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. 129no_heap Disable heap-style segment allocation which finds free 130 segments for data from the beginning of main area, while 131 for node from the end of main area. 132nouser_xattr Disable Extended User Attributes. Note: xattr is enabled 133 by default if CONFIG_F2FS_FS_XATTR is selected. 134noacl Disable POSIX Access Control List. Note: acl is enabled 135 by default if CONFIG_F2FS_FS_POSIX_ACL is selected. 136active_logs=%u Support configuring the number of active logs. In the 137 current design, f2fs supports only 2, 4, and 6 logs. 138 Default number is 6. 139disable_ext_identify Disable the extension list configured by mkfs, so f2fs 140 is not aware of cold files such as media files. 141inline_xattr Enable the inline xattrs feature. 142noinline_xattr Disable the inline xattrs feature. 143inline_xattr_size=%u Support configuring inline xattr size, it depends on 144 flexible inline xattr feature. 145inline_data Enable the inline data feature: Newly created small (<~3.4k) 146 files can be written into inode block. 147inline_dentry Enable the inline dir feature: data in newly created 148 directory entries can be written into inode block. The 149 space of inode block which is used to store inline 150 dentries is limited to ~3.4k. 151noinline_dentry Disable the inline dentry feature. 152flush_merge Merge concurrent cache_flush commands as much as possible 153 to eliminate redundant command issues. If the underlying 154 device handles the cache_flush command relatively slowly, 155 recommend to enable this option. 156nobarrier This option can be used if underlying storage guarantees 157 its cached data should be written to the novolatile area. 158 If this option is set, no cache_flush commands are issued 159 but f2fs still guarantees the write ordering of all the 160 data writes. 161barrier If this option is set, cache_flush commands are allowed to be 162 issued. 163fastboot This option is used when a system wants to reduce mount 164 time as much as possible, even though normal performance 165 can be sacrificed. 166extent_cache Enable an extent cache based on rb-tree, it can cache 167 as many as extent which map between contiguous logical 168 address and physical address per inode, resulting in 169 increasing the cache hit ratio. Set by default. 170noextent_cache Disable an extent cache based on rb-tree explicitly, see 171 the above extent_cache mount option. 172noinline_data Disable the inline data feature, inline data feature is 173 enabled by default. 174data_flush Enable data flushing before checkpoint in order to 175 persist data of regular and symlink. 176reserve_root=%d Support configuring reserved space which is used for 177 allocation from a privileged user with specified uid or 178 gid, unit: 4KB, the default limit is 0.2% of user blocks. 179resuid=%d The user ID which may use the reserved blocks. 180resgid=%d The group ID which may use the reserved blocks. 181fault_injection=%d Enable fault injection in all supported types with 182 specified injection rate. 183fault_type=%d Support configuring fault injection type, should be 184 enabled with fault_injection option, fault type value 185 is shown below, it supports single or combined type. 186 187 =========================== =========== 188 Type_Name Type_Value 189 =========================== =========== 190 FAULT_KMALLOC 0x000000001 191 FAULT_KVMALLOC 0x000000002 192 FAULT_PAGE_ALLOC 0x000000004 193 FAULT_PAGE_GET 0x000000008 194 FAULT_ALLOC_BIO 0x000000010 (obsolete) 195 FAULT_ALLOC_NID 0x000000020 196 FAULT_ORPHAN 0x000000040 197 FAULT_BLOCK 0x000000080 198 FAULT_DIR_DEPTH 0x000000100 199 FAULT_EVICT_INODE 0x000000200 200 FAULT_TRUNCATE 0x000000400 201 FAULT_READ_IO 0x000000800 202 FAULT_CHECKPOINT 0x000001000 203 FAULT_DISCARD 0x000002000 204 FAULT_WRITE_IO 0x000004000 205 FAULT_SLAB_ALLOC 0x000008000 206 FAULT_DQUOT_INIT 0x000010000 207 FAULT_LOCK_OP 0x000020000 208 FAULT_BLKADDR_VALIDITY 0x000040000 209 FAULT_BLKADDR_CONSISTENCE 0x000080000 210 =========================== =========== 211mode=%s Control block allocation mode which supports "adaptive" 212 and "lfs". In "lfs" mode, there should be no random 213 writes towards main area. 214 "fragment:segment" and "fragment:block" are newly added here. 215 These are developer options for experiments to simulate filesystem 216 fragmentation/after-GC situation itself. The developers use these 217 modes to understand filesystem fragmentation/after-GC condition well, 218 and eventually get some insights to handle them better. 