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 12.5% of user blocks. 177reserve_node=%d Support configuring reserved nodes which are used for 178 allocation from a privileged user with specified uid or 179 gid, the default limit is 12.5% of all nodes. 180resuid=%d The user ID which may use the reserved blocks and nodes. 181resgid=%d The group ID which may use the reserved blocks and nodes. 182fault_injection=%d Enable fault injection in all supported types with 183 specified injection rate. 184fault_type=%d Support configuring fault injection type, should be 185 enabled with fault_injection option, fault type value 186 is shown below, it supports single or combined type. 187 188 =========================== =========== 189 Type_Name Type_Value 190 =========================== =========== 191 FAULT_KMALLOC 0x000000001 192 FAULT_KVMALLOC 0x000000002 193 FAULT_PAGE_ALLOC 0x000000004 194 FAULT_PAGE_GET 0x000000008 195 FAULT_ALLOC_BIO 0x000000010 (obsolete) 196 FAULT_ALLOC_NID 0x000000020 197 FAULT_ORPHAN 0x000000040 198 FAULT_BLOCK 0x000000080 199 FAULT_DIR_DEPTH 0x000000100 200 FAULT_EVICT_INODE 0x000000200 201 FAULT_TRUNCATE 0x000000400 202 FAULT_READ_IO 0x000000800 203 FAULT_CHECKPOINT 0x000001000 204 FAULT_DISCARD 0x000002000 205 FAULT_WRITE_IO 0x000004000 206 FAULT_SLAB_ALLOC 0x000008000 207 FAULT_DQUOT_INIT 0x000010000 208 FAULT_LOCK_OP 0x000020000 209 FAULT_BLKADDR_VALIDITY 0x000040000 210 FAULT_BLKADDR_CONSISTENCE 0x000080000 211 FAULT_NO_SEGMENT 0x000100000 212 FAULT_INCONSISTENT_FOOTER 0x000200000 213 =========================== =========== 214mode=%s Control block allocation mode which supports "adaptive" 215 and "lfs". In "lfs" mode, there should be no random 216 writes towards main area. 217 "fragment:segment" and "fragment:block" are newly added here. 218 These are developer options for experiments to simulate filesystem 219 fragmentation/after-GC situation itself. The developers use these 220 modes to understand filesystem fragmentation/after-GC condition well, 221 and eventually get some insights to handle them better. 222 In "fragment:segment", f2fs allocates a new segment in ramdom 223 position. With this, we can simulate the after-GC condition. 224 In "fragment:block", we can scatter block allocation with 225 "max_fragment_chunk" and "max_fragment_hole" sysfs nodes. 226 We added some randomness to both chunk and hole size to make 227 it close to realistic IO pattern. So, in this mode, f2fs will allocate 228 1..<max_fragment_chunk> blocks in a chunk and make a hole in the 229 length of 1..<max_fragment_hole> by turns. With this, the newly 230 allocated blocks will be scattered throughout the whole partition. 231 Note that "fragment:block" implicitly enables "fragment:segment" 232 option for more randomness. 233 Please, use these options for your experiments and we strongly 234 recommend to re-format the filesystem after using these options. 235usrquota Enable plain user disk quota accounting. 236grpquota Enable plain group disk quota accounting. 237prjquota Enable plain project quota accounting. 238usrjquota=<file> Appoint specified file and type during mount, so that quota 239grpjquota=<file> information can be properly updated during recovery flow, 240prjjquota=<file> <quota file>: must be in root directory; 241jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1]. 242usrjquota= Turn off user journalled quota. 243grpjquota= Turn off group journalled quota. 244prjjquota= Turn off project journalled quota. 245quota Enable plain user disk quota accounting. 246noquota Disable all plain disk quota option. 247alloc_mode=%s Adjust block allocation policy, which supports "reuse" 248 and "default". 249fsync_mode=%s Control the policy of fsync. Currently supports "posix", 250 "strict", and "nobarrier". In "posix" mode, which is 251 default, fsync will follow POSIX semantics and does a 252 light operation to improve the filesystem performance. 253 In "strict" mode, fsync will be heavy and behaves in line 254 with xfs, ext4 and btrfs, where xfstest generic/342 will 255 pass, but the performance will regress. "nobarrier" is 256 based on "posix", but doesn't issue flush command for 257 non-atomic files likewise "nobarrier" mount option. 258test_dummy_encryption 259test_dummy_encryption=%s 260 Enable dummy encryption, which provides a fake fscrypt 261 context. The fake fscrypt context is used by xfstests. 