1.. SPDX-License-Identifier: GPL-2.0 2 3================================= 4Flash-Friendly File System (F2FS) 5================================= 6 7Overview 8======== 9 10NAND flash memory-based storage devices, such as SSD, eMMC, and SD cards, have 11been equipped on a variety systems ranging from mobile to server systems. Since 12they are known to have different characteristics from the conventional rotating 13disks, a file system, an upper layer to the storage device, should adapt to the 14changes from the sketch in the design level. 15 16F2FS is a file system exploiting NAND flash memory-based storage devices, which 17is based on Log-structured File System (LFS). The design has been focused on 18addressing the fundamental issues in LFS, which are snowball effect of wandering 19tree and high cleaning overhead. 20 21Since a NAND flash memory-based storage device shows different characteristic 22according to its internal geometry or flash memory management scheme, namely FTL, 23F2FS and its tools support various parameters not only for configuring on-disk 24layout, but also for selecting allocation and cleaning algorithms. 25 26The following git tree provides the file system formatting tool (mkfs.f2fs), 27a consistency checking tool (fsck.f2fs), and a debugging tool (dump.f2fs). 28 29- git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs-tools.git 30 31For sending patches, please use the following mailing list: 32 33- linux-f2fs-devel@lists.sourceforge.net 34 35For reporting bugs, please use the following f2fs bug tracker link: 36 37- https://bugzilla.kernel.org/enter_bug.cgi?product=File%20System&component=f2fs 38 39Background and Design issues 40============================ 41 42Log-structured File System (LFS) 43-------------------------------- 44"A log-structured file system writes all modifications to disk sequentially in 45a log-like structure, thereby speeding up both file writing and crash recovery. 46The log is the only structure on disk; it contains indexing information so that 47files can be read back from the log efficiently. In order to maintain large free 48areas on disk for fast writing, we divide the log into segments and use a 49segment cleaner to compress the live information from heavily fragmented 50segments." from Rosenblum, M. and Ousterhout, J. K., 1992, "The design and 51implementation of a log-structured file system", ACM Trans. Computer Systems 5210, 1, 26–52. 53 54Wandering Tree Problem 55---------------------- 56In LFS, when a file data is updated and written to the end of log, its direct 57pointer block is updated due to the changed location. Then the indirect pointer 58block is also updated due to the direct pointer block update. In this manner, 59the upper index structures such as inode, inode map, and checkpoint block are 60also updated recursively. This problem is called as wandering tree problem [1], 61and in order to enhance the performance, it should eliminate or relax the update 62propagation as much as possible. 63 64[1] Bityutskiy, A. 2005. JFFS3 design issues. http://www.linux-mtd.infradead.org/ 65 66Cleaning Overhead 67----------------- 68Since LFS is based on out-of-place writes, it produces so many obsolete blocks 69scattered across the whole storage. In order to serve new empty log space, it 70needs to reclaim these obsolete blocks seamlessly to users. This job is called 71as a cleaning process. 72 73The process consists of three operations as follows. 74 751. A victim segment is selected through referencing segment usage table. 762. It loads parent index structures of all the data in the victim identified by 77 segment summary blocks. 783. It checks the cross-reference between the data and its parent index structure. 794. It moves valid data selectively. 80 81This cleaning job may cause unexpected long delays, so the most important goal 82is to hide the latencies to users. And also definitely, it should reduce the 83amount of valid data to be moved, and move them quickly as well. 84 85Key Features 86============ 87 88Flash Awareness 89--------------- 90- Enlarge the random write area for better performance, but provide the high 91 spatial locality 92- Align FS data structures to the operational units in FTL as best efforts 93 94Wandering Tree Problem 95---------------------- 96- Use a term, “node”, that represents inodes as well as various pointer blocks 97- Introduce Node Address Table (NAT) containing the locations of all the “node” 98 blocks; this will cut off the update propagation. 99 100Cleaning Overhead 101----------------- 102- Support a background cleaning process 103- Support greedy and cost-benefit algorithms for victim selection policies 104- Support multi-head logs for static/dynamic hot and cold data separation 105- Introduce adaptive logging for efficient block allocation 106 107Mount Options 108============= 109 110 111======================== ============================================================ 112background_gc=%s Turn on/off cleaning operations, namely garbage 113 collection, triggered in background when I/O subsystem is 114 idle. If background_gc=on, it will turn on the garbage 115 collection and if background_gc=off, garbage collection 116 will be turned off. If background_gc=sync, it will turn 117 on synchronous garbage collection running in background. 118 Default value for this option is on. So garbage 119 collection is on by default. 120gc_merge When background_gc is on, this option can be enabled to 121 let background GC thread to handle foreground GC requests, 122 it can eliminate the sluggish issue caused by slow foreground 123 GC operation when GC is triggered from a process with limited 124 I/O and CPU resources. 