高效 eBPF 文件路径缓存的实现与挑战
问题概述
在 eBPF 程序中频繁解析文件路径会产生显著的 CPU 开销，通过缓存路径映射可降低约 90% 的 CPU 消耗，但这会带来额外的内存开销、缓存失效难题以及对多种文件系统语义（如硬链接、目录移动、绑定挂载）的支持需求。
痛点标签
- 缓存导致的额外内存消耗
- 缓存失效策略在权限变更、目录移动、硬链接、删除等情况下的处理
- 缓存大小限制及潜在的 OOM 风险
- 对多种路径表示（如 bind mounts）的兼容性
待验证问题
- 典型的缓存命中率和对应的内存占用比例如何？
- 在生产环境中，缓存失效策略的最佳实现是什么？
- 不同工作负载下应如何调整缓存大小和失效阈值？
证据
原话
Memoization trades compute time for memory: you reduce CPU cost by 90% _at the cost_ of some memory. The author measured one, but it doesn't look like they measured the other. It's an important detail to take into account.
原话
How is the cache invalidated when: - The permissions change? - Directories are moved? - Hard-links are added? - Things are deleted? Also: Is the cache limited in size?
原话
How do your path-only rules handle the many approaches for loading a file but making it appear to have a different path, such as bind mounts for one example? Simple memoization yields massive wins. Had similar experience with a hot loop, sometimes the basics are just magic.
免责声明：本卡片由 AI 从社区讨论中提炼，属于「痛点证据」，是需求分析与验证的起点，不等同于「已验证的市场需求」。AI 存在分析局限，关键决策前请打开上方原始讨论人工复核。