Superblock

The superblock holds filesystem-wide metadata: the first thing read on mount.

Typical fields:

struct superblock {
    uint32_t magic;          // FS magic number (e.g. 0xEF53 for ext)
    uint64_t total_blocks;
    uint64_t free_blocks;
    uint64_t total_inodes;
    uint64_t free_inodes;
    uint32_t block_size;
    uint32_t inode_size;
    uint32_t blocks_per_group;
    uint64_t root_inode;     // typically 1 or 2
    uint32_t mkfs_time;
    uint32_t mount_count;
    uint32_t max_mount_count;
    uint32_t state;          // clean, error, etc.
    uint8_t  uuid[16];
    char     volume_name[32];
    ...
};

Position: typically at fixed offset (e.g. block 1 in ext4; offset 1024).

Multiple copies:

• ext4 keeps backup superblocks in some block groups (e.g., every 8th).
• If primary corrupted: e2fsck -b <backup_offset> recovers.

Magic number:

• Identifies FS type at mount time.
• ext: 0xEF53.
• xfs: "XFSB" ASCII.
• btrfs: "_BHRfS_M".

Mount sequence:

1. Read superblock. Check magic.
2. Verify state (clean? Otherwise replay journal / fsck).
3. Increment mount_count.
4. Read root inode.
5. Mounted; ready for syscalls.

Unmount:

1. Flush all dirty buffers.
2. Update superblock state to "clean".
3. Write superblock to disk.

Mount errors → kernel rejects, returns EIO/EROFS.

tune2fs -l <device> shows ext superblock fields.

mkfs.<fs> creates a fresh filesystem: writes superblock, allocates bitmaps, sets up root directory.