219 In "fragment:segment", f2fs allocates a new segment in ramdom 220 position. With this, we can simulate the after-GC condition. 221 In "fragment:block", we can scatter block allocation with 222 "max_fragment_chunk" and "max_fragment_hole" sysfs nodes. 223 We added some randomness to both chunk and hole size to make 224 it close to realistic IO pattern. So, in this mode, f2fs will allocate 225 1..<max_fragment_chunk> blocks in a chunk and make a hole in the 226 length of 1..<max_fragment_hole> by turns. With this, the newly 227 allocated blocks will be scattered throughout the whole partition. 228 Note that "fragment:block" implicitly enables "fragment:segment" 229 option for more randomness. 230 Please, use these options for your experiments and we strongly 231 recommend to re-format the filesystem after using these options. 232io_bits=%u Set the bit size of write IO requests. It should be set 233 with "mode=lfs". 234usrquota Enable plain user disk quota accounting. 235grpquota Enable plain group disk quota accounting. 236prjquota Enable plain project quota accounting. 237usrjquota=<file> Appoint specified file and type during mount, so that quota 238grpjquota=<file> information can be properly updated during recovery flow, 239prjjquota=<file> <quota file>: must be in root directory; 240jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1]. 241offusrjquota Turn off user journalled quota. 242offgrpjquota Turn off group journalled quota. 243offprjjquota Turn off project journalled quota. 244quota Enable plain user disk quota accounting. 245noquota Disable all plain disk quota option. 246alloc_mode=%s Adjust block allocation policy, which supports "reuse" 247 and "default". 248fsync_mode=%s Control the policy of fsync. Currently supports "posix", 249 "strict", and "nobarrier". In "posix" mode, which is 250 default, fsync will follow POSIX semantics and does a 251 light operation to improve the filesystem performance. 252 In "strict" mode, fsync will be heavy and behaves in line 253 with xfs, ext4 and btrfs, where xfstest generic/342 will 254 pass, but the performance will regress. "nobarrier" is 255 based on "posix", but doesn't issue flush command for 256 non-atomic files likewise "nobarrier" mount option. 257test_dummy_encryption 258test_dummy_encryption=%s 259 Enable dummy encryption, which provides a fake fscrypt 260 context. The fake fscrypt context is used by xfstests. 261 The argument may be either "v1" or "v2", in order to 262 select the corresponding fscrypt policy version. 263checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable" 264 to reenable checkpointing. Is enabled by default. While 265 disabled, any unmounting or unexpected shutdowns will cause 266 the filesystem contents to appear as they did when the 267 filesystem was mounted with that option. 268 While mounting with checkpoint=disable, the filesystem must 269 run garbage collection to ensure that all available space can 270 be used. If this takes too much time, the mount may return 271 EAGAIN. You may optionally add a value to indicate how much 272 of the disk you would be willing to temporarily give up to 273 avoid additional garbage collection. This can be given as a 274 number of blocks, or as a percent. For instance, mounting 275 with checkpoint=disable:100% would always succeed, but it may 276 hide up to all remaining free space. The actual space that 277 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable 278 This space is reclaimed once checkpoint=enable. 279checkpoint_merge When checkpoint is enabled, this can be used to create a kernel 280 daemon and make it to merge concurrent checkpoint requests as 281 much as possible to eliminate redundant checkpoint issues. Plus, 282 we can eliminate the sluggish issue caused by slow checkpoint 283 operation when the checkpoint is done in a process context in 284 a cgroup having low i/o budget and cpu shares. To make this 285 do better, we set the default i/o priority of the kernel daemon 286 to "3", to give one higher priority than other kernel threads. 287 This is the same way to give a I/O priority to the jbd2 288 journaling thread of ext4 filesystem. 289nocheckpoint_merge Disable checkpoint merge feature. 290compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo", 291 "lz4", "zstd" and "lzo-rle" algorithm. 292compress_algorithm=%s:%d Control compress algorithm and its compress level, now, only 293 "lz4" and "zstd" support compress level config. 294 algorithm level range 295 lz4 3 - 16 296 zstd 1 - 22 297compress_log_size=%u Support configuring compress cluster size. The size will 298 be 4KB * (1 << %u). The default and minimum sizes are 16KB. 299compress_extension=%s Support adding specified extension, so that f2fs can enable 300 compression on those corresponding files, e.g. if all files 301 with '.ext' has high compression rate, we can set the '.ext' 302 on compression extension list and enable compression on 303 these file by default rather than to enable it via ioctl. 304 For other files, we can still enable compression via ioctl. 305 Note that, there is one reserved special extension '*', it 306 can be set to enable compression for all files. 307nocompress_extension=%s Support adding specified extension, so that f2fs can disable 308 compression on those corresponding files, just contrary to compression extension. 309 If you know exactly which files cannot be compressed, you can use this. 310 The same extension name can't appear in both compress and nocompress 311 extension at the same time. 312 If the compress extension specifies all files, the types specified by the 313 nocompress extension will be treated as special cases and will not be compressed. 314 Don't allow use '*' to specifie all file in nocompress extension. 315 After add nocompress_extension, the priority should be: 316 dir_flag < comp_extention,nocompress_extension < comp_file_flag,no_comp_file_flag. 317 See more in compression sections. 318 319compress_chksum Support verifying chksum of raw data in compressed cluster. 320compress_mode=%s Control file compression mode. This supports "fs" and "user" 321 modes. In "fs" mode (default), f2fs does automatic compression 322 on the compression enabled files. In "user" mode, f2fs disables 323 the automaic compression and gives the user discretion of 324 choosing the target file and the timing. The user can do manual 325 compression/decompression on the compression enabled files using 326 ioctls. 327compress_cache Support to use address space of a filesystem managed inode to 328 cache compressed block, in order to improve cache hit ratio of 329 random read. 330inlinecrypt When possible, encrypt/decrypt the contents of encrypted 331 files using the blk-crypto framework rather than 332 filesystem-layer encryption. This allows the use of 333 inline encryption hardware. The on-disk format is 334 unaffected. For more details, see 335 Documentation/block/inline-encryption.rst. 336atgc Enable age-threshold garbage collection, it provides high 337 effectiveness and efficiency on background GC. 338discard_unit=%s Control discard unit, the argument can be "block", "segment" 339 and "section", issued discard command's offset/size will be 340 aligned to the unit, by default, "discard_unit=block" is set, 341 so that small discard functionality is enabled. 342 For blkzoned device, "discard_unit=section" will be set by 343 default, it is helpful for large sized SMR or ZNS devices to 344 reduce memory cost by getting rid of fs metadata supports small 345 discard. 346memory=%s Control memory mode. This supports "normal" and "low" modes. 347 "low" mode is introduced to support low memory devices. 348 Because of the nature of low memory devices, in this mode, f2fs 349 will try to save memory sometimes by sacrificing performance. 350 "normal" mode is the default mode and same as before. 351age_extent_cache Enable an age extent cache based on rb-tree. It records 352 data block update frequency of the extent per inode, in 353 order to provide better temperature hints for data block 354 allocation. 