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