* Bidirectional DShot
Co-authored-by: Julian Oes <julian@oes.ch>
* f4/f1 support, not supported
* fix f1 build target
* sanity check timer_channel value, fix CCxNP ifdef, debug stuff
* removed debug code, added define for H7 HAVE_GTIM_CCXNP
* round robin sampling for less than 4 DMA
* unlimited esc_status logging
* dshot: fix formatting
* dshot: add define for number of DMA channels to use
This allows individual boards to override the number of DShot channels
and hence avoid round robin capture of the RPM feedback.
* ARK: enable 4 DMA channels for DShot on 6X
* dshot: publish when all channels are updated
This slows down the ESC_STATUS publication in the case of round robin
capture. E.g. for 800 Hz output with one DMA channel, the ESC_STATUS is
now published at 200 Hz.
* dshot: avoid duplicate publications for bidir and telem
Instead of publishing both bidirectional dshot updates as well as
telemetry updates, we now combine the data from both streams, and
publish whenever we get RPM updates, as the latter arrives with higher
rate, e.g. 200 Hz with round robin, or faster otherwise.
When combining the data, we take RPM from bidirectional dshot, and the
rest from telemetry.
When we have only one of the two, either telemetry or bidirectional
dshot, we just publish that one.
* boards: add ark fpv and pi6x BOARD_DMA_NUM_DSHOT_CHANNELS
* dshot: turn off debug build
---------
Co-authored-by: Julian Oes <julian@oes.ch>
Co-authored-by: alexklimaj <alex@arkelectron.com>
- As it is always only used for the vehicle command ack message
- It was a duplicate, hence making it error prone for maintainment
- The uORB message comments were updated to make the relationship with
the MAVLink message / enum definitions clear
This fixes an invalid memory access when exiting the module:
exit_and_cleanup destroys the object, but lock_guard is destructed after
and accesses the lock.
- fixes the deadlock in px4io ioctl mixer reset
- px4io Run() locks (CDev semaphore)
- mixer load goes through px4io ioctl MIXERIOCRESET which calls MixingOutput::resetMixerThreadSafe()
- MixingOutput::resetMixerThreadSafe() stores a Command::Type::resetMixer command in an atomic variable, schedules a work queue cycle, then sleep spins until the command is cleared
- the execution of the cycle eventually calls back into PX4IO::updateOutputs(), which tries to lock (and waits forever)
The current practice of adding topics to the default set isn't scalable,
as it affects all setups.
By making sure topics are advertised on init, logger can just discard
topics that don't exist. This does not work for all topics, so topics are
specifically marked as optional. It can be extended to more topics later
on though.
This reduces the list of topics by ~35 on a pixracer configured as quad,
and reduces RAM usage by ~1KB.
This allows modules to do a first-come-first-serve pin/timer reservation
on bootup.
E.g. camera trigger reserves any of the pins, and then PWM/DShot output
will just use the rest of the available pins.
- generate PWM_MAIN 1-14
- generate PWM_AUX 1-8
- generate PWM_EXTRA 1-8
- px4io and pwm_out directly read configuration parameters
- only available and active physical outputs are actually shown for configuration
- overall saves flash despite adding many new parameters
- parameter updates can be quite expensive because they trigger nearly all modules to reload all of their parameters immediately
- limit modules from updating faster than once per second
* Break out dshot.h from dshot.cpp and rename files to reflect the class name.
Move DShot class specific #defines from drv_pwm_output.h to dshot.h.
Replace c-style casts with static_casts, standardize method naming, single line comment styles, and formatting.
Rename update_dshot_out_state() to enable_dshot_outputs().
Fixes
src/drivers/dshot/dshot.cpp:1408:9: error: declaration of
'struct DShotOutput::custom_command(int, char**)::Command' shadows
a previous local [-Werror=shadow]
Jacob Dahl