1.. SPDX-License-Identifier: GPL-2.0 2 3======================== 4ext4 General Information 5======================== 6 7Ext4 is an advanced level of the ext3 filesystem which incorporates 8scalability and reliability enhancements for supporting large filesystems 9(64 bit) in keeping with increasing disk capacities and state-of-the-art 10feature requirements. 11 12Mailing list: linux-ext4@vger.kernel.org 13Web site: http://ext4.wiki.kernel.org 14 15 16Quick usage instructions 17======================== 18 19Note: More extensive information for getting started with ext4 can be 20found at the ext4 wiki site at the URL: 21http://ext4.wiki.kernel.org/index.php/Ext4_Howto 22 23 - The latest version of e2fsprogs can be found at: 24 25 https://www.kernel.org/pub/linux/kernel/people/tytso/e2fsprogs/ 26 27 or 28 29 http://sourceforge.net/project/showfiles.php?group_id=2406 30 31 or grab the latest git repository from: 32 33 https://git.kernel.org/pub/scm/fs/ext2/e2fsprogs.git 34 35 - Create a new filesystem using the ext4 filesystem type: 36 37 # mke2fs -t ext4 /dev/hda1 38 39 Or to configure an existing ext3 filesystem to support extents: 40 41 # tune2fs -O extents /dev/hda1 42 43 If the filesystem was created with 128 byte inodes, it can be 44 converted to use 256 byte for greater efficiency via: 45 46 # tune2fs -I 256 /dev/hda1 47 48 - Mounting: 49 50 # mount -t ext4 /dev/hda1 /wherever 51 52 - When comparing performance with other filesystems, it's always 53 important to try multiple workloads; very often a subtle change in a 54 workload parameter can completely change the ranking of which 55 filesystems do well compared to others. When comparing versus ext3, 56 note that ext4 enables write barriers by default, while ext3 does 57 not enable write barriers by default. So it is useful to use 58 explicitly specify whether barriers are enabled or not when via the 59 '-o barriers=[0|1]' mount option for both ext3 and ext4 filesystems 60 for a fair comparison. When tuning ext3 for best benchmark numbers, 61 it is often worthwhile to try changing the data journaling mode; '-o 62 data=writeback' can be faster for some workloads. (Note however that 63 running mounted with data=writeback can potentially leave stale data 64 exposed in recently written files in case of an unclean shutdown, 65 which could be a security exposure in some situations.) Configuring 66 the filesystem with a large journal can also be helpful for 67 metadata-intensive workloads. 68 69Features 70======== 71 72Currently Available 73------------------- 74 75* ability to use filesystems > 16TB (e2fsprogs support not available yet) 76* extent format reduces metadata overhead (RAM, IO for access, transactions) 77* extent format more robust in face of on-disk corruption due to magics, 78* internal redundancy in tree 79* improved file allocation (multi-block alloc) 80* lift 32000 subdirectory limit imposed by i_links_count[1] 81* nsec timestamps for mtime, atime, ctime, create time 82* inode version field on disk (NFSv4, Lustre) 83* reduced e2fsck time via uninit_bg feature 84* journal checksumming for robustness, performance 85* persistent file preallocation (e.g for streaming media, databases) 86* ability to pack bitmaps and inode tables into larger virtual groups via the 87 flex_bg feature 88* large file support 89* inode allocation using large virtual block groups via flex_bg 90* delayed allocation 91* large block (up to pagesize) support 92* efficient new ordered mode in JBD2 and ext4 (avoid using buffer head to force 93 the ordering) 94* Case-insensitive file name lookups 95 96[1] Filesystems with a block size of 1k may see a limit imposed by the 97directory hash tree having a maximum depth of two. 98 99case-insensitive file name lookups 100====================================================== 101 102The case-insensitive file name lookup feature is supported on a 103per-directory basis, allowing the user to mix case-insensitive and 104case-sensitive directories in the same filesystem. It is enabled by 105flipping the +F inode attribute of an empty directory. The 106case-insensitive string match operation is only defined when we know how 107text in encoded in a byte sequence. For that reason, in order to enable 108case-insensitive directories, the filesystem must have the 109casefold feature, which stores the filesystem-wide encoding 110model used. By default, the charset adopted is the latest version of 111Unicode (12.1.0, by the time of this writing), encoded in the UTF-8 112form. The comparison algorithm is implemented by normalizing the 113strings to the Canonical decomposition form, as defined by Unicode, 114followed by a byte per byte comparison. 