125nogc_merge Disable GC merge feature. 126disable_roll_forward Disable the roll-forward recovery routine 127norecovery Disable the roll-forward recovery routine, mounted read- 128 only (i.e., -o ro,disable_roll_forward) 129discard/nodiscard Enable/disable real-time discard in f2fs, if discard is 130 enabled, f2fs will issue discard/TRIM commands when a 131 segment is cleaned. 132heap/no_heap Deprecated. 133nouser_xattr Disable Extended User Attributes. Note: xattr is enabled 134 by default if CONFIG_F2FS_FS_XATTR is selected. 135noacl Disable POSIX Access Control List. Note: acl is enabled 136 by default if CONFIG_F2FS_FS_POSIX_ACL is selected. 137active_logs=%u Support configuring the number of active logs. In the 138 current design, f2fs supports only 2, 4, and 6 logs. 139 Default number is 6. 140disable_ext_identify Disable the extension list configured by mkfs, so f2fs 141 is not aware of cold files such as media files. 142inline_xattr Enable the inline xattrs feature. 143noinline_xattr Disable the inline xattrs feature. 144inline_xattr_size=%u Support configuring inline xattr size, it depends on 145 flexible inline xattr feature. 146inline_data Enable the inline data feature: Newly created small (<~3.4k) 147 files can be written into inode block. 148inline_dentry Enable the inline dir feature: data in newly created 149 directory entries can be written into inode block. The 150 space of inode block which is used to store inline 151 dentries is limited to ~3.4k. 152noinline_dentry Disable the inline dentry feature. 153flush_merge Merge concurrent cache_flush commands as much as possible 154 to eliminate redundant command issues. If the underlying 155 device handles the cache_flush command relatively slowly, 156 recommend to enable this option. 157nobarrier This option can be used if underlying storage guarantees 158 its cached data should be written to the novolatile area. 159 If this option is set, no cache_flush commands are issued 160 but f2fs still guarantees the write ordering of all the 161 data writes. 162barrier If this option is set, cache_flush commands are allowed to be 163 issued. 164fastboot This option is used when a system wants to reduce mount 165 time as much as possible, even though normal performance 166 can be sacrificed. 167extent_cache Enable an extent cache based on rb-tree, it can cache 168 as many as extent which map between contiguous logical 169 address and physical address per inode, resulting in 170 increasing the cache hit ratio. Set by default. 171noextent_cache Disable an extent cache based on rb-tree explicitly, see 172 the above extent_cache mount option. 173noinline_data Disable the inline data feature, inline data feature is 174 enabled by default. 175data_flush Enable data flushing before checkpoint in order to 176 persist data of regular and symlink. 177reserve_root=%d Support configuring reserved space which is used for 178 allocation from a privileged user with specified uid or 179 gid, unit: 4KB, the default limit is 12.5% of user blocks. 180reserve_node=%d Support configuring reserved nodes which are used for 181 allocation from a privileged user with specified uid or 182 gid, the default limit is 12.5% of all nodes. 183resuid=%d The user ID which may use the reserved blocks and nodes. 184resgid=%d The group ID which may use the reserved blocks and nodes. 185fault_injection=%d Enable fault injection in all supported types with 186 specified injection rate. 187fault_type=%d Support configuring fault injection type, should be 188 enabled with fault_injection option, fault type value 189 is shown below, it supports single or combined type. 190 191 =========================== ========== 192 Type_Name Type_Value 193 =========================== ========== 194 FAULT_KMALLOC 0x00000001 195 FAULT_KVMALLOC 0x00000002 196 FAULT_PAGE_ALLOC 0x00000004 197 FAULT_PAGE_GET 0x00000008 198 FAULT_ALLOC_BIO 0x00000010 (obsolete) 199 FAULT_ALLOC_NID 0x00000020 200 FAULT_ORPHAN 0x00000040 201 FAULT_BLOCK 0x00000080 202 FAULT_DIR_DEPTH 0x00000100 203 FAULT_EVICT_INODE 0x00000200 204 FAULT_TRUNCATE 0x00000400 205 FAULT_READ_IO 0x00000800 206 FAULT_CHECKPOINT 0x00001000 207 FAULT_DISCARD 0x00002000 208 FAULT_WRITE_IO 0x00004000 209 FAULT_SLAB_ALLOC 0x00008000 210 FAULT_DQUOT_INIT 0x00010000 211 FAULT_LOCK_OP 0x00020000 212 FAULT_BLKADDR_VALIDITY 0x00040000 213 FAULT_BLKADDR_CONSISTENCE 0x00080000 214 FAULT_NO_SEGMENT 0x00100000 215 FAULT_INCONSISTENT_FOOTER 0x00200000 216 FAULT_TIMEOUT 0x00400000 (1000ms) 217 FAULT_VMALLOC 0x00800000 218 =========================== ========== 219mode=%s Control block allocation mode which supports "adaptive" 220 and "lfs". In "lfs" mode, there should be no random 221 writes towards main area. 222 "fragment:segment" and "fragment:block" are newly added here. 223 These are developer options for experiments to simulate filesystem 224 fragmentation/after-GC situation itself. The developers use these 225 modes to understand filesystem fragmentation/after-GC condition well, 226 and eventually get some insights to handle them better. 227 In "fragment:segment", f2fs allocates a new segment in random 228 position. With this, we can simulate the after-GC condition. 229 In "fragment:block", we can scatter block allocation with 230 "max_fragment_chunk" and "max_fragment_hole" sysfs nodes. 231 We added some randomness to both chunk and hole size to make 232 it close to realistic IO pattern. So, in this mode, f2fs will allocate 233 1..<max_fragment_chunk> blocks in a chunk and make a hole in the 234 length of 1..