1.. SPDX-License-Identifier: GPL-2.0 2 3========================================== 4WHAT IS Flash-Friendly File System (F2FS)? 5========================================== 6 7NAND flash memory-based storage devices, such as SSD, eMMC, and SD cards, have 8been equipped on a variety systems ranging from mobile to server systems. Since 9they are known to have different characteristics from the conventional rotating 10disks, a file system, an upper layer to the storage device, should adapt to the 11changes from the sketch in the design level. 12 13F2FS is a file system exploiting NAND flash memory-based storage devices, which 14is based on Log-structured File System (LFS). The design has been focused on 15addressing the fundamental issues in LFS, which are snowball effect of wandering 16tree and high cleaning overhead. 17 18Since a NAND flash memory-based storage device shows different characteristic 19according to its internal geometry or flash memory management scheme, namely FTL, 20F2FS and its tools support various parameters not only for configuring on-disk 21layout, but also for selecting allocation and cleaning algorithms. 22 23The following git tree provides the file system formatting tool (mkfs.f2fs), 24a consistency checking tool (fsck.f2fs), and a debugging tool (dump.f2fs). 25 26- git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs-tools.git 27 28For sending patches, please use the following mailing list: 29 30- linux-f2fs-devel@lists.sourceforge.net 31 32For reporting bugs, please use the following f2fs bug tracker link: 33 34- https://bugzilla.kernel.org/enter_bug.cgi?product=File%20System&component=f2fs 35 36Background and Design issues 37============================ 38 39Log-structured File System (LFS) 40-------------------------------- 41"A log-structured file system writes all modifications to disk sequentially in 42a log-like structure, thereby speeding up both file writing and crash recovery. 43The log is the only structure on disk; it contains indexing information so that 44files can be read back from the log efficiently. In order to maintain large free 45areas on disk for fast writing, we divide the log into segments and use a 46segment cleaner to compress the live information from heavily fragmented 47segments." from Rosenblum, M. and Ousterhout, J. K., 1992, "The design and 48implementation of a log-structured file system", ACM Trans. Computer Systems 4910, 1, 26–52. 50 51Wandering Tree Problem 52---------------------- 53In LFS, when a file data is updated and written to the end of log, its direct 54pointer block is updated due to the changed location. Then the indirect pointer 55block is also updated due to the direct pointer block update. In this manner, 56the upper index structures such as inode, inode map, and checkpoint block are 57also updated recursively. This problem is called as wandering tree problem [1], 58and in order to enhance the performance, it should eliminate or relax the update 59propagation as much as possible. 60 61[1] Bityutskiy, A. 2005. JFFS3 design issues. http://www.linux-mtd.infradead.org/ 62 63Cleaning Overhead 64----------------- 65Since LFS is based on out-of-place writes, it produces so many obsolete blocks 66scattered across the whole storage. In order to serve new empty log space, it 67needs to reclaim these obsolete blocks seamlessly to users. This job is called 68as a cleaning process. 69 70The process consists of three operations as follows. 71 721. A victim segment is selected through referencing segment usage table. 732. It loads parent index structures of all the data in the victim identified by 74 segment summary blocks. 753. It checks the cross-reference between the data and its parent index structure. 764. It moves valid data selectively. 77 78This cleaning job may cause unexpected long delays, so the most important goal 79is to hide the latencies to users. And also definitely, it should reduce the 80amount of valid data to be moved, and move them quickly as well. 81 82Key Features 83============ 84 85Flash Awareness 86--------------- 87- Enlarge the random write area for better performance, but provide the high 88 spatial locality 89- Align FS data structures to the operational units in FTL as best efforts 90 91Wandering Tree Problem 92---------------------- 93- Use a term, “node”, that represents inodes as well as various pointer blocks 94- Introduce Node Address Table (NAT) containing the locations of all the “node” 95 blocks; this will cut off the update propagation. 96 97Cleaning Overhead 98----------------- 99- Support a background cleaning process 100- Support greedy and cost-benefit algorithms for victim selection policies 101- Support multi-head logs for static/dynamic hot and cold data separation 102- Introduce adaptive logging for efficient block allocation 103 104Mount Options 105============= 106 107 108======================== ============================================================ 109background_gc=%s Turn on/off cleaning operations, namely garbage 110 collection, triggered in background when I/O subsystem is 111 idle. If background_gc=on, it will turn on the garbage 112 collection and if background_gc=off, garbage collection 113 will be turned off. If background_gc=sync, it will turn 114 on synchronous garbage collection running in background. 