355errors=%s Specify f2fs behavior on critical errors. This supports modes: 356 "panic", "continue" and "remount-ro", respectively, trigger 357 panic immediately, continue without doing anything, and remount 358 the partition in read-only mode. By default it uses "continue" 359 mode. 360 ====================== =============== =============== ======== 361 mode continue remount-ro panic 362 ====================== =============== =============== ======== 363 access ops normal normal N/A 364 syscall errors -EIO -EROFS N/A 365 mount option rw ro N/A 366 pending dir write keep keep N/A 367 pending non-dir write drop keep N/A 368 pending node write drop keep N/A 369 pending meta write keep keep N/A 370 ====================== =============== =============== ======== 371======================== ============================================================ 372 373Debugfs Entries 374=============== 375 376/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as 377f2fs. Each file shows the whole f2fs information. 378 379/sys/kernel/debug/f2fs/status includes: 380 381 - major file system information managed by f2fs currently 382 - average SIT information about whole segments 383 - current memory footprint consumed by f2fs. 384 385Sysfs Entries 386============= 387 388Information about mounted f2fs file systems can be found in 389/sys/fs/f2fs. Each mounted filesystem will have a directory in 390/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda). 391The files in each per-device directory are shown in table below. 392 393Files in /sys/fs/f2fs/<devname> 394(see also Documentation/ABI/testing/sysfs-fs-f2fs) 395 396Usage 397===== 398 3991. Download userland tools and compile them. 400 4012. Skip, if f2fs was compiled statically inside kernel. 402 Otherwise, insert the f2fs.ko module:: 403 404 # insmod f2fs.ko 405 4063. Create a directory to use when mounting:: 407 408 # mkdir /mnt/f2fs 409 4104. Format the block device, and then mount as f2fs:: 411 412 # mkfs.f2fs -l label /dev/block_device 413 # mount -t f2fs /dev/block_device /mnt/f2fs 414 415mkfs.f2fs 416--------- 417The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem, 418which builds a basic on-disk layout. 419 420The quick options consist of: 421 422=============== =========================================================== 423``-l [label]`` Give a volume label, up to 512 unicode name. 424``-a [0 or 1]`` Split start location of each area for heap-based allocation. 425 426 1 is set by default, which performs this. 427``-o [int]`` Set overprovision ratio in percent over volume size. 428 429 5 is set by default. 430``-s [int]`` Set the number of segments per section. 431 432 1 is set by default. 433``-z [int]`` Set the number of sections per zone. 434 435 1 is set by default. 436``-e [str]`` Set basic extension list. e.g. "mp3,gif,mov" 437``-t [0 or 1]`` Disable discard command or not. 438 439 1 is set by default, which conducts discard. 440=============== =========================================================== 441 442Note: please refer to the manpage of mkfs.f2fs(8) to get full option list. 443 444fsck.f2fs 445--------- 446The fsck.f2fs is a tool to check the consistency of an f2fs-formatted 447partition, which examines whether the filesystem metadata and user-made data 448are cross-referenced correctly or not. 449Note that, initial version of the tool does not fix any inconsistency. 450 451The quick options consist of:: 452 453 -d debug level [default:0] 454 455Note: please refer to the manpage of fsck.f2fs(8) to get full option list. 456 457dump.f2fs 458--------- 459The dump.f2fs shows the information of specific inode and dumps SSA and SIT to 460file. Each file is dump_ssa and dump_sit. 461 462The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem. 463It shows on-disk inode information recognized by a given inode number, and is 464able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and 465./dump_sit respectively. 466 467The options consist of:: 468 469 -d debug level [default:0] 470 -i inode no (hex) 471 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1] 472 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1] 473 474Examples:: 475 476 # dump.f2fs -i [ino] /dev/sdx 477 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 478 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 479 480Note: please refer to the manpage of dump.f2fs(8) to get full option list. 481 482sload.f2fs 483---------- 484The sload.f2fs gives a way to insert files and directories in the existing disk 485image. This tool is useful when building f2fs images given compiled files. 486 487Note: please refer to the manpage of sload.f2fs(8) to get full option list. 488 489resize.f2fs 490----------- 491The resize.f2fs lets a user resize the f2fs-formatted disk image, while preserving 492all the files and directories stored in the image. 