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_VALIDITY 0x000040000 207 FAULT_BLKADDR_CONSISTENCE 0x000080000 208 =========================== =========== 209mode=%s Control block allocation mode which supports "adaptive" 210 and "lfs". In "lfs" mode, there should be no random 211 writes towards main area. 212 "fragment:segment" and "fragment:block" are newly added here. 213 These are developer options for experiments to simulate filesystem 214 fragmentation/after-GC situation itself. The developers use these 215 modes to understand filesystem fragmentation/after-GC condition well, 216 and eventually get some insights to handle them better. 217 In "fragment:segment", f2fs allocates a new segment in ramdom 218 position. With this, we can simulate the after-GC condition. 219 In "fragment:block", we can scatter block allocation with 220 "max_fragment_chunk" and "max_fragment_hole" sysfs nodes. 221 We added some randomness to both chunk and hole size to make 222 it close to realistic IO pattern. So, in this mode, f2fs will allocate 223 1..<max_fragment_chunk> blocks in a chunk and make a hole in the 224 length of 1..<max_fragment_hole> by turns. With this, the newly 225 allocated blocks will be scattered throughout the whole partition. 226 Note that "fragment:block" implicitly enables "fragment:segment" 227 option for more randomness. 228 Please, use these options for your experiments and we strongly 229 recommend to re-format the filesystem after using these options. 230usrquota Enable plain user disk quota accounting. 231grpquota Enable plain group disk quota accounting. 232prjquota Enable plain project quota accounting. 233usrjquota=<file> Appoint specified file and type during mount, so that quota 234grpjquota=<file> information can be properly updated during recovery flow, 235prjjquota=<file> <quota file>: must be in root directory; 236jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1]. 237offusrjquota Turn off user journalled quota. 238offgrpjquota Turn off group journalled quota. 239offprjjquota Turn off project journalled quota. 240quota Enable plain user disk quota accounting. 241noquota Disable all plain disk quota option. 242alloc_mode=%s Adjust block allocation policy, which supports "reuse" 243 and "default". 244fsync_mode=%s Control the policy of fsync. Currently supports "posix", 245 "strict", and "nobarrier". In "posix" mode, which is 246 default, fsync will follow POSIX semantics and does a 247 light operation to improve the filesystem performance. 248 In "strict" mode, fsync will be heavy and behaves in line 249 with xfs, ext4 and btrfs, where xfstest generic/342 will 250 pass, but the performance will regress. "nobarrier" is 251 based on "posix", but doesn't issue flush command for 252 non-atomic files likewise "nobarrier" mount option. 253test_dummy_encryption 254test_dummy_encryption=%s 255 Enable dummy encryption, which provides a fake fscrypt 256 context. The fake fscrypt context is used by xfstests. 257 The argument may be either "v1" or "v2", in order to 258 select the corresponding fscrypt policy version. 259checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable" 260 to reenable checkpointing. Is enabled by default. While 261 disabled, any unmounting or unexpected shutdowns will cause 262 the filesystem contents to appear as they did when the 263 filesystem was mounted with that option. 264 While mounting with checkpoint=disable, the filesystem must 265 run garbage collection to ensure that all available space can 266 be used. If this takes too much time, the mount may return 267 EAGAIN. You may optionally add a value to indicate how much 268 of the disk you would be willing to temporarily give up to 269 avoid additional garbage collection. This can be given as a 270 number of blocks, or as a percent. For instance, mounting 271 with checkpoint=disable:100% would always succeed, but it may 272 hide up to all remaining free space. The actual space that 273 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable 274 This space is reclaimed once checkpoint=enable. 275checkpoint_merge When checkpoint is enabled, this can be used to create a kernel 276 daemon and make it to merge concurrent checkpoint requests as 277 much as possible to eliminate redundant checkpoint issues. Plus, 278 we can eliminate the sluggish issue caused by slow checkpoint 279 operation when the checkpoint is done in a process context in 280 a cgroup having low i/o budget and cpu shares. To make this 281 do better, we set the default i/o priority of the kernel daemon 282 to "3", to give one higher priority than other kernel threads. 