115 Default value for this option is on. So garbage 116 collection is on by default. 117gc_merge When background_gc is on, this option can be enabled to 118 let background GC thread to handle foreground GC requests, 119 it can eliminate the sluggish issue caused by slow foreground 120 GC operation when GC is triggered from a process with limited 121 I/O and CPU resources. 122nogc_merge Disable GC merge feature. 123disable_roll_forward Disable the roll-forward recovery routine 124norecovery Disable the roll-forward recovery routine, mounted read- 125 only (i.e., -o ro,disable_roll_forward) 126discard/nodiscard Enable/disable real-time discard in f2fs, if discard is 127 enabled, f2fs will issue discard/TRIM commands when a 128 segment is cleaned. 129heap/no_heap Deprecated. 130nouser_xattr Disable Extended User Attributes. Note: xattr is enabled 131 by default if CONFIG_F2FS_FS_XATTR is selected. 132noacl Disable POSIX Access Control List. Note: acl is enabled 133 by default if CONFIG_F2FS_FS_POSIX_ACL is selected. 134active_logs=%u Support configuring the number of active logs. In the 135 current design, f2fs supports only 2, 4, and 6 logs. 136 Default number is 6. 137disable_ext_identify Disable the extension list configured by mkfs, so f2fs 138 is not aware of cold files such as media files. 139inline_xattr Enable the inline xattrs feature. 140noinline_xattr Disable the inline xattrs feature. 141inline_xattr_size=%u Support configuring inline xattr size, it depends on 142 flexible inline xattr feature. 143inline_data Enable the inline data feature: Newly created small (<~3.4k) 144 files can be written into inode block. 145inline_dentry Enable the inline dir feature: data in newly created 146 directory entries can be written into inode block. The 147 space of inode block which is used to store inline 148 dentries is limited to ~3.4k. 149noinline_dentry Disable the inline dentry feature. 150flush_merge Merge concurrent cache_flush commands as much as possible 151 to eliminate redundant command issues. If the underlying 152 device handles the cache_flush command relatively slowly, 153 recommend to enable this option. 154nobarrier This option can be used if underlying storage guarantees 155 its cached data should be written to the novolatile area. 156 If this option is set, no cache_flush commands are issued 157 but f2fs still guarantees the write ordering of all the 158 data writes. 159barrier If this option is set, cache_flush commands are allowed to be 160 issued. 161fastboot This option is used when a system wants to reduce mount 162 time as much as possible, even though normal performance 163 can be sacrificed. 164extent_cache Enable an extent cache based on rb-tree, it can cache 165 as many as extent which map between contiguous logical 166 address and physical address per inode, resulting in 167 increasing the cache hit ratio. Set by default. 168noextent_cache Disable an extent cache based on rb-tree explicitly, see 169 the above extent_cache mount option. 170noinline_data Disable the inline data feature, inline data feature is 171 enabled by default. 172data_flush Enable data flushing before checkpoint in order to 173 persist data of regular and symlink. 174reserve_root=%d Support configuring reserved space which is used for 175 allocation from a privileged user with specified uid or 176 gid, unit: 4KB, the default limit is 0.2% of user blocks. 177resuid=%d The user ID which may use the reserved blocks. 178resgid=%d The group ID which may use the reserved blocks. 179fault_injection=%d Enable fault injection in all supported types with 180 specified injection rate. 181fault_type=%d Support configuring fault injection type, should be 182 enabled with fault_injection option, fault type value 183 is shown below, it supports single or combined type. 184 185 =================== =========== 186 Type_Name Type_Value 187 =================== =========== 188 FAULT_KMALLOC 0x000000001 189 FAULT_KVMALLOC 0x000000002 190 FAULT_PAGE_ALLOC 0x000000004 191 FAULT_PAGE_GET 0x000000008 192 FAULT_ALLOC_BIO 0x000000010 (obsolete) 193 FAULT_ALLOC_NID 0x000000020 194 FAULT_ORPHAN 0x000000040 195 FAULT_BLOCK 0x000000080 196 FAULT_DIR_DEPTH 0x000000100 197 FAULT_EVICT_INODE 0x000000200 198 FAULT_TRUNCATE 0x000000400 199 FAULT_READ_IO 0x000000800 200 FAULT_CHECKPOINT 0x000001000 201 FAULT_DISCARD 0x000002000 202 FAULT_WRITE_IO 0x000004000 203 FAULT_SLAB_ALLOC 0x000008000 204 FAULT_DQUOT_INIT 0x000010000 205 FAULT_LOCK_OP 0x000020000 206 FAULT_BLKADDR 0x000040000 207 =================== =========== 208mode=%s Control block allocation mode which supports "adaptive" 209 and "lfs". In "lfs" mode, there should be no random 210 writes towards main area. 211 "fragment:segment" and "fragment:block" are newly added here. 212 These are developer options for experiments to simulate filesystem 213 fragmentation/after-GC situation itself. The developers use these 214 modes to understand filesystem fragmentation/after-GC condition well, 215 and eventually get some insights to handle them better. 