isimodem: parse extra details from REG_STATUS_IND
PATCH 1/1 isimodem: parse extra details from REG_STATUS_IND
Extract additional operator information available in REG_STATUS_IND messages. These include mnc, mnc, and different variations of operator name.
Additionally, in case NET_OPER_NAME_READ_RESP is an error result, which happens with some operators, report the extracted information to core instead of bailing out.
--- drivers/isimodem/debug.c | 1 + drivers/isimodem/network-registration.c | 73 ++++++++++++++++++++++++- drivers/isimodem/network.h | 1 + 3 files changed, 74 insertions(+), 1 deletion(-)
drivers/isimodem/debug.c +1 -0
@@ -1190,6 +1190,7 @@ const char *net_subblock_name(enum net_subblock value)_ (NET_REGISTRATION_CONF1_INFO );_ (NET_ROAMING_CONF1_INFO );_ (NET_AVAIL_NETWORK_INFO_COMMON ); +_ (NET_REG_STATUS_IND_OPER_NAME );_ (NET_OPER_NAME_INFO ); }return "NET_<UNKNOWN>" ;
drivers/isimodem/network-registration.c +72 -1
119 lines not on this page · raw message
drivers/isimodem/network.h +1 -0
@@ -90,6 +90,7 @@ enum net_subblock {NET_REGISTRATION_CONF1_INFO =0x59 ,NET_ROAMING_CONF1_INFO =0x5A ,NET_AVAIL_NETWORK_INFO_COMMON =0xE1 , +NET_REG_STATUS_IND_OPER_NAME =0xE3 , NET_OPER_NAME_INFO = 0xE7, };
Signature · 1 line
2.42.0
Hi Sicelo,
Quoted 12 lines · Sicelo A. Mhlongo
Extract additional operator information available in
REG_STATUS_INDmessages. These include mnc, mnc, and different variations of operator name.Additionally, in case
NET_OPER_NAME_READ_RESPis an error result, which happens with some operators, report the extracted information to core instead of bailing out.--- drivers/isimodem/debug.c | 1 + drivers/isimodem/network-registration.c | 73 ++++++++++++++++++++++++- drivers/isimodem/network.h | 1 + 3 files changed, 74 insertions(+), 1 deletion(-)
I do get a compiler warning here:
drivers/isimodem/network-registration.c:537:3: error: misleading indentation; statement is not part of the previous 'if' [-Werror,-Wmisleading-indentation]
strncpy(op.name, nd->nitz_name, OFONO_MAX_OPERATOR_NAME_LENGTH);
^
drivers/isimodem/network-registration.c:532:2: note: previous statement is here
if (!check_response_status(msg, NET_OLD_OPER_NAME_READ_RESP) &&
^A few nits:
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 203
Quoted 10 lines
drivers/isimodem/network-registration.c +6
-0@@ -193,6 +195,12 @@ static void reg_status_ind_cb(const GIsiMessage *msg, void *data)struct netreg_data *nd =ofono_netreg_get_data (netreg);GIsiSubBlockIter iter; +uint8_t len =0 ; +uint8_t type =0 ; +char *abbrev =NULL ; +char *an =NULL ; +char *fn =NULL ; +
You set these to NULL here, but ...
Quoted 6 lines
drivers/isimodem/network-registration.c +0
-0if (netreg ==NULL || nd ==NULL )return ;@@ -209,16 +217,69 @@ static void reg_status_ind_cb(const GIsiMessage *msg, void *data)if (!parse_common_info (&iter, &nd->reg))
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 228
Quoted 10 lines
drivers/isimodem/network-registration.c +9
-0return ; + +/* Parse abbreviated operator name */ +if (!g_isi_sb_iter_get_byte (&iter, &len,7 )) +return ; + + len *=2 ; + +if (!g_isi_sb_iter_get_alpha_tag (&iter, &abbrev, len,8 )) +return ;
Is the intent really to ignore the whole message if this attribute can't be parsed?
