Change the driver event set to be more orthogonal.

This had the expected effect, the N command can now works on SiRFs
again because the subdevice probe is not called after every driver
switch.

One of the Trimbles has minor regression breakage on this change
(two fewer reports emitted just after packet lock) but I'm not
worrying about that as the TSIP driver needs to be upgraded to
do CYCLE_END, anyway.

1 files changed, 24 insertions, 11 deletions

diff --git a/drivers.c b/drivers.c
index 1cea6794..1c2c70cb 100644
--- a/drivers.c
+++ b/drivers.c
@@ -113,7 +113,11 @@ gps_mask_t nmea_parse_input(struct gps_device_t *session)
 
 static void nmea_event_hook(struct gps_device_t *session, event_t event)
 {
-    if (event == event_probe_subtype) {
+    /*
+     * This is where we try to tickle NMEA devices into erevrealing their
+     * inner natures.
+     */
+    if (event == event_configure) {
 	/* change this guard if the probe count goes up */ 
 	if (session->packet.counter <= 8)
 	    gpsd_report(LOG_WARN, "=> Probing device subtype %d\n", session->packet.counter);
@@ -255,11 +259,16 @@ static void garmin_mode_switch(struct gps_device_t *session, int mode)
     }
 }
 
+
 static void garmin_nmea_event_hook(struct gps_device_t *session, event_t event)
 {
-    if (event == event_configure) {
+    if (event == event_driver_switch) {
+	/* forces a reconfigure as the following packets come in */
+	session->packet.counter = 0;
+    } if (event == event_configure) {
 	/*
-	 * Receivers like the Garmin GPS-10 don't handle having having a lot of
+	 * And here's that reconfigure.  It's spplit up like this because
+	 * receivers like the Garmin GPS-10 don't handle having having a lot of
 	 * probes shoved at them very well.
 	 */
 	switch (session->packet.counter) {
@@ -336,7 +345,8 @@ static void ashtech_event_hook(struct gps_device_t *session, event_t event)
 {
     if (event == event_wakeup)
 	(void)nmea_send(session, "$PASHQ,RID");
-    if (event == event_configure && session->packet.counter == 0) {
+    /* FIXME: Do we need to do this on reactivate as well? */
+    if (event == event_identified) {
 	/* turn WAAS on. can't hurt... */
 	(void)nmea_send(session, "$PASHS,WAS,ON");
 	/* reset to known output state */
@@ -392,8 +402,10 @@ static void fv18_event_hook(struct gps_device_t *session, event_t event)
     /*
      * Tell an FV18 to send GSAs so we'll know if 3D is accurate.
      * Suppress GLL and VTG.  Enable ZDA so dates will be accurate for replay.
+     * It's possible we might not need to redo this on event_reactivate,
+     * but doing so is safe and cheap.
      */
-    if (event == event_configure && session->packet.counter == 0)
+    if (event == event_configure && event == event_reactivate)
 	(void)nmea_send(session,
 		    "$PFEC,GPint,GSA01,DTM00,ZDA01,RMC01,GLL00,VTG00,GSV05");
 }
@@ -438,7 +450,7 @@ static void gpsclock_event_hook(struct gps_device_t *session, event_t event)
      * Michael St. Laurent <mikes@hartwellcorp.com> reports that you have to
      * ignore the trailing PPS edge when extracting time from this chip.
      */
-    if (event == event_probe_subtype && session->packet.counter == 0) {
+    if (event == event_identified && event == event_reactivate) {
 	gpsd_report(LOG_INF, "PPS trailing edge will be ignored");
 	session->driver.nmea.ignore_trailing_edge = true;
     }
@@ -485,10 +497,10 @@ const struct gps_type_t gpsclock = {
 static void tripmate_event_hook(struct gps_device_t *session, event_t event)
 {
     /* TripMate requires this response to the ASTRAL it sends at boot time */
-    if (event == event_probe_subtype && session->packet.counter == 0)
+    if (event == event_identified)
 	(void)nmea_send(session, "$IIGPQ,ASTRAL");
     /* stop it sending PRWIZCH */
-    if (event == event_configure && session->packet.counter == 0)
+    if (event == event_identified || event == event_reactivate)
 	(void)nmea_send(session, "$PRWIILOG,ZCH,V,,");
 }
 #endif /* ALLOW_RECONFIGURE */
@@ -528,7 +540,7 @@ static const struct gps_type_t tripmate = {
 
 static void earthmate_event_hook(struct gps_device_t *session, event_t event)
 {
-    if (event == event_probe_subtype && session->packet.counter == 0) {
+    if (event == event_identified) {
 	(void)gpsd_write(session, "EARTHA\r\n", 8);
 	(void)usleep(10000);
 	(void)gpsd_switch_driver(session, "Zodiac Binary");
@@ -657,7 +669,7 @@ static int tnt_packet_sniff(struct gps_device_t *session)
 
 static void tnt_event_hook(struct gps_device_t *session, event_t event)
 {
-    if (event == event_probe_subtype && session->packet_counter == 0) {
+    if (event == event_identified) {
 	// Send codes to start the flow of data
 	//tnt_send(session->gpsdata.gps_fd, "@BA?"); // Query current rate
 	//tnt_send(session->gpsdata.gps_fd, "@BA=8"); // Start HTM packet at 1Hz
@@ -972,7 +984,8 @@ static void mkt3301_event_hook(struct gps_device_t *session, event_t event)
 "$PMTK314,1,1,1,1,1,5,1,1,0,0,0,0,0,0,0,0,0,1,0"
 
 */
-    if(event == event_configure && session->packet.counter == 0) {
+    /* FIXME: Do we need to resend this on reactivation? */
+    if(event == event_identified) {
 	(void)nmea_send(session,"$PMTK320,0"); /* power save off */
 	(void)nmea_send(session,"$PMTK300,1000,0,0,0.0,0.0"); /* Fix interval */
 	(void)nmea_send(session,"$PMTK314,0,1,0,1,1,5,1,1,0,0,0,0,0,0,0,0,0,1,0");