<max_fragment_hole> by turns. With this, the newly 235 allocated blocks will be scattered throughout the whole partition. 236 Note that "fragment:block" implicitly enables "fragment:segment" 237 option for more randomness. 238 Please, use these options for your experiments and we strongly 239 recommend to re-format the filesystem after using these options. 240usrquota Enable plain user disk quota accounting. 241grpquota Enable plain group disk quota accounting. 242prjquota Enable plain project quota accounting. 243usrjquota=<file> Appoint specified file and type during mount, so that quota 244grpjquota=<file> information can be properly updated during recovery flow, 245prjjquota=<file> <quota file>: must be in root directory; 246jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1]. 247usrjquota= Turn off user journalled quota. 248grpjquota= Turn off group journalled quota. 249prjjquota= Turn off project journalled quota. 250quota Enable plain user disk quota accounting. 251noquota Disable all plain disk quota option. 252alloc_mode=%s Adjust block allocation policy, which supports "reuse" 253 and "default". 254fsync_mode=%s Control the policy of fsync. Currently supports "posix", 255 "strict", and "nobarrier". In "posix" mode, which is 256 default, fsync will follow POSIX semantics and does a 257 light operation to improve the filesystem performance. 258 In "strict" mode, fsync will be heavy and behaves in line 259 with xfs, ext4 and btrfs, where xfstest generic/342 will 260 pass, but the performance will regress. "nobarrier" is 261 based on "posix", but doesn't issue flush command for 262 non-atomic files likewise "nobarrier" mount option. 263test_dummy_encryption 264test_dummy_encryption=%s 265 Enable dummy encryption, which provides a fake fscrypt 266 context. The fake fscrypt context is used by xfstests. 267 The argument may be either "v1" or "v2", in order to 268 select the corresponding fscrypt policy version. 269checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable" 270 to re-enable checkpointing. Is enabled by default. While 271 disabled, any unmounting or unexpected shutdowns will cause 272 the filesystem contents to appear as they did when the 273 filesystem was mounted with that option. 274 While mounting with checkpoint=disable, the filesystem must 275 run garbage collection to ensure that all available space can 276 be used. If this takes too much time, the mount may return 277 EAGAIN. You may optionally add a value to indicate how much 278 of the disk you would be willing to temporarily give up to 279 avoid additional garbage collection. This can be given as a 280 number of blocks, or as a percent. For instance, mounting 281 with checkpoint=disable:100% would always succeed, but it may 282 hide up to all remaining free space. The actual space that 283 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable 284 This space is reclaimed once checkpoint=enable. 285checkpoint_merge When checkpoint is enabled, this can be used to create a kernel 286 daemon and make it to merge concurrent checkpoint requests as 287 much as possible to eliminate redundant checkpoint issues. Plus, 288 we can eliminate the sluggish issue caused by slow checkpoint 289 operation when the checkpoint is done in a process context in 290 a cgroup having low i/o budget and cpu shares. To make this 291 do better, we set the default i/o priority of the kernel daemon 292 to "3", to give one higher priority than other kernel threads. 293 This is the same way to give a I/O priority to the jbd2 294 journaling thread of ext4 filesystem. 295nocheckpoint_merge Disable checkpoint merge feature. 296compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo", 297 "lz4", "zstd" and "lzo-rle" algorithm. 298compress_algorithm=%s:%d Control compress algorithm and its compress level, now, only 299 "lz4" and "zstd" support compress level config. 300 301 ========= =========== 302 algorithm level range 303 ========= =========== 304 lz4 3 - 16 305 zstd 1 - 22 306 ========= =========== 307compress_log_size=%u Support configuring compress cluster size. The size will 308 be 4KB * (1 << %u). The default and minimum sizes are 16KB. 309compress_extension=%s Support adding specified extension, so that f2fs can enable 310 compression on those corresponding files, e.g. if all files 311 with '.ext' has high compression rate, we can set the '.ext' 312 on compression extension list and enable compression on 313 these file by default rather than to enable it via ioctl. 314 For other files, we can still enable compression via ioctl. 315 Note that, there is one reserved special extension '*', it 316 can be set to enable compression for all files. 317nocompress_extension=%s Support adding specified extension, so that f2fs can disable 318 compression on those corresponding files, just contrary to compression extension. 319 If you know exactly which files cannot be compressed, you can use this. 320 The same extension name can't appear in both compress and nocompress 321 extension at the same time. 322 If the compress extension specifies all files, the types specified by the 323 nocompress extension will be treated as special cases and will not be compressed. 324 Don't allow use '*' to specifie all file in nocompress extension. 325 After add nocompress_extension, the priority should be: 326 dir_flag < comp_extention,nocompress_extension < comp_file_flag,no_comp_file_flag. 327 See more in compression sections. 328 329compress_chksum Support verifying chksum of raw data in compressed cluster. 