Byte content
The file-system API commonly presents a sequence of bytes even when applications interpret records, images or source code.
PART 3 • MEMORY & STORAGE
Turn Persistent Blocks into Named, Protected Files
Trace pathname lookup, separate names from metadata, compare block allocation and explain how journaling limits damage after a crash.
BY THE END, YOU CAN
Reason from name to block
- Explain file metadata and operations.
- Trace directories and inodes.
- Compare allocation methods.
- Manage free space.
- Explain consistency and journaling.
A File Combines Data, Metadata and an Interface
Facts about content
Size, owner, permissions, timestamps, type and block pointers support protection and lookup.
Per-use context
A descriptor references a kernel open-file entry containing mode, current offset and object identity.
read(fd, buffer, count)
write(fd, buffer, count)
seek changes current offset
close releases descriptor
A filename is not usually stored inside the inode. A directory maps a name to an object number; multiple hard links may map different names to the same inode. A symbolic link stores a pathname and can cross file systems.
Directories Form a Searchable Namespace
Choose start
An absolute path begins at root; a relative path begins at the process’s current directory.
Check execute/search permission
Permission is checked while traversing each directory, not only on the final file.
Map name to inode
Directory entries provide the next object identifier; caches may accelerate lookup.
Open and return descriptor
The kernel validates requested access and installs a descriptor in the process table.
| Allocation | Sequential access | Random access | Growth/fragmentation |
|---|---|---|---|
| Contiguous | Excellent | Direct | External fragmentation; hard growth |
| Linked | Good | Slow traversal | Easy growth; pointer overhead |
| Indexed | Good | Direct via index | Index overhead; scalable variants |
Place One File using Three Strategies
READY
Select a method
The block map and method trade-off will appear here.
Free-space structures
One bit per block
Fast to locate runs with word operations; bitmap itself consumes predictable space.
Chain free blocks
Simple allocation but finding a large contiguous run may require traversal.
Compress free ranges
Store groups or starting block plus run length when free blocks cluster.
Crash Consistency Is about Legal On-Disk States
Creating a file may update a directory entry, inode allocation bitmap, inode and data blocks. A crash between writes can leak blocks or expose incomplete metadata. Write ordering, journaling, copy-on-write trees and recovery checks address different parts of this problem.
Log intent/metadata
Record a transaction before applying home-location updates; replay committed work after a crash.
Scan and repair
Reconstruct consistency from on-disk structures, often slower on large file systems.
Publish new roots
Write changed blocks elsewhere, then atomically switch metadata to the new version.
Trace open("/home/venu/notes.txt")
Ten Concept Checks
Answered correctly: 0 of 10
Connect Names, Metadata and Blocks
- File versus inode?
- Hard versus symbolic link?
- What is a descriptor?
- Define mounting.
- What is journaling?
- Trace pathname resolution.
- Compare allocation methods.
- Explain free-space management.
- Explain crash consistency.
- Describe directory structures.
- Why can deleted files remain open?
- What does fsync promise?
- Why are sparse files useful?
- How does rename help atomic replace?
- What costs random linked access?
LEVEL 10 SUMMARY
You Can Trace a Name to Persistent Blocks
- Directories map names; inodes hold metadata and block mapping.
- Descriptors represent per-process access to open objects.
- Allocation trades locality, direct access, growth and metadata.
- Free-space structures make reusable blocks discoverable.
- Journaling and copy-on-write protect update consistency.
COURSE CHECKPOINT
Mark complete after you can compare all three allocation methods.
Saved only in this browser.