115 116The case-awareness is name-preserving on the disk, meaning that the file 117name provided by userspace is a byte-per-byte match to what is actually 118written in the disk. The Unicode normalization format used by the 119kernel is thus an internal representation, and not exposed to the 120userspace nor to the disk, with the important exception of disk hashes, 121used on large case-insensitive directories with DX feature. On DX 122directories, the hash must be calculated using the casefolded version of 123the filename, meaning that the normalization format used actually has an 124impact on where the directory entry is stored. 125 126When we change from viewing filenames as opaque byte sequences to seeing 127them as encoded strings we need to address what happens when a program 128tries to create a file with an invalid name. The Unicode subsystem 129within the kernel leaves the decision of what to do in this case to the 130filesystem, which select its preferred behavior by enabling/disabling 131the strict mode. When Ext4 encounters one of those strings and the 132filesystem did not require strict mode, it falls back to considering the 133entire string as an opaque byte sequence, which still allows the user to 134operate on that file, but the case-insensitive lookups won't work. 135 136Options 137======= 138 139When mounting an ext4 filesystem, the following option are accepted: 140(*) == default 141 142 ro 143 Mount filesystem read only. Note that ext4 will replay the journal (and 144 thus write to the partition) even when mounted "read only". The mount 145 options "ro,noload" can be used to prevent writes to the filesystem. 146 147 journal_checksum 148 Enable checksumming of the journal transactions. This will allow the 149 recovery code in e2fsck and the kernel to detect corruption in the 150 kernel. It is a compatible change and will be ignored by older 151 kernels. 152 153 journal_async_commit 154 Commit block can be written to disk without waiting for descriptor 155 blocks. If enabled older kernels cannot mount the device. This will 156 enable 'journal_checksum' internally. 157 158 journal_path=path, journal_dev=devnum 159 When the external journal device's major/minor numbers have changed, 160 these options allow the user to specify the new journal location. The 161 journal device is identified through either its new major/minor numbers 162 encoded in devnum, or via a path to the device. 163 164 norecovery, noload 165 Don't load the journal on mounting. Note that if the filesystem was 166 not unmounted cleanly, skipping the journal replay will lead to the 167 filesystem containing inconsistencies that can lead to any number of 168 problems. 169 170 data=journal 171 All data are committed into the journal prior to being written into the 172 main file system. Enabling this mode will disable delayed allocation 173 and O_DIRECT support. 174 175 data=ordered (*) 176 All data are forced directly out to the main file system prior to its 177 metadata being committed to the journal. 178 179 data=writeback 180 Data ordering is not preserved, data may be written into the main file 181 system after its metadata has been committed to the journal. 182 183 commit=nrsec (*) 184 Ext4 can be told to sync all its data and metadata every 'nrsec' 185 seconds. The default value is 5 seconds. This means that if you lose 186 your power, you will lose as much as the latest 5 seconds of work (your 187 filesystem will not be damaged though, thanks to the journaling). This 188 default value (or any low value) will hurt performance, but it's good 189 for data-safety. Setting it to 0 will have the same effect as leaving 190 it at the default (5 seconds). Setting it to very large values will 191 improve performance. 192 193 barrier=<0|1(*)>, barrier(*), nobarrier 194 This enables/disables the use of write barriers in the jbd code. 195 barrier=0 disables, barrier=1 enables. This also requires an IO stack 196 which can support barriers, and if jbd gets an error on a barrier 197 write, it will disable again with a warning. Write barriers enforce 198 proper on-disk ordering of journal commits, making volatile disk write 199 caches safe to use, at some performance penalty. If your disks are 200 battery-backed in one way or another, disabling barriers may safely 201 improve performance. The mount options "barrier" and "nobarrier" can 202 also be used to enable or disable barriers, for consistency with other 203 ext4 mount options. 