Advancing File System Model Checking: Coverage, Framework, and Scalability
Bibliographic record
Abstract
File systems serve as the foundation for data storage and access, making their reliability crucial to maintaining system correctness and data integrity. However, building robust file systems remains a significant challenge. Despite numerous testing and verification techniques, file system bugs continue to emerge. To detect file system bugs and improve reliability, we tackle three key aspects: new coverage metrics for testing, a novel model checking approach, and enhanced scalability for file system verification. We begin by introducing input and output coverage (IOCov) as metrics for evaluating and improving file system testing, along with IOCov to compute them. We integrated IOCov into existing file system testing workflows, achieving broader input coverage and improving the detection of crash consistency bugs. Next, we present Metis, a file system model checking framework designed to explore diverse inputs under different file system states. Using a reference file system (RefFS), Metis compares the behaviors of two file systems and reports any discrepancies as potential bugs. Metis leverages Swarm Verification (SV) to scale state exploration by distributing parallel verification tasks (VTs) across multiple cores and machines. Finally, we describe Containerized Swarm Verification (CoSV), in which each VT runs in a container and is managed by an orchestrator. CoSV enhances the scalability of SV by packaging each VT as a self-contained unit, allowing for easy adaptation to dynamic resource availability. In addition, CoSV ensures fault isolation across VTs to prevent faults in one task from interfering with the execution of others. Our thesis is that effective file system testing requires coverage metrics to guide evaluation, new techniques for thorough checking, and scalable parallelism to explore large state spaces. Overall, input/output coverage helps developers evaluate file system testing, while model checking systematically verifies states, and containerized swarm verification scales this process through efficient, fault-isolated parallelism.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.014 | 0.084 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.004 |
| Bibliometrics | 0.005 | 0.002 |
| Science and technology studies | 0.001 | 0.005 |
| Scholarly communication | 0.005 | 0.011 |
| Open science | 0.004 | 0.007 |
| Research integrity | 0.002 | 0.005 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".