July 2010 - Freeradius-Devel - lists.freeradius.org

Missing TLS Change Cipher Spec and TLS Finished in EAP-TLS exchanges
by arno＠natisbad.org 09 Jul '11

09 Jul '11

Hi, I upgraded an internal freeradius server (Linux, Debian unstable, libssl 0.9.8) used for EAP-TLS (802.11 WPA2 access) from 2.0.5 to 2.1.3. With 2.1.3, client authentication fails in the following way (wpasupplicant on Debian unstable, working with 2.0.5): - the TLS exchange is completed from freeradius point of view. It even sends the Access-Accept message to the NAS. - from the clients point of view, the TLS exchange is not over, it expects some more data from the server which are never sent. I took two pcap traces (sent to you privately, Alan), one with 2.0.5, one with 2.1.3. It looks like the TLS Change Cipher and TLS Tinished elements of the TLS exchange are not in the EAP encapsulated TLS packets sent by the server. Compare packet #23 of 2.1.3 trace with packet #27 of 2.0.5 trace. If you use wireshark, you will notice that the SSL layer is dissected; this is because #27 has a small brother: #29. It is not sent by 2.1.3. Alan, if you have an idea or a patch, I can test it on monday. Cheers, a+

5 15

eap2 in freeradius
by Chanthearith HUON 04 Apr '11

04 Apr '11

Hi everyone, I am testing eap2 in freeradius. I run "radiusd -X", I got following error message " radiusd: symbol lookup error: /usr/lib/rlm_eap2-2.1.0.so: undefined symbol: eap_server_register_methods " Anyone, have any idea what might be wrong with the configuration? Thank in advance, Thierry