330compress_mode=%s Control file compression mode. This supports "fs" and "user" 331 modes. In "fs" mode (default), f2fs does automatic compression 332 on the compression enabled files. In "user" mode, f2fs disables 333 the automaic compression and gives the user discretion of 334 choosing the target file and the timing. The user can do manual 335 compression/decompression on the compression enabled files using 336 ioctls. 337compress_cache Support to use address space of a filesystem managed inode to 338 cache compressed block, in order to improve cache hit ratio of 339 random read. 340inlinecrypt When possible, encrypt/decrypt the contents of encrypted 341 files using the blk-crypto framework rather than 342 filesystem-layer encryption. This allows the use of 343 inline encryption hardware. The on-disk format is 344 unaffected. For more details, see 345 Documentation/block/inline-encryption.rst. 346atgc Enable age-threshold garbage collection, it provides high 347 effectiveness and efficiency on background GC. 348discard_unit=%s Control discard unit, the argument can be "block", "segment" 349 and "section", issued discard command's offset/size will be 350 aligned to the unit, by default, "discard_unit=block" is set, 351 so that small discard functionality is enabled. 352 For blkzoned device, "discard_unit=section" will be set by 353 default, it is helpful for large sized SMR or ZNS devices to 354 reduce memory cost by getting rid of fs metadata supports small 355 discard. 356memory=%s Control memory mode. This supports "normal" and "low" modes. 357 "low" mode is introduced to support low memory devices. 358 Because of the nature of low memory devices, in this mode, f2fs 359 will try to save memory sometimes by sacrificing performance. 360 "normal" mode is the default mode and same as before. 361age_extent_cache Enable an age extent cache based on rb-tree. It records 362 data block update frequency of the extent per inode, in 363 order to provide better temperature hints for data block 364 allocation. 365errors=%s Specify f2fs behavior on critical errors. This supports modes: 366 "panic", "continue" and "remount-ro", respectively, trigger 367 panic immediately, continue without doing anything, and remount 368 the partition in read-only mode. By default it uses "continue" 369 mode. 370 371 ====================== =============== =============== ======== 372 mode continue remount-ro panic 373 ====================== =============== =============== ======== 374 access ops normal normal N/A 375 syscall errors -EIO -EROFS N/A 376 mount option rw ro N/A 377 pending dir write keep keep N/A 378 pending non-dir write drop keep N/A 379 pending node write drop keep N/A 380 pending meta write keep keep N/A 381 ====================== =============== =============== ======== 382nat_bits Enable nat_bits feature to enhance full/empty nat blocks access, 383 by default it's disabled. 384lookup_mode=%s Control the directory lookup behavior for casefolded 385 directories. This option has no effect on directories 386 that do not have the casefold feature enabled. 387 388 ================== ======================================== 389 Value Description 390 ================== ======================================== 391 perf (Default) Enforces a hash-only lookup. 392 The linear search fallback is always 393 disabled, ignoring the on-disk flag. 394 compat Enables the linear search fallback for 395 compatibility with directory entries 396 created by older kernel that used a 397 different case-folding algorithm. 398 This mode ignores the on-disk flag. 399 auto F2FS determines the mode based on the 400 on-disk `SB_ENC_NO_COMPAT_FALLBACK_FL` 401 flag. 402 ================== ======================================== 403======================== ============================================================ 404 405Debugfs Entries 406=============== 407 408/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as 409f2fs. Each file shows the whole f2fs information. 410 411/sys/kernel/debug/f2fs/status includes: 412 413 - major file system information managed by f2fs currently 414 - average SIT information about whole segments 415 - current memory footprint consumed by f2fs. 416 417Sysfs Entries 418============= 419 420Information about mounted f2fs file systems can be found in 421/sys/fs/f2fs. Each mounted filesystem will have a directory in 422/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda). 423The files in each per-device directory are shown in table below. 424 425Files in /sys/fs/f2fs/<devname> 426(see also Documentation/ABI/testing/sysfs-fs-f2fs) 427 428Usage 429===== 430 4311. Download userland tools and compile them. 432 4332. Skip, if f2fs was compiled statically inside kernel. 434 Otherwise, insert the f2fs.ko module:: 435 436 # insmod f2fs.ko 437 4383. Create a directory to use when mounting:: 439 440 # mkdir /mnt/f2fs 441 4424. Format the block device, and then mount as f2fs:: 443 444 # mkfs.f2fs -l label /dev/block_device 445 # mount -t f2fs /dev/block_device /mnt/f2fs 446 447mkfs.f2fs 448--------- 449The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem, 450which builds a basic on-disk layout. 451 452The quick options consist of: 453 454=============== =========================================================== 455``-l [label]`` Give a volume label, up to 512 unicode name. 456``-a [0 or 1]`` Split start location of each area for heap-based allocation. 457 458 1 is set by default, which performs this. 459``-o [int]`` Set overprovision ratio in percent over volume size. 460 461 5 is set by default. 462``-s [int]`` Set the number of segments per section. 463 464 1 is set by default. 465``-z [int]`` Set the number of sections per zone. 466 467 1 is set by default. 