283 This is the same way to give a I/O priority to the jbd2 284 journaling thread of ext4 filesystem. 285nocheckpoint_merge Disable checkpoint merge feature. 286compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo", 287 "lz4", "zstd" and "lzo-rle" algorithm. 288compress_algorithm=%s:%d Control compress algorithm and its compress level, now, only 289 "lz4" and "zstd" support compress level config. 290 algorithm level range 291 lz4 3 - 16 292 zstd 1 - 22 293compress_log_size=%u Support configuring compress cluster size. The size will 294 be 4KB * (1 << %u). The default and minimum sizes are 16KB. 295compress_extension=%s Support adding specified extension, so that f2fs can enable 296 compression on those corresponding files, e.g. if all files 297 with '.ext' has high compression rate, we can set the '.ext' 298 on compression extension list and enable compression on 299 these file by default rather than to enable it via ioctl. 300 For other files, we can still enable compression via ioctl. 301 Note that, there is one reserved special extension '*', it 302 can be set to enable compression for all files. 303nocompress_extension=%s Support adding specified extension, so that f2fs can disable 304 compression on those corresponding files, just contrary to compression extension. 305 If you know exactly which files cannot be compressed, you can use this. 306 The same extension name can't appear in both compress and nocompress 307 extension at the same time. 308 If the compress extension specifies all files, the types specified by the 309 nocompress extension will be treated as special cases and will not be compressed. 310 Don't allow use '*' to specifie all file in nocompress extension. 311 After add nocompress_extension, the priority should be: 312 dir_flag < comp_extention,nocompress_extension < comp_file_flag,no_comp_file_flag. 313 See more in compression sections. 314 315compress_chksum Support verifying chksum of raw data in compressed cluster. 316compress_mode=%s Control file compression mode. This supports "fs" and "user" 317 modes. In "fs" mode (default), f2fs does automatic compression 318 on the compression enabled files. In "user" mode, f2fs disables 319 the automaic compression and gives the user discretion of 320 choosing the target file and the timing. The user can do manual 321 compression/decompression on the compression enabled files using 322 ioctls. 323compress_cache Support to use address space of a filesystem managed inode to 324 cache compressed block, in order to improve cache hit ratio of 325 random read. 326inlinecrypt When possible, encrypt/decrypt the contents of encrypted 327 files using the blk-crypto framework rather than 328 filesystem-layer encryption. This allows the use of 329 inline encryption hardware. The on-disk format is 330 unaffected. For more details, see 331 Documentation/block/inline-encryption.rst. 332atgc Enable age-threshold garbage collection, it provides high 333 effectiveness and efficiency on background GC. 334discard_unit=%s Control discard unit, the argument can be "block", "segment" 335 and "section", issued discard command's offset/size will be 336 aligned to the unit, by default, "discard_unit=block" is set, 337 so that small discard functionality is enabled. 338 For blkzoned device, "discard_unit=section" will be set by 339 default, it is helpful for large sized SMR or ZNS devices to 340 reduce memory cost by getting rid of fs metadata supports small 341 discard. 342memory=%s Control memory mode. This supports "normal" and "low" modes. 343 "low" mode is introduced to support low memory devices. 344 Because of the nature of low memory devices, in this mode, f2fs 345 will try to save memory sometimes by sacrificing performance. 346 "normal" mode is the default mode and same as before. 347age_extent_cache Enable an age extent cache based on rb-tree. It records 348 data block update frequency of the extent per inode, in 349 order to provide better temperature hints for data block 350 allocation. 