snd.sb_hiwat / 2) { /* * convert the limit from user data bytes to * packets then to packet data bytes. */ ssthresh = (ssthresh + tp->t_maxseg / 2) / tp->t_maxseg; if (ssthresh < 2) ssthresh = 2; ssthresh *= (u_long)(tp->t_maxseg + #ifdef INET6 (isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr) : #endif sizeof (struct tcpiphdr) #ifdef INET6 ) #endif ); } else ssthresh = 0; metrics.rmx_ssthresh = ssthresh; metrics.rmx_rtt = tp->t_srtt; metrics.rmx_rttvar = tp->t_rttvar; metrics.rmx_cwnd = tp->snd_cwnd; metrics.rmx_sendpipe = 0; metrics.rmx_recvpipe = 0; tcp_hc_update(&inp->inp_inc, &metrics); } /* free the reassembly queue, if any */ tcp_reass_flush(tp); #ifdef TCP_OFFLOAD /* Disconnect offload device, if any. */ if (tp->t_flags & TF_TOE) tcp_offload_detach(tp); #endif tcp_free_sackholes(tp); /* Allow the CC algorithm to clean up after itself. */ if (CC_ALGO(tp)->cb_destroy != NULL) CC_ALGO(tp)->cb_destroy(tp->ccv); khelp_destroy_osd(tp->osd); CC_ALGO(tp) = NULL; inp->inp_ppcb = NULL; tp->t_inpcb = NULL; uma_zfree(V_tcpcb_zone, tp); } /* * Attempt to close a TCP control block, marking it as dropped, and freeing * the socket if we hold the only reference. */ struct tcpcb * tcp_close(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); #ifdef TCP_OFFLOAD if (tp->t_state == TCPS_LISTEN) tcp_offload_listen_stop(tp); #endif in_pcbdrop(inp); TCPSTAT_INC(tcps_closed); KASSERT(inp->inp_socket != NULL, ("tcp_close: inp_socket NULL")); so = inp->inp_socket; soisdisconnected(so); if (inp->inp_flags & INP_SOCKREF) { KASSERT(so->so_state & SS_PROTOREF, ("tcp_close: !SS_PROTOREF")); inp->inp_flags &= ~INP_SOCKREF; INP_WUNLOCK(inp); ACCEPT_LOCK(); SOCK_LOCK(so); so->so_state &= ~SS_PROTOREF; sofree(so); return (NULL); } return (tp); } void tcp_drain(void) { VNET_ITERATOR_DECL(vnet_iter); if (!do_tcpdrain) return; VNET_LIST_RLOCK_NOSLEEP(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); struct inpcb *inpb; struct tcpcb *tcpb; /* * Walk the tcpbs, if existing, and flush the reassembly queue, * if there is one... * XXX: The "Net/3" implementation doesn't imply that the TCP * reassembly queue should be flushed, but in a situation * where we're really low on mbufs, this is potentially * usefull. */ INP_INFO_RLOCK(&V_tcbinfo); LIST_FOREACH(inpb, V_tcbinfo.ipi_listhead, inp_list) { if (inpb->inp_flags & INP_TIMEWAIT) continue; INP_WLOCK(inpb); if ((tcpb = intotcpcb(inpb)) != NULL) { tcp_reass_flush(tcpb); tcp_clean_sackreport(tcpb); } INP_WUNLOCK(inpb); } INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK_NOSLEEP(); } /* * Notify a tcp user of an asynchronous error; * store error as soft error, but wake up user * (for now, won't do anything until can select for soft error). * * Do not wake up user since there currently is no mechanism for * reporting soft errors (yet - a kqueue filter may be added). */ static struct inpcb * tcp_notify(struct inpcb *inp, int error) { struct tcpcb *tp; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_notify: tp == NULL")); /* * Ignore some errors if we are hooked up. * If connection hasn't completed, has retransmitted several times, * and receives a second error, give up now. This is better * than waiting a long time to establish a connection that * can never complete. */ if (tp->t_state == TCPS_ESTABLISHED && (error == EHOSTUNREACH || error == ENETUNREACH || error == EHOSTDOWN)) { return (inp); } else if (tp->t_state < TCPS_ESTABLISHED && tp->t_rxtshift > 3 && tp->t_softerror) { tp = tcp_drop(tp, error); if (tp != NULL) return (inp); else return (NULL); } else { tp->t_softerror = error; return (inp); } #if 0 wakeup( &so->so_timeo); sorwakeup(so); sowwakeup(so); #endif } static int tcp_pcblist(SYSCTL_HANDLER_ARGS) { int error, i, m, n, pcb_count; struct inpcb *inp, **inp_list; inp_gen_t gencnt; struct xinpgen xig; /* * The process of preparing the TCB list is too time-consuming and * resource-intensive to repeat twice on every request. */ if (req->oldptr == NULL) { n = V_tcbinfo.ipi_count + syncache_pcbcount(); n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xtcpcb); return (0); } if (req->newptr != NULL) return (EPERM); /* * OK, now we're committed to doing something. */ INP_INFO_RLOCK(&V_tcbinfo); gencnt = V_tcbinfo.ipi_gencnt; n = V_tcbinfo.ipi_count; INP_INFO_RUNLOCK(&V_tcbinfo); m = syncache_pcbcount(); error = sysctl_wire_old_buffer(req, 2 * (sizeof xig) + (n + m) * sizeof(struct xtcpcb)); if (error != 0) return (error); xig.xig_len = sizeof xig; xig.xig_count = n + m; xig.xig_gen = gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return (error); error = syncache_pcblist(req, m, &pcb_count); if (error) return (error); inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); if (inp_list == NULL) return (ENOMEM); INP_INFO_RLOCK(&V_tcbinfo); for (inp = LIST_FIRST(V_tcbinfo.ipi_listhead), i = 0; inp != NULL && i < n; inp = LIST_NEXT(inp, inp_list)) { INP_WLOCK(inp); if (inp->inp_gencnt <= gencnt) { /* * XXX: This use of cr_cansee(), introduced with * TCP state changes, is not quite right, but for * now, better than nothing. */ if (inp->inp_flags & INP_TIMEWAIT) { if (intotw(inp) != NULL) error = cr_cansee(req->td->td_ucred, intotw(inp)->tw_cred); else error = EINVAL; /* Skip this inp. */ } else error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) { in_pcbref(inp); inp_list[i++] = inp; } } INP_WUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); n = i; error = 0; for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (inp->inp_gencnt <= gencnt) { struct xtcpcb xt; void *inp_ppcb; bzero(&xt, sizeof(xt)); xt.xt_len = sizeof xt; /* XXX should avoid extra copy */ bcopy(inp, &xt.xt_inp, sizeof *inp); inp_ppcb = inp->inp_ppcb; if (inp_ppcb == NULL) bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); else if (inp->inp_flags & INP_TIMEWAIT) { bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); xt.xt_tp.t_state = TCPS_TIME_WAIT; } else { bcopy(inp_ppcb, &xt.xt_tp, sizeof xt.xt_tp); if (xt.xt_tp.t_timers) tcp_timer_to_xtimer(&xt.xt_tp, xt.xt_tp.t_timers, &xt.xt_timer); } if (inp->inp_socket != NULL) sotoxsocket(inp->inp_socket, &xt.xt_socket); else { bzero(&xt.xt_socket, sizeof xt.xt_socket); xt.xt_socket.xso_protocol = IPPROTO_TCP; } xt.xt_inp.inp_gencnt = inp->inp_gencnt; INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xt, sizeof xt); } else INP_RUNLOCK(inp); } INP_INFO_WLOCK(&V_tcbinfo); for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (!in_pcbrele_rlocked(inp)) INP_RUNLOCK(inp); } INP_INFO_WUNLOCK(&V_tcbinfo); if (!error) { /* * Give the user an updated idea of our state. * If the generation differs from what we told * her before, she knows that something happened * while we were processing this request, and it * might be necessary to retry. */ INP_INFO_RLOCK(&V_tcbinfo); xig.xig_gen = V_tcbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_tcbinfo.ipi_count + pcb_count; INP_INFO_RUNLOCK(&V_tcbinfo); error = SYSCTL_OUT(req, &xig, sizeof xig); } free(inp_list, M_TEMP); return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_PCBLIST, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0, tcp_pcblist, "S,xtcpcb", "List of active TCP connections"); #ifdef INET static int tcp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; struct