468``-e [str]`` Set basic extension list. e.g. "mp3,gif,mov" 469``-t [0 or 1]`` Disable discard command or not. 470 471 1 is set by default, which conducts discard. 472=============== =========================================================== 473 474Note: please refer to the manpage of mkfs.f2fs(8) to get full option list. 475 476fsck.f2fs 477--------- 478The fsck.f2fs is a tool to check the consistency of an f2fs-formatted 479partition, which examines whether the filesystem metadata and user-made data 480are cross-referenced correctly or not. 481Note that, initial version of the tool does not fix any inconsistency. 482 483The quick options consist of:: 484 485 -d debug level [default:0] 486 487Note: please refer to the manpage of fsck.f2fs(8) to get full option list. 488 489dump.f2fs 490--------- 491The dump.f2fs shows the information of specific inode and dumps SSA and SIT to 492file. Each file is dump_ssa and dump_sit. 493 494The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem. 495It shows on-disk inode information recognized by a given inode number, and is 496able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and 497./dump_sit respectively. 498 499The options consist of:: 500 501 -d debug level [default:0] 502 -i inode no (hex) 503 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1] 504 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1] 505 506Examples:: 507 508 # dump.f2fs -i [ino] /dev/sdx 509 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 510 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 511 512Note: please refer to the manpage of dump.f2fs(8) to get full option list. 513 514sload.f2fs 515---------- 516The sload.f2fs gives a way to insert files and directories in the existing disk 517image. This tool is useful when building f2fs images given compiled files. 518 519Note: please refer to the manpage of sload.f2fs(8) to get full option list. 520 521resize.f2fs 522----------- 523The resize.f2fs lets a user resize the f2fs-formatted disk image, while preserving 524all the files and directories stored in the image. 525 526Note: please refer to the manpage of resize.f2fs(8) to get full option list. 527 528defrag.f2fs 529----------- 530The defrag.f2fs can be used to defragment scattered written data as well as 531filesystem metadata across the disk. This can improve the write speed by giving 532more free consecutive space. 533 534Note: please refer to the manpage of defrag.f2fs(8) to get full option list. 535 536f2fs_io 537------- 538The f2fs_io is a simple tool to issue various filesystem APIs as well as 539f2fs-specific ones, which is very useful for QA tests. 540 541Note: please refer to the manpage of f2fs_io(8) to get full option list. 542 543Design 544====== 545 546On-disk Layout 547-------------- 548 549F2FS divides the whole volume into a number of segments, each of which is fixed 550to 2MB in size. A section is composed of consecutive segments, and a zone 551consists of a set of sections. By default, section and zone sizes are set to one 552segment size identically, but users can easily modify the sizes by mkfs. 553 554F2FS splits the entire volume into six areas, and all the areas except superblock 555consist of multiple segments as described below:: 556 557 align with the zone size <-| 558 |-> align with the segment size 559 _________________________________________________________________________ 560 | | | Segment | Node | Segment | | 561 | Superblock | Checkpoint | Info. | Address | Summary | Main | 562 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | | 563 |____________|_____2______|______N______|______N______|______N_____|__N___| 564 . . 565 . . 566 . . 567 ._________________________________________. 568 |_Segment_|_..._|_Segment_|_..._|_Segment_| 569 . . 570 ._________._________ 571 |_section_|__...__|_ 572 . . 573 .________. 574 |__zone__| 575 576- Superblock (SB) 577 It is located at the beginning of the partition, and there exist two copies 578 to avoid file system crash. It contains basic partition information and some 579 default parameters of f2fs. 580 581- Checkpoint (CP) 582 It contains file system information, bitmaps for valid NAT/SIT sets, orphan 583 inode lists, and summary entries of current active segments. 584 585- Segment Information Table (SIT) 586 It contains segment information such as valid block count and bitmap for the 587 validity of all the blocks. 588 589- Node Address Table (NAT) 590 It is composed of a block address table for all the node blocks stored in 591 Main area. 592 593- Segment Summary Area (SSA) 594 It contains summary entries which contains the owner information of all the 595 data and node blocks stored in Main area. 596 597- Main Area 598 It contains file and directory data including their indices. 599 600In order to avoid misalignment between file system and flash-based storage, F2FS 601aligns the start block address of CP with the segment size. Also, it aligns the 602start block address of Main area with the zone size by reserving some segments 603in SSA area. 604 605Reference the following survey for additional technical details. 606https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey 607 608File System Metadata Structure 609------------------------------ 610 611F2FS adopts the checkpointing scheme to maintain file system consistency. At 612mount time, F2FS first tries to find the last valid checkpoint data by scanning 613CP area. In order to reduce the scanning time, F2FS uses only two copies of CP. 614One of them always indicates the last valid data, which is called as shadow copy 615mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism. 616 617For file system consistency, each CP points to which NAT and SIT copies are 618valid, as shown as below:: 619 620 +--------+----------+---------+ 621 | CP | SIT | NAT | 622 +--------+----------+---------+ 623 . . . . 