216 In "fragment:segment", f2fs allocates a new segment in ramdom 217 position. With this, we can simulate the after-GC condition. 218 In "fragment:block", we can scatter block allocation with 219 "max_fragment_chunk" and "max_fragment_hole" sysfs nodes. 220 We added some randomness to both chunk and hole size to make 221 it close to realistic IO pattern. So, in this mode, f2fs will allocate 222 1..<max_fragment_chunk> blocks in a chunk and make a hole in the 223 length of 1..<max_fragment_hole> by turns. With this, the newly 224 allocated blocks will be scattered throughout the whole partition. 225 Note that "fragment:block" implicitly enables "fragment:segment" 226 option for more randomness. 227 Please, use these options for your experiments and we strongly 228 recommend to re-format the filesystem after using these options. 229usrquota Enable plain user disk quota accounting. 230grpquota Enable plain group disk quota accounting. 231prjquota Enable plain project quota accounting. 232usrjquota=<file> Appoint specified file and type during mount, so that quota 233grpjquota=<file> information can be properly updated during recovery flow, 234prjjquota=<file> <quota file>: must be in root directory; 235jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1]. 236offusrjquota Turn off user journalled quota. 237offgrpjquota Turn off group journalled quota. 238offprjjquota Turn off project journalled quota. 239quota Enable plain user disk quota accounting. 240noquota Disable all plain disk quota option. 241alloc_mode=%s Adjust block allocation policy, which supports "reuse" 242 and "default". 243fsync_mode=%s Control the policy of fsync. Currently supports "posix", 244 "strict", and "nobarrier". In "posix" mode, which is 245 default, fsync will follow POSIX semantics and does a 246 light operation to improve the filesystem performance. 247 In "strict" mode, fsync will be heavy and behaves in line 248 with xfs, ext4 and btrfs, where xfstest generic/342 will 249 pass, but the performance will regress. "nobarrier" is 250 based on "posix", but doesn't issue flush command for 251 non-atomic files likewise "nobarrier" mount option. 252test_dummy_encryption 253test_dummy_encryption=%s 254 Enable dummy encryption, which provides a fake fscrypt 255 context. The fake fscrypt context is used by xfstests. 256 The argument may be either "v1" or "v2", in order to 257 select the corresponding fscrypt policy version. 258checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable" 259 to reenable checkpointing. Is enabled by default. While 260 disabled, any unmounting or unexpected shutdowns will cause 261 the filesystem contents to appear as they did when the 262 filesystem was mounted with that option. 263 While mounting with checkpoint=disable, the filesystem must 264 run garbage collection to ensure that all available space can 265 be used. If this takes too much time, the mount may return 266 EAGAIN. You may optionally add a value to indicate how much 267 of the disk you would be willing to temporarily give up to 268 avoid additional garbage collection. This can be given as a 269 number of blocks, or as a percent. For instance, mounting 270 with checkpoint=disable:100% would always succeed, but it may 271 hide up to all remaining free space. The actual space that 272 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable 273 This space is reclaimed once checkpoint=enable. 274checkpoint_merge When checkpoint is enabled, this can be used to create a kernel 275 daemon and make it to merge concurrent checkpoint requests as 276 much as possible to eliminate redundant checkpoint issues. Plus, 277 we can eliminate the sluggish issue caused by slow checkpoint 278 operation when the checkpoint is done in a process context in 279 a cgroup having low i/o budget and cpu shares. To make this 280 do better, we set the default i/o priority of the kernel daemon 281 to "3", to give one higher priority than other kernel threads. 282 This is the same way to give a I/O priority to the jbd2 283 journaling thread of ext4 filesystem. 284nocheckpoint_merge Disable checkpoint merge feature. 285compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo", 286 "lz4", "zstd" and "lzo-rle" algorithm. 287compress_algorithm=%s:%d Control compress algorithm and its compress level, now, only 288 "lz4" and "zstd" support compress level config. 289 algorithm level range 290 lz4 3 - 16 291 zstd 1 - 22 292compress_log_size=%u Support configuring compress cluster size. The size will 293 be 4KB * (1 << %u). The default and minimum sizes are 16KB. 294compress_extension=%s Support adding specified extension, so that f2fs can enable 295 compression on those corresponding files, e.g. if all files 296 with '.ext' has high compression rate, we can set the '.ext' 297 on compression extension list and enable compression on 298 these file by default rather than to enable it via ioctl. 