351errors=%s Specify f2fs behavior on critical errors. This supports modes: 352 "panic", "continue" and "remount-ro", respectively, trigger 353 panic immediately, continue without doing anything, and remount 354 the partition in read-only mode. By default it uses "continue" 355 mode. 356 ====================== =============== =============== ======== 357 mode continue remount-ro panic 358 ====================== =============== =============== ======== 359 access ops normal normal N/A 360 syscall errors -EIO -EROFS N/A 361 mount option rw ro N/A 362 pending dir write keep keep N/A 363 pending non-dir write drop keep N/A 364 pending node write drop keep N/A 365 pending meta write keep keep N/A 366 ====================== =============== =============== ======== 367======================== ============================================================ 368 369Debugfs Entries 370=============== 371 372/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as 373f2fs. Each file shows the whole f2fs information. 374 375/sys/kernel/debug/f2fs/status includes: 376 377 - major file system information managed by f2fs currently 378 - average SIT information about whole segments 379 - current memory footprint consumed by f2fs. 380 381Sysfs Entries 382============= 383 384Information about mounted f2fs file systems can be found in 385/sys/fs/f2fs. Each mounted filesystem will have a directory in 386/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda). 387The files in each per-device directory are shown in table below. 388 389Files in /sys/fs/f2fs/<devname> 390(see also Documentation/ABI/testing/sysfs-fs-f2fs) 391 392Usage 393===== 394 3951. Download userland tools and compile them. 396 3972. Skip, if f2fs was compiled statically inside kernel. 398 Otherwise, insert the f2fs.ko module:: 399 400 # insmod f2fs.ko 401 4023. Create a directory to use when mounting:: 403 404 # mkdir /mnt/f2fs 405 4064. Format the block device, and then mount as f2fs:: 407 408 # mkfs.f2fs -l label /dev/block_device 409 # mount -t f2fs /dev/block_device /mnt/f2fs 410 411mkfs.f2fs 412--------- 413The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem, 414which builds a basic on-disk layout. 415 416The quick options consist of: 417 418=============== =========================================================== 419``-l [label]`` Give a volume label, up to 512 unicode name. 420``-a [0 or 1]`` Split start location of each area for heap-based allocation. 421 422 1 is set by default, which performs this. 423``-o [int]`` Set overprovision ratio in percent over volume size. 424 425 5 is set by default. 426``-s [int]`` Set the number of segments per section. 427 428 1 is set by default. 429``-z [int]`` Set the number of sections per zone. 430 431 1 is set by default. 432``-e [str]`` Set basic extension list. e.g. "mp3,gif,mov" 433``-t [0 or 1]`` Disable discard command or not. 434 435 1 is set by default, which conducts discard. 436=============== =========================================================== 437 438Note: please refer to the manpage of mkfs.f2fs(8) to get full option list. 439 440fsck.f2fs 441--------- 442The fsck.f2fs is a tool to check the consistency of an f2fs-formatted 443partition, which examines whether the filesystem metadata and user-made data 444are cross-referenced correctly or not. 445Note that, initial version of the tool does not fix any inconsistency. 446 447The quick options consist of:: 448 449 -d debug level [default:0] 450 451Note: please refer to the manpage of fsck.f2fs(8) to get full option list. 452 453dump.f2fs 454--------- 455The dump.f2fs shows the information of specific inode and dumps SSA and SIT to 456file. Each file is dump_ssa and dump_sit. 457 458The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem. 459It shows on-disk inode information recognized by a given inode number, and is 460able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and 461./dump_sit respectively. 462 463The options consist of:: 464 465 -d debug level [default:0] 466 -i inode no (hex) 467 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1] 468 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1] 469 470Examples:: 471 472 # dump.f2fs -i [ino] /dev/sdx 473 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 474 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 475 476Note: please refer to the manpage of dump.f2fs(8) to get full option list. 477 478sload.f2fs 479---------- 480The sload.f2fs gives a way to insert files and directories in the existing disk 481image. This tool is useful when building f2fs images given compiled files. 