Regular periodic GPS Watchdog timeout

[dojomouse]

Firmware was current on Oct 5th.

GPS is detected on startup, position is accurate.

I get regular mavlink messages (every 28 seconds) saying "UBX Module stopped responding", followed by immediate reconnection "UBX Module found, status ok".

Looking at the related code in ubx_watchdog_loop, I wonder if there's a bug - ubx_fault_count is only reset when ubx_healthy changes from False to True. Once the system is healthy, any errors (instances of !all_ok) are integrated up until eventually they equal the ubx_health_fail_counter_limit, at which point the "stopped responding" error is given. As coded this is inevitable unless all_okay is never false.

If this is the problem, a solution would be to move:

else {
/* gps healthy */
ubx_success_count++;
ubx_fail_count = 0; // TO HERE
if (!ubx_healthy && ubx_success_count == UBX_HEALTH_SUCCESS_COUNTER_LIMIT) {
//printf("[gps] ublox UBX module status ok (baud=%d)\r\n", current_gps_speed);
// global_data_send_subsystem_info(&ubx_present_enabled_healthy);
mavlink_log_info(mavlink_fd, "[gps] UBX module found, status ok\n");
ubx_healthy = true;
ubx_fail_count = 0; // FROM HERE
once_ok = true;
}
}

[LorenzMeier]

This has been fixed in the current master - please pull and verify. Thanks!

[dojomouse]

Finally got around to checking it - this issue is fixed, but a new bug maybe introduced in the process; the last mavlink message I get is the "trying to connect to ubx module" one... never any success notification. GPS works fine with no dropouts, but status is messed up. Should I file a new bug report?

[px4dev]

The answer to "should I file a bug report" is almost always "yes" - if you file and it's a dupe or we know about it, no big deal. But if you don't, we may never learn of your problem.

Please always file a bug if you've run into something you think is wrong. Bugs are good; lots of bug reports means lots of information we can use to improve the software.