5 10

Additional EAP-TLS Logging Option
by Ross, Michael 08 Sep '10

08 Sep '10

Hello, I customized a freeradius installation to include an option to log information from the client certificates (subject, issuer, serial, expiration) when doing eap-tls authentication. This is done through a new option, log_certificates, in the tls section of the eap.conf file. Is there any interest in adding this capability into the baseline freeradius? I'm happy to make any changes that would be required to do that. Below are the updated rlm_eap_tls.h and rlm_eap_tls.c. Thanks, Mike Ross /* * rlm_eap_tls.h * * Version: $Id: rlm_eap_tls.h,v 1.13 2007/04/25 14:19:26 aland Exp $ * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA * * Copyright 2001 hereUare Communications, Inc. <raghud(a)hereuare.com> * Copyright 2003 Alan DeKok <aland(a)freeradius.org> * Copyright 2006 The FreeRADIUS server project * *----------------------------------------------------------------------------- * Modified by The Boeing Company, August 26, 2008 * * Added int log_certificates to the eap_tls_conf struct * * Copyright 2008 The Boeing Company <Michael.Ross2(a)boeing.com> *----------------------------------------------------------------------------- */ #ifndef _RLM_EAP_TLS_H #define _RLM_EAP_TLS_H #include <freeradius-devel/ident.h> RCSIDH(rlm_eap_tls_h, "$Id: rlm_eap_tls.h,v 1.13 2007/04/25 14:19:26 aland Exp $") #include "eap_tls.h" #include <freeradius-devel/radiusd.h> #include <freeradius-devel/modules.h> /* configured values goes right here */ typedef struct eap_tls_conf { char *private_key_password; char *private_key_file; char *certificate_file; char *random_file; char *ca_path; char *ca_file; char *dh_file; char *rsa_file; char *make_cert_command; int rsa_key; int dh_key; int rsa_key_length; int dh_key_length; int verify_depth; int file_type; int include_length; /* * Always < 4096 (due to radius limit), 0 by default = 2048 */ int fragment_size; int check_crl; char *check_cert_cn; char *cipher_list; char *check_cert_issuer; int log_certificates; /* log_certificates added by Boeing, 9/26/2008 */ } EAP_TLS_CONF; /* This structure gets stored in arg */ typedef struct _eap_tls_t { EAP_TLS_CONF *conf; SSL_CTX *ctx; } eap_tls_t; #endif /* _RLM_EAP_TLS_H */ /* * rlm_eap_tls.c contains the interfaces that are called from eap * * Version: $Id: rlm_eap_tls.c,v 1.54 2008/03/22 08:31:03 aland Exp $ * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA * * Copyright 2001 hereUare Communications, Inc. <raghud(a)hereuare.com> * Copyright 2003 Alan DeKok <aland(a)freeradius.org> * Copyright 2006 The FreeRADIUS server project * *----------------------------------------------------------------------------- * Modified by The Boeing Company, August 26, 2008 * * Added log_certificates to module_config CONF_PARSER * Added additional authentication logging into cbtls_verify * Added log_full_cert function to write certificate into log file * * Copyright 2008 The Boeing Company <Michael.Ross2(a)boeing.com> *----------------------------------------------------------------------------- */ #include <freeradius-devel/ident.h> RCSID("$Id: rlm_eap_tls.c,v 1.54 2008/03/22 08:31:03 aland Exp $") #include <freeradius-devel/autoconf.h> #ifdef HAVE_OPENSSL_RAND_H #include <openssl/rand.h> #endif #include "rlm_eap_tls.h" #ifdef HAVE_SYS_STAT_H #include <sys/stat.h> #endif static CONF_PARSER module_config[] = { { "rsa_key_exchange", PW_TYPE_BOOLEAN, offsetof(EAP_TLS_CONF, rsa_key), NULL, "no" }, { "dh_key_exchange", PW_TYPE_BOOLEAN, offsetof(EAP_TLS_CONF, dh_key), NULL, "yes" }, { "rsa_key_length", PW_TYPE_INTEGER, offsetof(EAP_TLS_CONF, rsa_key_length), NULL, "512" }, { "dh_key_length", PW_TYPE_INTEGER, offsetof(EAP_TLS_CONF, dh_key_length), NULL, "512" }, { "verify_depth", PW_TYPE_INTEGER, offsetof(EAP_TLS_CONF, verify_depth), NULL, "0" }, { "CA_path", PW_TYPE_FILENAME, offsetof(EAP_TLS_CONF, ca_path), NULL, NULL }, { "pem_file_type", PW_TYPE_BOOLEAN, offsetof(EAP_TLS_CONF, file_type), NULL, "yes" }, { "private_key_file", PW_TYPE_FILENAME, offsetof(EAP_TLS_CONF, private_key_file), NULL, NULL }, { "certificate_file", PW_TYPE_FILENAME, offsetof(EAP_TLS_CONF, certificate_file), NULL, NULL }, { "CA_file", PW_TYPE_FILENAME, offsetof(EAP_TLS_CONF, ca_file), NULL, NULL }, { "private_key_password", PW_TYPE_STRING_PTR, offsetof(EAP_TLS_CONF, private_key_password), NULL, NULL }, { "dh_file", PW_TYPE_STRING_PTR, offsetof(EAP_TLS_CONF, dh_file), NULL, NULL }, { "random_file", PW_TYPE_STRING_PTR, offsetof(EAP_TLS_CONF, random_file), NULL, NULL }, { "fragment_size", PW_TYPE_INTEGER, offsetof(EAP_TLS_CONF, fragment_size), NULL, "1024" }, { "include_length", PW_TYPE_BOOLEAN, offsetof(EAP_TLS_CONF, include_length), NULL, "yes" }, { "check_crl", PW_TYPE_BOOLEAN, offsetof(EAP_TLS_CONF, check_crl), NULL, "no"}, { "check_cert_cn", PW_TYPE_STRING_PTR, offsetof(EAP_TLS_CONF, check_cert_cn), NULL, NULL}, { "cipher_list", PW_TYPE_STRING_PTR, offsetof(EAP_TLS_CONF, cipher_list), NULL, NULL}, { "check_cert_issuer", PW_TYPE_STRING_PTR, offsetof(EAP_TLS_CONF, check_cert_issuer), NULL, NULL}, { "make_cert_command", PW_TYPE_STRING_PTR, offsetof(EAP_TLS_CONF, make_cert_command), NULL, NULL}, /* log_certificates added by Boeing, 9/26/2008 */ { "log_certificates", PW_TYPE_BOOLEAN, offsetof(EAP_TLS_CONF, log_certificates), NULL, "no" }, { NULL, -1, 0, NULL, NULL } /* end the list */ }; /* * log_full_cert added by Boeing, 9/26/2008 * * Write the certificate serial number, expiration * date, issuer, and subject to the log file */ static int log_full_cert(int ok, X509 *cert) { const int ub_serial = 20; const int ub_time = 15; char serialNumber[2*ub_serial+1]; char expiration[ub_time+1]; char issuer[ub_name+1]; char subject[ub_name+1]; char *pTemp = NULL; int local_ok = ok; int i = 0; ASN1_INTEGER *pSn = NULL; ASN1_TIME *pTime = NULL; /* * Get the Serial Number and format for display */ serialNumber[0] = '\0'; pSn = X509_get_serialNumber(cert); pTemp = serialNumber; if ( pSn == NULL || pSn->length > ub_serial ) { radlog(L_AUTH, "rlm_eap_tls: Malformed Certificate Serial Number"); local_ok = 0; } else { for (i = 0; i < pSn->length; i++) { sprintf(pTemp, "%02x", (int)pSn->data[i]); pTemp += 2; } } /* * Get the Expiration Date and format for display */ expiration[0] = '\0'; pTime = X509_get_notAfter(cert); if ( pTime == NULL || pTime->length > ub_time ) { radlog(L_AUTH, "rlm_eap_tls: Malformed Certificate Expiration"); local_ok = 0; } else { strncpy(expiration, pTime->data, pTime->length); expiration[pTime->length] = '\0'; } /* * Get the Issuer */ issuer[0] = '\0'; X509_NAME_oneline(X509_get_issuer_name(cert), issuer, ub_name); issuer[0] = ' '; issuer[ub_name] = '\0'; for (i = 0; i < ub_name; i++) { if (issuer[i] == '/') { issuer[i] = ','; } } /* * Get the Subject */ subject[0] = '\0'; X509_NAME_oneline(X509_get_subject_name(cert), subject, ub_name); subject[0] = ' '; subject[ub_name] = '\0'; for (i = 0; i < ub_name; i++) { if (subject[i] == '/') { subject[i] = ','; } } /* * Log required attributes */ radlog(L_AUTH, "Certificate: Serial=%s; Expiration=%s; Issuer:%s; Subject:%s", serialNumber, expiration, issuer, subject); return local_ok; } /* * TODO: Check for the type of key exchange * like conf->dh_key */ static int load_dh_params(SSL_CTX *ctx, char *file) { DH *dh = NULL; BIO *bio; if ((bio = BIO_new_file(file, "r")) == NULL) { radlog(L_ERR, "rlm_eap_tls: Unable to open DH file - %s", file); return -1; } dh = PEM_read_bio_DHparams(bio, NULL, NULL, NULL); BIO_free(bio); if (!dh) { DEBUG2("WARNING: rlm_eap_tls: Unable to set DH parameters. DH cipher suites may not work!"); DEBUG2("WARNING: Fix this by running the OpenSSL command listed in eap.conf"); return 0; } if (SSL_CTX_set_tmp_dh(ctx, dh) < 0) { radlog(L_ERR, "rlm_eap_tls: Unable to set DH parameters"); DH_free(dh); return -1; } DH_free(dh); return 0; } /* * Before trusting a certificate, you must make sure that the * certificate is 'valid'. There are several steps that your * application can take in determining if a certificate is * valid. Commonly used steps are: * * 1.Verifying the certificate's signature, and verifying that * the certificate has been issued by a trusted Certificate * Authority. * * 2.Verifying that the certificate is valid for the present date * (i.e. it is being presented within its validity dates). * * 3.Verifying that the certificate has not been revoked by its * issuing Certificate Authority, by checking with respect to a * Certificate Revocation List (CRL). * * 4.Verifying that the credentials presented by the certificate * fulfill additional requirements specific to the application, * such as with respect to access control lists or with respect * to OCSP (Online Certificate Status Processing). * * NOTE: This callback will be called multiple times based on the * depth of the root certificate chain */ static int cbtls_verify(int ok, X509_STORE_CTX *ctx) { char subject[1024]; /* Used for the subject name */ char issuer[1024]; /* Used for the issuer name */ char common_name[1024]; char cn_str[1024]; EAP_HANDLER *handler = NULL; X509 *client_cert; SSL *ssl; int err, depth; EAP_TLS_CONF *conf; int my_ok = ok; client_cert = X509_STORE_CTX_get_current_cert(ctx); err = X509_STORE_CTX_get_error(ctx); depth = X509_STORE_CTX_get_error_depth(ctx); /* * Moved below the conf check by Boeing, 9/26/2008 */ /* *if (!my_ok) { * radlog(L_ERR,"--> verify error:num=%d:%s\n",err, * X509_verify_cert_error_string(err)); * return my_ok; *} */ /* * Retrieve the pointer to the SSL of the connection currently treated * and the application specific data stored into the SSL object. */ ssl = X509_STORE_CTX_get_ex_data(ctx, SSL_get_ex_data_X509_STORE_CTX_idx()); handler = (EAP_HANDLER *)SSL_get_ex_data(ssl, 0); conf = (EAP_TLS_CONF *)SSL_get_ex_data(ssl, 1); /* * Copied from above by Boeing, 9/26/2008 */ if (!my_ok) { radlog(L_ERR,"--> verify error:num=%d:%s\n",err, X509_verify_cert_error_string(err)); /* * Added by Boeing, 9/26/2008 * If the config file is set to log certificates * log the certificate if there has been an authentication * error */ if (conf->log_certificates) { log_full_cert(my_ok, client_cert); } return my_ok; } /* * Get the Subject & Issuer */ subject[0] = issuer[0] = '\0'; X509_NAME_oneline(X509_get_subject_name(client_cert), subject, sizeof(subject)); X509_NAME_oneline(X509_get_issuer_name(ctx->current_cert), issuer, sizeof(issuer)); subject[sizeof(subject) - 1] = '\0'; issuer[sizeof(issuer) - 1] = '\0'; /* * Get the Common Name */ X509_NAME_get_text_by_NID(X509_get_subject_name(client_cert), NID_commonName, common_name, sizeof(common_name)); common_name[sizeof(common_name) - 1] = '\0'; switch (ctx->error) { case X509_V_ERR_UNABLE_TO_GET_ISSUER_CERT: radlog(L_ERR, "issuer= %s\n", issuer); break; case X509_V_ERR_CERT_NOT_YET_VALID: case X509_V_ERR_ERROR_IN_CERT_NOT_BEFORE_FIELD: radlog(L_ERR, "notBefore="); #if 0 ASN1_TIME_print(bio_err, X509_get_notBefore(ctx->current_cert)); #endif break; case X509_V_ERR_CERT_HAS_EXPIRED: case X509_V_ERR_ERROR_IN_CERT_NOT_AFTER_FIELD: radlog(L_ERR, "notAfter="); #if 0 ASN1_TIME_print(bio_err, X509_get_notAfter(ctx->current_cert)); #endif break; } /* * If we're at the actual client cert, apply additional * checks. */ if (depth == 0) { /* * If the conf tells us to, check cert issuer * against the specified value and fail * verification if they don't match. */ if (conf->check_cert_issuer && (strcmp(issuer, conf->check_cert_issuer) != 0)) { radlog(L_AUTH, "rlm_eap_tls: Certificate issuer (%s) does not match specified value (%s)!", issuer, conf->check_cert_issuer); my_ok = 0; } /* * If the conf tells us to, check the CN in the * cert against xlat'ed value, but only if the * previous checks passed. */ if (my_ok && conf->check_cert_cn) { if (!radius_xlat(cn_str, sizeof(cn_str), conf->check_cert_cn, handler->request, NULL)) { radlog(L_ERR, "rlm_eap_tls (%s): xlat failed.", conf->check_cert_cn); /* if this fails, fail the verification */ my_ok = 0; } else { DEBUG2(" rlm_eap_tls: checking certificate CN (%s) with xlat'ed value (%s)", common_name, cn_str); if (strcmp(cn_str, common_name) != 0) { radlog(L_AUTH, "rlm_eap_tls: Certificate CN (%s) does not match specified value (%s)!", common_name, cn_str); my_ok = 0; } } } /* check_cert_cn */ /* * Added by Boeing, 9/26/2008 * If the config file is set to log certificates * log the certificate */ if (conf->log_certificates) { my_ok = log_full_cert(my_ok, client_cert); } } /* depth == 0 */ if (debug_flag > 0) { DEBUG2("chain-depth=%d, ", depth); DEBUG2("error=%d", err); DEBUG2("--> User-Name = %s", handler->identity); DEBUG2("--> BUF-Name = %s", common_name); DEBUG2("--> subject = %s", subject); DEBUG2("--> issuer = %s", issuer); DEBUG2("--> verify return:%d", my_ok); } return my_ok; } /* * Create Global context SSL and use it in every new session * * - Load the trusted CAs * - Load the Private key & the certificate * - Set the Context options & Verify options */ static SSL_CTX *init_tls_ctx(EAP_TLS_CONF *conf) { SSL_METHOD *meth; SSL_CTX *ctx; X509_STORE *certstore; int verify_mode = SSL_VERIFY_NONE; int ctx_options = 0; int type; /* * Add all the default ciphers and message digests * Create our context. */ SSL_library_init(); SSL_load_error_strings(); meth = TLSv1_method(); ctx = SSL_CTX_new(meth); /* * Identify the type of certificates that needs to be loaded */ if (conf->file_type) { type = SSL_FILETYPE_PEM; } else { type = SSL_FILETYPE_ASN1; } /* * Set the password to load private key */ if (conf->private_key_password) { SSL_CTX_set_default_passwd_cb_userdata(ctx, conf->private_key_password); SSL_CTX_set_default_passwd_cb(ctx, cbtls_password); } /* * Load our keys and certificates * * If certificates are of type PEM then we can make use * of cert chain authentication using openssl api call * SSL_CTX_use_certificate_chain_file. Please see how * the cert chain needs to be given in PEM from * openSSL.org */ if (type == SSL_FILETYPE_PEM) { if (!(SSL_CTX_use_certificate_chain_file(ctx, conf->certificate_file))) { radlog(L_ERR, "rlm_eap: SSL error %s", ERR_error_string(ERR_get_error(), NULL)); radlog(L_ERR, "rlm_eap_tls: Error reading certificate file %s", conf->certificate_file); return NULL; } } else if (!(SSL_CTX_use_certificate_file(ctx, conf->certificate_file, type))) { radlog(L_ERR, "rlm_eap: SSL error %s", ERR_error_string(ERR_get_error(), NULL)); radlog(L_ERR, "rlm_eap_tls: Error reading certificate file %s", conf->certificate_file); return NULL; } /* Load the CAs we trust */ if (!SSL_CTX_load_verify_locations(ctx, conf->ca_file, conf->ca_path)) { radlog(L_ERR, "rlm_eap: SSL error %s", ERR_error_string(ERR_get_error(), NULL)); radlog(L_ERR, "rlm_eap_tls: Error reading Trusted root CA list %s",conf->ca_file ); return NULL; } if (conf->ca_file && *conf->ca_file) SSL_CTX_set_client_CA_list(ctx, SSL_load_client_CA_file(conf->ca_file)); if (!(SSL_CTX_use_PrivateKey_file(ctx, conf->private_key_file, type))) { radlog(L_ERR, "rlm_eap: SSL error %s", ERR_error_string(ERR_get_error(), NULL)); radlog(L_ERR, "rlm_eap_tls: Error reading private key file %s", conf->private_key_file); return NULL; } /* * Check if the loaded private key is the right one */ if (!SSL_CTX_check_private_key(ctx)) { radlog(L_ERR, "rlm_eap_tls: Private key does not match the certificate public key"); return NULL; } /* * Set ctx_options */ ctx_options |= SSL_OP_NO_SSLv2; ctx_options |= SSL_OP_NO_SSLv3; /* * SSL_OP_SINGLE_DH_USE must be used in order to prevent * small subgroup attacks and forward secrecy. Always * using * * SSL_OP_SINGLE_DH_USE has an impact on the computer * time needed during negotiation, but it is not very * large. */ ctx_options |= SSL_OP_SINGLE_DH_USE; /* * SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS to work around issues * in Windows Vista client. * http://www.openssl.org/~bodo/tls-cbc.txt * http://www.nabble.com/(RADIATOR)-Radiator-Version-3.16-released-t2600070.ht… */ ctx_options |= SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS; SSL_CTX_set_options(ctx, ctx_options); /* * TODO: Set the RSA & DH * SSL_CTX_set_tmp_rsa_callback(ctx, cbtls_rsa); * SSL_CTX_set_tmp_dh_callback(ctx, cbtls_dh); */ /* * set the message callback to identify the type of * message. For every new session, there can be a * different callback argument. * * SSL_CTX_set_msg_callback(ctx, cbtls_msg); */ /* Set Info callback */ SSL_CTX_set_info_callback(ctx, cbtls_info); /* * Check the certificates for revocation. */ #ifdef X509_V_FLAG_CRL_CHECK if (conf->check_crl) { certstore = SSL_CTX_get_cert_store(ctx); if (certstore == NULL) { radlog(L_ERR, "rlm_eap: SSL error %s", ERR_error_string(ERR_get_error(), NULL)); radlog(L_ERR, "rlm_eap_tls: Error reading Certificate Store"); return NULL; } X509_STORE_set_flags(certstore, X509_V_FLAG_CRL_CHECK); } #endif /* * Set verify modes * Always verify the peer certificate */ verify_mode |= SSL_VERIFY_PEER; verify_mode |= SSL_VERIFY_FAIL_IF_NO_PEER_CERT; verify_mode |= SSL_VERIFY_CLIENT_ONCE; SSL_CTX_set_verify(ctx, verify_mode, cbtls_verify); if (conf->verify_depth) { SSL_CTX_set_verify_depth(ctx, conf->verify_depth); } /* Load randomness */ if (!(RAND_load_file(conf->random_file, 1024*1024))) { radlog(L_ERR, "rlm_eap: SSL error %s", ERR_error_string(ERR_get_error(), NULL)); radlog(L_ERR, "rlm_eap_tls: Error loading randomness"); return NULL; } /* * Set the cipher list if we were told to */ if (conf->cipher_list) { if (!SSL_CTX_set_cipher_list(ctx, conf->cipher_list)) { radlog(L_ERR, "rlm_eap_tls: Error setting cipher list"); return NULL; } } return ctx; } /* * Detach the EAP-TLS module. */ static int eaptls_detach(void *arg) { EAP_TLS_CONF *conf; eap_tls_t *inst; inst = (eap_tls_t *) arg; conf = inst->conf; if (conf) { memset(conf, 0, sizeof(*conf)); free(inst->conf); inst->conf = NULL; } if (inst->ctx) SSL_CTX_free(inst->ctx); inst->ctx = NULL; free(inst); return 0; } /* * Attach the EAP-TLS module. */ static int eaptls_attach(CONF_SECTION *cs, void **instance) { EAP_TLS_CONF *conf; eap_tls_t *inst; /* Store all these values in the data structure for later references */ inst = (eap_tls_t *)malloc(sizeof(*inst)); if (!inst) { radlog(L_ERR, "rlm_eap_tls: out of memory"); return -1; } memset(inst, 0, sizeof(*inst)); /* * Parse the config file & get all the configured values */ conf = (EAP_TLS_CONF *)malloc(sizeof(*conf)); if (conf == NULL) { free(inst); radlog(L_ERR, "rlm_eap_tls: out of memory"); return -1; } memset(conf, 0, sizeof(*conf)); inst->conf = conf; if (cf_section_parse(cs, conf, module_config) < 0) { eaptls_detach(inst); return -1; } /* * The EAP RFC's say 1020, but we're less picky. */ if (conf->fragment_size < 100) { radlog(L_ERR, "rlm_eap_tls: Fragment size is too small."); eaptls_detach(inst); return -1; } /* * The maximum size for a RADIUS packet is 4096, * minus the header (20), Message-Authenticator (18), * and State (18), etc. results in about 4000 bytes of data * that can be devoted *solely* to EAP. */ if (conf->fragment_size > 4000) { radlog(L_ERR, "rlm_eap_tls: Fragment size is too large."); eaptls_detach(inst); return -1; } /* * Account for the EAP header (4), and the EAP-TLS header * (6), as per Section 4.2 of RFC 2716. What's left is * the maximum amount of data we read from a TLS buffer. */ conf->fragment_size -= 10; /* * This magic makes the administrators life HUGELY easier * on initial deployments. * * If the server starts up in debugging mode, AND the * bootstrap command is configured, AND it