inpcb *inp; int error; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); inp = in_pcblookup(&V_tcbinfo, addrs[1].sin_addr, addrs[1].sin_port, addrs[0].sin_addr, addrs[0].sin_port, INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else error = ENOENT; if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet_tcp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp_getcred, "S,xucred", "Get the xucred of a TCP connection"); #endif /* INET */ #ifdef INET6 static int tcp6_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in6 addrs[2]; struct inpcb *inp; int error; #ifdef INET int mapped = 0; #endif error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); if ((error = sa6_embedscope(&addrs[0], V_ip6_use_defzone)) != 0 || (error = sa6_embedscope(&addrs[1], V_ip6_use_defzone)) != 0) { return (error); } if (IN6_IS_ADDR_V4MAPPED(&addrs[0].sin6_addr)) { #ifdef INET if (IN6_IS_ADDR_V4MAPPED(&addrs[1].sin6_addr)) mapped = 1; else #endif return (EINVAL); } #ifdef INET if (mapped == 1) inp = in_pcblookup(&V_tcbinfo, *(struct in_addr *)&addrs[1].sin6_addr.s6_addr[12], addrs[1].sin6_port, *(struct in_addr *)&addrs[0].sin6_addr.s6_addr[12], addrs[0].sin6_port, INPLOOKUP_RLOCKPCB, NULL); else #endif inp = in6_pcblookup(&V_tcbinfo, &addrs[1].sin6_addr, addrs[1].sin6_port, &addrs[0].sin6_addr, addrs[0].sin6_port, INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else error = ENOENT; if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet6_tcp6, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp6_getcred, "S,xucred", "Get the xucred of a TCP6 connection"); #endif /* INET6 */ #ifdef INET void tcp_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct ip *ip = vip; struct tcphdr *th; struct in_addr faddr; struct inpcb *inp; struct tcpcb *tp; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct icmp *icp; struct in_conninfo inc; tcp_seq icmp_tcp_seq; int mtu; faddr = ((struct sockaddr_in *)sa)->sin_addr; if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc_notify; else if (V_icmp_may_rst && (cmd == PRC_UNREACH_ADMIN_PROHIB || cmd == PRC_UNREACH_PORT || cmd == PRC_TIMXCEED_INTRANS) && ip) notify = tcp_drop_syn_sent; /* * Redirects don't need to be handled up here. */ else if (PRC_IS_REDIRECT(cmd)) return; /* * Source quench is depreciated. */ else if (cmd == PRC_QUENCH) return; /* * Hostdead is ugly because it goes linearly through all PCBs. * XXX: We never get this from ICMP, otherwise it makes an * excellent DoS attack on machines with many connections. */ else if (cmd == PRC_HOSTDEAD) ip = NULL; else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) return; if (ip != NULL) { icp = (struct icmp *)((caddr_t)ip - offsetof(struct icmp, icmp_ip)); th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); INP_INFO_WLOCK(&V_tcbinfo); inp = in_pcblookup(&V_tcbinfo, faddr, th->th_dport, ip->ip_src, th->th_sport, INPLOOKUP_WLOCKPCB, NULL); if (inp != NULL) { if (!(inp->inp_flags & INP_TIMEWAIT) && !(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket == NULL)) { icmp_tcp_seq = htonl(th->th_seq); tp = intotcpcb(inp); if (SEQ_GEQ(icmp_tcp_seq, tp->snd_una) && SEQ_LT(icmp_tcp_seq, tp->snd_max)) { if (cmd == PRC_MSGSIZE) { /* * MTU discovery: * If we got a needfrag set the MTU * in the route to the suggested new * value (if given) and then notify. */ bzero(&inc, sizeof(inc)); inc.inc_faddr = faddr; inc.inc_fibnum = inp->inp_inc.inc_fibnum; mtu = ntohs(icp->icmp_nextmtu); /* * If no alternative MTU was * proposed, try the next smaller * one. ip->ip_len has already * been swapped in icmp_input(). */ if (!mtu) mtu = ip_next_mtu(ip->ip_len, 1); if (mtu < V_tcp_minmss + sizeof(struct tcpiphdr)) mtu = V_tcp_minmss + sizeof(struct tcpiphdr); /* * Only cache the MTU if it * is smaller than the interface * or route MTU. tcp_mtudisc() * will do right thing by itself. */ if (mtu <= tcp_maxmtu(&inc, NULL)) tcp_hc_updatemtu(&inc, mtu); tcp_mtudisc(inp, mtu); } else inp = (*notify)(inp, inetctlerrmap[cmd]); } } if (inp != NULL) INP_WUNLOCK(inp); } else { bzero(&inc, sizeof(inc)); inc.inc_fport = th->th_dport; inc.inc_lport = th->th_sport; inc.inc_faddr = faddr; inc.inc_laddr = ip->ip_src; syncache_unreach(&inc, th); } INP_INFO_WUNLOCK(&V_tcbinfo); } else in_pcbnotifyall(&V_tcbinfo, faddr, inetctlerrmap[cmd], notify); } #endif /* INET */ #ifdef INET6 void tcp6_ctlinput(int cmd, struct sockaddr *sa, void *d) { struct tcphdr th; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct ip6_hdr *ip6; struct mbuf *m; struct ip6ctlparam *ip6cp = NULL; const struct sockaddr_in6 *sa6_src = NULL; int off; struct tcp_portonly { u_int16_t th_sport; u_int16_t th_dport; } *thp; if (sa->sa_family != AF_INET6 || sa->sa_len != sizeof(struct sockaddr_in6)) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc_notify; else if (!PRC_IS_REDIRECT(cmd) && ((unsigned)cmd >= PRC_NCMDS || inet6ctlerrmap[cmd] == 0)) return; /* Source quench is depreciated. */ else if (cmd == PRC_QUENCH) return; /* if the parameter is from icmp6, decode it. */ if (d != NULL) { ip6cp = (struct ip6ctlparam *)d; m = ip6cp->ip6c_m; ip6 = ip6cp->ip6c_ip6; off = ip6cp->ip6c_off; sa6_src = ip6cp->ip6c_src; } else { m = NULL; ip6 = NULL; off = 0; /* fool gcc */ sa6_src = &sa6_any; } if (ip6 != NULL) { struct in_conninfo inc; /* * XXX: We assume that when IPV6 is non NULL, * M and OFF are valid. */ /* check if we can safely examine src and dst ports */ if (m->m_pkthdr.len < off + sizeof(*thp)) return; bzero(&th, sizeof(th)); m_copydata(m, off, sizeof(*thp), (caddr_t)&th); in6_pcbnotify(&V_tcbinfo, sa, th.th_dport, (struct sockaddr *)ip6cp->ip6c_src, th.th_sport, cmd, NULL, notify); bzero(&inc, sizeof(inc)); inc.inc_fport = th.th_dport; inc.inc_lport = th.th_sport; inc.inc6_faddr = ((struct sockaddr_in6 *)sa)->sin6_addr; inc.inc6_laddr = ip6cp->ip6c_src->sin6_addr; inc.inc_flags |= INC_ISIPV6; INP_INFO_WLOCK(&V_tcbinfo); syncache_unreach(&inc, &th); INP_INFO_WUNLOCK(&V_tcbinfo); } else in6_pcbnotify(&V_tcbinfo, sa, 0, (const struct sockaddr *)sa6_src, 0, cmd, NULL, notify); } #endif /* INET6 */ /* * Following is where TCP initial sequence number generation occurs. * * There are two places where we must use initial sequence numbers: * 1. In SYN-ACK packets. * 2. In SYN packets. * * All ISNs for SYN-ACK packets are generated by the syncache. See * tcp_syncache.c for details. * * The ISNs in SYN packets must be monotonic; TIME_WAIT recycling * depends on this property. In addition, these ISNs should be * unguessable so as to prevent connection hijacking. To satisfy * the requirements of this situation, the algorithm outlined in * RFC 1948 is used, with only small modifications. * * Implementation details: * * Time is based off the system timer, and is corrected so that it * increases by one megabyte per second. This allows for proper * recycling on high speed LANs while still leaving over an hour * before rollover. * * As reading the *exact* system time is too expensive to be done * whenever setting up a TCP connection, we increment the time * offset in two ways. First, a small random positive increment * is added to isn_offset for each connection that is set up. * Second, the function tcp_isn_tick fires once per clock tick * and increments isn_offset as necessary so that sequence numbers * are incremented at approximately ISN_BYTES_PER_SECOND. The * random positive increments serve only to ensure that the same * exact sequence number is never sent out twice (as could otherwise * happen when a port is recycled in less than the system tick * interval.) * * net.inet.tcp.isn_reseed_interval controls the number of seconds * between seeding of isn_secret. This is normally set to zero, * as reseeding should not be necessary. * * Locking of the global variables isn_secret, isn_last_reseed, isn_offset, * isn_offset_old, and isn_ctx is performed using the TCP pcbinfo lock. In * general, this means holding an exclusive (write) lock. */ #define ISN_BYTES_PER_SECOND 1048576 #define ISN_STATIC_INCREMENT 4096 #define ISN_RANDOM_INCREMENT (4096 - 1) static VNET_DEFINE(u_char, isn_secret[32]); static VNET_DEFINE(int, isn_last); static VNET_DEFINE(int, isn_last_reseed); static VNET_DEFINE(u_int32_t, isn_offset); static VNET_DEFINE(u_int32_t, isn_offset_old); #define V_isn_secret VNET(isn_secret) #define V_isn_last VNET(isn_last) #define V_isn_last_reseed VNET(isn_last_reseed) #define V_isn_offset VNET(isn_offset) #define V_isn_offset_old VNET(isn_offset_old) tcp_seq tcp_new_isn(struct tcpcb *tp) { MD5_CTX isn_ctx; u_int32_t md5_buffer[4]; tcp_seq new_isn; u_int32_t projected_offset; INP_WLOCK_ASSERT(tp->t_inpcb); ISN_LOCK(); /* Seed if this is the first use, reseed if requested. */ if ((V_isn_last_reseed == 0) || ((V_tcp_isn_reseed_interval > 0) && (((u_int)V_isn_last_reseed + (u_int)V_tcp_isn_reseed_interval*hz) < (u_int)ticks))) { read_random(&V_isn_secret, sizeof(V_isn_secret)); V_isn_last_reseed = ticks; } /* Compute the md5 hash and return the ISN. */ MD5Init(&isn_ctx); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_fport, sizeof(u_short)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_lport, sizeof(u_short)); #ifdef INET6 if ((tp->t_inpcb->inp_vflag & INP_IPV6) != 0) { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_faddr, sizeof(struct in6_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_laddr, sizeof(struct in6_addr)); } else #endif { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_faddr, sizeof(struct in_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_laddr, sizeof(struct in_addr)); } MD5Update(&isn_ctx, (u_char *) &V_isn_secret, sizeof(V_isn_secret)); MD5Final((u_char *) &md5_buffer, &isn_ctx); new_isn = (tcp_seq) md5_buffer[0]; V_isn_offset += ISN_STATIC_INCREMENT + (arc4random() & ISN_RANDOM_INCREMENT); if (ticks != V_isn_last) { projected_offset = V_isn_offset_old + ISN_BYTES_PER_SECOND / hz * (ticks - V_isn_last); if (SEQ_GT(projected_offset, V_isn_offset)) V_isn_offset = projected_offset; V_isn_offset_old = V_isn_offset; V_isn_last = ticks; } new_isn += V_isn_offset; ISN_UNLOCK(); return (new_isn); } /* * When a specific ICMP unreachable message is received and the * connection state is SYN-SENT, drop the connection. This behavior * is controlled by the icmp_may_rst sysctl. */ struct inpcb * tcp_drop_syn_sent(struct inpcb *inp, int errno) { struct tcpcb *tp; INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); if (tp->t_state != TCPS_SYN_SENT) return (inp); tp = tcp_drop(tp, errno); if (tp != NULL) return (inp); else return (NULL); } /* * When `need fragmentation' ICMP is received, update our idea of the MSS * based on the new value. Also nudge TCP to send something, since we * know the packet we just sent was dropped. * This duplicates some code in the tcp_mss() function in tcp_input.c. */ static struct inpcb * tcp_mtudisc_notify(struct inpcb *inp, int error) { return (tcp_mtudisc(inp, -1)); } struct inpcb * tcp_mtudisc(struct inpcb *inp, int mtuoffer) { struct tcpcb *tp; struct socket *so; INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_mtudisc: tp == NULL")); tcp_mss_update(tp, -1, mtuoffer, NULL, NULL); so = inp->inp_socket; SOCKBUF_LOCK(&so->so_snd); /* If the mss is larger than the socket buffer, decrease the mss. */ if (so->so_snd.sb_hiwat < tp->t_maxseg) tp->t_maxseg = so->so_snd.sb_hiwat; SOCKBUF_UNLOCK(&so->so_snd); TCPSTAT_INC(tcps_mturesent); tp->t_rtttime = 0; tp->snd_nxt = tp->snd_una; tcp_free_sackholes(tp); tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_SACK_PERMIT) EXIT_FASTRECOVERY(tp->t_flags); tcp_output(tp); return (inp); } #ifdef INET /* * Look-up the routing entry to the peer of this inpcb. If no route * is found and it cannot be allocated, then return 0. This routine * is called by TCP routines that access the rmx structure and by * tcp_mss_update to get the peer/interface MTU. */ u_long tcp_maxmtu(struct in_conninfo *inc, struct tcp_ifcap *cap) { struct route sro; struct sockaddr_in *dst; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu with NULL in_conninfo pointer")); bzero(&sro, sizeof(sro)); if (inc->inc_faddr.s_addr != INADDR_ANY) { dst = (struct sockaddr_in *)&sro.ro_dst; dst->sin_family = AF_INET; dst->sin_len = sizeof(*dst); dst->sin_addr = inc->inc_faddr; in_rtalloc_ign(&sro, 0, inc->inc_fibnum); } if (sro.ro_rt != NULL) { ifp = sro.ro_rt->rt_ifp; if (sro.ro_rt->rt_rmx.rmx_mtu == 0) maxmtu = ifp->if_mtu; else maxmtu = min(sro.ro_rt->rt_rmx.rmx_mtu, ifp->if_mtu); /* Report additional interface capabilities. */ if (cap != NULL) { if (ifp->if_capenable & IFCAP_TSO4 && ifp->if_hwassist & CSUM_TSO) { cap->ifcap |= CSUM_TSO; cap->tsomax = ifp->if_hw_tsomax; cap->tsomaxsegcount = ifp->if_hw_tsomaxsegcount; cap->tsomaxsegsize = ifp->if_hw_tsomaxsegsize; } } RTFREE(sro.ro_rt); } return (maxmtu); } #endif /* INET */ #ifdef INET6 u_long tcp_maxmtu6(struct in_conninfo *inc, struct tcp_ifcap *cap) { struct route_in6 sro6; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu6 with NULL in_conninfo pointer")); bzero(&sro6, sizeof(sro6)); if (!IN6_IS_ADDR_UNSPECIFIED(&inc->inc6_faddr)) { sro6.ro_dst.sin6_family = AF_INET6; sro6.ro_dst.sin6_len = sizeof(struct sockaddr_in6); sro6.ro_dst.sin6_addr = inc->inc6_faddr; in6_rtalloc_ign(&sro6, 0, inc->inc_fibnum); } if (sro6.ro_rt != NULL) { ifp = sro6.ro_rt->rt_ifp; if (sro6.ro_rt->rt_rmx.rmx_mtu == 0) maxmtu = IN6_LINKMTU(sro6.ro_rt->rt_ifp); else maxmtu = min(sro6.ro_rt->rt_rmx.rmx_mtu, IN6_LINKMTU(sro6.ro_rt->rt_ifp)); /* Report additional interface capabilities. */ if (cap != NULL) { if (ifp->if_capenable & IFCAP_TSO6 && ifp->if_hwassist & CSUM_TSO) { cap->ifcap |= CSUM_TSO; cap->tsomax = ifp->if_hw_tsomax; cap->tsomaxsegcount = ifp->if_hw_tsomaxsegcount; cap->tsomaxsegsize = ifp->if_hw_tsomaxsegsize; } } RTFREE(sro6.ro_rt); } return (maxmtu); } #endif /* INET6 */ #ifdef IPSEC /* compute ESP/AH header size for TCP, including outer IP header. */ size_t ipsec_hdrsiz_tcp(struct tcpcb *tp) { struct inpcb *inp; struct mbuf *m; size_t hdrsiz; struct ip *ip; #ifdef INET6 struct ip6_hdr *ip6; #endif struct tcphdr *th; if ((tp == NULL) || ((inp = tp->t_inpcb) == NULL)) return (0); MGETHDR(m, M_DONTWAIT, MT_DATA); if (!m) return (0); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)(ip6 + 1); m->m_pkthdr.len = m->m_len = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); tcpip_fillheaders(inp, ip6, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } else #endif /* INET6 */ { ip = mtod(m, struct