299 For other files, we can still enable compression via ioctl. 300 Note that, there is one reserved special extension '*', it 301 can be set to enable compression for all files. 302nocompress_extension=%s Support adding specified extension, so that f2fs can disable 303 compression on those corresponding files, just contrary to compression extension. 304 If you know exactly which files cannot be compressed, you can use this. 305 The same extension name can't appear in both compress and nocompress 306 extension at the same time. 307 If the compress extension specifies all files, the types specified by the 308 nocompress extension will be treated as special cases and will not be compressed. 309 Don't allow use '*' to specifie all file in nocompress extension. 310 After add nocompress_extension, the priority should be: 311 dir_flag < comp_extention,nocompress_extension < comp_file_flag,no_comp_file_flag. 312 See more in compression sections. 313 314compress_chksum Support verifying chksum of raw data in compressed cluster. 315compress_mode=%s Control file compression mode. This supports "fs" and "user" 316 modes. In "fs" mode (default), f2fs does automatic compression 317 on the compression enabled files. In "user" mode, f2fs disables 318 the automaic compression and gives the user discretion of 319 choosing the target file and the timing. The user can do manual 320 compression/decompression on the compression enabled files using 321 ioctls. 322compress_cache Support to use address space of a filesystem managed inode to 323 cache compressed block, in order to improve cache hit ratio of 324 random read. 325inlinecrypt When possible, encrypt/decrypt the contents of encrypted 326 files using the blk-crypto framework rather than 327 filesystem-layer encryption. This allows the use of 328 inline encryption hardware. The on-disk format is 329 unaffected. For more details, see 330 Documentation/block/inline-encryption.rst. 331atgc Enable age-threshold garbage collection, it provides high 332 effectiveness and efficiency on background GC. 333discard_unit=%s Control discard unit, the argument can be "block", "segment" 334 and "section", issued discard command's offset/size will be 335 aligned to the unit, by default, "discard_unit=block" is set, 336 so that small discard functionality is enabled. 337 For blkzoned device, "discard_unit=section" will be set by 338 default, it is helpful for large sized SMR or ZNS devices to 339 reduce memory cost by getting rid of fs metadata supports small 340 discard. 341memory=%s Control memory mode. This supports "normal" and "low" modes. 342 "low" mode is introduced to support low memory devices. 343 Because of the nature of low memory devices, in this mode, f2fs 344 will try to save memory sometimes by sacrificing performance. 345 "normal" mode is the default mode and same as before. 346age_extent_cache Enable an age extent cache based on rb-tree. It records 347 data block update frequency of the extent per inode, in 348 order to provide better temperature hints for data block 349 allocation. 350errors=%s Specify f2fs behavior on critical errors. This supports modes: 351 "panic", "continue" and "remount-ro", respectively, trigger 352 panic immediately, continue without doing anything, and remount 353 the partition in read-only mode. By default it uses "continue" 354 mode. 355 ====================== =============== =============== ======== 356 mode continue remount-ro panic 357 ====================== =============== =============== ======== 358 access ops normal normal N/A 359 syscall errors -EIO -EROFS N/A 360 mount option rw ro N/A 361 pending dir write keep keep N/A 362 pending non-dir write drop keep N/A 363 pending node write drop keep N/A 364 pending meta write keep keep N/A 365 ====================== =============== =============== ======== 366======================== ============================================================ 367 368Debugfs Entries 369=============== 370 371/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as 372f2fs. Each file shows the whole f2fs information. 373 374/sys/kernel/debug/f2fs/status includes: 375 376 - major file system information managed by f2fs currently 377 - average SIT information about whole segments 378 - current memory footprint consumed by f2fs. 379 380Sysfs Entries 381============= 382 383Information about mounted f2fs file systems can be found in 384/sys/fs/f2fs. Each mounted filesystem will have a directory in 385/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda). 386The files in each per-device directory are shown in table below. 387 388Files in /sys/fs/f2fs/<devname> 389(see also Documentation/ABI/testing/sysfs-fs-f2fs) 390 391Usage 392===== 393 3941. Download userland tools and compile them. 395 3962. Skip, if f2fs was compiled statically inside kernel. 397 Otherwise, insert the f2fs.ko module:: 398 399 # insmod f2fs.ko 400 4013. Create a directory to use when mounting:: 402 403 # mkdir /mnt/f2fs 404 4054. Format the block device, and then mount as f2fs:: 406 407 # mkfs.f2fs -l label /dev/block_device 408 # mount -t f2fs /dev/block_device /mnt/f2fs 409 410mkfs.f2fs 411--------- 412The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem, 413which builds a basic on-disk layout. 