Sorry, was a Copy/Paste mistake. Fixed.

diff --git a/src/host/layer23/include/osmocom/bb/common/sap_interface.h b/src/host/layer23/include/osmocom/bb/common/sap_interface.h
index bf19356..e4e64ce 100644
--- a/src/host/layer23/include/osmocom/bb/common/sap_interface.h
+++ b/src/host/layer23/include/osmocom/bb/common/sap_interface.h
@@ -11,6 +11,7 @@ int osmosap_sapsocket(struct osmocom_ms *ms, const char *path);
 int osmosap_init(struct osmocom_ms *ms);
 
 enum osmosap_state {
+    SAP_SOCKET_ERROR,
     SAP_NOT_CONNECTED,
     SAP_IDLE,
     SAP_CONNECTION_UNDER_NEGOTIATION,
diff --git a/src/host/layer23/src/common/sap_interface.c b/src/host/layer23/src/common/sap_interface.c
index a56f4f2..d5e8d1d 100644
--- a/src/host/layer23/src/common/sap_interface.c
+++ b/src/host/layer23/src/common/sap_interface.c
@@ -515,7 +515,7 @@ int sap_open(struct osmocom_ms *ms, const char *socket_path)
     rc = connect(ms->sap_wq.bfd.fd, (struct sockaddr *) &local, sizeof(local));
     if (rc < 0) {
         fprintf(stderr, "Failed to connect to '%s'\n", local.sun_path);
-        set->sap_socket_path[0] = 0;
+        ms->sap_entity.sap_state = SAP_SOCKET_ERROR;

         close(ms->sap_wq.bfd.fd);
         return rc;
     }
diff --git a/src/host/layer23/src/common/sim.c b/src/host/layer23/src/common/sim.c
index 8e8d7bf..df9fbd2 100644
--- a/src/host/layer23/src/common/sim.c
+++ b/src/host/layer23/src/common/sim.c
@@ -188,7 +188,7 @@ static int sim_apdu_send(struct osmocom_ms *ms, uint8_t *data, uint16_t length)
 
     /* adding SAP client support
      * it makes more sense to do it here then in L1CTL */
-    if(ms->settings.sap_socket_path[0] == 0) {
+    if(ms->sap_entity.sap_state == SAP_SOCKET_ERROR) {
         LOGP(DSIM, LOGL_INFO, "Using built-in SIM reader\n");
         l1ctl_tx_sim_req(ms, data, length);
     } else {
diff --git a/src/host/layer23/src/mobile/vty_interface.c b/src/host/layer23/src/mobile/vty_interface.c
index 5782a17..fd0fd0c 100644
--- a/src/host/layer23/src/mobile/vty_interface.c
+++ b/src/host/layer23/src/mobile/vty_interface.c
@@ -1067,7 +1067,7 @@ DEFUN(cfg_no_gps_enable, cfg_no_gps_enable_cmd, "no gps enable",
 }
 
 #ifdef _HAVE_GPSD
-DEFUN(cfg_gps_host, cfg_gps_host_cmd, "gps host HOST:PORT",
+DEFUN(cfg_gpsd_host, cfg_gpsd_host_cmd, "gpsd host HOST:PORT",
     "GPS receiver\nSelect gpsd host and port\n"
     "IP and port (optional) of the host running gpsd")
 {
@@ -1075,7 +1075,7 @@ DEFUN(cfg_gps_host, cfg_gps_host_cmd, "gps host HOST:PORT",
     if (colon != NULL) {
         memcpy(g.gpsd_host, argv[0], colon - argv[0] - 1);
         g.gpsd_host[colon - argv[0]] = '\0';
-        memcpy(g.gpsd_port, colon, strlen(colon));
+        memcpy(g.gpsd_port, colon + 1, strlen(colon));
         g.gpsd_port[strlen(colon)] = '\0';
     } else {
         snprintf(g.gpsd_host, ARRAY_SIZE(g.gpsd_host), "%s", argv[0]);
@@ -1499,8 +1499,7 @@ static int config_write(struct vty *vty)
     struct osmocom_ms *ms;
 
 #ifdef _HAVE_GPSD
-    vty_out(vty, "gpsd host %s%s", g.gpsd_host, VTY_NEWLINE);
-    vty_out(vty, "gpsd port %s%s", g.gpsd_port, VTY_NEWLINE);
+    vty_out(vty, "gpsd host %s:%s%s", g.gpsd_host, g.gpsd_port, VTY_NEWLINE);
 #endif
     vty_out(vty, "gps device %s%s", g.device, VTY_NEWLINE);
     if (g.baud)
@@ -2822,7 +2821,7 @@ int ms_vty_init(void)
     install_element(ENABLE_NODE, &delete_forbidden_plmn_cmd);
 
 #ifdef _HAVE_GPSD
-    install_element(CONFIG_NODE, &cfg_gps_host_cmd);
+    install_element(CONFIG_NODE, &cfg_gpsd_host_cmd);
 #endif
     install_element(CONFIG_NODE, &cfg_gps_device_cmd);
     install_element(CONFIG_NODE, &cfg_gps_baud_cmd);

2016-01-03 1:19 GMT+06:00 Harald Welte <laforge@gnumonks.org>:

Hi Vadim,

thanks for your patch, but..

On Sun, Jan 03, 2016 at 01:14:23AM +0600, Вадим Яницкий wrote:
> -        set->sap_socket_path[0] = 0;
> +        // set->sap_socket_path[0] = 0;
> +        ms->sap_entity.sap_state == SAP_SOCKET_ERROR;

a) if you remove code, remove it for good, not just comment it out

b) the double-equals will not cause an assignment. I don't think you
   tested this code...

Regards,
        Harald
--
- Harald Welte <laforge@gnumonks.org>           http://laforge.gnumonks.org/
============================================================================
"Privacy in residential applications is a desirable marketing option."
                                                  (ETSI EN 300 175-7 Ch. A6)