ip *); th = (struct tcphdr *)(ip + 1); m->m_pkthdr.len = m->m_len = sizeof(struct tcpiphdr); tcpip_fillheaders(inp, ip, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } m_free(m); return (hdrsiz); } #endif /* IPSEC */ #ifdef TCP_SIGNATURE /* * Callback function invoked by m_apply() to digest TCP segment data * contained within an mbuf chain. */ static int tcp_signature_apply(void *fstate, void *data, u_int len) { MD5Update(fstate, (u_char *)data, len); return (0); } /* * Compute TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * _unused * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * direction direction of flow (IPSEC_DIR_INBOUND or OUTBOUND) * * We do this over ip, tcphdr, segment data, and the key in the SADB. * When called from tcp_input(), we can be sure that th_sum has been * zeroed out and verified already. * * Return 0 if successful, otherwise return -1. * * XXX The key is retrieved from the system's PF_KEY SADB, by keying a * search with the destination IP address, and a 'magic SPI' to be * determined by the application. This is hardcoded elsewhere to 1179 * right now. Another branch of this code exists which uses the SPD to * specify per-application flows but it is unstable. */ int tcp_signature_compute(struct mbuf *m, int _unused, int len, int optlen, u_char *buf, u_int direction) { union sockaddr_union dst; #ifdef INET struct ippseudo ippseudo; #endif MD5_CTX ctx; int doff; struct ip *ip; #ifdef INET struct ipovly *ipovly; #endif struct secasvar *sav; struct tcphdr *th; #ifdef INET6 struct ip6_hdr *ip6; struct in6_addr in6; char ip6buf[INET6_ADDRSTRLEN]; uint32_t plen; uint16_t nhdr; #endif u_short savecsum; KASSERT(m != NULL, ("NULL mbuf chain")); KASSERT(buf != NULL, ("NULL signature pointer")); /* Extract the destination from the IP header in the mbuf. */ bzero(&dst, sizeof(union sockaddr_union)); ip = mtod(m, struct ip *); #ifdef INET6 ip6 = NULL; /* Make the compiler happy. */ #endif switch (ip->ip_v) { #ifdef INET case IPVERSION: dst.sa.sa_len = sizeof(struct sockaddr_in); dst.sa.sa_family = AF_INET; dst.sin.sin_addr = (direction == IPSEC_DIR_INBOUND) ? ip->ip_src : ip->ip_dst; break; #endif #ifdef INET6 case (IPV6_VERSION >> 4): ip6 = mtod(m, struct ip6_hdr *); dst.sa.sa_len = sizeof(struct sockaddr_in6); dst.sa.sa_family = AF_INET6; dst.sin6.sin6_addr = (direction == IPSEC_DIR_INBOUND) ? ip6->ip6_src : ip6->ip6_dst; break; #endif default: return (EINVAL); /* NOTREACHED */ break; } /* Look up an SADB entry which matches the address of the peer. */ sav = KEY_ALLOCSA(&dst, IPPROTO_TCP, htonl(TCP_SIG_SPI)); if (sav == NULL) { ipseclog((LOG_ERR, "%s: SADB lookup failed for %s\n", __func__, (ip->ip_v == IPVERSION) ? inet_ntoa(dst.sin.sin_addr) : #ifdef INET6 (ip->ip_v == (IPV6_VERSION >> 4)) ? ip6_sprintf(ip6buf, &dst.sin6.sin6_addr) : #endif "(unsupported)")); return (EINVAL); } MD5Init(&ctx); /* * Step 1: Update MD5 hash with IP(v6) pseudo-header. * * XXX The ippseudo header MUST be digested in network byte order, * or else we'll fail the regression test. Assume all fields we've * been doing arithmetic on have been in host byte order. * XXX One cannot depend on ipovly->ih_len here. When called from * tcp_output(), the underlying ip_len member has not yet been set. */ switch (ip->ip_v) { #ifdef INET case IPVERSION: ipovly = (struct ipovly *)ip; ippseudo.ippseudo_src = ipovly->ih_src; ippseudo.ippseudo_dst = ipovly->ih_dst; ippseudo.ippseudo_pad = 0; ippseudo.ippseudo_p = IPPROTO_TCP; ippseudo.ippseudo_len = htons(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&ippseudo, sizeof(struct ippseudo)); th = (struct tcphdr *)((u_char *)ip + sizeof(struct ip)); doff = sizeof(struct ip) + sizeof(struct tcphdr) + optlen; break; #endif #ifdef INET6 /* * RFC 2385, 2.0 Proposal * For IPv6, the pseudo-header is as described in RFC 2460, namely the * 128-bit source IPv6 address, 128-bit destination IPv6 address, zero- * extended next header value (to form 32 bits), and 32-bit segment * length. * Note: Upper-Layer Packet Length comes before Next Header. */ case (IPV6_VERSION >> 4): in6 = ip6->ip6_src; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); in6 = ip6->ip6_dst; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); plen = htonl(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&plen, sizeof(uint32_t)); nhdr = 0; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); nhdr = IPPROTO_TCP; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); th = (struct tcphdr *)((u_char *)ip6 + sizeof(struct ip6_hdr)); doff = sizeof(struct ip6_hdr) + sizeof(struct tcphdr) + optlen; break; #endif default: return (EINVAL); /* NOTREACHED */ break; } /* * Step 2: Update MD5 hash with TCP header, excluding options. * The TCP checksum must be set to zero. */ savecsum = th->th_sum; th->th_sum = 0; MD5Update(&ctx, (char *)th, sizeof(struct tcphdr)); th->th_sum = savecsum; /* * Step 3: Update MD5 hash with TCP segment data. * Use m_apply() to avoid an early m_pullup(). */ if (len > 0) m_apply(m, doff, len, tcp_signature_apply, &ctx); /* * Step 4: Update MD5 hash with shared secret. */ MD5Update(&ctx, sav->key_auth->key_data, _KEYLEN(sav->key_auth)); MD5Final(buf, &ctx); key_sa_recordxfer(sav, m); KEY_FREESAV(&sav); return (0); } /* * Verify the TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * direction direction of flow (IPSEC_DIR_INBOUND or OUTBOUND) * * Return 1 if successful, otherwise return 0. */ int tcp_signature_verify(struct mbuf *m, int off0, int tlen, int optlen, struct tcpopt *to, struct tcphdr *th, u_int tcpbflag) { char tmpdigest[TCP_SIGLEN]; if (tcp_sig_checksigs == 0) return (1); if ((tcpbflag & TF_SIGNATURE) == 0) { if ((to->to_flags & TOF_SIGNATURE) != 0) { /* * If this socket is not expecting signature but * the segment contains signature just fail. */ TCPSTAT_INC(tcps_sig_err_sigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } /* Signature is not expected, and not present in segment. */ return (1); } /* * If this socket is expecting signature but the segment does not * contain any just fail. */ if ((to->to_flags & TOF_SIGNATURE) == 0) { TCPSTAT_INC(tcps_sig_err_nosigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (tcp_signature_compute(m, off0, tlen, optlen, &tmpdigest[0], IPSEC_DIR_INBOUND) == -1) { TCPSTAT_INC(tcps_sig_err_buildsig); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (bcmp(to->to_signature, &tmpdigest[0], TCP_SIGLEN) != 0) { TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } TCPSTAT_INC(tcps_sig_rcvgoodsig); return (1); } #endif /* TCP_SIGNATURE */ static int sysctl_drop(SYSCTL_HANDLER_ARGS) { /* addrs[0] is a foreign socket, addrs[1] is a local one. */ struct sockaddr_storage addrs[2]; struct inpcb *inp; struct tcpcb *tp; struct tcptw *tw; struct sockaddr_in *fin, *lin; #ifdef INET6 struct sockaddr_in6 *fin6, *lin6; #endif int error; inp = NULL; fin = lin = NULL; #ifdef INET6 fin6 = lin6 = NULL; #endif error = 0; if (req->oldptr != NULL || req->oldlen != 0) return (EINVAL); if (req->newptr == NULL) return (EPERM); if (req->newlen < sizeof(addrs)) return (ENOMEM); error = SYSCTL_IN(req, &addrs, sizeof(addrs)); if (error) return (error); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: fin6 = (struct