624 . . . . 625 . . . . 626 +-------+-------+--------+--------+--------+--------+ 627 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 | 628 +-------+-------+--------+--------+--------+--------+ 629 | ^ ^ 630 | | | 631 `----------------------------------------' 632 633Index Structure 634--------------- 635 636The key data structure to manage the data locations is a "node". Similar to 637traditional file structures, F2FS has three types of node: inode, direct node, 638indirect node. F2FS assigns 4KB to an inode block which contains 923 data block 639indices, two direct node pointers, two indirect node pointers, and one double 640indirect node pointer as described below. One direct node block contains 1018 641data blocks, and one indirect node block contains also 1018 node blocks. Thus, 642one inode block (i.e., a file) covers:: 643 644 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB. 645 646 Inode block (4KB) 647 |- data (923) 648 |- direct node (2) 649 | `- data (1018) 650 |- indirect node (2) 651 | `- direct node (1018) 652 | `- data (1018) 653 `- double indirect node (1) 654 `- indirect node (1018) 655 `- direct node (1018) 656 `- data (1018) 657 658Note that all the node blocks are mapped by NAT which means the location of 659each node is translated by the NAT table. In the consideration of the wandering 660tree problem, F2FS is able to cut off the propagation of node updates caused by 661leaf data writes. 662 663Directory Structure 664------------------- 665 666A directory entry occupies 11 bytes, which consists of the following attributes. 667 668- hash hash value of the file name 669- ino inode number 670- len the length of file name 671- type file type such as directory, symlink, etc 672 673A dentry block consists of 214 dentry slots and file names. Therein a bitmap is 674used to represent whether each dentry is valid or not. A dentry block occupies 6754KB with the following composition. 676 677:: 678 679 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) + 680 dentries(11 * 214 bytes) + file name (8 * 214 bytes) 681 682 [Bucket] 683 +--------------------------------+ 684 |dentry block 1 | dentry block 2 | 685 +--------------------------------+ 686 . . 687 . . 688 . [Dentry Block Structure: 4KB] . 689 +--------+----------+----------+------------+ 690 | bitmap | reserved | dentries | file names | 691 +--------+----------+----------+------------+ 692 [Dentry Block: 4KB] . . 693 . . 694 . . 695 +------+------+-----+------+ 696 | hash | ino | len | type | 697 +------+------+-----+------+ 698 [Dentry Structure: 11 bytes] 699 700F2FS implements multi-level hash tables for directory structure. Each level has 701a hash table with dedicated number of hash buckets as shown below. Note that 702"A(2B)" means a bucket includes 2 data blocks. 703 704:: 705 706 ---------------------- 707 A : bucket 708 B : block 709 N : MAX_DIR_HASH_DEPTH 710 ---------------------- 711 712 level #0 | A(2B) 713 | 714 level #1 | A(2B) - A(2B) 715 | 716 level #2 | A(2B) - A(2B) - A(2B) - A(2B) 717 . | . . . . 718 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) 719 . | . . . . 720 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) 721 722The number of blocks and buckets are determined by:: 723 724 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2, 725 # of blocks in level #n = | 726 `- 4, Otherwise 727 728 ,- 2^(n + dir_level), 729 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2, 730 # of buckets in level #n = | 731 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), 732 Otherwise 733 734When F2FS finds a file name in a directory, at first a hash value of the file 735name is calculated. Then, F2FS scans the hash table in level #0 to find the 736dentry consisting of the file name and its inode number. If not found, F2FS 737scans the next hash table in level #1. In this way, F2FS scans hash tables in 738each levels incrementally from 1 to N. In each level F2FS needs to scan only 739one bucket determined by the following equation, which shows O(log(# of files)) 740complexity:: 741 742 bucket number to scan in level #n = (hash value) % (# of buckets in level #n) 743 744In the case of file creation, F2FS finds empty consecutive slots that cover the 745file name. F2FS searches the empty slots in the hash tables of whole levels from 7461 to N in the same way as the lookup operation. 747 748The following figure shows an example of two cases holding children:: 749 750 --------------> Dir <-------------- 751 | | 752 child child 753 754 child - child [hole] - child 755 756 child - child - child [hole] - [hole] - child 757 758 Case 1: Case 2: 759 Number of children = 6, Number of children = 3, 760 File size = 7 File size = 7 761 762Default Block Allocation 763------------------------ 764 765At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node 766and Hot/Warm/Cold data. 767 768- Hot node contains direct node blocks of directories. 769- Warm node contains direct node blocks except hot node blocks. 