482 483Note: please refer to the manpage of sload.f2fs(8) to get full option list. 484 485resize.f2fs 486----------- 487The resize.f2fs lets a user resize the f2fs-formatted disk image, while preserving 488all the files and directories stored in the image. 489 490Note: please refer to the manpage of resize.f2fs(8) to get full option list. 491 492defrag.f2fs 493----------- 494The defrag.f2fs can be used to defragment scattered written data as well as 495filesystem metadata across the disk. This can improve the write speed by giving 496more free consecutive space. 497 498Note: please refer to the manpage of defrag.f2fs(8) to get full option list. 499 500f2fs_io 501------- 502The f2fs_io is a simple tool to issue various filesystem APIs as well as 503f2fs-specific ones, which is very useful for QA tests. 504 505Note: please refer to the manpage of f2fs_io(8) to get full option list. 506 507Design 508====== 509 510On-disk Layout 511-------------- 512 513F2FS divides the whole volume into a number of segments, each of which is fixed 514to 2MB in size. A section is composed of consecutive segments, and a zone 515consists of a set of sections. By default, section and zone sizes are set to one 516segment size identically, but users can easily modify the sizes by mkfs. 517 518F2FS splits the entire volume into six areas, and all the areas except superblock 519consist of multiple segments as described below:: 520 521 align with the zone size <-| 522 |-> align with the segment size 523 _________________________________________________________________________ 524 | | | Segment | Node | Segment | | 525 | Superblock | Checkpoint | Info. | Address | Summary | Main | 526 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | | 527 |____________|_____2______|______N______|______N______|______N_____|__N___| 528 . . 529 . . 530 . . 531 ._________________________________________. 532 |_Segment_|_..._|_Segment_|_..._|_Segment_| 533 . . 534 ._________._________ 535 |_section_|__...__|_ 536 . . 537 .________. 538 |__zone__| 539 540- Superblock (SB) 541 It is located at the beginning of the partition, and there exist two copies 542 to avoid file system crash. It contains basic partition information and some 543 default parameters of f2fs. 544 545- Checkpoint (CP) 546 It contains file system information, bitmaps for valid NAT/SIT sets, orphan 547 inode lists, and summary entries of current active segments. 548 549- Segment Information Table (SIT) 550 It contains segment information such as valid block count and bitmap for the 551 validity of all the blocks. 552 553- Node Address Table (NAT) 554 It is composed of a block address table for all the node blocks stored in 555 Main area. 556 557- Segment Summary Area (SSA) 558 It contains summary entries which contains the owner information of all the 559 data and node blocks stored in Main area. 560 561- Main Area 562 It contains file and directory data including their indices. 563 564In order to avoid misalignment between file system and flash-based storage, F2FS 565aligns the start block address of CP with the segment size. Also, it aligns the 566start block address of Main area with the zone size by reserving some segments 567in SSA area. 568 569Reference the following survey for additional technical details. 570https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey 571 572File System Metadata Structure 573------------------------------ 574 575F2FS adopts the checkpointing scheme to maintain file system consistency. At 576mount time, F2FS first tries to find the last valid checkpoint data by scanning 577CP area. In order to reduce the scanning time, F2FS uses only two copies of CP. 578One of them always indicates the last valid data, which is called as shadow copy 579mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism. 580 581For file system consistency, each CP points to which NAT and SIT copies are 582valid, as shown as below:: 583 584 +--------+----------+---------+ 585 | CP | SIT | NAT | 586 +--------+----------+---------+ 587 . . . . 588 . . . . 589 . . . . 