# man > ext4(5) --- type: CommandReference command: ext4 mode: man section: 5 source: man-pages --- ## Quick Reference - `mke2fs -t ext4 /dev/sda1` — create an ext4 filesystem - `mount -t ext4 -o data=ordered /dev/sda1 /mnt` — mount with ordered journaling mode - `tune2fs -j /dev/sda1` — enable journal on an ext2 filesystem - `tune2fs -O ext_attr /dev/sda1` — enable extended attributes - `mount -t ext4 -o noatime /dev/sda1 /mnt` — mount with noatime - `e2fsck /dev/sda1` — check filesystem consistency - `resize2fs /dev/sda1 10G` — resize filesystem to 10 GB - `chattr +i file` — make a file immutable ## Name ext2, ext3, ext4 — the second, third, and fourth extended file systems for Linux. ## Synopsis The ext4 filesystem driver is used by the `mount` command and managed by utilities like `mke2fs`, `tune2fs`, `e2fsck`, `resize2fs`, and `chattr`. Mount syntax: `mount -t ext4 [options] device mountpoint` Create syntax: `mke2fs -t ext4 [options] device` ## Options ### File System Features Features are enabled via `mke2fs -O` or `tune2fs -O`. Kernel version indicates first stable support. - `64bit` — allow filesystem > 2^32 blocks (ext4, 2.6.28) - `bigalloc` — clustered block allocation (ext4, 3.2; experimental, requires `extent`) - `casefold` — case-insensitive directory lookups (ext4, 5.2) - `dir_index` — hashed b-tree directories for fast lookups (ext3, 2.6.0; ignored by ext2) - `dir_nlink` — unlimited hard links for directories (ext4, 2.6.28) - `ea_inode` — store extended attributes in separate inodes (ext4, 4.13) - `encrypt` — per-file encryption of data and names (ext4, 4.1; metadata not encrypted) - `ext_attr` — extended attributes (ext2/ext3, 2.6.0) - `extent` — extent tree for efficient block mapping (ext4, 2.6.28) - `extra_isize` — reserve space in inodes for extended metadata (ext4, 2.6.28; requires inode size >= 256) - `filetype` — store file type in directory entries (ext2, 2.2.0) - `flex_bg` — flexible block group placement (ext4, 2.6.28) - `has_journal` — create a journal (ext3, 2.4.15) - `huge_file` — allow files > 2 TB (ext4, 2.6.28) - `inline_data` — store small data in inode (ext4, 3.8) - `journal_dev` — external journal device (requires matching block size) - `large_dir` — increase max files per directory (ext4, 4.13) - `large_file` — automatically set for files > 2 GB (ext2, 2.2.0) - `metadata_csum` — checksums for all metadata (ext4, 3.18; incompatible with `uninit_bg`) - `metadata_csum_seed` — store checksum seed in superblock (ext4, 4.4) - `meta_bg` — online resize without reserved space (ext4, 2.6.28; slows mount) - `mmp` — multiple mount protection (ext4, 3.0) - `project` — project quota support (ext4, 4.5) - `quota` — create quota inodes (ext4, 3.6) - `resize_inode` — reserve space for online resize (ext3, 2.6.10; requires `sparse_super` or `sparse_super2`) - `sparse_super` — backup superblocks only in some block groups (ext2, 2.2.0) - `sparse_super2` — only two backup superblocks (ext4, 3.16) - `stable_inodes` — inode numbers and UUID are stable (ext4, 5.5) - `uninit_bg` — lazy initialization of block groups, protected by checksums (ext4, 2.6.28) - `verity` — transparent integrity verification via Merkle tree (ext4, 5.4) ### Mount Options for ext2 - `acl` / `noacl` — POSIX ACL support - `bsddf` / `minixdf` — statfs reporting style (default bsddf) - `check=none` / `nocheck` — no fsck at mount (default) - `debug` — print debugging info on remount - `errors=continue|remount-ro|panic` — behavior on error (default set in superblock) - `grpid` / `bsdgroups` and `nogrpid` / `sysvgroups` — group ID assignment for new files - `grpquota` / `noquota` / `quota` / `usrquota` — enable quota support - `nouid32` — disable 32-bit UIDs/GIDs - `oldalloc` / `orlov` — inode allocation strategy (orlov is default) - `resgid=n` / `resuid=n` — allow reserved blocks to specific user/group - `sb=n` — use alternative superblock (e.g., after corruption) - `user_xattr` / `nouser_xattr` — extended user attributes ### Mount Options for ext3 In addition to ext2 options: - `journal_dev=devnum` / `journal_path=path` — specify external journal device - `norecovery` / `noload` — skip journal replay (dangerous) - `data=journal|ordered|writeback` — journaling mode (default ordered) - `data_err=ignore|abort` — action on data buffer error in ordered mode - `barrier=0` / `barrier=1` — write barriers (default on) - `commit=nrsec` — journal commit interval (default 5 seconds) - `jqfmt=vfsold|vfsv0|vfsv1` — quota format (journaled quotas with vfsv0/vfsv1) - `usrjquota=file` / `grpjquota=file` — quota database files (ignored if `quota` feature enabled) ### Mount Options for ext4 In addition to ext3 options: - `journal_checksum` / `nojournal_checksum` — checksum journal transactions - `journal_async_commit` — commit block without waiting for descriptors (implies journal_checksum) - `barrier` / `nobarrier` — same as ext3 barrier options (default on) - `inode_readahead_blks=n` — inode table readahead (power of 2, default 32) - `stripe=n` — RAID stripe size in filesystem blocks - `delalloc` — defer block allocation until write-out (default on) - `nodelalloc` — disable delayed allocation - `max_batch_time=usec` — max time to batch sync writes (default 15000 µs) - `min_batch_time=usec` — commit time minimum (default 0) - `journal_ioprio=prio` — I/O priority for kjournald2 (0-7, default 3) - `abort` — simulate ext4_abort() for debugging - `auto_da_alloc` / `noauto_da_alloc` — detect replace-via-rename patterns and force allocation before commit - `noinit_itable` — skip background inode table initialization - `init_itable=n` — wait multiplier for lazy inode init (n times previous zeroing time) - `discard` / `nodiscard` — issue TRIM on free (default off) - `block_validity` / `noblock_validity` — track metadata blocks for debugging (default off) - `dioread_lock` / `dioread_nolock` — DIO read locking (dioread_lock default) - `max_dir_size_kb=n` — limit directory size (prevents ENOSPC on large dirs) - `i_version` — enable 64-bit inode version (default off) - `nombcache` — disable extended attribute deduplication cache - `prjquota` — enable project quota (requires `project` feature) ### File Attributes Set via `chattr`: - `a` — append only - `A` — no atime updates - `d` — no dump - `D` — synchronous directory updates - `i` — immutable - `S` — synchronous updates - `u` — undeletable - `j` — data journaling (ext3/ext4 only) - `e` — extents format (ext4 only) ## Examples **bsddf vs minixdf (statfs behavior):** shell # With minixdf (includes overhead blocks) mount /k -o minixdf df /k # Output: Filesystem 1024-blocks Used Available Capacity Mounted on # /dev/sda6 2630655 86954 2412169 3% /k # With bsddf (subtracts overhead, default) mount /k -o bsddf df /k # Output: Filesystem 1024-blocks Used Available Capacity Mounted on # /dev/sda6 2543714 13 2412169 0% /k **Mount with data=ordered and noatime:** shell mount -t ext4 -o data=ordered,noatime /dev/sda1 /mnt **Enable journal on an existing ext2 filesystem:** shell tune2fs -j /dev/sda1 ## See Also - [mke2fs(8)](http://localhost/phpMan.php/man/mke2fs/8/markdown) - [mke2fs.conf(5)](http://localhost/phpMan.php/man/mke2fs.conf/5/markdown) - [e2fsck(8)](http://localhost/phpMan.php/man/e2fsck/8/markdown) - [dumpe2fs(8)](http://localhost/phpMan.php/man/dumpe2fs/8/markdown) - [tune2fs(8)](http://localhost/phpMan.php/man/tune2fs/8/markdown) - [debugfs(8)](http://localhost/phpMan.php/man/debugfs/8/markdown) - [mount(8)](http://localhost/phpMan.php/man/mount/8/markdown) - [chattr(1)](http://localhost/phpMan.php/man/chattr/1/markdown) - [acl(5)](http://localhost/phpMan.php/man/acl/5/markdown) - [resize2fs(8)](http://localhost/phpMan.php/man/resize2fs/8/markdown)