When Predicate-Negation Controls Are Vacuous: A Reproducible PX4 Autopilot Case Study of an Added Command-Rejection Guard
Abstract In modular flight and robotic software, ground-control links often route dozens of telecommand types through one shared ingestion routine before dispatching each command to its subsystem. When a maintenance patch inserts a command-agnostic early-return validation check at that entry point, testing only inputs meant to trigger or negate the check can miss an interface-level integration failure: valid status and configuration requests get rejected alongside malformed navigation commands. Under a single-insertion handler model, any test built by negating the added guard condition executes the exact same instructions before and after the patch and is therefore vacuous for checking guard scope. GuardScope addresses this gap through a conservative static walk of the production C + + diff. Given an explicit field-to-payload map, it isolates dispatch branches that never read the guarded payload as exterior invariance controls and abstains whenever the full command structure is passed to external helpers. Evaluated on a historical patch in unmodified PX4 autopilot software-in-the-loop simulation, GuardScope identifies six exterior controls. All six keep acknowledgement code 0 under the global integer frame, whereas all 12 requests across two non-global frames shift from 0 to 9 across repeat and clean-clone runs. A 169-command differential sweep and eight-year PX4 and ArduPilot mining bound both baseline recall and idiom prevalence.
Authors
- Manisha Khulbe (ORCID: https://orcid.org/0000-0002-6114-8303)
- Satish Kumar Mishra
Institutions
- Netaji Subhas University of Technology (IN)
Publication Details
- Journal
- Research Square
- Published
- 2026-11-13
- DOI
- https://doi.org/10.21203/rs.3.rs-11703294/v1
- Primary Topic
- Software Testing and Debugging Techniques
- Type
- preprint