Quoted 26 lines
drivers/isimodem/network-registration.c +20
-0break ;case NET_GSM_REG_INFO :if (!parse_gsm_info (&iter, &nd->gsm))return ; + +/* Parse mcc and mnc */ +if (!g_isi_sb_iter_get_oper_code (&iter, nd->mcc, + nd->mnc,8 )) +return ; +break ; + +case NET_REG_STATUS_IND_OPER_NAME : + +if (!g_isi_sb_iter_get_byte (&iter, &type,2 )) +return ; + +if (!g_isi_sb_iter_get_byte (&iter, &len,5 )) +return ; + +/* Name is UCS-2 encoded */ + len *=2 ; + +switch (type) { +case 2 :
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 264
Quoted 10 lines
drivers/isimodem/network-registration.c +10
-0+/* +* Alternate full name, which is similar to +* long name, but may have less whitespace +*/ +g_isi_sb_iter_get_alpha_tag (&iter, &an, len,6 ); +break ; +case 3 : +/* Full name */ +g_isi_sb_iter_get_alpha_tag (&iter, &fn, len,6 ); +break ;
But you don't check the error returns of these two calls
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 273
Quoted 9 lines
drivers/isimodem/network-registration.c +5
-0+ }break ; } } +if (fn[0 ] !='\0' ) +strncpy (nd->nitz_name, fn,OFONO_MAX_OPERATOR_NAME_LENGTH ); +else if (an[0 ] !='\0' ) +strncpy (nd->nitz_name, an,OFONO_MAX_OPERATOR_NAME_LENGTH );
So how do you know fn/an aren't NULL?
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 275
Quoted 2 lines
drivers/isimodem/network-registration.c +2
-0+else if (abbrev[0 ] !='\0' ) +strncpy (nd->nitz_name, abbrev,OFONO_MAX_OPERATOR_NAME_LENGTH );
Use g_strlcpy instead of strncpy since it takes care of setting the terminating NULL.
Quoted 21 lines
drivers/isimodem/network-registration.c +9
-0+ + nd->nitz_name[OFONO_MAX_OPERATOR_NAME_LENGTH ] ='\0' ; + +g_free (an); +g_free (fn); +g_free (abbrev); +ofono_netreg_status_notify (netreg,isi_status_to_at_status (&nd->reg), nd->gsm.lac, nd->gsm.ci,isi_to_at_tech (&nd->rat, &nd->gsm));@@ -458,6 +519,8 @@ static void name_get_resp_cb(const GIsiMessage *msg, void *data) {struct isi_cb_data *cbd = data;ofono_netreg_operator_cb_t cb = cbd->cb; +struct ofono_netreg *netreg = cbd->user; +struct netreg_data *nd =ofono_netreg_get_data (netreg);struct ofono_network_operator op;GIsiSubBlockIter iter;@@ -468,7 +531,14 @@ static void name_get_resp_cb(const GIsiMessage *msg, void *data)
drivers/isimodem/network-registration.c static void name_get_resp_cb(const GIsiMessage *msg, void *data) · line 540
Quoted 10 lines
drivers/isimodem/network-registration.c +7
-1if (!check_response_status (msg,NET_OLD_OPER_NAME_READ_RESP ) && !check_response_status (msg,NET_OPER_NAME_READ_RESP )) - goto error; +if (nd->nitz_name[0 ] =='\0' ) + goto error; + +strncpy (op.name, nd->nitz_name,OFONO_MAX_OPERATOR_NAME_LENGTH ); + op.name[OFONO_MAX_OPERATOR_NAME_LENGTH ] ='\0' ; +strncpy (op.mcc, nd->mcc,OFONO_MAX_MCC_LENGTH ); +strncpy (op.mnc, nd->mnc,OFONO_MAX_MNC_LENGTH );
ditto here, g_strlcpy.
drivers/isimodem/network-registration.c static void name_get_resp_cb(const GIsiMessage *msg, void *data) · line 576
Quoted 12 lines
drivers/isimodem/network-registration.c +2
-0+ goto success; for (g_isi_sb_iter_init(&iter, msg, 6); g_isi_sb_iter_is_valid(&iter);@@ -500,6 +570,7 @@ static void name_get_resp_cb(const GIsiMessage *msg, void *data) } } +success:CALLBACK_WITH_SUCCESS (cb, &op, cbd->data);return ;
Regards,
-Denis
Hello Denis
Quoted 10 lines · Denis Kenzior
I do get a compiler warning here:
drivers/isimodem/network-registration.c:537:3: error: misleading indentation; statement is not part of the previous 'if' [-Werror,-Wmisleading-indentation]strncpy(op.name, nd->nitz_name, OFONO_MAX_OPERATOR_NAME_LENGTH); ^ drivers/isimodem/network-registration.c:532:2: note: previous statement is here if (!check_response_status(msg, NET_OLD_OPER_NAME_READ_RESP) && ^
Interestingly this one was not caught on my compilation runs. I will fix in v2.