493 494Note: please refer to the manpage of resize.f2fs(8) to get full option list. 495 496defrag.f2fs 497----------- 498The defrag.f2fs can be used to defragment scattered written data as well as 499filesystem metadata across the disk. This can improve the write speed by giving 500more free consecutive space. 501 502Note: please refer to the manpage of defrag.f2fs(8) to get full option list. 503 504f2fs_io 505------- 506The f2fs_io is a simple tool to issue various filesystem APIs as well as 507f2fs-specific ones, which is very useful for QA tests. 508 509Note: please refer to the manpage of f2fs_io(8) to get full option list. 510 511Design 512====== 513 514On-disk Layout 515-------------- 516 517F2FS divides the whole volume into a number of segments, each of which is fixed 518to 2MB in size. A section is composed of consecutive segments, and a zone 519consists of a set of sections. By default, section and zone sizes are set to one 520segment size identically, but users can easily modify the sizes by mkfs. 521 522F2FS splits the entire volume into six areas, and all the areas except superblock 523consist of multiple segments as described below:: 524 525 align with the zone size <-| 526 |-> align with the segment size 527 _________________________________________________________________________ 528 | | | Segment | Node | Segment | | 529 | Superblock | Checkpoint | Info. | Address | Summary | Main | 530 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | | 531 |____________|_____2______|______N______|______N______|______N_____|__N___| 532 . . 533 . . 534 . . 535 ._________________________________________. 536 |_Segment_|_..._|_Segment_|_..._|_Segment_| 537 . . 538 ._________._________ 539 |_section_|__...__|_ 540 . . 541 .________. 542 |__zone__| 543 544- Superblock (SB) 545 It is located at the beginning of the partition, and there exist two copies 546 to avoid file system crash. It contains basic partition information and some 547 default parameters of f2fs. 548 549- Checkpoint (CP) 550 It contains file system information, bitmaps for valid NAT/SIT sets, orphan 551 inode lists, and summary entries of current active segments. 552 553- Segment Information Table (SIT) 554 It contains segment information such as valid block count and bitmap for the 555 validity of all the blocks. 556 557- Node Address Table (NAT) 558 It is composed of a block address table for all the node blocks stored in 559 Main area. 560 561- Segment Summary Area (SSA) 562 It contains summary entries which contains the owner information of all the 563 data and node blocks stored in Main area. 564 565- Main Area 566 It contains file and directory data including their indices. 567 568In order to avoid misalignment between file system and flash-based storage, F2FS 569aligns the start block address of CP with the segment size. Also, it aligns the 570start block address of Main area with the zone size by reserving some segments 571in SSA area. 572 573Reference the following survey for additional technical details. 574https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey 575 576File System Metadata Structure 577------------------------------ 578 579F2FS adopts the checkpointing scheme to maintain file system consistency. At 580mount time, F2FS first tries to find the last valid checkpoint data by scanning 581CP area. In order to reduce the scanning time, F2FS uses only two copies of CP. 582One of them always indicates the last valid data, which is called as shadow copy 583mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism. 584 585For file system consistency, each CP points to which NAT and SIT copies are 586valid, as shown as below:: 587 588 +--------+----------+---------+ 589 | CP | SIT | NAT | 590 +--------+----------+---------+ 591 . . . . 592 . . . . 593 . . . . 