204 205 inode_readahead_blks=n 206 This tuning parameter controls the maximum number of inode table blocks 207 that ext4's inode table readahead algorithm will pre-read into the 208 buffer cache. The default value is 32 blocks. 209 210 nouser_xattr 211 Disables Extended User Attributes. See the attr(5) manual page for 212 more information about extended attributes. 213 214 noacl 215 This option disables POSIX Access Control List support. If ACL support 216 is enabled in the kernel configuration (CONFIG_EXT4_FS_POSIX_ACL), ACL 217 is enabled by default on mount. See the acl(5) manual page for more 218 information about acl. 219 220 bsddf (*) 221 Make 'df' act like BSD. 222 223 minixdf 224 Make 'df' act like Minix. 225 226 debug 227 Extra debugging information is sent to syslog. 228 229 abort 230 Simulate the effects of calling ext4_abort() for debugging purposes. 231 This is normally used while remounting a filesystem which is already 232 mounted. 233 234 errors=remount-ro 235 Remount the filesystem read-only on an error. 236 237 errors=continue 238 Keep going on a filesystem error. 239 240 errors=panic 241 Panic and halt the machine if an error occurs. (These mount options 242 override the errors behavior specified in the superblock, which can be 243 configured using tune2fs) 244 245 data_err=ignore(*) 246 Just print an error message if an error occurs in a file data buffer in 247 ordered mode. 248 data_err=abort 249 Abort the journal if an error occurs in a file data buffer in ordered 250 mode. 251 252 grpid | bsdgroups 253 New objects have the group ID of their parent. 254 255 nogrpid (*) | sysvgroups 256 New objects have the group ID of their creator. 257 258 resgid=n 259 The group ID which may use the reserved blocks. 260 261 resuid=n 262 The user ID which may use the reserved blocks. 263 264 sb= 265 Use alternate superblock at this location. 266 267 quota, noquota, grpquota, usrquota 268 These options are ignored by the filesystem. They are used only by 269 quota tools to recognize volumes where quota should be turned on. See 270 documentation in the quota-tools package for more details 271 (http://sourceforge.net/projects/linuxquota). 272 273 jqfmt=<quota type>, usrjquota=<file>, grpjquota=<file> 274 These options tell filesystem details about quota so that quota 275 information can be properly updated during journal replay. They replace 276 the above quota options. See documentation in the quota-tools package 277 for more details (http://sourceforge.net/projects/linuxquota). 278 279 stripe=n 280 Number of filesystem blocks that mballoc will try to use for allocation 281 size and alignment. For RAID5/6 systems this should be the number of 282 data disks * RAID chunk size in file system blocks. 283 284 delalloc (*) 285 Defer block allocation until just before ext4 writes out the block(s) 286 in question. This allows ext4 to better allocation decisions more 287 efficiently. 288 289 nodelalloc 290 Disable delayed allocation. Blocks are allocated when the data is 291 copied from userspace to the page cache, either via the write(2) system 292 call or when an mmap'ed page which was previously unallocated is 293 written for the first time. 294 295 max_batch_time=usec 296 Maximum amount of time ext4 should wait for additional filesystem 297 operations to be batch together with a synchronous write operation. 298 Since a synchronous write operation is going to force a commit and then 299 a wait for the I/O complete, it doesn't cost much, and can be a huge 300 throughput win, we wait for a small amount of time to see if any other 301 transactions can piggyback on the synchronous write. The algorithm 302 used is designed to automatically tune for the speed of the disk, by 303 measuring the amount of time (on average) that it takes to finish 304 committing a transaction. Call this time the "commit time". If the 305 time that the transaction has been running is less than the commit 306 time, ext4 will try sleeping for the commit time to see if other 307 operations will join the transaction. The commit time is capped by 308 the max_batch_time, which defaults to 15000us (15ms). This 309 optimization can be turned off entirely by setting max_batch_time to 0. 