It turns out ICM20602 and ICM20948 are not actually started/probed on
the px4/v6x platform. We should be able to remove them without
repercussions and save some flash.
- Initialize W25N01GV NAND flash on SPI1 with littlefs.
- Enable BOARD_SMALL_FLASH_LOGGING for small flash handling.
- Increase logger buffer size for flash write performance.
This cherry-picks upstream commits adding NAND flash support for the
Winbond W25N specifically W25N01GV chip.
NAND flash is used together with littlefs, hence I updated that to 2.5.1
to match upstream NuttX.
For flash constrained builds, the component metadata files are
downloaded directly from s3. For the main branch, these files are
currently still uploaded to master. Therefore, we also need to reference
master and not main.
This fixes the actuators tab in QQC for KakuteH743-Wing which is one of
the boards that did not exist in source when we briefly uploaded to main
before reverting back to master, presumably for QGC compatibility.
* Tools: rewrite uploader script
This adds a script called px4_uploader.py which is a complete rewrite
using Claude Code of px_uploader.py.
The main improvements over the previous px_uploader.py script are:
- Separate smaller classes instead of one big uploader class.
- Easier debugging with --verbose or --debug flags.
- Nicer progress animation.
- No more hard to debug Exception swallowing over multiple levels.
- Auto-detection and looping of devices, removing the functionality from
cmake.
- Auto-detection of PX4 devices by USB VID/PID.
- Add noninteractive mode
- Add JSON output mode
Commit 5fe82aa added mutex protection in ~Mavlink() to fix a race
condition when start_helper() deleted an instance without holding the
mutex. However, this caused a deadlock because stop_command() and
destroy_all_instances() already hold mavlink_module_mutex when calling
delete, and the mutex is non-recursive.
Fix by moving instance cleanup to the callers:
- All callers now hold the mutex and remove the instance from
mavlink_module_instances BEFORE calling delete
- The destructor no longer touches mavlink_module_instances
- Event handoff remains in destructor (works because `this` is already
removed from the list when destructor runs)
This hopefully fixes the original race condition while avoiding the
deadlock that caused USB mavlink to hang on reconnect.
I don't think we should change the logging profile based on the type of
airframe configured. Instead, this is an option you set based on the
phase of development/testing you're in.
This came up because the KakuteH7v2 which is 4050 by default would log
excessively which is not a good idea with only 128 MB flash storage.
This prevents PX4 from sending out the GPS_GLOBAL_ORIGIN message
immediately when a SET_GPS_GLOBAL_ORIGIN message arrives.
Instead, we apply the new origin in the EKF, and only then send out
the new origin, which is much more intuitive and doesn't confuse a user
of the API.
* mavlink: fix potential use-after-free
If a mavlink instance is force stopped, the main thread might be out of
scope and the receiver thread would be doing a use-after-free.
Instead the receiver thread needs to check its own _should_exit flag.
* mavlink: protect shared data by mutex in dtor
I'm not sure if this potentially fixes any of the segfaults we have seen
on stopping mavlink instances but it potentially could matter if the
mavlink_receiver thread is killed after a timeout and tries to send any
messages as a zombie.
I don't think it makes sense to ignore required streams that easily.
If we do use some streams that are only in the MAVLink development.xml
dialect, then we will have to properly and explicitly ifdef them
everywhere that we use them. Otherwise, this basically means that we
will just swallow this warning on most (non mavlink-dev) platforms which
can mask issues.
We should not build and release with the MAVLink development dialect
because messages in development.xml can change at any time and break
things.
Instead we should prototype and test things using specific mavlink-dev
targets.
According to the mavlink spec we should be publishing the home attitude
as a quaternion rather than just the yaw/heading.
Additionally, this allows setting the landing roll and pitch angle using
DO_SET_HOME (this yet needs to go into the MAVLink spec though).
This time including the message versioning and translation.
According to the mavlink spec we should be publishing the home attitude
as a quaternion rather than just the yaw/heading.
Additionally, this allows setting the landing roll and pitch angle using
DO_SET_HOME (this yet needs to go into the MAVLink spec though).
The define is used in dataman, not navigator, so this config variable
needs to be moved to the dataman module, otherwise it breaks the build
if navigator is not included in the build.
When a mission finishes with an RTL command, there's a race condition
between:
1. RTL command setting user_mode_intention RTL
2. Mission completion logic forcing LOITER
The auto-loiter transition was checking current nav_state (which is
still AUTO_MISSION) instead of any pending user_mode_intention,
causing it to override the RTL request.