Quoted 8 lines
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 264
Quoted 6 lines
drivers/isimodem/network-registration.c +6
-0+g_isi_sb_iter_get_alpha_tag (&iter, &an, len,6 ); +break ; +case 3 : +/* Full name */ +g_isi_sb_iter_get_alpha_tag (&iter, &fn, len,6 ); +break ;But you don't check the error returns of these two calls
I left out the checking here because in both cases, we are breaking at the next statement, regardless of the success or failure of the calls. I will see if I can implement checking.
Quoted 6 lines
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 273
Quoted 4 lines
drivers/isimodem/network-registration.c +4
-0+if (fn[0 ] !='\0' ) +strncpy (nd->nitz_name, fn,OFONO_MAX_OPERATOR_NAME_LENGTH ); +else if (an[0 ] !='\0' ) +strncpy (nd->nitz_name, an,OFONO_MAX_OPERATOR_NAME_LENGTH );So how do you know fn/an aren't NULL?
Please excuse my confusion here - I thought the if conditions do check whether an or fn are NULL. I will appreciate assistance or suggestion for an alternative way to ensure this, since I definitely do not want to copy over garbage into nd->nitz_name at this point in the code.
Quoted 2 lines
Use
g_strlcpyinstead of strncpy since it takes care of setting the terminating NULL.
I will do this in v2. While going through some older commits, I also found use of l_strlcpy. Should I prefer g_strlcpy in this instance?
Thank you very much for the review, and looking forward to hearing from you soon.
Sincerely
Sicelo
Hi Sicelo,
Quoted 10 lines
Quoted 9 lines
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 264
Quoted 7 lines
drivers/isimodem/network-registration.c +6
-0+g_isi_sb_iter_get_alpha_tag (&iter, &an, len,6 ); +break ; +case 3 : +/* Full name */ +g_isi_sb_iter_get_alpha_tag (&iter, &fn, len,6 ); +break ;But you don't check the error returns of these two calls
Quoted 3 lines
I left out the checking here because in both cases, we are breaking at the next statement, regardless of the success or failure of the calls. I will see if I can implement checking.
You break out of the outer switch, but you still don't know whether obtaining the alpha tag succeeded or not. So fn/an (or both) can still be NULL.
Quoted 10 lines
Quoted 7 lines
drivers/isimodem/network-registration.c static void reg_status_ind_cb(const GIsiMessage *msg, void *data) · line 273
Quoted 5 lines
drivers/isimodem/network-registration.c +4
-0+if (fn[0 ] !='\0' ) +strncpy (nd->nitz_name, fn,OFONO_MAX_OPERATOR_NAME_LENGTH ); +else if (an[0 ] !='\0' ) +strncpy (nd->nitz_name, an,OFONO_MAX_OPERATOR_NAME_LENGTH );So how do you know fn/an aren't NULL?
Please excuse my confusion here - I thought the
ifconditions do check whether an or fn are NULL. I will appreciate assistance or suggestion
No, you're checking that fn/an are not equal to an empty string, not that fn/an are NULL.
You want something like:
if (fn && fn[0] != '\0')
...
Quoted 3 lines
for an alternative way to ensure this, since I definitely do not want to copy over garbage into
nd->nitz_nameat this point in the code.
Same thing for abbrev. You're checking whether or not abbrev is an empty string "", but the whole TLV might be absent and g_isi_sb_iter_get_alpha_tag would never be executed.
Quoted 5 lines
Quoted 2 lines
Use
g_strlcpyinstead of strncpy since it takes care of setting the terminating NULL.I will do this in v2. While going through some older commits, I also found use of
l_strlcpy. Should I preferg_strlcpyin this instance?
They're equivalent, you can use either. g_strlcpy is the GLib version. l_strlcpy is the ell version. oFono will slowly be migrating away from GLib to use ell.
Since you're already using l_utf8_from_ucs2be inside gisi/*, you might as well use the ell version.
Regards,
-Denis