310 311 min_batch_time=usec 312 This parameter sets the commit time (as described above) to be at least 313 min_batch_time. It defaults to zero microseconds. Increasing this 314 parameter may improve the throughput of multi-threaded, synchronous 315 workloads on very fast disks, at the cost of increasing latency. 316 317 journal_ioprio=prio 318 The I/O priority (from 0 to 7, where 0 is the highest priority) which 319 should be used for I/O operations submitted by kjournald2 during a 320 commit operation. This defaults to 3, which is a slightly higher 321 priority than the default I/O priority. 322 323 auto_da_alloc(*), noauto_da_alloc 324 Many broken applications don't use fsync() when replacing existing 325 files via patterns such as fd = open("foo.new")/write(fd,..)/close(fd)/ 326 rename("foo.new", "foo"), or worse yet, fd = open("foo", 327 O_TRUNC)/write(fd,..)/close(fd). If auto_da_alloc is enabled, ext4 328 will detect the replace-via-rename and replace-via-truncate patterns 329 and force that any delayed allocation blocks are allocated such that at 330 the next journal commit, in the default data=ordered mode, the data 331 blocks of the new file are forced to disk before the rename() operation 332 is committed. This provides roughly the same level of guarantees as 333 ext3, and avoids the "zero-length" problem that can happen when a 334 system crashes before the delayed allocation blocks are forced to disk. 335 336 noinit_itable 337 Do not initialize any uninitialized inode table blocks in the 338 background. This feature may be used by installation CD's so that the 339 install process can complete as quickly as possible; the inode table 340 initialization process would then be deferred until the next time the 341 file system is unmounted. 342 343 init_itable=n 344 The lazy itable init code will wait n times the number of milliseconds 345 it took to zero out the previous block group's inode table. This 346 minimizes the impact on the system performance while file system's 347 inode table is being initialized. 348 349 discard, nodiscard(*) 350 Controls whether ext4 should issue discard/TRIM commands to the 351 underlying block device when blocks are freed. This is useful for SSD 352 devices and sparse/thinly-provisioned LUNs, but it is off by default 353 until sufficient testing has been done. 354 355 nouid32 356 Disables 32-bit UIDs and GIDs. This is for interoperability with 357 older kernels which only store and expect 16-bit values. 358 359 block_validity(*), noblock_validity 360 These options enable or disable the in-kernel facility for tracking 361 filesystem metadata blocks within internal data structures. This 362 allows multi- block allocator and other routines to notice bugs or 363 corrupted allocation bitmaps which cause blocks to be allocated which 364 overlap with filesystem metadata blocks. 365 366 dioread_lock, dioread_nolock 367 Controls whether or not ext4 should use the DIO read locking. If the 368 dioread_nolock option is specified ext4 will allocate uninitialized 369 extent before buffer write and convert the extent to initialized after 370 IO completes. This approach allows ext4 code to avoid using inode 371 mutex, which improves scalability on high speed storages. However this 372 does not work with data journaling and dioread_nolock option will be 373 ignored with kernel warning. Note that dioread_nolock code path is only 374 used for extent-based files. Because of the restrictions this options 375 comprises it is off by default (e.g. dioread_lock). 376 377 max_dir_size_kb=n 378 This limits the size of directories so that any attempt to expand them 379 beyond the specified limit in kilobytes will cause an ENOSPC error. 380 This is useful in memory constrained environments, where a very large 381 directory can cause severe performance problems or even provoke the Out 382 Of Memory killer. (For example, if there is only 512mb memory 383 available, a 176mb directory may seriously cramp the system's style.) 384 385 i_version 386 Enable 64-bit inode version support. This option is off by default. 387 388 dax 389 Use direct access (no page cache). See 390 Documentation/filesystems/dax.txt. Note that this option is 391 incompatible with data=journal. 392 393Data Mode 394========= 395There are 3 different data modes: 396 397* writeback mode 398 399 In data=writeback mode, ext4 does not journal data at all. This mode provides 400 a similar level of journaling as that of XFS, JFS, and ReiserFS in its default 401 mode - metadata journaling. A crash+recovery can cause incorrect data to 402 appear in files which were written shortly before the crash. This mode will 403 typically provide the best ext4 performance. 