590 +-------+-------+--------+--------+--------+--------+ 591 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 | 592 +-------+-------+--------+--------+--------+--------+ 593 | ^ ^ 594 | | | 595 `----------------------------------------' 596 597Index Structure 598--------------- 599 600The key data structure to manage the data locations is a "node". Similar to 601traditional file structures, F2FS has three types of node: inode, direct node, 602indirect node. F2FS assigns 4KB to an inode block which contains 923 data block 603indices, two direct node pointers, two indirect node pointers, and one double 604indirect node pointer as described below. One direct node block contains 1018 605data blocks, and one indirect node block contains also 1018 node blocks. Thus, 606one inode block (i.e., a file) covers:: 607 608 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB. 609 610 Inode block (4KB) 611 |- data (923) 612 |- direct node (2) 613 | `- data (1018) 614 |- indirect node (2) 615 | `- direct node (1018) 616 | `- data (1018) 617 `- double indirect node (1) 618 `- indirect node (1018) 619 `- direct node (1018) 620 `- data (1018) 621 622Note that all the node blocks are mapped by NAT which means the location of 623each node is translated by the NAT table. In the consideration of the wandering 624tree problem, F2FS is able to cut off the propagation of node updates caused by 625leaf data writes. 626 627Directory Structure 628------------------- 629 630A directory entry occupies 11 bytes, which consists of the following attributes. 631 632- hash hash value of the file name 633- ino inode number 634- len the length of file name 635- type file type such as directory, symlink, etc 636 637A dentry block consists of 214 dentry slots and file names. Therein a bitmap is 638used to represent whether each dentry is valid or not. A dentry block occupies 6394KB with the following composition. 640 641:: 642 643 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) + 644 dentries(11 * 214 bytes) + file name (8 * 214 bytes) 645 646 [Bucket] 647 +--------------------------------+ 648 |dentry block 1 | dentry block 2 | 649 +--------------------------------+ 650 . . 651 . . 652 . [Dentry Block Structure: 4KB] . 653 +--------+----------+----------+------------+ 654 | bitmap | reserved | dentries | file names | 655 +--------+----------+----------+------------+ 656 [Dentry Block: 4KB] . . 657 . . 658 . . 659 +------+------+-----+------+ 660 | hash | ino | len | type | 661 +------+------+-----+------+ 662 [Dentry Structure: 11 bytes] 663 664F2FS implements multi-level hash tables for directory structure. Each level has 665a hash table with dedicated number of hash buckets as shown below. Note that 666"A(2B)" means a bucket includes 2 data blocks. 667 668:: 669 670 ---------------------- 671 A : bucket 672 B : block 673 N : MAX_DIR_HASH_DEPTH 674 ---------------------- 675 676 level #0 | A(2B) 677 | 678 level #1 | A(2B) - A(2B) 679 | 680 level #2 | A(2B) - A(2B) - A(2B) - A(2B) 681 . | . . . . 682 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) 683 . | . . . . 684 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) 685 686The number of blocks and buckets are determined by:: 687 688 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2, 689 # of blocks in level #n = | 690 `- 4, Otherwise 691 692 ,- 2^(n + dir_level), 693 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2, 694 # of buckets in level #n = | 695 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), 696 Otherwise 697 698When F2FS finds a file name in a directory, at first a hash value of the file 699name is calculated. Then, F2FS scans the hash table in level #0 to find the 700dentry consisting of the file name and its inode number. If not found, F2FS 701scans the next hash table in level #1. In this way, F2FS scans hash tables in 702each levels incrementally from 1 to N. In each level F2FS needs to scan only 703one bucket determined by the following equation, which shows O(log(# of files)) 704complexity:: 705 706 bucket number to scan in level #n = (hash value) % (# of buckets in level #n) 707 708In the case of file creation, F2FS finds empty consecutive slots that cover the 709file name. F2FS searches the empty slots in the hash tables of whole levels from 7101 to N in the same way as the lookup operation. 711 712The following figure shows an example of two cases holding children:: 713 714 --------------> Dir <-------------- 715 | | 716 child child 717 718 child - child [hole] - child 719 720 child - child - child [hole] - [hole] - child 721 722 Case 1: Case 2: 723 Number of children = 6, Number of children = 3, 724 File size = 7 File size = 7 725 726Default Block Allocation 727------------------------ 728 729At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node 730and Hot/Warm/Cold data. 731 732- Hot node contains direct node blocks of directories. 