594 +-------+-------+--------+--------+--------+--------+ 595 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 | 596 +-------+-------+--------+--------+--------+--------+ 597 | ^ ^ 598 | | | 599 `----------------------------------------' 600 601Index Structure 602--------------- 603 604The key data structure to manage the data locations is a "node". Similar to 605traditional file structures, F2FS has three types of node: inode, direct node, 606indirect node. F2FS assigns 4KB to an inode block which contains 923 data block 607indices, two direct node pointers, two indirect node pointers, and one double 608indirect node pointer as described below. One direct node block contains 1018 609data blocks, and one indirect node block contains also 1018 node blocks. Thus, 610one inode block (i.e., a file) covers:: 611 612 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB. 613 614 Inode block (4KB) 615 |- data (923) 616 |- direct node (2) 617 | `- data (1018) 618 |- indirect node (2) 619 | `- direct node (1018) 620 | `- data (1018) 621 `- double indirect node (1) 622 `- indirect node (1018) 623 `- direct node (1018) 624 `- data (1018) 625 626Note that all the node blocks are mapped by NAT which means the location of 627each node is translated by the NAT table. In the consideration of the wandering 628tree problem, F2FS is able to cut off the propagation of node updates caused by 629leaf data writes. 630 631Directory Structure 632------------------- 633 634A directory entry occupies 11 bytes, which consists of the following attributes. 635 636- hash hash value of the file name 637- ino inode number 638- len the length of file name 639- type file type such as directory, symlink, etc 640 641A dentry block consists of 214 dentry slots and file names. Therein a bitmap is 642used to represent whether each dentry is valid or not. A dentry block occupies 6434KB with the following composition. 644 645:: 646 647 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) + 648 dentries(11 * 214 bytes) + file name (8 * 214 bytes) 649 650 [Bucket] 651 +--------------------------------+ 652 |dentry block 1 | dentry block 2 | 653 +--------------------------------+ 654 . . 655 . . 656 . [Dentry Block Structure: 4KB] . 657 +--------+----------+----------+------------+ 658 | bitmap | reserved | dentries | file names | 659 +--------+----------+----------+------------+ 660 [Dentry Block: 4KB] . . 661 . . 662 . . 663 +------+------+-----+------+ 664 | hash | ino | len | type | 665 +------+------+-----+------+ 666 [Dentry Structure: 11 bytes] 667 668F2FS implements multi-level hash tables for directory structure. Each level has 669a hash table with dedicated number of hash buckets as shown below. Note that 670"A(2B)" means a bucket includes 2 data blocks. 671 672:: 673 674 ---------------------- 675 A : bucket 676 B : block 677 N : MAX_DIR_HASH_DEPTH 678 ---------------------- 679 680 level #0 | A(2B) 681 | 682 level #1 | A(2B) - A(2B) 683 | 684 level #2 | A(2B) - A(2B) - A(2B) - A(2B) 685 . | . . . . 686 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) 687 . | . . . . 688 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) 689 690The number of blocks and buckets are determined by:: 691 692 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2, 693 # of blocks in level #n = | 694 `- 4, Otherwise 695 696 ,- 2^(n + dir_level), 697 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2, 698 # of buckets in level #n = | 699 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), 700 Otherwise 701 702When F2FS finds a file name in a directory, at first a hash value of the file 703name is calculated. Then, F2FS scans the hash table in level #0 to find the 704dentry consisting of the file name and its inode number. If not found, F2FS 705scans the next hash table in level #1. In this way, F2FS scans hash tables in 706each levels incrementally from 1 to N. In each level F2FS needs to scan only 707one bucket determined by the following equation, which shows O(log(# of files)) 708complexity:: 709 710 bucket number to scan in level #n = (hash value) % (# of buckets in level #n) 711 712In the case of file creation, F2FS finds empty consecutive slots that cover the 713file name. F2FS searches the empty slots in the hash tables of whole levels from 7141 to N in the same way as the lookup operation. 715 716The following figure shows an example of two cases holding children:: 717 718 --------------> Dir <-------------- 719 | | 720 child child 721 722 child - child [hole] - child 723 724 child - child - child [hole] - [hole] - child 725 726 Case 1: Case 2: 727 Number of children = 6, Number of children = 3, 728 File size = 7 File size = 7 729 730Default Block Allocation 731------------------------ 732 733At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node 734and Hot/Warm/Cold data. 735 736- Hot node contains direct node blocks of directories. 737- Warm node contains direct node blocks except hot node blocks. 