this is a partial backport of #24132 which fixes RTK injection when
the 1st GPS module is a DroneCAN RTK rover. Without this change RTCM
injection for RTK fix on the base will only work if it happens to come
up as the first module
this fixes an issue where a lua library function triggers an exception
after it does a math operation which changes the floating point
registers on M7 MCUs (such as STM32H7). An example is math.random()
which calls math_random(), which pre-calculates a double value before
checking if the arguments to the call are valid. When it then checks
and finds invalid values the exception longjmp does not restore the
floating point registers.
filter status was initially set to zero then updated. This interacts
with the IMU filtering code which checks filter status from a
different thread to determine active_EKF_type(). When the race
condition is hit then the IMU we are running notch filters on changes
for a single sample, causing a notch filter glitch
if you have a mission item for engine control with delayed start at
height and the engine is already running them it would put the ICE
subsystem into a state where it would no longer start the engine
It was actually 2 bugs:
- an engine control to do a height delayed start should be ignored if
the engine is already running. This prevents an engine control to
start the engine from stopping the engine
- a start_chan high should always try to start the engine
immediately, even if in the wait state
we need to tell the IO firmware that a byte was consumed when we first
detect a protocol as otherwise the next bad byte on DSM will lock us
on the DSM port
this fixes an issue where when you lose R/C input on IOMCU that you
may not regain it when R/C comes back.
The issue stems from us still processing the DSM uart when we are
using the SD3 "SBUS" uart for RC input, and still doing the switch of
the SD3 config every 2 seconds.
When we are not searching for a new protocol we should not be changing
UART config
set_message_interval takes a port number, not a channel number (it was originally written as an internal function, after all). Before this patch we were double-mapping from one to the other. That works so long as the port you are mapping also corresponds to the chanel number - which it will, for example, if you are using serial2 with both serial0 and serial1 also set to mavlink. If you set serial5_protocol to 2 an attempt to use it for controlling a gremsy it will *not* work because we map into backwards twice.
this fixes a pre-arm failure "GPS 1 failing configuration checks" on
non-M10 GPS modules, including AP_Periph
it also adds the ublox unconfigured msgs to the DroneCAN GNSS.Status
errors field for easier diagnosis of this type of issue in the future
if the IOMCU resets twice in quick succession then the code that
restores the safety state while flying can fail, leading to the
aircraft trying to continue flying with safety on
This results from two issues:
- a race in handling the last_safety_off variable
- the fact that plane sets the soft_armed state based on safety state
Failing due to being out of time meant we wouldn't incremement the counter, even though we'd emitted the item.
it is important we try to send something, so move this check to be after we increment whichever counter we are using.
RemoteID modules are required to set EMERGENCY status on uncontrolled
descent or crash. This fixes our implementation to do that, either via
existing vehicle crash checking code or via a parachute release
this fixes a regression from 4.2 to 4.3.
previously we automatically set the diagnoals to 1,1,1 if they were
0,0,0. We don't do that any more. I was helping a user who had copied
an old config with 0,0,0 for diagonals and did not understand two
things:
- the compass did not work in 4.3
- large vehicle mag cal didn't work
