From Descriptors to Directories 掌握文件描述符到目录遍历的全链路

System call pipelines, metadata, hard/soft links, and directory walking aligned with lab & exam scenarios 系统调用流程、元数据、硬/软链接与目录遍历，紧贴实验与考试题型

🎯 Learning Objectives学习目标

Explain how file descriptors abstract open files and how they relate to stdio streams.解释文件描述符如何抽象打开的文件，并与 stdio 流建立联系。
Use low-level system calls (open, read, write, close) safely for streaming I/O.安全使用底层系统调用（open、read、write、close）进行流式 I/O。
Inspect metadata with stat and interpret permissions, link counts, and special file types.通过 stat 检查文件元数据，解释权限、链接计数和特殊文件类型。
Compare hard vs symbolic links and reason about inode sharing.比较硬链接与符号链接，分析 inode 共享机制。
Traverse directories recursively while avoiding cycles and respecting sandbox rules.递归遍历目录，避免循环并遵守沙箱规则。

🧭 Exam Alignment考试对齐

Descriptor tracing — 22T3 Q7 variants require mapping duplicates and inherited fds.描述符追踪 — 22T3 Q7 衍生题需映射重复和继承的 fd。
Metadata Q&A — Lab09 «stat_report» prepares for questions about permissions & link counts.元数据问答 — Lab09 “stat_report” 预演权限与链接计数题。
Directory walkers — Past finals ask to detect cycles and skip hidden entries.目录遍历 — 历届期末要求检测循环并跳过隐藏文件。
Link semantics — 20T2 Q6 discuss rename/link/unlink effects on inodes.链接语义 — 20T2 Q6 探讨 rename/link/unlink 对 inode 的影响。
Streaming vs buffered I/O — Tutorial09 compares read/write with stdio buffering.流式 vs 缓存 I/O — Tutorial09 对比 read/write 与 stdio 缓冲。

Coverage goals: assemble ≥3 file-descriptor diagrams, catalogue metadata fields, and document at least two directory traversal patterns with cycle guards. 覆盖目标：绘制 ≥3 个文件描述符示意，整理关键元数据字段，并记录至少两种带循环防护的目录遍历模式。

📚 Core Concepts核心概念

File Descriptor Basics文件描述符基础

Descriptors are integers indexing kernel file tables. 0/1/2 map to stdin/stdout/stderr. Duplication via dup2 enables redirection.描述符是索引内核文件表的整数。0/1/2 对应标准输入/输出/错误。通过 dup2 可实现重定向。

Low-level System Calls底层系统调用

open returns descriptor; read/write transfer bytes; close releases kernel resources. Always check return values.open 返回描述符；read/write 传输字节；close 释放资源。必须检查返回值。

Metadata & Permissions元数据与权限

stat reveals size, ownership, timestamps, mode bits, link count, and device IDs. Understand octal permissions and special bits (setuid, sticky).stat 展示大小、所有者、时间戳、模式位、链接数与设备号。需掌握八进制权限及特殊位（setuid、sticky）。

Hard vs Symbolic Links硬链接 vs 软链接

Hard links share the same inode, increasing link count; symlinks store path text and can span filesystems.硬链接共享同一 inode 并增加链接计数；符号链接存储路径文本，可跨文件系统。

Directory Traversal目录遍历

Use opendir/readdir/closedir or ftw. Skip “.” and “..”, detect cycles via inode+device pairs.通过 opendir/readdir/closedir 或 ftw 遍历目录。忽略 “.” 和 “..”，并用 inode+device 组合检测循环。

Streams vs Syscalls流式 I/O 与系统调用

stdio buffers data for convenience; mixing stdio with raw descriptors requires flush/sync steps to stay consistent.stdio 为方便会缓存数据；在 stdio 与底层描述符混用时需刷新/同步以保持一致。

🧪 Worked Examples示例串讲

Example 1 — Implementing tee with dup2示例 1 — 使用 dup2 实现 tee

Duplicate stdout into a log file while still printing to terminal. Use dup2(log_fd, STDOUT_FILENO) after opening file.将 stdout 复制到日志文件同时输出终端。打开文件后执行 dup2(log_fd, STDOUT_FILENO)。

Remember to save original stdout via dup if you need to restore later.如需恢复标准输出，先用 dup 备份原 stdout。

Example 2 — Summarising metadata from stat示例 2 — stat 元数据摘要

Collect size, mode, UID/GID, link count, and modification time; format as table for Lab09 «stat_report».收集文件大小、模式、UID/GID、链接数与修改时间，并以表格展示，配合 Lab09 “stat_report”。

