Dear NetBSD kernel developers, I hope this email finds you well. You can call me Mahno, and I am very interested in contributing to the "Asynchronous I/O" project for NetBSD’s GSoC 2025. After carefully reviewing the project description, I have a few questions about the technical expectations for contributors, and I would be grateful for your insights.
Questions About Project Requirements: 1. Prior Contributions to Industrial-Grade Kernels: - The project involves significant revisions to NetBSD’s I/O subsystem. Is prior experience contributing to production-grade kernels (e.g., NetBSD, Linux, FreeBSD) a strict requirement, or can familiarity with academic/research kernels (e.g., xv6, Minix) suffice for initial contributions? 2. Skill Priorities: - Would familiarity with thread synchronization models (e.g., lock-free queues, work-stealing) be essential for addressing the current single-threaded bottleneck in the AIO implementation? 3. NetBSD-Specific Knowledge: - To diagnose undocumented edge cases in the current AIO implementation, how much familiarity with NetBSD’s VFS layer or driver architecture is expected upfront? My Current Skills: - Kernel Development: - Completed all labs in MIT 6.1810 (Operating System Engineering), including implementing system calls, CoW virtual memory, and thread scheduling in xv6. - Experimented with debugging in kernel code using gdb. - Algorithmic Foundations: - Awarded an International Collegiate Programming Contest Regional Bronze Medal (2023, 2024), with rigorous training in optimizing algorithms and data structures—a skill I believe could help improve parallelism in the AIO subsystem. - Systems Familarity: - Daily use of BSD/Linux systems for 3+ years, including writing small kernel modules and profiling performance bottlenecks with tools like `perf` and `dtrace`. To align with the project’s goals, I am currently: - Auditing NetBSD’s existing AIO implementation to understand its threading model. - Researching async I/O paradigms (e.g., Linux `io_uring`, Windows IOCP) to compare design tradeoffs. I would deeply appreciate any advice on how to strengthen my preparation or specific areas to focus on. Thank you for your time — I am eager to learn and contribute to NetBSD’s future. Best regards, ma...@libertarian.dev