733- Warm node contains direct node blocks except hot node blocks. 734- Cold node contains indirect node blocks 735- Hot data contains dentry blocks 736- Warm data contains data blocks except hot and cold data blocks 737- Cold data contains multimedia data or migrated data blocks 738 739LFS has two schemes for free space management: threaded log and copy-and-compac- 740tion. The copy-and-compaction scheme which is known as cleaning, is well-suited 741for devices showing very good sequential write performance, since free segments 742are served all the time for writing new data. However, it suffers from cleaning 743overhead under high utilization. Contrarily, the threaded log scheme suffers 744from random writes, but no cleaning process is needed. F2FS adopts a hybrid 745scheme where the copy-and-compaction scheme is adopted by default, but the 746policy is dynamically changed to the threaded log scheme according to the file 747system status. 748 749In order to align F2FS with underlying flash-based storage, F2FS allocates a 750segment in a unit of section. F2FS expects that the section size would be the 751same as the unit size of garbage collection in FTL. Furthermore, with respect 752to the mapping granularity in FTL, F2FS allocates each section of the active 753logs from different zones as much as possible, since FTL can write the data in 754the active logs into one allocation unit according to its mapping granularity. 755 756Cleaning process 757---------------- 758 759F2FS does cleaning both on demand and in the background. On-demand cleaning is 760triggered when there are not enough free segments to serve VFS calls. Background 761cleaner is operated by a kernel thread, and triggers the cleaning job when the 762system is idle. 763 764F2FS supports two victim selection policies: greedy and cost-benefit algorithms. 765In the greedy algorithm, F2FS selects a victim segment having the smallest number 766of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment 767according to the segment age and the number of valid blocks in order to address 768log block thrashing problem in the greedy algorithm. F2FS adopts the greedy 769algorithm for on-demand cleaner, while background cleaner adopts cost-benefit 770algorithm. 771 772In order to identify whether the data in the victim segment are valid or not, 773F2FS manages a bitmap. Each bit represents the validity of a block, and the 774bitmap is composed of a bit stream covering whole blocks in main area. 775 776Fallocate(2) Policy 777------------------- 778 779The default policy follows the below POSIX rule. 780 781Allocating disk space 782 The default operation (i.e., mode is zero) of fallocate() allocates 783 the disk space within the range specified by offset and len. The 784 file size (as reported by stat(2)) will be changed if offset+len is 785 greater than the file size. Any subregion within the range specified 786 by offset and len that did not contain data before the call will be 787 initialized to zero. This default behavior closely resembles the 788 behavior of the posix_fallocate(3) library function, and is intended 789 as a method of optimally implementing that function. 790 791However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to 792fallocate(fd, DEFAULT_MODE), it allocates on-disk block addresses having 793zero or random data, which is useful to the below scenario where: 794 795 1. create(fd) 796 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 797 3. fallocate(fd, 0, 0, size) 798 4. address = fibmap(fd, offset) 799 5. open(blkdev) 800 6. write(blkdev, address) 801 802Compression implementation 803-------------------------- 804 805- New term named cluster is defined as basic unit of compression, file can 806 be divided into multiple clusters logically. One cluster includes 4 << n 807 (n >= 0) logical pages, compression size is also cluster size, each of 808 cluster can be compressed or not. 809 810- In cluster metadata layout, one special block address is used to indicate 811 a cluster is a compressed one or normal one; for compressed cluster, following 812 metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs 813 stores data including compress header and compressed data. 814 815- In order to eliminate write amplification during overwrite, F2FS only 816 support compression on write-once file, data can be compressed only when 817 all logical blocks in cluster contain valid data and compress ratio of 818 cluster data is lower than specified threshold. 