论文提出 CoEvoBench:LLM 能否为 Linux 内核接口变更适配驱动
Keeping Up with the Kernel: Can LLMs Adapt Linux Drivers to Interface Changes?
做内核开发或系统级编程的会感兴趣:316 个真实内核提交改写任务测下来,强模型也只有在编译报错场景能对上一小半,接口语义变更基本抓瞎。数据翔实,能看清 LLM 在系统代码维护上的真实水平。
arXiv 论文 2611.24269 研究了 Linux 内核的 collateral evolution(驱动随内核接口变更需协调更新)问题,覆盖 v3.0 到 v7.2 共 20 年历史。数据显示自 v2.6.13 以来驱动文件和支持库规模增长约 8 倍,对最常用库的一次修改现在可波及约 5.3k 个驱动文件。
论文构建了 CoEvoBench 基准,包含来自 202 个主线提交、268 个模块的 316 个人工确认任务,覆盖内核版本 v6.19 至 v7.2,并区分出 134 个编译失败任务和 182 个编译兼容任务。作者还开发了 CoEvoAgent 框架,评测 Direct、SmPL、mini-SWE-agent 三种方式配合 DeepSeek Flash 的表现,并将 Direct 方式与 Claude Sonnet 5.5、GPT-6.1 Sol 对比。
结果显示 DeepSeek 默认 Direct 生成在编译失败任务上达到 41.0% 的严格参考补丁一致率,但编译兼容任务仅 3.3%,说明对接口变更语义的理解仍是难点。
Keeping Up with the Kernel: Can LLMs Adapt Linux Drivers to Interface Changes?
We analyze collateral evolution in Linux, the coordinated updates that drivers require when the kernel interfaces they depend on change, over the 20 years since it was first studied, covering releases from v3.0 to v7.2. Since v2.6.13, the number of driver files and support libraries has grown about eightfold, a single change to the most widely used library can now reach about 5.3k driver files, and such changes arrive either in bursts or spread over many releases. We introduce CoEvoBench, a benchmark that asks models to adapt an old driver module to a target kernel and evaluates the result against the developer's reference patch. The benchmark contains 316 manually confirmed tasks from 202 mainline commits and 268 modules, spanning kernel releases v6.19 to v7.2. It distinguishes 134 build-breaking tasks, where the old driver fails to compile, from 182 build-compatible tasks, where the old usage still builds. Using CoEvoAgent, we evaluate Direct, SmPL, and mini-SWE-agent with DeepSeek Flash, and compare Direct with Claude Sonnet 5.5 and GPT-6.1 Sol. With DeepSeek, default Direct generation achieves strict reference agreement on 41.0% of build-breaking tasks but only 3.3% of build-compatible tasks.
This report is categorized under AITOP's "Papers" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Linux, CoEvoBench, DeepSeek, Claude.
Primary source verification: published directly by arXiv: DeepSeek with first-party authority.