414 415The quick options consist of: 416 417=============== =========================================================== 418``-l [label]`` Give a volume label, up to 512 unicode name. 419``-a [0 or 1]`` Split start location of each area for heap-based allocation. 420 421 1 is set by default, which performs this. 422``-o [int]`` Set overprovision ratio in percent over volume size. 423 424 5 is set by default. 425``-s [int]`` Set the number of segments per section. 426 427 1 is set by default. 428``-z [int]`` Set the number of sections per zone. 429 430 1 is set by default. 431``-e [str]`` Set basic extension list. e.g. "mp3,gif,mov" 432``-t [0 or 1]`` Disable discard command or not. 433 434 1 is set by default, which conducts discard. 435=============== =========================================================== 436 437Note: please refer to the manpage of mkfs.f2fs(8) to get full option list. 438 439fsck.f2fs 440--------- 441The fsck.f2fs is a tool to check the consistency of an f2fs-formatted 442partition, which examines whether the filesystem metadata and user-made data 443are cross-referenced correctly or not. 444Note that, initial version of the tool does not fix any inconsistency. 445 446The quick options consist of:: 447 448 -d debug level [default:0] 449 450Note: please refer to the manpage of fsck.f2fs(8) to get full option list. 451 452dump.f2fs 453--------- 454The dump.f2fs shows the information of specific inode and dumps SSA and SIT to 455file. Each file is dump_ssa and dump_sit. 456 457The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem. 458It shows on-disk inode information recognized by a given inode number, and is 459able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and 460./dump_sit respectively. 461 462The options consist of:: 463 464 -d debug level [default:0] 465 -i inode no (hex) 466 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1] 467 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1] 468 469Examples:: 470 471 # dump.f2fs -i [ino] /dev/sdx 472 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 473 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 474 475Note: please refer to the manpage of dump.f2fs(8) to get full option list. 476 477sload.f2fs 478---------- 479The sload.f2fs gives a way to insert files and directories in the existing disk 480image. This tool is useful when building f2fs images given compiled files. 481 482Note: please refer to the manpage of sload.f2fs(8) to get full option list. 483 484resize.f2fs 485----------- 486The resize.f2fs lets a user resize the f2fs-formatted disk image, while preserving 487all the files and directories stored in the image. 488 489Note: please refer to the manpage of resize.f2fs(8) to get full option list. 490 491defrag.f2fs 492----------- 493The defrag.f2fs can be used to defragment scattered written data as well as 494filesystem metadata across the disk. This can improve the write speed by giving 495more free consecutive space. 496 497Note: please refer to the manpage of defrag.f2fs(8) to get full option list. 498 499f2fs_io 500------- 501The f2fs_io is a simple tool to issue various filesystem APIs as well as 502f2fs-specific ones, which is very useful for QA tests. 503 504Note: please refer to the manpage of f2fs_io(8) to get full option list. 505 506Design 507====== 508 509On-disk Layout 510-------------- 511 512F2FS divides the whole volume into a number of segments, each of which is fixed 513to 2MB in size. A section is composed of consecutive segments, and a zone 514consists of a set of sections. By default, section and zone sizes are set to one 515segment size identically, but users can easily modify the sizes by mkfs. 516 517F2FS splits the entire volume into six areas, and all the areas except superblock 518consist of multiple segments as described below:: 519 520 align with the zone size <-| 521 |-> align with the segment size 522 _________________________________________________________________________ 523 | | | Segment | Node | Segment | | 524 | Superblock | Checkpoint | Info. | Address | Summary | Main | 525 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | | 526 |____________|_____2______|______N______|______N______|______N_____|__N___| 527 . . 528 . . 529 . . 530 ._________________________________________. 531 |_Segment_|_..._|_Segment_|_..._|_Segment_| 532 . . 533 ._________._________ 534 |_section_|__...__|_ 535 . . 536 .________. 537 |__zone__| 538 539- Superblock (SB) 540 It is located at the beginning of the partition, and there exist two copies 541 to avoid file system crash. It contains basic partition information and some 542 default parameters of f2fs. 543 544- Checkpoint (CP) 545 It contains file system information, bitmaps for valid NAT/SIT sets, orphan 546 inode lists, and summary entries of current active segments. 547 548- Segment Information Table (SIT) 549 It contains segment information such as valid block count and bitmap for the 550 validity of all the blocks. 