262 The argument may be either "v1" or "v2", in order to 263 select the corresponding fscrypt policy version. 264checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable" 265 to reenable checkpointing. Is enabled by default. While 266 disabled, any unmounting or unexpected shutdowns will cause 267 the filesystem contents to appear as they did when the 268 filesystem was mounted with that option. 269 While mounting with checkpoint=disable, the filesystem must 270 run garbage collection to ensure that all available space can 271 be used. If this takes too much time, the mount may return 272 EAGAIN. You may optionally add a value to indicate how much 273 of the disk you would be willing to temporarily give up to 274 avoid additional garbage collection. This can be given as a 275 number of blocks, or as a percent. For instance, mounting 276 with checkpoint=disable:100% would always succeed, but it may 277 hide up to all remaining free space. The actual space that 278 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable 279 This space is reclaimed once checkpoint=enable. 280checkpoint_merge When checkpoint is enabled, this can be used to create a kernel 281 daemon and make it to merge concurrent checkpoint requests as 282 much as possible to eliminate redundant checkpoint issues. Plus, 283 we can eliminate the sluggish issue caused by slow checkpoint 284 operation when the checkpoint is done in a process context in 285 a cgroup having low i/o budget and cpu shares. To make this 286 do better, we set the default i/o priority of the kernel daemon 287 to "3", to give one higher priority than other kernel threads. 288 This is the same way to give a I/O priority to the jbd2 289 journaling thread of ext4 filesystem. 290nocheckpoint_merge Disable checkpoint merge feature. 291compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo", 292 "lz4", "zstd" and "lzo-rle" algorithm. 293compress_algorithm=%s:%d Control compress algorithm and its compress level, now, only 294 "lz4" and "zstd" support compress level config. 295 algorithm level range 296 lz4 3 - 16 297 zstd 1 - 22 298compress_log_size=%u Support configuring compress cluster size. The size will 299 be 4KB * (1 << %u). The default and minimum sizes are 16KB. 300compress_extension=%s Support adding specified extension, so that f2fs can enable 301 compression on those corresponding files, e.g. if all files 302 with '.ext' has high compression rate, we can set the '.ext' 303 on compression extension list and enable compression on 304 these file by default rather than to enable it via ioctl. 305 For other files, we can still enable compression via ioctl. 306 Note that, there is one reserved special extension '*', it 307 can be set to enable compression for all files. 308nocompress_extension=%s Support adding specified extension, so that f2fs can disable 309 compression on those corresponding files, just contrary to compression extension. 310 If you know exactly which files cannot be compressed, you can use this. 311 The same extension name can't appear in both compress and nocompress 312 extension at the same time. 313 If the compress extension specifies all files, the types specified by the 314 nocompress extension will be treated as special cases and will not be compressed. 315 Don't allow use '*' to specifie all file in nocompress extension. 316 After add nocompress_extension, the priority should be: 317 dir_flag < comp_extention,nocompress_extension < comp_file_flag,no_comp_file_flag. 318 See more in compression sections. 319 320compress_chksum Support verifying chksum of raw data in compressed cluster. 321compress_mode=%s Control file compression mode. This supports "fs" and "user" 322 modes. In "fs" mode (default), f2fs does automatic compression 323 on the compression enabled files. In "user" mode, f2fs disables 324 the automaic compression and gives the user discretion of 325 choosing the target file and the timing. The user can do manual 326 compression/decompression on the compression enabled files using 327 ioctls. 