Keywords
Boot Misconfigurations; Linux Kernel; Boot Repair; LLM Agents
Abstract
The Linux kernel is an actively developed, highly configurable system software that supports a diverse range of platforms. Kernel fuzzers, like Google’s syzkaller, have been embraced to find bugs, but they require bootable configurations that work in their environment. However, these tools rely on a set of configurations that exclude most code changes. Each option controls which code is included when the kernel compiles. Making modifications requires extensive knowledge, as there are over 15,000 interrelated configuration options. While prior work has used formal methods to guarantee configuration validity, this doesn’t guarantee bootability. In this paper, we introduce KBootRepair, an iterative agent-based approach that repairs Linux kernel boot misconfigurations. It combines LLM boot suggestions with klocalizer, a formal reasoning tool, to generate bootable configurations. We evaluated KBootRepair on 50 random x86_64 configurations and achieved a 90% success rate within 20 attempts.
Thesis Completion Year
2026
Thesis Completion Semester
Spring
Thesis Chair
Gazzillo, Paul
College
College of Engineering and Computer Science
Department
Computer Science
Thesis Discipline
Computer Science
Language
English
Access Status
Open Access
Length of Campus Access
None
Campus Location
Orlando (Main) Campus
STARS Citation
Garfinkel, Ryan P., "KBootRepair: Automated LLM Guided Repair of Linux Kernel Boot Failures" (2026). Honors Undergraduate Theses. 502.
https://stars.library.ucf.edu/hut2024/502
Accessibility Statement
This item was created or digitized prior to April 24, 2027, or is a reproduction of legacy media created before that date. It is preserved in its original, unmodified state specifically for research, reference, or historical recordkeeping. In accordance with the ADA Title II Final Rule, the University Libraries provides accessible versions of archival materials upon request. To request an accommodation for this item, please submit an accessibility request form.