exists, AND * there is no server certificate */ if (conf->make_cert_command && (debug_flag >= 2)) { struct stat buf; if ((stat(conf->make_cert_command, &buf) == 0) && (stat(conf->certificate_file, &buf) < 0) && (errno == ENOENT) && (radius_exec_program(conf->make_cert_command, NULL, 1, NULL, 0, NULL, NULL, 0) != 0)) { eaptls_detach(inst); return -1; } } /* * Initialize TLS */ inst->ctx = init_tls_ctx(conf); if (inst->ctx == NULL) { eaptls_detach(inst); return -1; } if (load_dh_params(inst->ctx, conf->dh_file) < 0) { eaptls_detach(inst); return -1; } *instance = inst; return 0; } /* * Send an initial eap-tls request to the peer. * * Frame eap reply packet. * len = header + type + tls_typedata * tls_typedata = flags(Start (S) bit set, and no data) * * Once having received the peer's Identity, the EAP server MUST * respond with an EAP-TLS/Start packet, which is an * EAP-Request packet with EAP-Type=EAP-TLS, the Start (S) bit * set, and no data. The EAP-TLS conversation will then begin, * with the peer sending an EAP-Response packet with * EAP-Type = EAP-TLS. The data field of that packet will * be the TLS data. * * Fragment length is Framed-MTU - 4. * * http://mail.frascone.com/pipermail/public/eap/2003-July/001426.html */ static int eaptls_initiate(void *type_arg, EAP_HANDLER *handler) { int status; tls_session_t *ssn; eap_tls_t *inst; VALUE_PAIR *vp; int client_cert = TRUE; int verify_mode = 0; inst = (eap_tls_t *)type_arg; /* * If we're TTLS or PEAP, then do NOT require a client * certificate. * * FIXME: This should be more configurable. */ if (handler->eap_type != PW_EAP_TLS) { vp = pairfind(handler->request->config_items, PW_EAP_TLS_REQUIRE_CLIENT_CERT); if (!vp) { client_cert = FALSE; } else { client_cert = vp->vp_integer; } } /* * Every new session is started only from EAP-TLS-START. * Before Sending EAP-TLS-START, open a new SSL session. * Create all the required data structures & store them * in Opaque. So that we can use these data structures * when we get the response */ ssn = eaptls_new_session(inst->ctx, client_cert); if (!ssn) { return 0; } /* * Verify the peer certificate, if asked. */ if (client_cert) { DEBUG2(" rlm_eap_tls: Requiring client certificate"); verify_mode = SSL_VERIFY_PEER; verify_mode |= SSL_VERIFY_FAIL_IF_NO_PEER_CERT; verify_mode |= SSL_VERIFY_CLIENT_ONCE; } SSL_set_verify(ssn->ssl, verify_mode, cbtls_verify); /* * Create a structure for all the items required to be * verified for each client and set that as opaque data * structure. * * NOTE: If we want to set each item sepearately then * this index should be global. */ SSL_set_ex_data(ssn->ssl, 0, (void *)handler); SSL_set_ex_data(ssn->ssl, 1, (void *)inst->conf); ssn->length_flag = inst->conf->include_length; /* * We use default fragment size, unless the Framed-MTU * tells us it's too big. Note that we do NOT account * for the EAP-TLS headers if conf->fragment_size is * large, because that config item looks to be confusing. * * i.e. it should REALLY be called MTU, and the code here * should figure out what that means for TLS fragment size. * asking the administrator to know the internal details * of EAP-TLS in order to calculate fragment sizes is * just too much. */ ssn->offset = inst->conf->fragment_size; vp = pairfind(handler->request->packet->vps, PW_FRAMED_MTU); if (vp && ((vp->vp_integer - 14) < ssn->offset)) { /* * Discount the Framed-MTU by: * 4 : EAPOL header * 4 : EAP header (code + id + length) * 1 : EAP type == EAP-TLS * 1 : EAP-TLS Flags * 4 : EAP-TLS Message length * (even if conf->include_length == 0, * just to be lazy). * --- * 14 */ ssn->offset = vp->vp_integer - 14; } handler->opaque = ((void *)ssn); handler->free_opaque = session_free; DEBUG2(" rlm_eap_tls: Initiate"); /* * PEAP-specific breakage. */ if (handler->eap_type == PW_EAP_PEAP) { /* * As it is a poorly designed protocol, PEAP uses * bits in the TLS header to indicate PEAP * version numbers. For now, we only support * PEAP version 0, so it doesn't matter too much. * However, if we support later versions of PEAP, * we will need this flag to indicate which * version we're currently dealing with. */ ssn->peap_flag = 0x00; /* * PEAP version 0 requires 'include_length = no', * so rather than hoping the user figures it out, * we force it here. */ ssn->length_flag = 0; } /* * TLS session initialization is over. Now handle TLS * related handshaking or application data. */ status = eaptls_start(handler->eap_ds, ssn->peap_flag); DEBUG2(" rlm_eap_tls: Start returned %d", status); if (status == 0) return 0; /* * The next stage to process the packet. */ handler->stage = AUTHENTICATE; return 1; } /* * Do authentication, by letting EAP-TLS do most of the work. */ static int eaptls_authenticate(void *arg UNUSED, EAP_HANDLER *handler) { eaptls_status_t status; tls_session_t *tls_session = (tls_session_t *) handler->opaque; DEBUG2(" rlm_eap_tls: Authenticate"); status = eaptls_process(handler); DEBUG2(" eaptls_process returned %d\n", status); switch (status) { /* * EAP-TLS handshake was successful, return an * EAP-TLS-Success packet here. */ case EAPTLS_SUCCESS: break; /* * The TLS code is still working on the TLS * exchange, and it's a valid TLS request. * do nothing. */ case EAPTLS_HANDLED: return 1; /* * Handshake is done, proceed with decoding tunneled * data. */ case EAPTLS_OK: DEBUG2(" rlm_eap_tls: Received unexpected tunneled data after successful handshake."); #ifndef NDEBUG if ((debug_flag > 2) && fr_log_fp) { unsigned int i; unsigned int data_len; unsigned char buffer[1024]; data_len = (tls_session->record_minus)(&tls_session->dirty_in, buffer, sizeof(buffer)); log_debug(" Tunneled data (%u bytes)\n", data_len); for (i = 0; i < data_len; i++) { if ((i & 0x0f) == 0x00) fprintf(fr_log_fp, " %x: ", i); if ((i & 0x0f) == 0x0f) fprintf(fr_log_fp, "\n"); fprintf(fr_log_fp, "%02x ", buffer[i]); } fprintf(fr_log_fp, "\n"); } #endif eaptls_fail(handler->eap_ds, 0); return 0; break; /* * Anything else: fail. */ default: return 0; } /* * Success: Return MPPE keys. */ eaptls_success(handler->eap_ds, 0); eaptls_gen_mppe_keys(&handler->request->reply->vps, tls_session->ssl, "client EAP encryption"); return 1; } /* * The module name should be the only globally exported symbol. * That is, everything else should be 'static'. */ EAP_TYPE rlm_eap_tls = { "eap_tls", eaptls_attach, /* attach */ eaptls_initiate, /* Start the initial request */ NULL, /* authorization */ eaptls_authenticate, /* authentication */ eaptls_detach /* detach */ };