404 405* ordered mode 406 407 In data=ordered mode, ext4 only officially journals metadata, but it logically 408 groups metadata information related to data changes with the data blocks into 409 a single unit called a transaction. When it's time to write the new metadata 410 out to disk, the associated data blocks are written first. In general, this 411 mode performs slightly slower than writeback but significantly faster than 412 journal mode. 413 414* journal mode 415 416 data=journal mode provides full data and metadata journaling. All new data is 417 written to the journal first, and then to its final location. In the event of 418 a crash, the journal can be replayed, bringing both data and metadata into a 419 consistent state. This mode is the slowest except when data needs to be read 420 from and written to disk at the same time where it outperforms all others 421 modes. Enabling this mode will disable delayed allocation and O_DIRECT 422 support. 423 424/proc entries 425============= 426 427Information about mounted ext4 file systems can be found in 428/proc/fs/ext4. Each mounted filesystem will have a directory in 429/proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or 430/proc/fs/ext4/dm-0). The files in each per-device directory are shown 431in table below. 432 433Files in /proc/fs/ext4/<devname> 434 435 mb_groups 436 details of multiblock allocator buddy cache of free blocks 437 438/sys entries 439============ 440 441Information about mounted ext4 file systems can be found in 442/sys/fs/ext4. Each mounted filesystem will have a directory in 443/sys/fs/ext4 based on its device name (i.e., /sys/fs/ext4/hdc or 444/sys/fs/ext4/dm-0). The files in each per-device directory are shown 445in table below. 446 447Files in /sys/fs/ext4/<devname>: 448 449(see also Documentation/ABI/testing/sysfs-fs-ext4) 450 451 delayed_allocation_blocks 452 This file is read-only and shows the number of blocks that are dirty in 453 the page cache, but which do not have their location in the filesystem 454 allocated yet. 455 456 inode_goal 457 Tuning parameter which (if non-zero) controls the goal inode used by 458 the inode allocator in preference to all other allocation heuristics. 459 This is intended for debugging use only, and should be 0 on production 460 systems. 461 462 inode_readahead_blks 463 Tuning parameter which controls the maximum number of inode table 464 blocks that ext4's inode table readahead algorithm will pre-read into 465 the buffer cache. 466 467 lifetime_write_kbytes 468 This file is read-only and shows the number of kilobytes of data that 469 have been written to this filesystem since it was created. 470 471 max_writeback_mb_bump 472 The maximum number of megabytes the writeback code will try to write 473 out before move on to another inode. 474 475 mb_group_prealloc 476 The multiblock allocator will round up allocation requests to a 477 multiple of this tuning parameter if the stripe size is not set in the 478 ext4 superblock 479 480 mb_max_to_scan 481 The maximum number of extents the multiblock allocator will search to 482 find the best extent. 483 484 mb_min_to_scan 485 The minimum number of extents the multiblock allocator will search to 486 find the best extent. 487 488 mb_order2_req 489 Tuning parameter which controls the minimum size for requests (as a 490 power of 2) where the buddy cache is used. 491 492 mb_stats 493 Controls whether the multiblock allocator should collect statistics, 494 which are shown during the unmount. 1 means to collect statistics, 0 495 means not to collect statistics. 496 497 mb_stream_req 498 Files which have fewer blocks than this tunable parameter will have 499 their blocks allocated out of a block group specific preallocation 500 pool, so that small files are packed closely together. Each large file 501 will have its blocks allocated out of its own unique preallocation 502 pool. 503 504 session_write_kbytes 505 This file is read-only and shows the number of kilobytes of data that 506 have been written to this filesystem since it was mounted. 507 508 reserved_clusters 509 This is RW file and contains number of reserved clusters in the file 510 system which will be used in the specific situations to avoid costly 511 zeroout, unexpected ENOSPC, or possible data loss. The default is 2% or 512 4096 clusters, whichever is smaller and this can be changed however it 513 can never exceed number of clusters in the file system. If there is not 514 enough space for the reserved space when mounting the file mount will 515 _not_ fail. 516 517Ioctls 518====== 519 520There is some Ext4 specific functionality which can be accessed by applications 521through the system call interfaces. The list of all Ext4 specific ioctls are 522shown in the table below. 