770- Cold node contains indirect node blocks 771- Hot data contains dentry blocks 772- Warm data contains data blocks except hot and cold data blocks 773- Cold data contains multimedia data or migrated data blocks 774 775LFS has two schemes for free space management: threaded log and copy-and-compac- 776tion. The copy-and-compaction scheme which is known as cleaning, is well-suited 777for devices showing very good sequential write performance, since free segments 778are served all the time for writing new data. However, it suffers from cleaning 779overhead under high utilization. Contrarily, the threaded log scheme suffers 780from random writes, but no cleaning process is needed. F2FS adopts a hybrid 781scheme where the copy-and-compaction scheme is adopted by default, but the 782policy is dynamically changed to the threaded log scheme according to the file 783system status. 784 785In order to align F2FS with underlying flash-based storage, F2FS allocates a 786segment in a unit of section. F2FS expects that the section size would be the 787same as the unit size of garbage collection in FTL. Furthermore, with respect 788to the mapping granularity in FTL, F2FS allocates each section of the active 789logs from different zones as much as possible, since FTL can write the data in 790the active logs into one allocation unit according to its mapping granularity. 791 792Cleaning process 793---------------- 794 795F2FS does cleaning both on demand and in the background. On-demand cleaning is 796triggered when there are not enough free segments to serve VFS calls. Background 797cleaner is operated by a kernel thread, and triggers the cleaning job when the 798system is idle. 799 800F2FS supports two victim selection policies: greedy and cost-benefit algorithms. 801In the greedy algorithm, F2FS selects a victim segment having the smallest number 802of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment 803according to the segment age and the number of valid blocks in order to address 804log block thrashing problem in the greedy algorithm. F2FS adopts the greedy 805algorithm for on-demand cleaner, while background cleaner adopts cost-benefit 806algorithm. 807 808In order to identify whether the data in the victim segment are valid or not, 809F2FS manages a bitmap. Each bit represents the validity of a block, and the 810bitmap is composed of a bit stream covering whole blocks in main area. 811 812Write-hint Policy 813----------------- 814 815F2FS sets the whint all the time with the below policy. 816 817===================== ======================== =================== 818User F2FS Block 819===================== ======================== =================== 820N/A META WRITE_LIFE_NONE|REQ_META 821N/A HOT_NODE WRITE_LIFE_NONE 822N/A WARM_NODE WRITE_LIFE_MEDIUM 823N/A COLD_NODE WRITE_LIFE_LONG 824ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME 825extension list " " 826 827-- buffered io 828------------------------------------------------------------------ 829N/A COLD_DATA WRITE_LIFE_EXTREME 830N/A HOT_DATA WRITE_LIFE_SHORT 831N/A WARM_DATA WRITE_LIFE_NOT_SET 832 833-- direct io 834------------------------------------------------------------------ 835WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 836WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 837WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 838WRITE_LIFE_NONE " WRITE_LIFE_NONE 839WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM 840WRITE_LIFE_LONG " WRITE_LIFE_LONG 841===================== ======================== =================== 842 843Fallocate(2) Policy 844------------------- 845 846The default policy follows the below POSIX rule. 847 848Allocating disk space 849 The default operation (i.e., mode is zero) of fallocate() allocates 850 the disk space within the range specified by offset and len. The 851 file size (as reported by stat(2)) will be changed if offset+len is 852 greater than the file size. Any subregion within the range specified 853 by offset and len that did not contain data before the call will be 854 initialized to zero. This default behavior closely resembles the 855 behavior of the posix_fallocate(3) library function, and is intended 856 as a method of optimally implementing that function. 857 858However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to 859fallocate(fd, DEFAULT_MODE), it allocates on-disk block addresses having 860zero or random data, which is useful to the below scenario where: 861 862 1. create(fd) 863 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 864 3. fallocate(fd, 0, 0, size) 865 4. address = fibmap(fd, offset) 866 5. open(blkdev) 867 6. write(blkdev, address) 868 869Compression implementation 870-------------------------- 871 872- New term named cluster is defined as basic unit of compression, file can 873 be divided into multiple clusters logically. One cluster includes 4 << n 874 (n >= 0) logical pages, compression size is also cluster size, each of 875 cluster can be compressed or not. 876 877- In cluster metadata layout, one special block address is used to indicate 878 a cluster is a compressed one or normal one; for compressed cluster, following 879 metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs 880 stores data including compress header and compressed data. 881 882- In order to eliminate write amplification during overwrite, F2FS only 883 support compression on write-once file, data can be compressed only when 884 all logical blocks in cluster contain valid data and compress ratio of 885 cluster data is lower than specified threshold. 886 887- To enable compression on regular inode, there are four ways: 888 889 * chattr +c file 890 * chattr +c dir; touch dir/file 891 * mount w/ -o compress_extension=ext; touch file.ext 892 * mount w/ -o compress_extension=*; touch any_file 893 894- To disable compression on regular inode, there are two ways: 895 896 * chattr -c file 897 * mount w/ -o nocompress_extension=ext; touch file.ext 898 899- Priority in between FS_COMPR_FL, FS_NOCOMP_FS, extensions: 900 901 * compress_extension=so; nocompress_extension=zip; chattr +c dir; touch 902 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so and baz.txt 903 should be compresse, bar.zip should be non-compressed. chattr +c dir/bar.zip 904 can enable compress on bar.zip. 905 * compress_extension=so; nocompress_extension=zip; chattr -c dir; touch 906 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so should be 907 compresse, bar.zip and baz.txt should be non-compressed. 