551 552- Node Address Table (NAT) 553 It is composed of a block address table for all the node blocks stored in 554 Main area. 555 556- Segment Summary Area (SSA) 557 It contains summary entries which contains the owner information of all the 558 data and node blocks stored in Main area. 559 560- Main Area 561 It contains file and directory data including their indices. 562 563In order to avoid misalignment between file system and flash-based storage, F2FS 564aligns the start block address of CP with the segment size. Also, it aligns the 565start block address of Main area with the zone size by reserving some segments 566in SSA area. 567 568Reference the following survey for additional technical details. 569https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey 570 571File System Metadata Structure 572------------------------------ 573 574F2FS adopts the checkpointing scheme to maintain file system consistency. At 575mount time, F2FS first tries to find the last valid checkpoint data by scanning 576CP area. In order to reduce the scanning time, F2FS uses only two copies of CP. 577One of them always indicates the last valid data, which is called as shadow copy 578mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism. 579 580For file system consistency, each CP points to which NAT and SIT copies are 581valid, as shown as below:: 582 583 +--------+----------+---------+ 584 | CP | SIT | NAT | 585 +--------+----------+---------+ 586 . . . . 587 . . . . 588 . . . . 589 +-------+-------+--------+--------+--------+--------+ 590 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 | 591 +-------+-------+--------+--------+--------+--------+ 592 | ^ ^ 593 | | | 594 `----------------------------------------' 595 596Index Structure 597--------------- 598 599The key data structure to manage the data locations is a "node". Similar to 600traditional file structures, F2FS has three types of node: inode, direct node, 601indirect node. F2FS assigns 4KB to an inode block which contains 923 data block 602indices, two direct node pointers, two indirect node pointers, and one double 603indirect node pointer as described below. One direct node block contains 1018 604data blocks, and one indirect node block contains also 1018 node blocks. Thus, 605one inode block (i.e., a file) covers:: 606 607 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB. 608 609 Inode block (4KB) 610 |- data (923) 611 |- direct node (2) 612 | `- data (1018) 613 |- indirect node (2) 614 | `- direct node (1018) 615 | `- data (1018) 616 `- double indirect node (1) 617 `- indirect node (1018) 618 `- direct node (1018) 619 `- data (1018) 620 621Note that all the node blocks are mapped by NAT which means the location of 622each node is translated by the NAT table. In the consideration of the wandering 623tree problem, F2FS is able to cut off the propagation of node updates caused by 624leaf data writes. 625 626Directory Structure 627------------------- 628 629A directory entry occupies 11 bytes, which consists of the following attributes. 630 631- hash hash value of the file name 632- ino inode number 633- len the length of file name 634- type file type such as directory, symlink, etc 635 636A dentry block consists of 214 dentry slots and file names. Therein a bitmap is 637used to represent whether each dentry is valid or not. A dentry block occupies 6384KB with the following composition. 639 640:: 641 642 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) + 643 dentries(11 * 214 bytes) + file name (8 * 214 bytes) 644 645 [Bucket] 646 +--------------------------------+ 647 |dentry block 1 | dentry block 2 | 648 +--------------------------------+ 649 . . 650 . . 651 . [Dentry Block Structure: 4KB] . 652 +--------+----------+----------+------------+ 653 | bitmap | reserved | dentries | file names | 654 +--------+----------+----------+------------+ 655 [Dentry Block: 4KB] . . 656 . . 657 . . 658 +------+------+-----+------+ 659 | hash | ino | len | type | 660 +------+------+-----+------+ 661 [Dentry Structure: 11 bytes] 662 663F2FS implements multi-level hash tables for directory structure. Each level has 664a hash table with dedicated number of hash buckets as shown below. Note that 665"A(2B)" means a bucket includes 2 data blocks. 666 667:: 668 669 ---------------------- 670 A : bucket 671 B : block 672 N : MAX_DIR_HASH_DEPTH 673 ---------------------- 674 675 level #0 | A(2B) 676 | 677 level #1 | A(2B) - A(2B) 678 | 679 level #2 | A(2B) - A(2B) - A(2B) - A(2B) 680 . | . . . . 681 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) 682 . | . . . . 