3 16

freeradius-client thread safety & memleaks patch
by Alex Massover 26 Jul '10

26 Jul '10

Hello, I want to sync changes I did to CVS. No new functionality only fixes for thread-safety and fixing memory leaks. I use this code with static (shared among threads) rc_handle and experience no segfaults and no memory leaks :) Some parts of this patch we posted before, but this one combines them all and is prepared against current CVS. Changes: freeradius-client.h: - adding rad_hostent (will be used by ip_util.c) - changing functions signatures due to changes in ip_utils.c avpair.c: - using localtime_r() instead of non-reentrant localtime() - fixing minor memory leak posted by somebody else (not me) to this mail list, but not applied yet - inet_ntoa() is not reentrant, using inet_ntop() instead - gmtime_r() instead of gmtime() buildreq.c: - rc_get_id() is now simple, fast and reentrant. config.c: - getservbyname_r() instead of getservbyname() - using rad_hostent in rc_gethostby* - fixing minor memleak - serv->secret[j] is not freed ip_util.c: - using rad_hostent instead of thread-local buffer. Thread local storage is tricky and might leak in thread per request scenario - implementation of rc_ip_hostname() is copied from freeradius server. (something is wrong with current one) - using getservbyname_r() instead of getservbyname() sendserver.c: - using new rc_ip_hostname() - using rand_r() instead of rand() -- Best Regards, Alex Massover VoIP R&D TL Jajah Inc. This mail was sent via Mail-SeCure System.

1 2

small patch pathc postgresql trigger function
by Jasen Betts 25 Jul '10

25 Jul '10

>From jasen(a)treshna.com Thu Jul 22 13:02:19 2010 Date: Thu, 22 Jul 2010 13:02:19 +1200 From: Jasen Betts <jasen(a)treshna.com> To: freeradius-devel(a)lists.freeradius.org Subject: small patch pathc postgresql trigger function Message-ID: <20100722010219.GA7516(a)treshna.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.18 (2008-05-17) Status: RO X-Status: A here's a small patch, not sure where to submit it. This version should prove faster as it halves the number of writes, however it still does a lookup. this code replaces similar code in etc/raddb/sql/postgresql/update_radacct_group_trigger.sql ---- CREATE OR REPLACE FUNCTION upd_radgroups() RETURNS trigger AS' BEGIN IF NEW.groupname IS NULL THEN NEW.groupname = groupname FROM radusergroup WHERE calledstationid = NEW.calledstationid AND username = NEW.username; END IF; RETURN NEW; END 'LANGUAGE plpgsql; DROP TRIGGER upd_radgroups ON radacct; CREATE TRIGGER upd_radgroups BEFORE INSERT ON radacct FOR EACH ROW EXECUTE PROCEDURE upd_radgroups(); ---- -- Jasen Betts treshna Enterprises Ltd