sockaddr_in6 *)&addrs[0]; lin6 = (struct sockaddr_in6 *)&addrs[1]; if (fin6->sin6_len != sizeof(struct sockaddr_in6) || lin6->sin6_len != sizeof(struct sockaddr_in6)) return (EINVAL); if (IN6_IS_ADDR_V4MAPPED(&fin6->sin6_addr)) { if (!IN6_IS_ADDR_V4MAPPED(&lin6->sin6_addr)) return (EINVAL); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[0]); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[1]); fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; break; } error = sa6_embedscope(fin6, V_ip6_use_defzone); if (error) return (error); error = sa6_embedscope(lin6, V_ip6_use_defzone); if (error) return (error); break; #endif #ifdef INET case AF_INET: fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; if (fin->sin_len != sizeof(struct sockaddr_in) || lin->sin_len != sizeof(struct sockaddr_in)) return (EINVAL); break; #endif default: return (EINVAL); } INP_INFO_WLOCK(&V_tcbinfo); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: inp = in6_pcblookup(&V_tcbinfo, &fin6->sin6_addr, fin6->sin6_port, &lin6->sin6_addr, lin6->sin6_port, INPLOOKUP_WLOCKPCB, NULL); break; #endif #ifdef INET case AF_INET: inp = in_pcblookup(&V_tcbinfo, fin->sin_addr, fin->sin_port, lin->sin_addr, lin->sin_port, INPLOOKUP_WLOCKPCB, NULL); break; #endif } if (inp != NULL) { if (inp->inp_flags & INP_TIMEWAIT) { /* * XXXRW: There currently exists a state where an * inpcb is present, but its timewait state has been * discarded. For now, don't allow dropping of this * type of inpcb. */ tw = intotw(inp); if (tw != NULL) tcp_twclose(tw, 0); else INP_WUNLOCK(inp); } else if (!(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket->so_options & SO_ACCEPTCONN)) { tp = intotcpcb(inp); tp = tcp_drop(tp, ECONNABORTED); if (tp != NULL) INP_WUNLOCK(inp); } else INP_WUNLOCK(inp); } else error = ESRCH; INP_INFO_WUNLOCK(&V_tcbinfo); return (error); } SYSCTL_VNET_PROC(_net_inet_tcp, TCPCTL_DROP, drop, CTLTYPE_STRUCT|CTLFLAG_WR|CTLFLAG_SKIP, NULL, 0, sysctl_drop, "", "Drop TCP connection"); /* * Generate a standardized TCP log line for use throughout the * tcp subsystem. Memory allocation is done with M_NOWAIT to * allow use in the interrupt context. * * NB: The caller MUST free(s, M_TCPLOG) the returned string. * NB: The function may return NULL if memory allocation failed. * * Due to header inclusion and ordering limitations the struct ip * and ip6_hdr pointers have to be passed as void pointers. */ char * tcp_log_vain(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_in_vain == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } char * tcp_log_addrs(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_debug == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { char *s, *sp; size_t size; struct ip *ip; #ifdef INET6 const struct ip6_hdr *ip6; ip6 = (const struct ip6_hdr *)ip6hdr; #endif /* INET6 */ ip = (struct ip *)ip4hdr; /* * The log line looks like this: * "TCP: [1.2.3.4]:50332 to [1.2.3.4]:80 tcpflags 0x2" */ size = sizeof("TCP: []:12345 to []:12345 tcpflags 0x2<>") + sizeof(PRINT_TH_FLAGS) + 1 + #ifdef INET6 2 * INET6_ADDRSTRLEN; #else 2 * INET_ADDRSTRLEN; #endif /* INET6 */ s = malloc(size, M_TCPLOG, M_ZERO|M_NOWAIT); if (s == NULL) return (NULL); strcat(s, "TCP: ["); sp = s + strlen(s); if (inc && ((inc->inc_flags & INC_ISIPV6) == 0)) { inet_ntoa_r(inc->inc_faddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); inet_ntoa_r(inc->inc_laddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); #ifdef INET6 } else if (inc) { ip6_sprintf(sp, &inc->inc6_faddr); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); ip6_sprintf(sp, &inc->inc6_laddr); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); } else if (ip6 && th) { ip6_sprintf(sp, &ip6->ip6_src); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); ip6_sprintf(sp, &ip6->ip6_dst); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); #endif /* INET6 */ #ifdef INET } else if (ip && th) { inet_ntoa_r(ip->ip_src, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); inet_ntoa_r(ip->ip_dst, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); #endif /* INET */ } else { free(s, M_TCPLOG); return (NULL); } sp = s + strlen(s); if (th) sprintf(sp, " tcpflags 0x%b", th->th_flags, PRINT_TH_FLAGS); if (*(s + size - 1) != '\0') panic("%s: string too long", __func__); return (s); } Index: stable/9/sys/netinet/tcp_var.h =================================================================== --- stable/9/sys/netinet/tcp_var.h (revision 285976) +++ stable/9/sys/netinet/tcp_var.h (revision 285977) @@ -1,758 +1,755 @@ /*- * Copyright (c) 1982, 1986, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 * $FreeBSD$ */ #ifndef _NETINET_TCP_VAR_H_ #define _NETINET_TCP_VAR_H_ #include #ifdef _KERNEL #include /* * Kernel variables for tcp. */ VNET_DECLARE(int, tcp_do_rfc1323); #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) #endif /* _KERNEL */ /* TCP segment queue entry */ struct tseg_qent { LIST_ENTRY(tseg_qent) tqe_q; int tqe_len; /* TCP segment data length */ struct tcphdr *tqe_th; /* a pointer to tcp header */ struct mbuf *tqe_m; /* mbuf contains packet */ }; LIST_HEAD(tsegqe_head, tseg_qent); struct sackblk { tcp_seq start; /* start seq no. of sack block */ tcp_seq end; /* end seq no. */ }; struct sackhole { tcp_seq start; /* start seq no. of hole */ tcp_seq end; /* end seq no. */ tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ }; struct sackhint { struct sackhole *nexthole; int sack_bytes_rexmit; tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ int ispare; /* explicit pad for 64bit alignment */ uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */ }; struct tcptemp { u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ struct tcphdr tt_t; }; #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ /* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ #ifdef INET6 #define ND6_HINT(tp) \ do { \ if ((tp) && (tp)->t_inpcb && \ ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ nd6_nud_hint(NULL, NULL, 0); \ } while (0) #else #define ND6_HINT(tp) #endif /* * Tcp control block, one per tcp; fields: * Organized for 16 byte cacheline efficiency. */ struct tcpcb { struct tsegqe_head t_segq; /* segment reassembly queue */ void *t_pspare[2]; /* new reassembly queue */ int t_segqlen; /* segment reassembly queue length */ int t_dupacks; /* consecutive dup acks recd */ struct tcp_timer *t_timers; /* All the TCP timers in one struct */ struct inpcb *t_inpcb; /* back pointer to internet pcb */ int t_state; /* state of this connection */ u_int t_flags; struct vnet *t_vnet; /* back pointer to parent vnet */ tcp_seq snd_una; /* send unacknowledged */ tcp_seq snd_max; /* highest sequence number sent; * used to recognize retransmits */ tcp_seq snd_nxt; /* send next */ tcp_seq snd_up; /* send urgent pointer */ tcp_seq snd_wl1; /* window update seg seq number */ tcp_seq snd_wl2; /* window update seg ack number */ tcp_seq iss; /* initial send sequence number */ tcp_seq irs; /* initial receive sequence number */ tcp_seq rcv_nxt; /* receive next */ tcp_seq rcv_adv; /* advertised window */ u_long rcv_wnd; /* receive window */ tcp_seq rcv_up; /* receive urgent pointer */ u_long snd_wnd; /* send window */ u_long snd_cwnd; /* congestion-controlled window */ u_long