Skip to content

Add clang-tidy config and CI job from cymon - #9

Merged
Tecnologic merged 1 commit into
mainfrom
copilot/add-clang-tidy-check-to-ci
Jun 26, 2026
Merged

Add clang-tidy config and CI job from cymon#9
Tecnologic merged 1 commit into
mainfrom
copilot/add-clang-tidy-check-to-ci

Conversation

Copilot AI commented May 27, 2026

Copy link
Copy Markdown
Contributor

Ports the clang-tidy configuration from Tecnologic/cymon into this repo and wires it into CI as a dedicated analysis job.

Changes

  • .clang-tidy — enables clang-diagnostic-*, clang-analyzer-*, cppcoreguidelines-*, modernize-*, performance-*, readability-* with suppressions for magic-numbers, reinterpret-cast, pointer-arithmetic, and union-access; enforces CamelCase for classes/functions/constants (k-prefixed), lower_case with _ suffix for members
  • .github/workflows/ci.yml — new clang-tidy job installs clang-tidy, configures the hosted build with -DENABLE_CLANG_TIDY=ON (hooked via the existing cmake/StaticAnalyzers.cmake), and builds the system_test target to run analysis across all compiled sources

@Tecnologic
Tecnologic marked this pull request as ready for review May 27, 2026 05:33
@Tecnologic
Tecnologic merged commit bf6ec2e into main Jun 26, 2026
3 checks passed
Tecnologic added a commit that referenced this pull request Aug 4, 2026
@Tecnologic
Resolve main-to-develop back-merge conflicts (#11)
a198ff1 
* updated modm to current release

* added small not on python version

* ignore py cache files

* fix the build after modm update

* feat: add SVM, dead-time compensation, MTPA, field weakening, back-EMF observer, HFI

- lib/control/Svm.hpp: centered SVPWM with duty clamped to 5-95%
- lib/control/DeadTimeCompensation.hpp: per-phase dead-time correction in α/β
- lib/control/Mtpa.hpp: closed-form MTPA for IPMSM
- lib/control/FieldWeakening.hpp: PI field weakening via i_d control
- lib/observer/MechanicalObserver.hpp: full-order back-EMF Luenberger observer
  + PLL; supports flying start; inject_angle_error() for HFI/hall feedback
- lib/observer/Hfi.hpp: VESC-style 4-step HFI with angle error injection
- lib/CMakeLists.txt: expose control/ and observer/ include paths
- Fix pre-existing bugs: sin_cos<T> template arg in park/inverse_park,
  missing <cmath> in StatorReference/ThreePhase, SinCos abstract class,
  SinCos this.length() → this->length(), wrong sinCos2 test values

* feat: add MotorType enum, ASM flux observer, and ASM flux controller

- lib/system/MotorType.hpp: enum class MotorType { PMSM, ASM }
- lib/observer/AsmFluxObserver.hpp: full-order Luenberger rotor-flux observer
  for induction motors; corrects stator current and rotor flux estimates from
  the current prediction error; injects flux angle into MechanicalObserver PLL
  via inject_angle_error() so both share the same speed/angle estimates
- lib/control/AsmFluxController.hpp: PI flux controller (psi_r* -> i_d*) with
  integrator anti-windup and slip-frequency feedforward omega_slip

* feat: add EESM rotor excitation control and absolute position control with homing

- Add EESM to MotorType enum with full doc comment
- Add ExcitationController: PI loop for rotor I_f or flux ψ_f (CurrentMode / FluxMode)
- Add ExcitationObserver: first-order LPF on rotor current → psi_f estimate
- Add PositionTracker: multi-turn absolute position (±4096+ rev), homing support
- Add PositionController: cascaded P+PI position/speed loops, homing state machine
- Add ControlMode enum (TORQUE / SPEED / POSITION)
- Fix pre-existing SinCos.hpp Wshadow build error
- Fix CMakeLists.txt unconditional target_link_libraries for test builds
- Add 34 new unit tests (47 total, all passing)

* Add comprehensive README with CI badge, Cyphal register table, NVM and architecture docs

* ci: add BatteryCase Controller firmware build job

* fix: address PR review feedback on homing identity build setup

* Add UDRAL servo control mode selection helper

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

* fix(SinCos): update class doc to reflect plain-value-type API

* fix: make SinCos::normToOne non-throwing for -fno-exceptions firmware

* feat: add smessmer/gitversion submodule and Cyphal heartbeat with version VSSC

- Add lib/gitversion as a git submodule (smessmer/gitversion)
- Replace manual git-describe version detection in CMakeLists.txt with
  TARGET_GIT_VERSION_INIT from gitversion's cmake.cmake
- Add src/cyphal/heartbeat.hpp: header-only uavcan.node.Heartbeat.1.0
  serializer that reads SW_VERSION_MAJOR/MINOR from gitversion's
  VERSION_STRING at compile time and encodes them in the 19-bit VSSC field

* Add current control ISR, SlowUpdate, double-buffer, and register map updates (#6)

* Add Kalman mechanical observer, PMSM flux observer, and hardware limits

* Add current control ISR, SlowUpdate, double-buffer, and register map updates

* chore: remove duplicate NvmSettings heading block

* fix: flip double buffer before slow-update snapshot

* fix: use real CCR values in ISR boundary guard

* Fix ISR settings mapping to use single motor parameter set

* fix: express CurrentController voltage limit in volts via v_dc parameter

Add `v_dc` parameter to `CurrentController::update()` so the voltage limit
is computed as `v_max * v_dc` [V] instead of comparing against the raw
fractional `v_max` value. This fixes the bug where `v_max=0.9` was being
compared against raw rotor-frame voltage demands in volts, causing immediate
saturation on any real bus.

