telemetry: guard CRSF ESC RPM/temp against stale data

[sensei-hacker]

Summary

ESC RPM and temperature forwarded over CRSF showed random/garbage values when the FC was disarmed and when individual ESCs lost contact in multi-motor setups. The previous null-guard on getEscTelemetry() was dead code — the function always returns a valid pointer into a static array — so stale last-seen values were unconditionally forwarded regardless of data age.

Changes

• crsfRpm(): check escState->dataAge < ESC_DATA_INVALID before forwarding RPM; send 0 for stale/uninitialized slots
• crsfTemperature(): same check before forwarding temperature; send TEMPERATURE_INVALID_VALUE for stale/uninitialized slots

ESC_DATA_INVALID (255) is already the initialization value set by escSensorInitialize() and is the value used consistently throughout esc_sensor.c for this purpose.

Testing

• SITL build confirmed clean (no warnings)
• Code analysis confirmed all four bug scenarios from the issue map directly to the missing dataAge guard:
  • Single ESC disarmed → stale last-seen RPM forwarded (now sends 0)
  • Multi-ESC in flight → divergent per-motor dataAge causes mixed fresh/stale values (now stale slots send 0)
  • RPM stays elevated after disarm → dataAge saturates at 255 but was never checked (now checked)
  • Extra CRSF data points on octocopter → all motor slots emitted regardless of age (now filtered)
• Hardware testing with physical ESCs not performed; requesting confirmation from reporter

Related Issues

Fixes #11517

[qodo-code-review]

Review Summary by Qodo

Guard CRSF ESC telemetry against stale sensor data

🐞 Bug fix

Walkthroughs

Description

• Guard CRSF ESC RPM/temperature against stale data
• Check dataAge < ESC_DATA_INVALID before forwarding values
• Send 0 for stale RPM; send TEMPERATURE_INVALID_VALUE for stale temperature
• Fixes random/garbage values on disarm and multi-motor contact loss

Diagram

flowchart LR
  A["ESC Sensor Data"] --> B["dataAge Check"]
  B -->|Fresh| C["Forward RPM/Temp"]
  B -->|Stale| D["Send Default Values"]
  C --> E["CRSF Frame"]
  D --> E

Loading

File Changes

1. src/main/telemetry/crsf.c 🐞 Bug fix +2/-2

Add dataAge validation to ESC telemetry functions

• Modified crsfRpm() to check escState->dataAge < ESC_DATA_INVALID before forwarding RPM values
• Modified crsfTemperature() to check escState->dataAge < ESC_DATA_INVALID before forwarding
 temperature values
• Sends 0 for stale RPM slots instead of forwarding last-seen values
• Sends TEMPERATURE_INVALID_VALUE for stale temperature slots with explicit type cast

src/main/telemetry/crsf.c

---

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (1) 📎 Requirement gaps (0)

1. crsfRpm() uses 0 sentinel 📘 Rule violation ☼ Reliability

Description

When ESC telemetry is stale (dataAge >= ESC_DATA_INVALID), the new code serializes RPM as 0,
which can be interpreted as a valid stopped-motor RPM rather than “no data”. This violates the
requirement to use explicit invalid/no-data sentinels for outputs to avoid ambiguous interpretation.

Code

src/main/telemetry/crsf.c[353]

+            crsfSerialize24(dst, escState->dataAge < ESC_DATA_INVALID ? escState->rpm : 0);

Evidence

PR Compliance ID 5 requires using explicit invalid/no-data sentinels instead of potentially-valid
defaults like zero; the changed line explicitly emits 0 RPM when ESC data is stale.

src/main/telemetry/crsf.c[353-353]
Best Practice: Learned patterns

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`crsfRpm()` currently emits `0` RPM when `escState->dataAge >= ESC_DATA_INVALID`. `0` can be a valid value (motor stopped), so it is an ambiguous “no data” indicator.
## Issue Context
Compliance requires explicit invalid/no-data sentinels for outputs where possible, instead of potentially-valid defaults like `0`.
## Fix Focus Areas
- src/main/telemetry/crsf.c[351-354]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Stale RPM still forwarded ☑ 🐞 Bug ≡ Correctness

Description

crsfRpm() considers any ESC telemetry with dataAge < ESC_DATA_INVALID (255) valid, so after an ESC
stops responding CRSF can continue forwarding last-seen RPM until dataAge saturates at 255
(potentially tens of seconds+ with multi-motor polling). This is inconsistent with other
telemetry/UI paths that treat ESC data stale after ESC_DATA_MAX_AGE (10), and can keep showing
incorrect RPM long after disarm/ESC dropout.

Code

src/main/telemetry/crsf.c[353]

+            crsfSerialize24(dst, escState->dataAge < ESC_DATA_INVALID ? escState->rpm : 0);

Evidence

ESC telemetry ages per motor on each request timeout/failure, with a 50ms timeout and round-robin
selection across motors, so reaching ESC_DATA_INVALID (255) can take 255 * 50ms * motorCount. Other
parts of the codebase gate ESC-derived values at ESC_DATA_MAX_AGE (10), indicating that is the
intended freshness threshold for consumers.

src/main/telemetry/crsf.c[351-355]
src/main/sensors/esc_sensor.h[42-45]
src/main/sensors/esc_sensor.c[54-57]
src/main/sensors/esc_sensor.c[97-113]
src/main/sensors/esc_sensor.c[252-289]
src/main/telemetry/sbus2.c[79-88]
src/main/io/osd.c[3920-3929]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`crsfRpm()` currently forwards RPM for any ESC slot with `dataAge < ESC_DATA_INVALID (255)`. That suppresses never-initialized slots, but still forwards very stale last-seen values until `dataAge` saturates at 255, which can take a long time with round-robin polling.
### Issue Context
Other telemetry/UI paths treat ESC telemetry as valid only when `dataAge <= ESC_DATA_MAX_AGE`.
### Fix Focus Areas
- src/main/telemetry/crsf.c[335-358]
### Suggested change
Gate RPM with `escState->dataAge <= ESC_DATA_MAX_AGE` (or an equivalent freshness threshold) instead of `< ESC_DATA_INVALID`.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[github-actions]

Test firmware build ready — commit 76a6748

Download firmware for PR #11536

234 targets built. Find your board's .hex file by name on that page (e.g. MATEKF405SE.hex). Files are individually downloadable — no GitHub login required.

Development build for testing only. Use Full Chip Erase when flashing.

[Jetrell]

I gave it a quick flight test. It seems to occur less. However it still happens on the occasion.

RPM.Telemetry.glitch.mp4

[sensei-hacker]

I think this is part of the problem, perhaps, but there is also a different problem that still needs to be investigated. Also, I will need to check if there is better sentinel for invalid value rather than 0.