Fix: Only auto-transition to loiter if no mode change is already
pending.
I don't think we have a way to explicitly detect if BAT1 or BAT2
"bricks" are correct, so we have to assume they are, and rely on the
voltage/current shown.
Additionally, we can now power cycle sensor power.
PX4 needs a bit of time to process an uploaded mission before it is
ready to accept the mission mode.
Therefore, we need to wait a bit.
Alternatively, we could wait on the mission progress arriving properly,
but this sleep is simple enough for now.
It turns out that we set the ADC range incorrectly leading to the
measured current being capped at a certain level as the ADC on the
sensor saturates.
Instead, we need to set the range according to the formula given in the
interface datasheet.
* hrt: Fix PPM input on channel 2
The CCMR1_PPM define for PPM input on channel 2 was incorrectly set to 2,
which was setting bits for channel 1 instead of channel 2. This prevented
PPM input from functioning properly on channel 2.
Changed CCMR1_PPM for channel 2 from 2 to (1 << 8), which correctly
configures the CC2S bits for input capture mode on TI2.
This fixes an issue noted in the existing code comment:
"FIXME! There is an interaction in the CCMR registers that prevents
using Chan 1 as the timer and chan 2 as the PPM"
Tested on STM32H743 with PPM input on PC7 (TIM8_CH2).
* rc_input: enable sharing serial and PPM pin
By setting RC_SERIAL_PORT_SHARED_PPM_PIN_GPIO_RX it is now possible to
use the same pin on the STM32 for PPM input as well as serial input.
* boards: Add support for Holybro KakuteH7-Wing
The older defines without the L for the APB1 Low domain errored for some
of the timers, such as 6 and 7.
I checked and it turns out the defines with and without L are identical.
We need to open the device later in the work queue and not in the
constructor during task_spawn.
There is already a lazy open in place, so just removing this fixes the
problem for me.
This adds RTL_TYPE 4 which means continue the mission or reverse back to
the takeoff location, whichever is closer in terms of mission items
in-between.
This would be nicer to have on a distance rather than mission item count
basis but that would require access to the dataman and make it more
complex.
This configures the RTC clock to use the HSE instead of the not existing
LSE clock which prevents boot waiting for the not existing LSE crystal
for a few seconds on startup.
The current approach was wrong because the gimbal protocol now
handles the case properly where the autopilot is in charge of a
non-MAVLink gimbal.
This means that we don't need to send message "as if we were a gimbal
device" and instead set thet gimbal_device_id to 1 (up to 6) to indicate
we are in charge or a non-MAVLink gimbal.
When we use a gimbal connected via "RC", so PWM via the Aux channels, we
need to set the gimbal_device_id to 1 as per the protocol.
This was missing for GIMBAL_DEVICE_ATTITUDE_STATUS, so I added that, and
fixed the name of that variable while at it.
* listener: only clear screen with multiple messages
* listener: fully clear before printing again
Otherwise, we end up seeing artifacts from the previous print.
This updates the GPS submodule which includes NMEA/Unicore fixes:
- Add correct return value for sat infos
- Only publish on position updates
- Request required topics at 5 Hz for Unicore
I suggest to clear the screen when using the listener with -n.
This way the updates don't just scroll by which isn't legible but
instead stay in place and update, similar to the top command.
It looks like the BMP280 was heavily filtered, leading to a step
response of around 3 seconds. Comparing it to baro logs of BMP388 and
MS5611, it looks very slow.
I therefore suggest to disable the on device IIR filter which makes the
log data look more like the other baros.
This fixes an issue where we dropped acks that we should have sent out
just because the queue was clogged with acks that have nothing to do
with us and just happen to be sent to us. We should just ignore them and
not publish them to uORB.
We should not send out acks for commands that are not targetted at us.
This can confuse other MAVLink components and MAVLink commands just
don't scale that way, and it unnecessarily fills up our ack queue.
This was likely introduced to debug when MAVLink components were
implemented incorrectly sending commands to a wrong target.
The most we should do in this case is to print/log an info.
A gimbal user needs to be able to receive gimbal device messages such as
GIMBAL_DEVICE_INFORMATION. Therefore, we need to forward this MAVLink
instance.
* boards: change board ID of KakuteH7 mini
Holybro would like to consolidate the board ID to what is used in
ArduPilot.
* boards: update KakuteH7mini bootloader
This is due to the board ID change.
When the px_uploader.py is used scripted, so without an interactive
shell, the progress bar doesn't update. Therefore, I suggest to print a
new line instead of a carriage return for the non-interactive shell
case.
Julian Oes