snd_spare1; /* unused */ u_long snd_ssthresh; /* snd_cwnd size threshold for * for slow start exponential to * linear switch */ u_long snd_spare2; /* unused */ tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ u_int t_maxopd; /* mss plus options */ u_int t_rcvtime; /* inactivity time */ u_int t_starttime; /* time connection was established */ u_int t_rtttime; /* RTT measurement start time */ tcp_seq t_rtseq; /* sequence number being timed */ u_int t_bw_spare1; /* unused */ tcp_seq t_bw_spare2; /* unused */ int t_rxtcur; /* current retransmit value (ticks) */ u_int t_maxseg; /* maximum segment size */ int t_srtt; /* smoothed round-trip time */ int t_rttvar; /* variance in round-trip time */ int t_rxtshift; /* log(2) of rexmt exp. backoff */ u_int t_rttmin; /* minimum rtt allowed */ u_int t_rttbest; /* best rtt we've seen */ u_long t_rttupdated; /* number of times rtt sampled */ u_long max_sndwnd; /* largest window peer has offered */ int t_softerror; /* possible error not yet reported */ /* out-of-band data */ char t_oobflags; /* have some */ char t_iobc; /* input character */ /* RFC 1323 variables */ u_char snd_scale; /* window scaling for send window */ u_char rcv_scale; /* window scaling for recv window */ u_char request_r_scale; /* pending window scaling */ u_int32_t ts_recent; /* timestamp echo data */ u_int ts_recent_age; /* when last updated */ u_int32_t ts_offset; /* our timestamp offset */ tcp_seq last_ack_sent; /* experimental */ u_long snd_cwnd_prev; /* cwnd prior to retransmit */ u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ int t_sndzerowin; /* zero-window updates sent */ u_int t_badrxtwin; /* window for retransmit recovery */ u_char snd_limited; /* segments limited transmitted */ /* SACK related state */ int snd_numholes; /* number of holes seen by sender */ TAILQ_HEAD(sackhole_head, sackhole) snd_holes; /* SACK scoreboard (sorted) */ tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ int rcv_numsacks; /* # distinct sack blks present */ struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ tcp_seq sack_newdata; /* New data xmitted in this recovery episode starts at this seq number */ struct sackhint sackhint; /* SACK scoreboard hint */ int t_rttlow; /* smallest observerved RTT */ u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ int rfbuf_cnt; /* recv buffer autoscaling byte count */ struct toedev *tod; /* toedev handling this connection */ int t_sndrexmitpack; /* retransmit packets sent */ int t_rcvoopack; /* out-of-order packets received */ void *t_toe; /* TOE pcb pointer */ int t_bytes_acked; /* # bytes acked during current RTT */ struct cc_algo *cc_algo; /* congestion control algorithm */ struct cc_var *ccv; /* congestion control specific vars */ struct osd *osd; /* storage for Khelp module data */ u_int t_keepinit; /* time to establish connection */ u_int t_keepidle; /* time before keepalive probes begin */ u_int t_keepintvl; /* interval between keepalives */ u_int t_keepcnt; /* number of keepalives before close */ u_int t_tsomax; /* tso burst length limit */ uint32_t t_ispare[7]; /* 5 UTO, 2 TBD */ void *t_pspare2[4]; /* 4 TBD */ uint64_t _pad[5]; /* 5 TBD (1-2 CC/RTT?) */ uint32_t t_tsomaxsegcount; /* TSO maximum segment count */ uint32_t t_tsomaxsegsize; /* TSO maximum segment size in bytes */ }; /* * Flags and utility macros for the t_flags field. */ #define TF_ACKNOW 0x000001 /* ack peer immediately */ #define TF_DELACK 0x000002 /* ack, but try to delay it */ #define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ #define TF_NOOPT 0x000008 /* don't use tcp options */ #define TF_SENTFIN 0x000010 /* have sent FIN */ #define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ #define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ #define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ #define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ #define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ #define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ #define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ #define TF_NOPUSH 0x001000 /* don't push */ #define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ #define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ #define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ #define TF_LASTIDLE 0x040000 /* connection was previously idle */ #define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ #define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ #define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ #define TF_FORCEDATA 0x800000 /* force out a byte */ #define TF_TSO 0x1000000 /* TSO enabled on this connection */ #define TF_TOE 0x2000000 /* this connection is offloaded */ #define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ #define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ #define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) /* * Flags for the t_oobflags field. */ #define TCPOOB_HAVEDATA 0x01 #define TCPOOB_HADDATA 0x02 #ifdef TCP_SIGNATURE /* * Defines which are needed by the xform_tcp module and tcp_[in|out]put * for SADB verification and lookup. */ #define TCP_SIGLEN 16 /* length of computed digest in bytes */ #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ /* * Only a single SA per host may be specified at this time. An SPI is * needed in order for the KEY_ALLOCSA() lookup to work. */ #define TCP_SIG_SPI 0x1000 #endif /* TCP_SIGNATURE */ /* * Structure to hold TCP options that are only used during segment * processing (in tcp_input), but not held in the tcpcb. * It's basically used to reduce the number of parameters * to tcp_dooptions and tcp_addoptions. * The binary order of the to_flags is relevant for packing of the * options in tcp_addoptions. */ struct tcpopt { u_int64_t to_flags; /* which options are present */ #define TOF_MSS 0x0001 /* maximum segment size */ #define TOF_SCALE 0x0002 /* window scaling */ #define TOF_SACKPERM 0x0004 /* SACK permitted */ #define TOF_TS 0x0010 /* timestamp */ #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ #define TOF_SACK 0x0080 /* Peer sent SACK option */ #define TOF_MAXOPT 0x0100 u_int32_t to_tsval; /* new timestamp */ u_int32_t to_tsecr; /* reflected timestamp */ u_char *to_sacks; /* pointer to the first SACK blocks */ u_char *to_signature; /* pointer to the TCP-MD5 signature */ u_int16_t to_mss; /* maximum segment size */ u_int8_t to_wscale; /* window scaling */ u_int8_t to_nsacks; /* number of SACK blocks */ u_int32_t to_spare; /* UTO */ }; /* * Flags for tcp_dooptions. */ #define TO_SYN 0x01 /* parse SYN-only options */ struct hc_metrics_lite { /* must stay in sync with hc_metrics */ u_long rmx_mtu; /* MTU for this path */ u_long rmx_ssthresh; /* outbound gateway buffer limit */ u_long rmx_rtt; /* estimated round trip time */ u_long rmx_rttvar; /* estimated rtt variance */ u_long rmx_bandwidth; /* estimated bandwidth */ u_long rmx_cwnd; /* congestion window */ u_long rmx_sendpipe; /* outbound delay-bandwidth product */ u_long rmx_recvpipe; /* inbound delay-bandwidth product */ }; /* * Used by tcp_maxmtu() to communicate interface specific features * and limits at the time of connection setup. */ struct tcp_ifcap { int ifcap; u_int tsomax; u_int