| Message ID | 59615bfbda0d1a899675ac86b8018412736186a6.1326658542.git.gmbnomis@gmail.com |
|---|---|
| State | Accepted |
| Commit | 8ab7a3708910c0f572084e7dabff10d7c8ee3734 |
| Headers | show |
On Sun, Jan 15, 2012 at 09:32:47PM +0100, Simon Baatz wrote: > The EF-AD (administrative data) file may contain information about the > length of the MNC (2 or 3 digits) in the IMSI. This can be used to > construct the realm according to 3GPP TS 23.003 during EAP-SIM or > EAP-AKA authentication. Thanks, applied. Though, I added a new return value for indicating that the optional MNC length field was not present in EF_AD to make the debug output somewhat cleaner. The first SIM card I tested (prepaid Elisa/Kolumbus) this with happened to not include the MNC length and the default assumption of 3-digit MNC in the patch 4/4 was not the correct one for that card.. Unfortunately, it looks like there would need to be a list of MCC/MNC values with the MNC length for the automatic detection to work reliably (what a silly design to make that field optional and not allow automatic way of figuring out the correct field length).
Hi Jouni, > > The EF-AD (administrative data) file may contain information about the > > length of the MNC (2 or 3 digits) in the IMSI. This can be used to > > construct the realm according to 3GPP TS 23.003 during EAP-SIM or > > EAP-AKA authentication. > > Thanks, applied. Though, I added a new return value for indicating that > the optional MNC length field was not present in EF_AD to make the debug > output somewhat cleaner. > > The first SIM card I tested (prepaid Elisa/Kolumbus) this with happened > to not include the MNC length and the default assumption of 3-digit MNC > in the patch 4/4 was not the correct one for that card.. Unfortunately, > it looks like there would need to be a list of MCC/MNC values with the > MNC length for the automatic detection to work reliably (what a silly > design to make that field optional and not allow automatic way of > figuring out the correct field length). this is not going to work. You can have operators that can mix and match 2 and 3 digit MNC as they please (within the same network). The field is most likely present, but the SIM filesystem access might be broken. The firmware for non-phone modems is really crappy in this area. Especially the Qualcomm based ones are broken all over the places. Regards Marcel
On Sun, Jan 22, 2012 at 06:48:16PM +0100, Marcel Holtmann wrote: > this is not going to work. You can have operators that can mix and match > 2 and 3 digit MNC as they please (within the same network). Great. Well, I can use the 3-digit MNC as a default and add exceptions as needed, I'd guess. The only SIM card that I found without the MNC length field was for MCC=244 and all networks in Finland use 2-digit MNC as far as I can tell.. > The field is > most likely present, but the SIM filesystem access might be broken. I don't think that that would have been the case in my tests. I used the same reader for all the SIM cards and the AF_AD contents from the Elisa/Kolumbus card returned three octets of data while the other cards returned four octets.
diff --git a/src/utils/pcsc_funcs.c b/src/utils/pcsc_funcs.c index c36193e..9c9f145 100644 --- a/src/utils/pcsc_funcs.c +++ b/src/utils/pcsc_funcs.c @@ -1024,6 +1024,58 @@ int scard_get_imsi(struct scard_data *scard, char *imsi, size_t *len) /** + * scard_get_mnc_len - Read length of MNC in the IMSI from SIM/USIM + * card + * @scard: Pointer to private data from scard_init() + * Returns: length (>0) on success, -1 if administrative data file cannot be + * selected, -2 if administrative data file selection returns invalid result + * code, -3 if parsing FSP template file fails (USIM only), -4 if length of + * the file is unexpected, -5 if reading file fails, -6 if MNC length is not + * in range (i.e. 2 or 3). + * + */ +int scard_get_mnc_len(struct scard_data *scard) +{ + unsigned char buf[100]; + size_t blen; + int file_size; + + wpa_printf(MSG_DEBUG, "SCARD: reading MNC len from (GSM) EF-AD"); + blen = sizeof(buf); + if (scard_select_file(scard, SCARD_FILE_GSM_EF_AD, buf, &blen)) + return -1; + if (blen < 4) { + wpa_printf(MSG_WARNING, "SCARD: too short (GSM) EF-AD " + "header (len=%ld)", (long) blen); + return -2; + } + + if (scard->sim_type == SCARD_GSM_SIM) { + file_size = (buf[2] << 8) | buf[3]; + } else { + if (scard_parse_fsp_templ(buf, blen, NULL, &file_size)) + return -3; + } + if (file_size < 4 || file_size > sizeof(buf)) { + wpa_printf(MSG_DEBUG, "SCARD: invalid file length=%ld", + (long) file_size); + return -4; + } + + if (scard_read_file(scard, buf, file_size)) + return -5; + buf[3] = buf[3] & 0x0f; /* upper nibble reserved for future use */ + if (buf[3] < 2 || buf[3] > 3) { + wpa_printf(MSG_DEBUG, "SCARD: invalid MNC length=%ld", + (long) buf[3]); + return -6; + } + wpa_printf(MSG_DEBUG, "SCARD: MNC length=%ld", (long) buf[3]); + return buf[3]; +} + + +/** * scard_gsm_auth - Run GSM authentication command on SIM card * @scard: Pointer to private data from scard_init() * @_rand: 16-byte RAND value from HLR/AuC diff --git a/src/utils/pcsc_funcs.h b/src/utils/pcsc_funcs.h index 543f7c5..483247b 100644 --- a/src/utils/pcsc_funcs.h +++ b/src/utils/pcsc_funcs.h @@ -26,6 +26,7 @@ #define SCARD_FILE_GSM_DF 0x7F20 #define SCARD_FILE_UMTS_DF 0x7F50 #define SCARD_FILE_GSM_EF_IMSI 0x6F07 +#define SCARD_FILE_GSM_EF_AD 0x6FAD #define SCARD_FILE_EF_DIR 0x2F00 #define SCARD_FILE_EF_ICCID 0x2FE2 #define SCARD_FILE_EF_CK 0x6FE1 @@ -47,6 +48,7 @@ void scard_deinit(struct scard_data *scard); int scard_set_pin(struct scard_data *scard, const char *pin); int scard_get_imsi(struct scard_data *scard, char *imsi, size_t *len); +int scard_get_mnc_len(struct scard_data *scard); int scard_gsm_auth(struct scard_data *scard, const unsigned char *_rand, unsigned char *sres, unsigned char *kc); int scard_umts_auth(struct scard_data *scard, const unsigned char *_rand,