328compress_cache Support to use address space of a filesystem managed inode to 329 cache compressed block, in order to improve cache hit ratio of 330 random read. 331inlinecrypt When possible, encrypt/decrypt the contents of encrypted 332 files using the blk-crypto framework rather than 333 filesystem-layer encryption. This allows the use of 334 inline encryption hardware. The on-disk format is 335 unaffected. For more details, see 336 Documentation/block/inline-encryption.rst. 337atgc Enable age-threshold garbage collection, it provides high 338 effectiveness and efficiency on background GC. 339discard_unit=%s Control discard unit, the argument can be "block", "segment" 340 and "section", issued discard command's offset/size will be 341 aligned to the unit, by default, "discard_unit=block" is set, 342 so that small discard functionality is enabled. 343 For blkzoned device, "discard_unit=section" will be set by 344 default, it is helpful for large sized SMR or ZNS devices to 345 reduce memory cost by getting rid of fs metadata supports small 346 discard. 347memory=%s Control memory mode. This supports "normal" and "low" modes. 348 "low" mode is introduced to support low memory devices. 349 Because of the nature of low memory devices, in this mode, f2fs 350 will try to save memory sometimes by sacrificing performance. 351 "normal" mode is the default mode and same as before. 352age_extent_cache Enable an age extent cache based on rb-tree. It records 353 data block update frequency of the extent per inode, in 354 order to provide better temperature hints for data block 355 allocation. 356errors=%s Specify f2fs behavior on critical errors. This supports modes: 357 "panic", "continue" and "remount-ro", respectively, trigger 358 panic immediately, continue without doing anything, and remount 359 the partition in read-only mode. By default it uses "continue" 360 mode. 361 ====================== =============== =============== ======== 362 mode continue remount-ro panic 363 ====================== =============== =============== ======== 364 access ops normal normal N/A 365 syscall errors -EIO -EROFS N/A 366 mount option rw ro N/A 367 pending dir write keep keep N/A 368 pending non-dir write drop keep N/A 369 pending node write drop keep N/A 370 pending meta write keep keep N/A 371 ====================== =============== =============== ======== 372nat_bits Enable nat_bits feature to enhance full/empty nat blocks access, 373 by default it's disabled. 374lookup_mode=%s Control the directory lookup behavior for casefolded 375 directories. This option has no effect on directories 376 that do not have the casefold feature enabled. 377 378 ================== ======================================== 379 Value Description 380 ================== ======================================== 381 perf (Default) Enforces a hash-only lookup. 382 The linear search fallback is always 383 disabled, ignoring the on-disk flag. 384 compat Enables the linear search fallback for 385 compatibility with directory entries 386 created by older kernel that used a 387 different case-folding algorithm. 388 This mode ignores the on-disk flag. 389 auto F2FS determines the mode based on the 390 on-disk `SB_ENC_NO_COMPAT_FALLBACK_FL` 391 flag. 392 ================== ======================================== 393======================== ============================================================ 394 395Debugfs Entries 396=============== 397 398/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as 399f2fs. Each file shows the whole f2fs information. 400 401/sys/kernel/debug/f2fs/status includes: 402 403 - major file system information managed by f2fs currently 404 - average SIT information about whole segments 405 - current memory footprint consumed by f2fs. 406 407Sysfs Entries 408============= 409 410Information about mounted f2fs file systems can be found in 411/sys/fs/f2fs. Each mounted filesystem will have a directory in 412/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda). 413The files in each per-device directory are shown in table below. 414 415Files in /sys/fs/f2fs/<devname> 416(see also Documentation/ABI/testing/sysfs-fs-f2fs) 417 418Usage 419===== 420 4211. Download userland tools and compile them. 422 4232. Skip, if f2fs was compiled statically inside kernel. 