⚠️ Common Pitfalls易错点

Neglecting error handling for read/write (short reads, EINTR).忽略 read/write 的错误处理（短读、EINTR）。
Following symbolic links blindly during recursion leads to infinite loops.递归遍历时盲目跟随符号链接会造成无限循环。
Assuming unlink removes file data immediately even when other hard links exist.误以为 unlink 会立即删除数据，忽略其他硬链接存在。
Mixing stdio and raw descriptors without flushing buffered data.混用 stdio 与底层描述符时未刷新缓冲区导致数据错乱。

🛠️ Practice Task实践任务

Build fs_inspector.c: recursively scan a directory tree, print metadata summary, detect cycles, and optionally copy regular files to a backup location.实现 fs_inspector.c：递归扫描目录树，输出元数据摘要，检测循环，并可选复制常规文件至备份目录。

Track visited (device, inode) pairs to avoid repeated traversal.通过记录 (device, inode) 对避免重复遍历。
Respect file permissions; fall back gracefully when lacking access.遵守权限限制；无权访问时优雅退回。
Optional Optional: add JSON export combining metadata for integration with dashboards.可选 Optional：导出 JSON 供仪表盘使用。

🧪 Tutorial & Lab Mapping教程与实验映射

Tutorial 09 HighlightsTutorial 09 精要

Descriptor duplication puzzles (dup2 behaviour).描述符复制谜题（dup2 行为）。
Permission matrix exercises translating mode bits to human-readable form.权限矩阵练习，将模式位转换为人类可读形式。
Link-count reasoning with rename/unlink scenarios.通过 rename/unlink 场景推理链接计数。

Lab 09 Programming TasksLab 09 编程任务

filecopy.c — robust copy using read/write loops.filecopy.c — 使用 read/write 循环完成稳健复制。
stat_report.c — summarise metadata for list of files.stat_report.c — 输出文件元数据汇总。
link_mirror.c — replicate directory tree using hard links.link_mirror.c — 使用硬链接镜像目录树。
walk_fs.c (challenge) — recursive walker with cycle detection.walk_fs.c（挑战） — 带循环检测的递归遍历器。

📝 Study Log学习记录

Inputs shared: files.pdf, Lab09 spec, Tutorial09 worksheet, sample 22T2/22T3 file-system questions.提供资料：files.pdf、Lab09 说明、Tutorial09 习题、22T2/22T3 文件系统题。
Prompt: “Create descriptor/link diagrams that double as exam answer templates.”提示词：“构建描述符/链接示意图，直接可用于考试答题。”
Breakthrough: Tracking device+inode pairs simplified cycle detection logic.收获：记录 device+inode 组合后，循环检测逻辑大幅简化。
Misconception fixed: Previously thought symlink increments target link count — corrected via stat experiments.修正误区：曾认为符号链接会增加目标链接计数，经 stat 实验纠正。
Action items: Practise translating mode bits to rwx string; rehearse unlink vs remove scenarios.后续行动：练习将模式位转换为 rwx 字符串；复盘 unlink 与 remove 场景差异。

Premium Quiz — 40 Questions on Unix FilesPremium 测验 — 40 道 Unix 文件题

28 basic (syscalls, metadata) · 8 intermediate (directory walking) · 4 advanced (link semantics & streaming design)基础28题（系统调用/元数据）· 中级8题（目录遍历）· 高级4题（链接语义与流式设计）

🔒 Open Week 9 Quiz (Premium) 打开第 9 周测验（会员）

🔭 Next Week Preview下周预告

Week 10 shifts to numeric representations and encoding (floating point & Unicode). Preview floating_point.pdf and unicode.pdf.第10周转向数值表示与编码（浮点与 Unicode）。请预习 floating_point.pdf 与 unicode.pdf。

📎 Resources & Checklist资源与检查表

PPT: files.pdf p1–68, focus on system call examples.PPT：files.pdf 第1–68页，重点关注系统调用示例。
Autotest: 1521 autotest lab09_filecopy, lab09_stat_report, lab09_link_mirror, lab09_walk_fs.自动测试：1521 autotest lab09_filecopy、lab09_stat_report、lab09_link_mirror、lab09_walk_fs。
Self-check: Can you explain difference between hard vs soft links? Can you safely interleave stdio and raw descriptors?自检：能否解释硬链接与软链接差异？能否安全混用 stdio 与底层描述符？