AuroraRAS opened a new pull request, #18527:
URL: https://github.com/apache/nuttx/pull/18527
## Summary
This PR introduces a comprehensive refactoring and stabilization of the
CC1101 wireless character driver. It aligns the driver's software logic with
the physical hardware state machine defined in the Texas Instruments
specifications (DN013/SWRA151A), ensuring safe cross-band operation and robust
IOCTL memory safety.
**Key Modifications:**
* **Modulation Support:** Extended `ioctl.h` to include `WLIOC_MSK` and
`WLIOC_4FSK`. Implemented strict hardware constraint checks in `WLIOC_SETMODU`
(e.g., rejecting MSK if baud rate <= 26 kBaud, enforcing Manchester coding
exclusions).
* **Dynamic PA Ramping:** Deprecated static `rfsettings->PA` logic.
Introduced a frequency-aware `g_cc1101_pa_calibration` matrix to dynamically
reconstruct the `PATABLE` via Burst Access. This natively resolves ASK/OOK
pulse truncation and minimizes FSK adjacent-channel splatter.
* **Standardized TX Power:** `WLIOC_SETTXPOWER` and `WLIOC_GETTXPOWER` now
strictly operate on `dBm` units (as defined by `ioctl.h`) using a quantized
nearest-match algorithm, abandoning the arbitrary 0-8 indexing.
* **Math & Overflow Safety:** Implemented hard clamping for mantissa
calculations in `WLIOC_FSK_SETBITRATE` (256-511) and `WLIOC_FSK_SETFDEV` (8-15)
to prevent silent register wrap-around under extreme input parameters.
* **Kernel Memory Safety:** Replaced uncompiled `DEBUGASSERT` checks with
strict `if (ptr == NULL) return -EFAULT;` validations across all IOCTL branches
to prevent user-space-triggered segmentation faults.
* **Low-Power Reliability:** Added a mandatory `PATABLE` reconstruction call
inside `cc1101_powerup()` to fix the known hardware issue where indices 1-7 are
lost during SLEEP state.
## Impact
- **Is new feature added?** Yes. MSK and 4-FSK modulation.
- **Is existing feature changed?** Yes. `WLIOC_SETTXPOWER` now accepts dBm
instead of raw index values.
- **Impact on Build:** None.
- **Impact on Hardware:** Significantly improves RF spectral purity and
communication stability across diverse frequency bands (315/433/868/915 MHz).
## Testing
I confirm that changes are verified locally.
* **Build Host:** Fedora 43
* **Target Architecture:** Xtensa / ESP32 (Hardware setup featuring dual
CC1101 transceivers and an nRF24L01, EvilCrowRF-V2 layout).
* **Testing Methods & Results:**
* **IOCTL Parameter & Clamping Verification:** Successfully validated the
full suite of standard IOCTLs (`WLIOC_SETRADIOFREQ`, `WLIOC_SETTXPOWER`,
`WLIOC_SETMODU`, `WLIOC_FSK_SETBITRATE`, `WLIOC_FSK_SETFDEV`) on both
`/dev/radio0` and `/dev/radio1`. The mathematical saturation and clamping logic
proved accurate against the hardware limits (e.g., configuring a 250,000 bps
target bitrate cleanly quantized to an actual 249,938 bps; 47,000 Hz target
FDEV quantized to 44,433 Hz).
* **Cross-Band RF Validation:** Executed bidirectional TX/RX payload
verification and 100-packet throughput stress tests across all four primary ISM
bands to guarantee the new dynamic PATABLE generator adapts correctly to
`WLIOC_SETRADIOFREQ` calls:
* **315 MHz & 433 MHz:** Achieved near-perfect reliability with 98%–100%
packet delivery rates and throughputs peaking at ~3.95 KB/s (31.5 kbps).
* **868 MHz & 915 MHz:** Payload verification passed successfully in
both directions. Stress tests completed with delivery rates ranging from 54% to
91%, handling the dynamic frequency shifts without state machine lockups.
* **Driver Coexistence / Regression:**
Confirmed that the SPI bus and IOCTL interface behavior for other
wireless drivers on the same hardware (`/dev/nrf24l01`) remain completely
unaffected by the CC1101 IOCTL architecture changes.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