424 Otherwise, insert the f2fs.ko module:: 425 426 # insmod f2fs.ko 427 4283. Create a directory to use when mounting:: 429 430 # mkdir /mnt/f2fs 431 4324. Format the block device, and then mount as f2fs:: 433 434 # mkfs.f2fs -l label /dev/block_device 435 # mount -t f2fs /dev/block_device /mnt/f2fs 436 437mkfs.f2fs 438--------- 439The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem, 440which builds a basic on-disk layout. 441 442The quick options consist of: 443 444=============== =========================================================== 445``-l [label]`` Give a volume label, up to 512 unicode name. 446``-a [0 or 1]`` Split start location of each area for heap-based allocation. 447 448 1 is set by default, which performs this. 449``-o [int]`` Set overprovision ratio in percent over volume size. 450 451 5 is set by default. 452``-s [int]`` Set the number of segments per section. 453 454 1 is set by default. 455``-z [int]`` Set the number of sections per zone. 456 457 1 is set by default. 458``-e [str]`` Set basic extension list. e.g. "mp3,gif,mov" 459``-t [0 or 1]`` Disable discard command or not. 460 461 1 is set by default, which conducts discard. 462=============== =========================================================== 463 464Note: please refer to the manpage of mkfs.f2fs(8) to get full option list. 465 466fsck.f2fs 467--------- 468The fsck.f2fs is a tool to check the consistency of an f2fs-formatted 469partition, which examines whether the filesystem metadata and user-made data 470are cross-referenced correctly or not. 471Note that, initial version of the tool does not fix any inconsistency. 472 473The quick options consist of:: 474 475 -d debug level [default:0] 476 477Note: please refer to the manpage of fsck.f2fs(8) to get full option list. 478 479dump.f2fs 480--------- 481The dump.f2fs shows the information of specific inode and dumps SSA and SIT to 482file. Each file is dump_ssa and dump_sit. 483 484The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem. 485It shows on-disk inode information recognized by a given inode number, and is 486able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and 487./dump_sit respectively. 488 489The options consist of:: 490 491 -d debug level [default:0] 492 -i inode no (hex) 493 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1] 494 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1] 495 496Examples:: 497 498 # dump.f2fs -i [ino] /dev/sdx 499 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 500 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 501 502Note: please refer to the manpage of dump.f2fs(8) to get full option list. 503 504sload.f2fs 505---------- 506The sload.f2fs gives a way to insert files and directories in the existing disk 507image. This tool is useful when building f2fs images given compiled files. 508 509Note: please refer to the manpage of sload.f2fs(8) to get full option list. 510 511resize.f2fs 512----------- 513The resize.f2fs lets a user resize the f2fs-formatted disk image, while preserving 514all the files and directories stored in the image. 515 516Note: please refer to the manpage of resize.f2fs(8) to get full option list. 517 518defrag.f2fs 519----------- 520The defrag.f2fs can be used to defragment scattered written data as well as 521filesystem metadata across the disk. This can improve the write speed by giving 522more free consecutive space. 523 524Note: please refer to the manpage of defrag.f2fs(8) to get full option list. 525 526f2fs_io 527------- 528The f2fs_io is a simple tool to issue various filesystem APIs as well as 529f2fs-specific ones, which is very useful for QA tests. 530 531Note: please refer to the manpage of f2fs_io(8) to get full option list. 532 533Design 534====== 535 536On-disk Layout 537-------------- 538 539F2FS divides the whole volume into a number of segments, each of which is fixed 540to 2MB in size. A section is composed of consecutive segments, and a zone 541consists of a set of sections. By default, section and zone sizes are set to one 542segment size identically, but users can easily modify the sizes by mkfs. 543 544F2FS splits the entire volume into six areas, and all the areas except superblock 545consist of multiple segments as described below:: 546 547 align with the zone size <-| 548 |-> align with the segment size 549 _________________________________________________________________________ 550 | | | Segment | Node | Segment | | 551 | Superblock | Checkpoint | Info. | Address | Summary | Main | 552 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | | 553 |____________|_____2______|______N______|______N______|______N_____|__N___| 554 . . 