683 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) 684 685The number of blocks and buckets are determined by:: 686 687 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2, 688 # of blocks in level #n = | 689 `- 4, Otherwise 690 691 ,- 2^(n + dir_level), 692 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2, 693 # of buckets in level #n = | 694 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), 695 Otherwise 696 697When F2FS finds a file name in a directory, at first a hash value of the file 698name is calculated. Then, F2FS scans the hash table in level #0 to find the 699dentry consisting of the file name and its inode number. If not found, F2FS 700scans the next hash table in level #1. In this way, F2FS scans hash tables in 701each levels incrementally from 1 to N. In each level F2FS needs to scan only 702one bucket determined by the following equation, which shows O(log(# of files)) 703complexity:: 704 705 bucket number to scan in level #n = (hash value) % (# of buckets in level #n) 706 707In the case of file creation, F2FS finds empty consecutive slots that cover the 708file name. F2FS searches the empty slots in the hash tables of whole levels from 7091 to N in the same way as the lookup operation. 710 711The following figure shows an example of two cases holding children:: 712 713 --------------> Dir <-------------- 714 | | 715 child child 716 717 child - child [hole] - child 718 719 child - child - child [hole] - [hole] - child 720 721 Case 1: Case 2: 722 Number of children = 6, Number of children = 3, 723 File size = 7 File size = 7 724 725Default Block Allocation 726------------------------ 727 728At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node 729and Hot/Warm/Cold data. 730 731- Hot node contains direct node blocks of directories. 732- Warm node contains direct node blocks except hot node blocks. 733- Cold node contains indirect node blocks 734- Hot data contains dentry blocks 735- Warm data contains data blocks except hot and cold data blocks 736- Cold data contains multimedia data or migrated data blocks 737 738LFS has two schemes for free space management: threaded log and copy-and-compac- 739tion. The copy-and-compaction scheme which is known as cleaning, is well-suited 740for devices showing very good sequential write performance, since free segments 741are served all the time for writing new data. However, it suffers from cleaning 742overhead under high utilization. Contrarily, the threaded log scheme suffers 743from random writes, but no cleaning process is needed. F2FS adopts a hybrid 744scheme where the copy-and-compaction scheme is adopted by default, but the 745policy is dynamically changed to the threaded log scheme according to the file 746system status. 747 748In order to align F2FS with underlying flash-based storage, F2FS allocates a 749segment in a unit of section. F2FS expects that the section size would be the 750same as the unit size of garbage collection in FTL. Furthermore, with respect 751to the mapping granularity in FTL, F2FS allocates each section of the active 752logs from different zones as much as possible, since FTL can write the data in 753the active logs into one allocation unit according to its mapping granularity. 754 755Cleaning process 756---------------- 757 758F2FS does cleaning both on demand and in the background. On-demand cleaning is 759triggered when there are not enough free segments to serve VFS calls. Background 760cleaner is operated by a kernel thread, and triggers the cleaning job when the 761system is idle. 762 763F2FS supports two victim selection policies: greedy and cost-benefit algorithms. 764In the greedy algorithm, F2FS selects a victim segment having the smallest number 765of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment 766according to the segment age and the number of valid blocks in order to address 767log block thrashing problem in the greedy algorithm. F2FS adopts the greedy 768algorithm for on-demand cleaner, while background cleaner adopts cost-benefit 769algorithm. 770 771In order to identify whether the data in the victim segment are valid or not, 772F2FS manages a bitmap. Each bit represents the validity of a block, and the 773bitmap is composed of a bit stream covering whole blocks in main area. 774 775Fallocate(2) Policy 776------------------- 777 778The default policy follows the below POSIX rule. 779 780Allocating disk space 781 The default operation (i.e., mode is zero) of fallocate() allocates 782 the disk space within the range specified by offset and len. The 783 file size (as reported by stat(2)) will be changed if offset+len is 784 greater than the file size. Any subregion within the range specified 785 by offset and len that did not contain data before the call will be 786 initialized to zero. This default behavior closely resembles the 787 behavior of the posix_fallocate(3) library function, and is intended 788 as a method of optimally implementing that function. 