- `v_max` remains a dimensionless fraction (0, 1] (e.g. 0.9 for SVM headroom)
- `update()` now takes `v_dc` [V] and computes `v_limit = v_max * v_dc`
- Call site in `CurrentControlIsr.cpp` passes the ADC-read `v_dc` value

* fix: resolve CI duplicate runs and build errors

- ci.yml: change push trigger from branches ["**"] to ["main"] to
  eliminate the duplicate push+pull_request run on every PR push
- ci.yml: add submodules: true to batterycase checkout so the
  lib/gitversion submodule is initialized before configure
- ci.yml: install python-is-python3 in batterycase job so cmake.cmake's
  `python -m gitversionbuilder` call resolves correctly on Ubuntu
- CMakeLists.txt: move INCLUDE(lib/gitversion/cmake.cmake) and
  TARGET_GIT_VERSION_INIT inside the if(HARDWARE_DIRECTORY) block so
  the hosted/test build never needs the submodule

* fix: replace freeze anti-windup with back-calculation in CurrentController

The previous integrator freeze strategy halted integration whenever the
total output (PI + feedforward) was saturated, including cases where the
feedforward terms alone pushed the output over the limit.  This caused
the integrator to freeze even with correct PI behaviour, producing
steady-state current error at high speed.

Replace with back-calculation anti-windup:

  integrator += (ki * e + kb * (u_limited - u_raw)) * dt

- When unsaturated: `u_limited == u_raw`, correction term is zero,
  integrator advances normally.
- When saturated: `(u_limited - u_raw) < 0`, so the integrator is
  wound back proportionally to the saturation error.
- The total output (PI + feedforward) is still always hard-clamped to
  the voltage circle regardless of what drives saturation.

Add `kb_d` / `kb_q` back-calculation gain parameters (default 100.0,
equal to ki/kp with default gains, giving tracking time-constant equal
to the integral time Ti = kp/ki).  Update class docstring and usage
example accordingly.

* fix: resolve ARM target build errors in RotorReference, StatorReference, ThreePhase

- Fix -Werror=shadow: rename constructor parameters (d->d_in, q->q_in,
  alpha->alpha_in, beta->beta_in, a->a_in, b->b_in, c->c_in) to avoid
  shadowing member variables
- Replace undefined sin_cos<T> with SinCos<T> in park() and inverse_park()
  signatures; add #include "SinCos.hpp" to RotorReference.hpp and
  StatorReference.hpp; remove stale sin_cos forward declarations
- Replace C++20 designated initializers (non-aggregate type errors) with
  positional constructor calls in park(), inverse_park(), and clark()

* fix: use arm-none-eabi-gcc as ASM compiler to support C-preprocessed assembly

* fix: install pyelftools for system Python in batterycase CI step

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>

* Add Cyphal ExecuteCommand interface for motor identification and telemetry (#7)

* feat: add cyphal execute command codes for all available measurements

* feat: add motor identification measurement commands and phase balance NVM fields

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>

* Hardware startup aid FSM for new UNIMOC boards (#8)

* feat: add hardware startup aid FSM (HwStartup)

* docs: add Hardware Bring-Up Guide to README

* fix: align NUM_STARTUP_STEPS comment with actual value

* fix: address all startup FSM review comments

- Update is_active() to return false for DONE and FAULT states so
  SlowUpdate resumes after the startup sequence completes or aborts
- Fix request_abort() to no-op when FSM is IDLE or DONE
- Gate ADC_OFFSET_CAL sampling with step_done_ to prevent log spam
  and accumulator drift after N_CAL samples are reached
- Gate ADC_NOISE_FLOOR sampling with step_done_ for the same reason
- Add step_done_ guard at the start of run_gate_driver_enable_check()
  to prevent gate_phase_disabled_ being reset and baseline re-collection
  on subsequent calls after the step is complete
- Fix DC_LINK_VOLTAGE_CHECK to accumulate raw_vdc directly (not via
  collect_sample() which accumulates raw_ia/raw_ib) and average N_CAL
  samples for measured_vdc instead of using a single last sample
- Fix PHASE_ADC_ALIGNMENT signed→uint32_t cast by keeping the
  intermediate value as int32_t and clamping to 0 before casting,
  preventing underflow for negative tick offsets
- Initialize base_trigger_offset_ from cc_.state.adc_trigger_offset
  at PHASE_ADC_ALIGNMENT sweep start so the sweep step-size is correct
  for any timer clock, not just 168 MHz
- Fix STARTUP_RESULTS Cyphal register type comment: uint8[<=14]
  (14 FSM states, not 13)

* fix: resolve double-promotion build errors in variance computations

* fix: remove hard-coded 168u default for base_trigger_offset_

* fix: DC_LINK_VOLTAGE_CHECK — use N_CAL average and compute gain_vdc when ext_vdc_V_ arrives late

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>

* Add clang-tidy config and CI job from cymon (#9)

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>

* fixed builds

* added host limits

* Fix hosted merge configuration

Co-authored-by: Tecnologic <1442404+Tecnologic@users.noreply.github.com>

---------

Co-authored-by: Alex Evers <tecnologic86@gmail.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Tecnologic <1442404+Tecnologic@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

1 more reviewer
@Tecnologic Tecnologic Tecnologic approved these changes
Reviewers whose approvals may not affect merge requirements

Assignees

@Tecnologic Tecnologic

Copilot code review Copilot

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@Tecnologic