523 524Table of Ext4 specific ioctls 525 526 EXT4_IOC_GETFLAGS 527 Get additional attributes associated with inode. The ioctl argument is 528 an integer bitfield, with bit values described in ext4.h. This ioctl is 529 an alias for FS_IOC_GETFLAGS. 530 531 EXT4_IOC_SETFLAGS 532 Set additional attributes associated with inode. The ioctl argument is 533 an integer bitfield, with bit values described in ext4.h. This ioctl is 534 an alias for FS_IOC_SETFLAGS. 535 536 EXT4_IOC_GETVERSION, EXT4_IOC_GETVERSION_OLD 537 Get the inode i_generation number stored for each inode. The 538 i_generation number is normally changed only when new inode is created 539 and it is particularly useful for network filesystems. The '_OLD' 540 version of this ioctl is an alias for FS_IOC_GETVERSION. 541 542 EXT4_IOC_SETVERSION, EXT4_IOC_SETVERSION_OLD 543 Set the inode i_generation number stored for each inode. The '_OLD' 544 version of this ioctl is an alias for FS_IOC_SETVERSION. 545 546 EXT4_IOC_GROUP_EXTEND 547 This ioctl has the same purpose as the resize mount option. It allows 548 to resize filesystem to the end of the last existing block group, 549 further resize has to be done with resize2fs, either online, or 550 offline. The argument points to the unsigned logn number representing 551 the filesystem new block count. 552 553 EXT4_IOC_MOVE_EXT 554 Move the block extents from orig_fd (the one this ioctl is pointing to) 555 to the donor_fd (the one specified in move_extent structure passed as 556 an argument to this ioctl). Then, exchange inode metadata between 557 orig_fd and donor_fd. This is especially useful for online 558 defragmentation, because the allocator has the opportunity to allocate 559 moved blocks better, ideally into one contiguous extent. 560 561 EXT4_IOC_GROUP_ADD 562 Add a new group descriptor to an existing or new group descriptor 563 block. The new group descriptor is described by ext4_new_group_input 564 structure, which is passed as an argument to this ioctl. This is 565 especially useful in conjunction with EXT4_IOC_GROUP_EXTEND, which 566 allows online resize of the filesystem to the end of the last existing 567 block group. Those two ioctls combined is used in userspace online 568 resize tool (e.g. resize2fs). 569 570 EXT4_IOC_MIGRATE 571 This ioctl operates on the filesystem itself. It converts (migrates) 572 ext3 indirect block mapped inode to ext4 extent mapped inode by walking 573 through indirect block mapping of the original inode and converting 574 contiguous block ranges into ext4 extents of the temporary inode. Then, 575 inodes are swapped. This ioctl might help, when migrating from ext3 to 576 ext4 filesystem, however suggestion is to create fresh ext4 filesystem 577 and copy data from the backup. Note, that filesystem has to support 578 extents for this ioctl to work. 579 580 EXT4_IOC_ALLOC_DA_BLKS 581 Force all of the delay allocated blocks to be allocated to preserve 582 application-expected ext3 behaviour. Note that this will also start 583 triggering a write of the data blocks, but this behaviour may change in 584 the future as it is not necessary and has been done this way only for 585 sake of simplicity. 586 587 EXT4_IOC_RESIZE_FS 588 Resize the filesystem to a new size. The number of blocks of resized 589 filesystem is passed in via 64 bit integer argument. The kernel 590 allocates bitmaps and inode table, the userspace tool thus just passes 591 the new number of blocks. 592 593 EXT4_IOC_SWAP_BOOT 594 Swap i_blocks and associated attributes (like i_blocks, i_size, 595 i_flags, ...) from the specified inode with inode EXT4_BOOT_LOADER_INO 596 (#5). This is typically used to store a boot loader in a secure part of 597 the filesystem, where it can't be changed by a normal user by accident. 598 The data blocks of the previous boot loader will be associated with the 599 given inode. 600 601References 602========== 603 604kernel source: <file:fs/ext4/> 605 <file:fs/jbd2/> 606 607programs: http://e2fsprogs.sourceforge.net/ 608 609useful links: http://fedoraproject.org/wiki/ext3-devel 610 http://www.bullopensource.org/ext4/ 611 http://ext4.wiki.kernel.org/index.php/Main_Page 612 http://fedoraproject.org/wiki/Features/Ext4 613