Hi Vasyl, On 02/23/2018 09:26 PM, Vasyl Vavrychuk wrote: > Previously for network without SSID g_supplicant_network_get_ssid > returned pointer to empty string. Then some naive users of it treated > such value as a valid SSID. > > This fixes "Network SSID is not set" for hidden networks. Can you elaborate a bit what is wrong? I tried to understand what you mean with 'previously'. Does a newer version of wpa_s behave differently compared to an older version? Also I found following commit: commit 91913c40054d9c55c6d38f0c626a78bdbcf4279a Author: Tomasz Bursztyka <tomasz.bursztyka(a)linux.intel.com&gt; Date: Thu Dec 12 15:57:05 2013 +0200 gsupplicant: A network ssid of length 0 is valid, it's an hidden one In case of an hidden AP around, its ssid has a length of 0, but it is obviously a valid network anyway. Fixes such regression from commit 363393cfb1a5f95f8892f40662486c87b80d0091 Reported by Sameer Naik diff --git a/gsupplicant/supplicant.c b/gsupplicant/supplicant.c index 23ea41a03642..559dfbc6fd2f 100644 --- a/gsupplicant/supplicant.c +++ b/gsupplicant/supplicant.c @@ -853,7 +853,7 @@ const char *g_supplicant_network_get_security(GSupplicantNetwork *network) const void *g_supplicant_network_get_ssid(GSupplicantNetwork *network, unsigned int *ssid_len) { - if (!network || network->ssid_len == 0) { + if (!network) { *ssid_len = 0; return NULL; } The report from Sameer was """ I am looking into the connman plugin/wifi.c to figure out the issue. so far i have found that in scan_callback_hidden function, g_supplicant_interface_get_max_scan_ssids(wifi->interface), returns 0 and therefore no attempt to connect to provisioned hidden wifi networks is performed. Will trace the calls further to figure out why g_supplicant_interface_get_max_scan_ssids returns 0. """ https://lists.01.org/pipermail/connman/2013-March/014020.html > --- > gsupplicant/supplicant.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/gsupplicant/supplicant.c b/gsupplicant/supplicant.c > index 5246c80b..ec2f4307 100644 > --- a/gsupplicant/supplicant.c > +++ b/gsupplicant/supplicant.c > @@ -1189,7 +1189,7 @@ const char *g_supplicant_network_get_security(GSupplicantNetwork *network) > const void *g_supplicant_network_get_ssid(GSupplicantNetwork *network, > unsigned int *ssid_len) > { > - if (!network) { > + if (!network || !network->ssid_len) { > *ssid_len = 0; > return NULL; > } Your patch is reverting commit 91913c40054d ("gsupplicant: A network ssid of length 0 is valid, it's an hidden one"). Thanks, Daniel

What about introducing a helper function which says, the SSID we have is a valig, something like this bool ssid_is_valid(const unsigned char *ssid, unsigned int ssid_len) { /* empty ssid, that is "" */ if (ssid_len == 0 && !ssid && ssid[0] == '\0') return true; if (ssid_len > 0) return true; return false; } and use it everyhwere we need. Open coding seems like a bad idea. From one side I like this function but from other side it bothers me

> I suggest go one of two ways. I prefer first if you are okay. > > 1. Simply remove noisy warning > --- a/src/config.c > +++ b/src/config.c > @@ -1219,8 +1219,7 @@ static int try_provision_service(struct > connman_config_service *config, > ssid = connman_network_get_blob(network, "WiFi.SSID", > &ssid_len); > if (!ssid) { > - connman_error("Network SSID not set"); > - return -EINVAL;> + return -ENOENT; There is only one user of this function and it just checks if the return value is != 0: static int find_and_provision_service_from_config(struct connman_service *service, struct connman_config *config) { GHashTableIter iter; gpointer value, key; g_hash_table_iter_init(&iter, config->service_table); while (g_hash_table_iter_next(&iter, &key, &value)) { if (!try_provision_service(value, service)) return 0; } return -ENOENT; } What's the point to change this line then?