738- Cold node contains indirect node blocks 739- Hot data contains dentry blocks 740- Warm data contains data blocks except hot and cold data blocks 741- Cold data contains multimedia data or migrated data blocks 742 743LFS has two schemes for free space management: threaded log and copy-and-compac- 744tion. The copy-and-compaction scheme which is known as cleaning, is well-suited 745for devices showing very good sequential write performance, since free segments 746are served all the time for writing new data. However, it suffers from cleaning 747overhead under high utilization. Contrarily, the threaded log scheme suffers 748from random writes, but no cleaning process is needed. F2FS adopts a hybrid 749scheme where the copy-and-compaction scheme is adopted by default, but the 750policy is dynamically changed to the threaded log scheme according to the file 751system status. 752 753In order to align F2FS with underlying flash-based storage, F2FS allocates a 754segment in a unit of section. F2FS expects that the section size would be the 755same as the unit size of garbage collection in FTL. Furthermore, with respect 756to the mapping granularity in FTL, F2FS allocates each section of the active 757logs from different zones as much as possible, since FTL can write the data in 758the active logs into one allocation unit according to its mapping granularity. 759 760Cleaning process 761---------------- 762 763F2FS does cleaning both on demand and in the background. On-demand cleaning is 764triggered when there are not enough free segments to serve VFS calls. Background 765cleaner is operated by a kernel thread, and triggers the cleaning job when the 766system is idle. 767 768F2FS supports two victim selection policies: greedy and cost-benefit algorithms. 769In the greedy algorithm, F2FS selects a victim segment having the smallest number 770of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment 771according to the segment age and the number of valid blocks in order to address 772log block thrashing problem in the greedy algorithm. F2FS adopts the greedy 773algorithm for on-demand cleaner, while background cleaner adopts cost-benefit 774algorithm. 775 776In order to identify whether the data in the victim segment are valid or not, 777F2FS manages a bitmap. Each bit represents the validity of a block, and the 778bitmap is composed of a bit stream covering whole blocks in main area. 779 780Fallocate(2) Policy 781------------------- 782 783The default policy follows the below POSIX rule. 784 785Allocating disk space 786 The default operation (i.e., mode is zero) of fallocate() allocates 787 the disk space within the range specified by offset and len. The 788 file size (as reported by stat(2)) will be changed if offset+len is 789 greater than the file size. Any subregion within the range specified 790 by offset and len that did not contain data before the call will be 791 initialized to zero. This default behavior closely resembles the 792 behavior of the posix_fallocate(3) library function, and is intended 793 as a method of optimally implementing that function. 794 795However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to 796fallocate(fd, DEFAULT_MODE), it allocates on-disk block addresses having 797zero or random data, which is useful to the below scenario where: 798 799 1. create(fd) 800 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 801 3. fallocate(fd, 0, 0, size) 802 4. address = fibmap(fd, offset) 803 5. open(blkdev) 804 6. write(blkdev, address) 805 806Compression implementation 807-------------------------- 808 809- New term named cluster is defined as basic unit of compression, file can 810 be divided into multiple clusters logically. One cluster includes 4 << n 811 (n >= 0) logical pages, compression size is also cluster size, each of 812 cluster can be compressed or not. 813 814- In cluster metadata layout, one special block address is used to indicate 815 a cluster is a compressed one or normal one; for compressed cluster, following 816 metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs 817 stores data including compress header and compressed data. 818 819- In order to eliminate write amplification during overwrite, F2FS only 820 support compression on write-once file, data can be compressed only when 821 all logical blocks in cluster contain valid data and compress ratio of 822 cluster data is lower than specified threshold. 