1 1

Master key and Pairwise Master Key
by Khan Ferdous Wahid 15 Jul '10

15 Jul '10

Hi, I want to know about the Master key (MK) encapsulation and Pairwise Master key (PMK) generation during EAP-TLS or EAP-TTLS methods. When the supplicant is authenticated, the server generates a MK and sends it in encrypted format to the supplicant. How this MK (I think it is a random number) is encrypted? Which algorithm is used and which parameters are included (input) to disguise the MK? Then how the PMK is generated independently inside server and supplicant? What algorithm and parameters are used to pass the PMK to authenticator (Access point)? Please tell me clearly because I am a newbie. Which source codes include these operations, where should I look? Thank you. /Khan

2 1

reject_delay
by Ben Wiechman 13 Jul '10

13 Jul '10

When an access-reject is being delayed due to the configuration of reject_delay, is the server core in any way aware of that? Problem: misbehaving clients that are not valid making many, many repeated network entry attempts in quick succession. Receiving a repeat request from the client appears to be causing a "discarding duplicate request" entry in the logs. Is there a (simple...) way to identify if a response is being delayed and update the discarding dups log entry accordingly?

3 3

HELP unsubscribe me please
by lanquan 12 Jul '10

12 Jul '10

Thank you > From: freeradius-devel-request(a)lists.freeradius.org > Subject: Freeradius-Devel Digest, Vol 63, Issue 1 > To: freeradius-devel(a)lists.freeradius.org > Date: Fri, 9 Jul 2010 12:00:15 +0200 > > Send Freeradius-Devel mailing list submissions to > freeradius-devel(a)lists.freeradius.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://lists.freeradius.org/mailman/listinfo/freeradius-devel > or, via email, send a message with subject or body 'help' to > freeradius-devel-request(a)lists.freeradius.org > > You can reach the person managing the list at > freeradius-devel-owner(a)lists.freeradius.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Freeradius-Devel digest..." > > > Today's Topics: > > 1. sql module and radgroup... (predrag balorda) > 2. Re: RES: Patch: avpair.c - empty reply VALUE_PAIR (0x0) for > replies withunknown vendor attributes (Adam Lewis) > 3. Re: sql module and radgroup... (Alan DeKok) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 8 Jul 2010 15:12:16 +0200 > From: predrag balorda <predrag.balorda(a)gmail.com> > Subject: sql module and radgroup... > To: freeradius-devel(a)lists.freeradius.org > Message-ID: > <AANLkTilbHng3bcXjYqFjZyoQQWAEfArp-pOvngSGHtx2(a)mail.gmail.com> > Content-Type: text/plain; charset=ISO-8859-1 > > Would it be so difficult to add a "groupinfo" table? Just like you > have /etc/passwords you have /etc/groups so should you have userinfo > and radgroup or groupinfo or somesuch. > Namely, there is an awful kludge in radgroupreply in that if a a reply > with a group name exists then the group exists, otherwise it doesn't. > Now that's funny, it's the same thing as saying if there is an entry > in radcheck then the user exists, otherwise they don't. Which is > silly. Not only in my opinion but in the the opinion of all database > people out there. Referential integrity is non-existent. So is > entity-integrity. > > A simple statement like this in your sql scripts would suffice (mysql > exaple here) > > DROP TABLE IF EXISTS `groupinfo`; > CREATE TABLE `groupinfo` ( > `id` bigint(20) NOT NULL auto_increment, > `groupname` varchar(200) default NULL, > PRIMARY KEY (`id`), > KEY name (`groupname`) > ); > > Then you can draw all sorts of relations back to this table from > radgroupcheck, radgroupreply, usergroup etc. > > Try and create a radgroupreply for group "Normal", say a bandwidth > limitation of some sort. Then add another limitation but this time > make an intentional typo and put in "nOrmal" in groupname instead. > What happens? If you had referential integrity the database wouldn't > let you make such mistaks. Just like this silly gmail webmail won't > let me make mistakes and underlines them with a red squiggly line. > > I'm saying all this because I'm writing a freeradius frontend (there, > that stupid red squiggly line again) - I'll probably call it > dialUpAdmin# as I'm using NHibernate, mono, asp.net2 and it's going > nice; I'll try and submit the initial version to the cvs (or is it > svn) repository sometime next week - and silliness like this pops up > everywhere. Some simple referential and entity integrity never hurt no > one. > > I'm probably talking crap here as I'll be switching to LDAP soon > enough for all this to go away, but still. It'd be nice. > > Best regards, > > Pele > > > ------------------------------ > > Message: 2 > Date: Thu, 8 Jul 2010 07:03:23 -0700 (PDT) > From: Adam Lewis <VNQM87(a)motorola.com> > Subject: Re: RES: Patch: avpair.c - empty reply VALUE_PAIR (0x0) for > replies withunknown vendor attributes > To: freeradius-devel(a)lists.freeradius.org > Message-ID: <29107373.post(a)talk.nabble.com> > Content-Type: text/plain; charset=us-ascii > > > Just a warning to anyone using the client on an embedded platform. The local > variable "buffer" in the function rc_avpair_gen() uses 507 bytes, so if the > function has many attributes to process, 507 bytes will be added to the > stack every time the function recurses. Perhaps another reason why this > function should be re-written. > > Regards, > Adam. > > > > Alan DeKok-2 wrote: > > > > Toledo, Luis Carlos wrote: > >> How to debug the unknown attributes? Put some prints in the code or have > >> clear way to do this? > > > > The code should print out which attributes it receives. If it > > doesn't, supply a patch to fix it. > > > > Alan DeKok. > > - > > List info/subscribe/unsubscribe? See > > http://www.freeradius.org/list/devel.html > > > > > > -- > View this message in context: http://old.nabble.com/Patch%3A-avpair.c---empty-reply-VALUE_PAIR-%280x0%29-… > Sent from the FreeRadius - Dev mailing list archive at Nabble.com. > > > > ------------------------------ > > Message: 3 > Date: Thu, 08 Jul 2010 16:16:54 +0200 > From: Alan DeKok <aland(a)deployingradius.com> > Subject: Re: sql module and radgroup... > To: FreeRadius developers mailing list > <freeradius-devel(a)lists.freeradius.org> > Message-ID: <4C35DDD6.2090009(a)deployingradius.com> > Content-Type: text/plain; charset=ISO-8859-1 > > predrag balorda wrote: > > Would it be so difficult to add a "groupinfo" table? Just like you > > have /etc/passwords you have /etc/groups so should you have userinfo > > and radgroup or groupinfo or somesuch. > > Namely, there is an awful kludge in radgroupreply in that if a a reply > > with a group name exists then the group exists, otherwise it doesn't. > > ? The reply is allowed to be empty. If it is empty, it has no effect > on group membership. > > > Now that's funny, it's the same thing as saying if there is an entry > > in radcheck then the user exists, otherwise they don't. Which is > > silly. Not only in my opinion but in the the opinion of all database > > people out there. Referential integrity is non-existent. So is > > entity-integrity. > > I have no idea what that means. > > > A simple statement like this in your sql scripts would suffice (mysql > > exaple here) > > > > DROP TABLE IF EXISTS `groupinfo`; > > CREATE TABLE `groupinfo` ( > > `id` bigint(20) NOT NULL auto_increment, > > `groupname` varchar(200) default NULL, > > PRIMARY KEY (`id`), > > KEY name (`groupname`) > > ); > > > > Then you can draw all sorts of relations back to this table from > > radgroupcheck, radgroupreply, usergroup etc. > > Sure. That's useful, but not required. > > I've seen 7 figure commercial solutions with *less* referential > integrity than the default FreeRADIUS schema. > > > Try and create a radgroupreply for group "Normal", say a bandwidth > > limitation of some sort. Then add another limitation but this time > > make an intentional typo and put in "nOrmal" in groupname instead. > > What happens? If you had referential integrity the database wouldn't > > let you make such mistaks. Just like this silly gmail webmail won't > > let me make mistakes and underlines them with a red squiggly line. > > You can supply a patch to the SQL schema && statements to add this > capability. The software *is* open source. > > > I'm saying all this because I'm writing a freeradius frontend (there, > > that stupid red squiggly line again) - I'll probably call it > > dialUpAdmin# as I'm using NHibernate, mono, asp.net2 and it's going > > nice; I'll try and submit the initial version to the cvs (or is it > > svn) repository sometime next week > > 'git'. > > > - and silliness like this pops up > > everywhere. Some simple referential and entity integrity never hurt no > > one. > > Sure. But it's one more requirement before people get the server up > and running. > > > I'm probably talking crap here as I'll be switching to LDAP soon > > enough for all this to go away, but still. It'd be nice. > > As always, patches are welcome. > > Alan DeKok. > > > ------------------------------ > > - > List info/subscribe/unsubscribe? See http://www.freeradius.org/list/devel.html > > > End of Freeradius-Devel Digest, Vol 63, Issue 1 > ***********************************************