tsomaxsegcount; u_int tsomaxsegsize; }; #ifndef _NETINET_IN_PCB_H_ struct in_conninfo; #endif /* _NETINET_IN_PCB_H_ */ struct tcptw { struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ tcp_seq snd_nxt; tcp_seq rcv_nxt; tcp_seq iss; tcp_seq irs; u_short last_win; /* cached window value */ u_short tw_so_options; /* copy of so_options */ struct ucred *tw_cred; /* user credentials */ u_int32_t t_recent; u_int32_t ts_offset; /* our timestamp offset */ u_int t_starttime; int tw_time; TAILQ_ENTRY(tcptw) tw_2msl; }; #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) /* * The smoothed round-trip time and estimated variance * are stored as fixed point numbers scaled by the values below. * For convenience, these scales are also used in smoothing the average * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). * With these scales, srtt has 3 bits to the right of the binary point, * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the * binary point, and is smoothed with an ALPHA of 0.75. */ #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ /* * The initial retransmission should happen at rtt + 4 * rttvar. * Because of the way we do the smoothing, srtt and rttvar * will each average +1/2 tick of bias. When we compute * the retransmit timer, we want 1/2 tick of rounding and * 1 extra tick because of +-1/2 tick uncertainty in the * firing of the timer. The bias will give us exactly the * 1.5 tick we need. But, because the bias is * statistical, we have to test that we don't drop below * the minimum feasible timer (which is 2 ticks). * This version of the macro adapted from a paper by Lawrence * Brakmo and Larry Peterson which outlines a problem caused * by insufficient precision in the original implementation, * which results in inappropriately large RTO values for very * fast networks. */ #define TCP_REXMTVAL(tp) \ max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) /* * TCP statistics. * Many of these should be kept per connection, * but that's inconvenient at the moment. */ struct tcpstat { u_long tcps_connattempt; /* connections initiated */ u_long tcps_accepts; /* connections accepted */ u_long tcps_connects; /* connections established */ u_long tcps_drops; /* connections dropped */ u_long tcps_conndrops; /* embryonic connections dropped */ u_long tcps_minmssdrops; /* average minmss too low drops */ u_long tcps_closed; /* conn. closed (includes drops) */ u_long tcps_segstimed; /* segs where we tried to get rtt */ u_long tcps_rttupdated; /* times we succeeded */ u_long tcps_delack; /* delayed acks sent */ u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ u_long tcps_rexmttimeo; /* retransmit timeouts */ u_long tcps_persisttimeo; /* persist timeouts */ u_long tcps_keeptimeo; /* keepalive timeouts */ u_long tcps_keepprobe; /* keepalive probes sent */ u_long tcps_keepdrops; /* connections dropped in keepalive */ u_long tcps_sndtotal; /* total packets sent */ u_long tcps_sndpack; /* data packets sent */ u_long tcps_sndbyte; /* data bytes sent */ u_long tcps_sndrexmitpack; /* data packets retransmitted */ u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ u_long tcps_sndrexmitbad; /* unnecessary packet retransmissions */ u_long tcps_sndacks; /* ack-only packets sent */ u_long tcps_sndprobe; /* window probes sent */ u_long tcps_sndurg; /* packets sent with URG only */ u_long tcps_sndwinup; /* window update-only packets sent */ u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ u_long tcps_rcvtotal; /* total packets received */ u_long tcps_rcvpack; /* packets received in sequence */ u_long tcps_rcvbyte; /* bytes received in sequence */ u_long tcps_rcvbadsum; /* packets received with ccksum errs */ u_long tcps_rcvbadoff; /* packets received with bad offset */ u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ u_long tcps_rcvshort; /* packets received too short */ u_long tcps_rcvduppack; /* duplicate-only packets received */ u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ u_long tcps_rcvpartduppack; /* packets with some duplicate data */ u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ u_long tcps_rcvoopack; /* out-of-order packets received */ u_long tcps_rcvoobyte; /* out-of-order bytes received */ u_long tcps_rcvpackafterwin; /* packets with data after window */ u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ u_long tcps_rcvafterclose; /* packets rcvd after "close" */ u_long tcps_rcvwinprobe; /* rcvd window probe packets */ u_long tcps_rcvdupack; /* rcvd duplicate acks */ u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ u_long tcps_rcvackpack; /* rcvd ack packets */ u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ u_long tcps_rcvwinupd; /* rcvd window update packets */ u_long tcps_pawsdrop; /* segments dropped due to PAWS */ u_long tcps_predack; /* times hdr predict ok for acks */ u_long tcps_preddat; /* times hdr predict ok for data pkts */ u_long tcps_pcbcachemiss; u_long tcps_cachedrtt; /* times cached RTT in route updated */ u_long tcps_cachedrttvar; /* times cached rttvar updated */ u_long tcps_cachedssthresh; /* times cached ssthresh updated */ u_long tcps_usedrtt; /* times RTT initialized from route */ u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */ u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/ u_long tcps_persistdrop; /* timeout in persist state */ u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */ u_long tcps_mturesent; /* resends due to MTU discovery */ u_long tcps_listendrop; /* listen queue overflows */ u_long tcps_badrst; /* ignored RSTs in the window */ u_long tcps_sc_added; /* entry added to syncache */ u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */ u_long tcps_sc_dupsyn; /* duplicate SYN packet */ u_long tcps_sc_dropped; /* could not reply to packet */ u_long tcps_sc_completed; /* successful extraction of entry */ u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */ u_long tcps_sc_reset; /* RST removed entry from syncache */ u_long tcps_sc_stale; /* timed out or listen socket gone */ u_long tcps_sc_aborted; /* syncache entry aborted */ u_long tcps_sc_badack; /* removed due to bad ACK */ u_long tcps_sc_unreach; /* ICMP unreachable received */ u_long tcps_sc_zonefail; /* zalloc() failed */ u_long tcps_sc_sendcookie; /* SYN cookie sent */ u_long tcps_sc_recvcookie; /* SYN cookie received */ u_long tcps_hc_added; /* entry added to hostcache */ u_long tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */ u_long tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ /* SACK related stats */ u_long tcps_sack_recovery_episode; /* SACK recovery episodes */ u_long tcps_sack_rexmits; /* SACK rexmit segments */ u_long tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ u_long tcps_sack_rcv_blocks; /* SACK blocks (options) received */ u_long tcps_sack_send_blocks; /* SACK blocks (options) sent */ u_long tcps_sack_sboverflow; /* times scoreboard overflowed */ /* ECN related stats */ u_long tcps_ecn_ce; /* ECN Congestion Experienced */ u_long tcps_ecn_ect0; /* ECN Capable Transport */ u_long tcps_ecn_ect1; /* ECN Capable Transport */ u_long tcps_ecn_shs; /* ECN successful handshakes */ u_long tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ /* TCP_SIGNATURE related stats */ u_long tcps_sig_rcvgoodsig; /* Total matching signature received */ u_long tcps_sig_rcvbadsig; /* Total bad signature received */ u_long tcps_sig_err_buildsig; /* Mismatching signature received */ u_long tcps_sig_err_sigopt; /* No signature expected by socket */ u_long tcps_sig_err_nosigopt; /* No signature provided by segment */ u_long _pad[12]; /* 6 UTO, 6 TBD */ }; #ifdef _KERNEL /* * In-kernel consumers can use these accessor macros directly to update * stats. */ #define TCPSTAT_ADD(name, val) V_tcpstat.name += (val) #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) /* * Kernel module consumers must use this accessor macro. */ void kmod_tcpstat_inc(int statnum); #define KMOD_TCPSTAT_INC(name) \ kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(u_long)) /* * TCP specific helper hook point identifiers. */ #define HHOOK_TCP_EST_IN 0 #define HHOOK_TCP_EST_OUT 1 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT struct tcp_hhook_data { struct tcpcb *tp; struct tcphdr *th; struct tcpopt *to; long len; int tso; tcp_seq curack; }; #endif /* * TCB structure exported to user-land via sysctl(3). * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been * included. Not all of our clients do. */ #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) struct xtcp_timer { int tt_rexmt; /* retransmit timer */ int tt_persist; /* retransmit persistence */ int tt_keep; /* keepalive */ int tt_2msl; /* 2*msl TIME_WAIT timer */ int tt_delack; /* delayed ACK timer */ int t_rcvtime; /* Time since last packet received */ }; struct xtcpcb { size_t xt_len; struct inpcb xt_inp; struct tcpcb xt_tp; struct xsocket xt_socket; struct xtcp_timer xt_timer; u_quad_t xt_alignment_hack; }; #endif /* * Names for TCP sysctl objects */ #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ #define TCPCTL_MSSDFLT 3 /* MSS default */ #define TCPCTL_STATS 4 /* statistics (read-only) */ #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ #define TCPCTL_SENDSPACE 8 /* send buffer space */ #define TCPCTL_RECVSPACE 9 /* receive buffer space */ #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ #define TCPCTL_DROP 15 /* drop tcp connection */ #define TCPCTL_MAXID 16 #define TCPCTL_FINWAIT2_TIMEOUT 17 #define TCPCTL_NAMES { \ { 0, 0 }, \ { "rfc1323", CTLTYPE_INT }, \ { "mssdflt", CTLTYPE_INT }, \ { "stats", CTLTYPE_STRUCT }, \ { "rttdflt", CTLTYPE_INT }, \ { "keepidle", CTLTYPE_INT }, \ { "keepintvl", CTLTYPE_INT }, \ { "sendspace", CTLTYPE_INT }, \ { "recvspace", CTLTYPE_INT }, \ { "keepinit", CTLTYPE_INT }, \ { "pcblist", CTLTYPE_STRUCT }, \ { "delacktime", CTLTYPE_INT }, \ { "v6mssdflt", CTLTYPE_INT }, \ { "maxid", CTLTYPE_INT }, \ } #ifdef _KERNEL #ifdef SYSCTL_DECL SYSCTL_DECL(_net_inet_tcp); SYSCTL_DECL(_net_inet_tcp_sack); MALLOC_DECLARE(M_TCPLOG); #endif VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ VNET_DECLARE(struct inpcbinfo, tcbinfo); VNET_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ extern int tcp_log_in_vain; VNET_DECLARE(int, tcp_mssdflt); /* XXX */ VNET_DECLARE(int, tcp_minmss); VNET_DECLARE(int, tcp_delack_enabled); VNET_DECLARE(int, tcp_do_rfc3390); VNET_DECLARE(int, tcp_do_initcwnd10); VNET_DECLARE(int, tcp_sendspace); VNET_DECLARE(int, tcp_recvspace); VNET_DECLARE(int, path_mtu_discovery); VNET_DECLARE(int, tcp_do_rfc3465); VNET_DECLARE(int, tcp_abc_l_var); #define V_tcb VNET(tcb) #define V_tcbinfo VNET(tcbinfo) #define V_tcpstat VNET(tcpstat) #define V_tcp_mssdflt VNET(tcp_mssdflt) #define V_tcp_minmss VNET(tcp_minmss) #define V_tcp_delack_enabled VNET(tcp_delack_enabled) #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) #define V_tcp_do_initcwnd10 VNET(tcp_do_initcwnd10) #define V_tcp_sendspace VNET(tcp_sendspace) #define V_tcp_recvspace VNET(tcp_recvspace) #define V_path_mtu_discovery VNET(path_mtu_discovery) #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) #define V_tcp_abc_l_var VNET(tcp_abc_l_var) VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ #define V_tcp_do_sack VNET(tcp_do_sack) #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ VNET_DECLARE(int, tcp_ecn_maxretries); #define V_tcp_do_ecn VNET(tcp_do_ecn) #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); #define V_tcp_hhh VNET(tcp_hhh) int tcp_addoptions(struct tcpopt *, u_char *); int tcp_ccalgounload(struct cc_algo *unload_algo); struct tcpcb * tcp_close(struct tcpcb *); void tcp_discardcb(struct tcpcb *); void tcp_twstart(struct tcpcb *); #if 0 int tcp_twrecycleable(struct tcptw *tw); #endif void tcp_twclose(struct tcptw *_tw, int _reuse); void tcp_ctlinput(int, struct sockaddr *, void *); int tcp_ctloutput(struct socket *, struct sockopt *); struct tcpcb * tcp_drop(struct tcpcb *, int); void tcp_drain(void); void tcp_init(void); #ifdef VIMAGE void tcp_destroy(void); #endif void tcp_fini(void *); char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, const void *); char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, const void *); int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); -void tcp_reass_init(void); +void tcp_reass_global_init(void); void tcp_reass_flush(struct tcpcb *); -#ifdef VIMAGE -void tcp_reass_destroy(void); -#endif void tcp_input(struct mbuf *, int); u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, struct tcp_ifcap *); void tcp_mss(struct tcpcb *, int); int tcp_mssopt(struct in_conninfo *); struct inpcb * tcp_drop_syn_sent(struct inpcb *, int); struct inpcb * tcp_mtudisc(struct inpcb *, int); struct tcpcb * tcp_newtcpcb(struct inpcb *); int tcp_output(struct tcpcb *); void tcp_respond(struct tcpcb *, void *, struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); void tcp_tw_init(void); #ifdef VIMAGE void tcp_tw_destroy(void); #endif void tcp_tw_zone_change(void); int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, struct mbuf *, int); int tcp_twrespond(struct tcptw *, int); void tcp_setpersist(struct tcpcb *); #ifdef TCP_SIGNATURE int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int); int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *, struct tcphdr *, u_int); #endif void tcp_slowtimo(void); struct tcptemp * tcpip_maketemplate(struct inpcb *); void tcpip_fillheaders(struct inpcb *, void *, void *); void tcp_timer_activate(struct tcpcb *, int, u_int); int tcp_timer_active(struct tcpcb *, int); void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); /* * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) */ void tcp_hc_init(void); #ifdef VIMAGE void tcp_hc_destroy(void); #endif void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); u_long tcp_hc_getmtu(struct in_conninfo *); void tcp_hc_updatemtu(struct in_conninfo *, u_long); void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); extern struct pr_usrreqs tcp_usrreqs; tcp_seq tcp_new_isn(struct tcpcb *); void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); void tcp_clean_sackreport(struct tcpcb *tp); void tcp_sack_adjust(struct tcpcb *tp); struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); void tcp_free_sackholes(struct tcpcb *tp); int tcp_newreno(struct tcpcb *, struct tcphdr *); u_long tcp_seq_subtract(u_long, u_long ); void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); #endif /* _KERNEL */ #endif /* _NETINET_TCP_VAR_H_ */