819 820- To enable compression on regular inode, there are four ways: 821 822 * chattr +c file 823 * chattr +c dir; touch dir/file 824 * mount w/ -o compress_extension=ext; touch file.ext 825 * mount w/ -o compress_extension=*; touch any_file 826 827- To disable compression on regular inode, there are two ways: 828 829 * chattr -c file 830 * mount w/ -o nocompress_extension=ext; touch file.ext 831 832- Priority in between FS_COMPR_FL, FS_NOCOMP_FS, extensions: 833 834 * compress_extension=so; nocompress_extension=zip; chattr +c dir; touch 835 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so and baz.txt 836 should be compresse, bar.zip should be non-compressed. chattr +c dir/bar.zip 837 can enable compress on bar.zip. 838 * compress_extension=so; nocompress_extension=zip; chattr -c dir; touch 839 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so should be 840 compresse, bar.zip and baz.txt should be non-compressed. 841 chattr+c dir/bar.zip; chattr+c dir/baz.txt; can enable compress on bar.zip 842 and baz.txt. 843 844- At this point, compression feature doesn't expose compressed space to user 845 directly in order to guarantee potential data updates later to the space. 846 Instead, the main goal is to reduce data writes to flash disk as much as 847 possible, resulting in extending disk life time as well as relaxing IO 848 congestion. Alternatively, we've added ioctl(F2FS_IOC_RELEASE_COMPRESS_BLOCKS) 849 interface to reclaim compressed space and show it to user after setting a 850 special flag to the inode. Once the compressed space is released, the flag 851 will block writing data to the file until either the compressed space is 852 reserved via ioctl(F2FS_IOC_RESERVE_COMPRESS_BLOCKS) or the file size is 853 truncated to zero. 854 855Compress metadata layout:: 856 857 [Dnode Structure] 858 +-----------------------------------------------+ 859 | cluster 1 | cluster 2 | ......... | cluster N | 860 +-----------------------------------------------+ 861 . . . . 862 . . . . 863 . Compressed Cluster . . Normal Cluster . 864 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 865 |compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 | 866 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 867 . . 868 . . 869 . . 870 +-------------+-------------+----------+----------------------------+ 871 | data length | data chksum | reserved | compressed data | 872 +-------------+-------------+----------+----------------------------+ 873 874Compression mode 875-------------------------- 876 877f2fs supports "fs" and "user" compression modes with "compression_mode" mount option. 878With this option, f2fs provides a choice to select the way how to compress the 879compression enabled files (refer to "Compression implementation" section for how to 880enable compression on a regular inode). 881 8821) compress_mode=fs 883This is the default option. f2fs does automatic compression in the writeback of the 884compression enabled files. 885 8862) compress_mode=user 887This disables the automatic compression and gives the user discretion of choosing the 888target file and the timing. The user can do manual compression/decompression on the 889compression enabled files using F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE 890ioctls like the below. 891 892To decompress a file, 893 894fd = open(filename, O_WRONLY, 0); 895ret = ioctl(fd, F2FS_IOC_DECOMPRESS_FILE); 896 897To compress a file, 898 899fd = open(filename, O_WRONLY, 0); 900ret = ioctl(fd, F2FS_IOC_COMPRESS_FILE); 901 902NVMe Zoned Namespace devices 903---------------------------- 904 905- ZNS defines a per-zone capacity which can be equal or less than the 906 zone-size. Zone-capacity is the number of usable blocks in the zone. 907 F2FS checks if zone-capacity is less than zone-size, if it is, then any 908 segment which starts after the zone-capacity is marked as not-free in 909 the free segment bitmap at initial mount time. These segments are marked 910 as permanently used so they are not allocated for writes and 911 consequently are not needed to be garbage collected. In case the 912 zone-capacity is not aligned to default segment size(2MB), then a segment 913 can start before the zone-capacity and span across zone-capacity boundary. 914 Such spanning segments are also considered as usable segments. All blocks 915 past the zone-capacity are considered unusable in these segments. 916