2 1

sql module and radgroup...
by predrag balorda 10 Jul '10

10 Jul '10

Would it be so difficult to add a "groupinfo" table? Just like you have /etc/passwords you have /etc/groups so should you have userinfo and radgroup or groupinfo or somesuch. Namely, there is an awful kludge in radgroupreply in that if a a reply with a group name exists then the group exists, otherwise it doesn't. Now that's funny, it's the same thing as saying if there is an entry in radcheck then the user exists, otherwise they don't. Which is silly. Not only in my opinion but in the the opinion of all database people out there. Referential integrity is non-existent. So is entity-integrity. A simple statement like this in your sql scripts would suffice (mysql exaple here) DROP TABLE IF EXISTS `groupinfo`; CREATE TABLE `groupinfo` ( `id` bigint(20) NOT NULL auto_increment, `groupname` varchar(200) default NULL, PRIMARY KEY (`id`), KEY name (`groupname`) ); Then you can draw all sorts of relations back to this table from radgroupcheck, radgroupreply, usergroup etc. Try and create a radgroupreply for group "Normal", say a bandwidth limitation of some sort. Then add another limitation but this time make an intentional typo and put in "nOrmal" in groupname instead. What happens? If you had referential integrity the database wouldn't let you make such mistaks. Just like this silly gmail webmail won't let me make mistakes and underlines them with a red squiggly line. I'm saying all this because I'm writing a freeradius frontend (there, that stupid red squiggly line again) - I'll probably call it dialUpAdmin# as I'm using NHibernate, mono, asp.net2 and it's going nice; I'll try and submit the initial version to the cvs (or is it svn) repository sometime next week - and silliness like this pops up everywhere. Some simple referential and entity integrity never hurt no one. I'm probably talking crap here as I'll be switching to LDAP soon enough for all this to go away, but still. It'd be nice. Best regards, Pele

4 11

Patch: avpair.c - empty reply VALUE_PAIR (0x0) for replies with unknown vendor attributes
by Stefan Karss 08 Jul '10

08 Jul '10

Hello again, when the RADIUS servers sends a reply packet with an attribute for an unknown vendor freeradius-client just skips over this when filling response VALUE_PAIR list. Unfortunately it doesn't for unknown attributes from known vendors - in that case rc_avpair_gen returns NULL in the first recursion which results in a NULL pointer for reply VALUE_PAIR. The attached patch just skips over unknown attributes for known vendors just like the lib skips over attributes from unknown vendors. Rgds, Stefan

4 13