Add a new parameter to map a MAVLink MANUAL_CONTROL button to turtle
mode activation. When the manual control data source is a MAVLink
instance, the driver uses the buttons field directly instead of aux
channels. When the source is RC, the existing aux channel behavior
via VOXL_ESC_MODE is preserved. Set to -1 (default) to disable.
So transitions between stopped/not stopped respect the slew rate.
- Remove previous stopped motor handling in publish_actuator_controls
- Replace with handle_stopped_motors, called in Run() before slew
- Introduce ApplyNanToActuators in ControlAllocation to stop
- Refactor ice shedding slightly to fit new structure
Add CI enforcement of conventional commit format for PR titles and
commit messages. Includes three Python scripts under Tools/ci/:
- conventional_commits.py: shared parsing/validation library
- check_pr_title.py: validates PR title format, suggests fixes
- check_commit_messages.py: checks commits for blocking errors
(fixup/squash/WIP leftovers) and advisory warnings (review-response,
formatter-only commits)
The workflow (.github/workflows/commit_checks.yml) posts concise
GitHub PR comments with actionable suggestions and auto-removes them
once issues are resolved.
Also updates CONTRIBUTING.md and docs with the conventional commits
convention.
Signed-off-by: Ramon Roche <mrpollo@gmail.com>
migrateFWFromRoot held the SD root directory open via opendir/readdir
while performing heavy file I/O (getFileInfo, copyFw, unlink) inside the
loop. Between readdir calls the FAT semaphore is released, allowing
other tasks (e.g. logger) to dirty the shared FAT sector cache. When
the next FAT operation needed a different sector, fat_fscacheflush
would write the dirty data followed by an immediate read — triggering
a write-busy to read transition in the SDMMC WRCOMPLETE path that
kills the SD card on STM32H7.
Split into two phases: first collect .bin filenames with the directory
open, then close it before doing any file I/O. This also fixes a
missing closedir on the mkdir error path and avoids modifying directory
entries (via unlink) while iterating them with readdir.
On Linux targets with high-rate external sensor data (>1000Hz), all
sensor calibrations (gyro, accel, mag) can freeze PX4 by starving
other threads of CPU. Normal flight is unaffected — only calibration
triggers the problem.
Two compounding issues in the calibration worker threads:
1. calibrate_cancel_check() creates a new uORB::Subscription on every
call, which triggers getDeviceNodeLocked() — an O(n) linear strcmp
scan through all uORB nodes. In gyro/mag calibration this was called
on every sensor sample, consuming the majority of CPU in strcmp alone.
2. SubscriptionBlocking::updatedBlocking() returns immediately when data
is already available (it only blocks when no data is pending). With
continuous high-rate sensor data, the calibration loops never yield,
spinning at 100% CPU.
These problems are addressed with this patch as follows:
- Throttle calibrate_cancel_check() to once per 200ms in gyro
and mag calibration loops.
- Add 1ms px4_usleep() yield before updatedBlocking()/updateBlocking()
in all calibration loops (gyro, accel, mag, orientation detection).
This caps the effective loop rate at ~1000Hz — still far above what
calibration needs (250-750 samples).
- Force Commander main loop to sleep during calibration so it does not
compete with calibration worker threads for CPU.
Tested under Linux (x64, aarch64) both with RT and non-RT scheduling,
with sensor data arriving at ~3600Hz. Calibration completes normally
and no longer results in a deadlocked process.
* refactor: use parseDefaultArguments
* style: reverted changes to docs (should be auto-generated)
* refactor: use parseDefaultArguments
* style: reverted changes to docs (should be auto-generated)
In the CA_ICE_PERIOD param description, a '>' was at the start of a
line. This is interpredeted by markdown as a quotation type indented
block which applies not only to that line, but to all following ones.
Changing the line breaks to only occur after full stops fixes it.
* SIH: explicitly use local velocity for all aerodynamic calculations
no functional change
* SIH: add param & vars for wind and apparent vel
no functional change
* SIH: replace all relevant vels with apparent vel
Only places where _v_E is remaining:
- ecefToNed to calculate _v_N
- generate_rover_ackermann_dynamics
- equations_of_motion
- parameters_updated, init_variables
and _v_N:
- ecefToNed
- print_status
- publish_ground_truth
- generate_rover_ackermann_dynamics
- equations_of_motion
- parameters_updated, init_variables
which are all not relevant for aerodynamics.
* sih: wind review suggestions
* sih wind: switch direction to global wind source direction convention
* SIH: clean up variable declarations
* SIH: rename variables for consistency
* docs: SIH: document new wind parameters
* Release notes: note for SIH wind settings
---------
Co-authored-by: Matthias Grob <maetugr@gmail.com>
When the PI controller computed zero on-time (e.g. temperature already
above target), the heater was still momentarily turned on every cycle
before being immediately turned off. Skip the on/off toggle entirely
when the computed on-time is zero.
* fix(dataman_client): fail fast when dataman is unavailable
Check client_id before every DatamanClient operation to avoid waiting
for a response timeout when dataman was never running.
- Adds CLIENT_ID_NOT_SET guard to all sync and async methods
- Avoids cross-module linkage between dataman_client lib and dataman module
Supersedes #26128
* better init