789 790However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to 791fallocate(fd, DEFAULT_MODE), it allocates on-disk block addresses having 792zero or random data, which is useful to the below scenario where: 793 794 1. create(fd) 795 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 796 3. fallocate(fd, 0, 0, size) 797 4. address = fibmap(fd, offset) 798 5. open(blkdev) 799 6. write(blkdev, address) 800 801Compression implementation 802-------------------------- 803 804- New term named cluster is defined as basic unit of compression, file can 805 be divided into multiple clusters logically. One cluster includes 4 << n 806 (n >= 0) logical pages, compression size is also cluster size, each of 807 cluster can be compressed or not. 808 809- In cluster metadata layout, one special block address is used to indicate 810 a cluster is a compressed one or normal one; for compressed cluster, following 811 metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs 812 stores data including compress header and compressed data. 813 814- In order to eliminate write amplification during overwrite, F2FS only 815 support compression on write-once file, data can be compressed only when 816 all logical blocks in cluster contain valid data and compress ratio of 817 cluster data is lower than specified threshold. 818 819- To enable compression on regular inode, there are four ways: 820 821 * chattr +c file 822 * chattr +c dir; touch dir/file 823 * mount w/ -o compress_extension=ext; touch file.ext 824 * mount w/ -o compress_extension=*; touch any_file 825 826- To disable compression on regular inode, there are two ways: 827 828 * chattr -c file 829 * mount w/ -o nocompress_extension=ext; touch file.ext 830 831- Priority in between FS_COMPR_FL, FS_NOCOMP_FS, extensions: 832 833 * compress_extension=so; nocompress_extension=zip; chattr +c dir; touch 834 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so and baz.txt 835 should be compresse, bar.zip should be non-compressed. chattr +c dir/bar.zip 836 can enable compress on bar.zip. 837 * compress_extension=so; nocompress_extension=zip; chattr -c dir; touch 838 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so should be 839 compresse, bar.zip and baz.txt should be non-compressed. 840 chattr+c dir/bar.zip; chattr+c dir/baz.txt; can enable compress on bar.zip 841 and baz.txt. 842 843- At this point, compression feature doesn't expose compressed space to user 844 directly in order to guarantee potential data updates later to the space. 845 Instead, the main goal is to reduce data writes to flash disk as much as 846 possible, resulting in extending disk life time as well as relaxing IO 847 congestion. Alternatively, we've added ioctl(F2FS_IOC_RELEASE_COMPRESS_BLOCKS) 848 interface to reclaim compressed space and show it to user after setting a 849 special flag to the inode. Once the compressed space is released, the flag 850 will block writing data to the file until either the compressed space is 851 reserved via ioctl(F2FS_IOC_RESERVE_COMPRESS_BLOCKS) or the file size is 852 truncated to zero. 853 854Compress metadata layout:: 855 856 [Dnode Structure] 857 +-----------------------------------------------+ 858 | cluster 1 | cluster 2 | ......... | cluster N | 859 +-----------------------------------------------+ 860 . . . . 861 . . . . 862 . Compressed Cluster . . Normal Cluster . 863 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 864 |compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 | 865 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 866 . . 867 . . 868 . . 869 +-------------+-------------+----------+----------------------------+ 870 | data length | data chksum | reserved | compressed data | 871 +-------------+-------------+----------+----------------------------+ 872 873Compression mode 874-------------------------- 875 876f2fs supports "fs" and "user" compression modes with "compression_mode" mount option. 877With this option, f2fs provides a choice to select the way how to compress the 878compression enabled files (refer to "Compression implementation" section for how to 879enable compression on a regular inode). 880 8811) compress_mode=fs 882This is the default option. f2fs does automatic compression in the writeback of the 883compression enabled files. 884 8852) compress_mode=user 886This disables the automatic compression and gives the user discretion of choosing the 887target file and the timing. The user can do manual compression/decompression on the 888compression enabled files using F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE 889ioctls like the below. 890 891To decompress a file, 892 893fd = open(filename, O_WRONLY, 0); 894ret = ioctl(fd, F2FS_IOC_DECOMPRESS_FILE); 895 896To compress a file, 897 898fd = open(filename, O_WRONLY, 0); 899ret = ioctl(fd, F2FS_IOC_COMPRESS_FILE); 900 901NVMe Zoned Namespace devices 902---------------------------- 903 904- ZNS defines a per-zone capacity which can be equal or less than the 905 zone-size. Zone-capacity is the number of usable blocks in the zone. 906 F2FS checks if zone-capacity is less than zone-size, if it is, then any 907 segment which starts after the zone-capacity is marked as not-free in 908 the free segment bitmap at initial mount time. These segments are marked 909 as permanently used so they are not allocated for writes and 910 consequently are not needed to be garbage collected. In case the 911 zone-capacity is not aligned to default segment size(2MB), then a segment 912 can start before the zone-capacity and span across zone-capacity boundary. 913 Such spanning segments are also considered as usable segments. All blocks 914 past the zone-capacity are considered unusable in these segments. 915