908 chattr+c dir/bar.zip; chattr+c dir/baz.txt; can enable compress on bar.zip 909 and baz.txt. 910 911- At this point, compression feature doesn't expose compressed space to user 912 directly in order to guarantee potential data updates later to the space. 913 Instead, the main goal is to reduce data writes to flash disk as much as 914 possible, resulting in extending disk life time as well as relaxing IO 915 congestion. Alternatively, we've added ioctl(F2FS_IOC_RELEASE_COMPRESS_BLOCKS) 916 interface to reclaim compressed space and show it to user after setting a 917 special flag to the inode. Once the compressed space is released, the flag 918 will block writing data to the file until either the compressed space is 919 reserved via ioctl(F2FS_IOC_RESERVE_COMPRESS_BLOCKS) or the file size is 920 truncated to zero. 921 922Compress metadata layout:: 923 924 [Dnode Structure] 925 +-----------------------------------------------+ 926 | cluster 1 | cluster 2 | ......... | cluster N | 927 +-----------------------------------------------+ 928 . . . . 929 . . . . 930 . Compressed Cluster . . Normal Cluster . 931 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 932 |compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 | 933 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 934 . . 935 . . 936 . . 937 +-------------+-------------+----------+----------------------------+ 938 | data length | data chksum | reserved | compressed data | 939 +-------------+-------------+----------+----------------------------+ 940 941Compression mode 942-------------------------- 943 944f2fs supports "fs" and "user" compression modes with "compression_mode" mount option. 945With this option, f2fs provides a choice to select the way how to compress the 946compression enabled files (refer to "Compression implementation" section for how to 947enable compression on a regular inode). 948 9491) compress_mode=fs 950 951 This is the default option. f2fs does automatic compression in the writeback of the 952 compression enabled files. 953 9542) compress_mode=user 955 956 This disables the automatic compression and gives the user discretion of choosing the 957 target file and the timing. The user can do manual compression/decompression on the 958 compression enabled files using F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE 959 ioctls like the below. 960 961To decompress a file:: 962 963 fd = open(filename, O_WRONLY, 0); 964 ret = ioctl(fd, F2FS_IOC_DECOMPRESS_FILE); 965 966To compress a file:: 967 968 fd = open(filename, O_WRONLY, 0); 969 ret = ioctl(fd, F2FS_IOC_COMPRESS_FILE); 970 971NVMe Zoned Namespace devices 972---------------------------- 973 974- ZNS defines a per-zone capacity which can be equal or less than the 975 zone-size. Zone-capacity is the number of usable blocks in the zone. 976 F2FS checks if zone-capacity is less than zone-size, if it is, then any 977 segment which starts after the zone-capacity is marked as not-free in 978 the free segment bitmap at initial mount time. These segments are marked 979 as permanently used so they are not allocated for writes and 980 consequently are not needed to be garbage collected. In case the 981 zone-capacity is not aligned to default segment size(2MB), then a segment 982 can start before the zone-capacity and span across zone-capacity boundary. 983 Such spanning segments are also considered as usable segments. All blocks 984 past the zone-capacity are considered unusable in these segments. 985 986Device aliasing feature 987----------------------- 988 989f2fs can utilize a special file called a "device aliasing file." This file allows 990the entire storage device to be mapped with a single, large extent, not using 991the usual f2fs node structures. This mapped area is pinned and primarily intended 992for holding the space. 993 994Essentially, this mechanism allows a portion of the f2fs area to be temporarily 995reserved and used by another filesystem or for different purposes. Once that 996external usage is complete, the device aliasing file can be deleted, releasing 997the reserved space back to F2FS for its own use. 998 999.. code-block:: 1000 1001 # ls /dev/vd* 1002 /dev/vdb (32GB) /dev/vdc (32GB) 1003 # mkfs.ext4 /dev/vdc 1004 # mkfs.f2fs -c /dev/vdc@vdc.file /dev/vdb 1005 # mount /dev/vdb /mnt/f2fs 1006 # ls -l /mnt/f2fs 1007 vdc.file 1008 # df -h 1009 /dev/vdb 64G 33G 32G 52% /mnt/f2fs 1010 1011 # mount -o loop /dev/vdc /mnt/ext4 1012 # df -h 1013 /dev/vdb 64G 33G 32G 52% /mnt/f2fs 1014 /dev/loop7 32G 24K 30G 1% /mnt/ext4 1015 # umount /mnt/ext4 1016 1017 # f2fs_io getflags /mnt/f2fs/vdc.file 1018 get a flag on /mnt/f2fs/vdc.file ret=0, flags=nocow(pinned),immutable 1019 # f2fs_io setflags noimmutable /mnt/f2fs/vdc.file 1020 get a flag on noimmutable ret=0, flags=800010 1021 set a flag on /mnt/f2fs/vdc.file ret=0, flags=noimmutable 1022 # rm /mnt/f2fs/vdc.file 1023 # df -h 1024 /dev/vdb 64G 753M 64G 2% /mnt/f2fs 1025 1026So, the key idea is, user can do any file operations on /dev/vdc, and 1027reclaim the space after the use, while the space is counted as /data. 1028That doesn't require modifying partition size and filesystem format. 1029