823 824- To enable compression on regular inode, there are four ways: 825 826 * chattr +c file 827 * chattr +c dir; touch dir/file 828 * mount w/ -o compress_extension=ext; touch file.ext 829 * mount w/ -o compress_extension=*; touch any_file 830 831- To disable compression on regular inode, there are two ways: 832 833 * chattr -c file 834 * mount w/ -o nocompress_extension=ext; touch file.ext 835 836- Priority in between FS_COMPR_FL, FS_NOCOMP_FS, extensions: 837 838 * 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 and baz.txt 840 should be compresse, bar.zip should be non-compressed. chattr +c dir/bar.zip 841 can enable compress on bar.zip. 842 * compress_extension=so; nocompress_extension=zip; chattr -c dir; touch 843 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so should be 844 compresse, bar.zip and baz.txt should be non-compressed. 845 chattr+c dir/bar.zip; chattr+c dir/baz.txt; can enable compress on bar.zip 846 and baz.txt. 847 848- At this point, compression feature doesn't expose compressed space to user 849 directly in order to guarantee potential data updates later to the space. 850 Instead, the main goal is to reduce data writes to flash disk as much as 851 possible, resulting in extending disk life time as well as relaxing IO 852 congestion. Alternatively, we've added ioctl(F2FS_IOC_RELEASE_COMPRESS_BLOCKS) 853 interface to reclaim compressed space and show it to user after setting a 854 special flag to the inode. Once the compressed space is released, the flag 855 will block writing data to the file until either the compressed space is 856 reserved via ioctl(F2FS_IOC_RESERVE_COMPRESS_BLOCKS) or the file size is 857 truncated to zero. 858 859Compress metadata layout:: 860 861 [Dnode Structure] 862 +-----------------------------------------------+ 863 | cluster 1 | cluster 2 | ......... | cluster N | 864 +-----------------------------------------------+ 865 . . . . 866 . . . . 867 . Compressed Cluster . . Normal Cluster . 868 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 869 |compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 | 870 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 871 . . 872 . . 873 . . 874 +-------------+-------------+----------+----------------------------+ 875 | data length | data chksum | reserved | compressed data | 876 +-------------+-------------+----------+----------------------------+ 877 878Compression mode 879-------------------------- 880 881f2fs supports "fs" and "user" compression modes with "compression_mode" mount option. 882With this option, f2fs provides a choice to select the way how to compress the 883compression enabled files (refer to "Compression implementation" section for how to 884enable compression on a regular inode). 885 8861) compress_mode=fs 887This is the default option. f2fs does automatic compression in the writeback of the 888compression enabled files. 889 8902) compress_mode=user 891This disables the automatic compression and gives the user discretion of choosing the 892target file and the timing. The user can do manual compression/decompression on the 893compression enabled files using F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE 894ioctls like the below. 895 896To decompress a file, 897 898fd = open(filename, O_WRONLY, 0); 899ret = ioctl(fd, F2FS_IOC_DECOMPRESS_FILE); 900 901To compress a file, 902 903fd = open(filename, O_WRONLY, 0); 904ret = ioctl(fd, F2FS_IOC_COMPRESS_FILE); 905 906NVMe Zoned Namespace devices 907---------------------------- 908 909- ZNS defines a per-zone capacity which can be equal or less than the 910 zone-size. Zone-capacity is the number of usable blocks in the zone. 911 F2FS checks if zone-capacity is less than zone-size, if it is, then any 912 segment which starts after the zone-capacity is marked as not-free in 913 the free segment bitmap at initial mount time. These segments are marked 914 as permanently used so they are not allocated for writes and 915 consequently are not needed to be garbage collected. In case the 916 zone-capacity is not aligned to default segment size(2MB), then a segment 917 can start before the zone-capacity and span across zone-capacity boundary. 918 Such spanning segments are also considered as usable segments. All blocks 919 past the zone-capacity are considered unusable in these segments. 920