555 . . 556 . . 557 ._________________________________________. 558 |_Segment_|_..._|_Segment_|_..._|_Segment_| 559 . . 560 ._________._________ 561 |_section_|__...__|_ 562 . . 563 .________. 564 |__zone__| 565 566- Superblock (SB) 567 It is located at the beginning of the partition, and there exist two copies 568 to avoid file system crash. It contains basic partition information and some 569 default parameters of f2fs. 570 571- Checkpoint (CP) 572 It contains file system information, bitmaps for valid NAT/SIT sets, orphan 573 inode lists, and summary entries of current active segments. 574 575- Segment Information Table (SIT) 576 It contains segment information such as valid block count and bitmap for the 577 validity of all the blocks. 578 579- Node Address Table (NAT) 580 It is composed of a block address table for all the node blocks stored in 581 Main area. 582 583- Segment Summary Area (SSA) 584 It contains summary entries which contains the owner information of all the 585 data and node blocks stored in Main area. 586 587- Main Area 588 It contains file and directory data including their indices. 589 590In order to avoid misalignment between file system and flash-based storage, F2FS 591aligns the start block address of CP with the segment size. Also, it aligns the 592start block address of Main area with the zone size by reserving some segments 593in SSA area. 594 595Reference the following survey for additional technical details. 596https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey 597 598File System Metadata Structure 599------------------------------ 600 601F2FS adopts the checkpointing scheme to maintain file system consistency. At 602mount time, F2FS first tries to find the last valid checkpoint data by scanning 603CP area. In order to reduce the scanning time, F2FS uses only two copies of CP. 604One of them always indicates the last valid data, which is called as shadow copy 605mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism. 606 607For file system consistency, each CP points to which NAT and SIT copies are 608valid, as shown as below:: 609 610 +--------+----------+---------+ 611 | CP | SIT | NAT | 612 +--------+----------+---------+ 613 . . . . 614 . . . . 615 . . . . 616 +-------+-------+--------+--------+--------+--------+ 617 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 | 618 +-------+-------+--------+--------+--------+--------+ 619 | ^ ^ 620 | | | 621 `----------------------------------------' 622 623Index Structure 624--------------- 625 626The key data structure to manage the data locations is a "node". Similar to 627traditional file structures, F2FS has three types of node: inode, direct node, 628indirect node. F2FS assigns 4KB to an inode block which contains 923 data block 629indices, two direct node pointers, two indirect node pointers, and one double 630indirect node pointer as described below. One direct node block contains 1018 631data blocks, and one indirect node block contains also 1018 node blocks. Thus, 632one inode block (i.e., a file) covers:: 633 634 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB. 635 636 Inode block (4KB) 637 |- data (923) 638 |- direct node (2) 639 | `- data (1018) 640 |- indirect node (2) 641 | `- direct node (1018) 642 | `- data (1018) 643 `- double indirect node (1) 644 `- indirect node (1018) 645 `- direct node (1018) 646 `- data (1018) 647 648Note that all the node blocks are mapped by NAT which means the location of 649each node is translated by the NAT table. In the consideration of the wandering 650tree problem, F2FS is able to cut off the propagation of node updates caused by 651leaf data writes. 652 653Directory Structure 654------------------- 655 656A directory entry occupies 11 bytes, which consists of the following attributes. 657 658- hash hash value of the file name 659- ino inode number 660- len the length of file name 661- type file type such as directory, symlink, etc 662 663A dentry block consists of 214 dentry slots and file names. Therein a bitmap is 664used to represent whether each dentry is valid or not. A dentry block occupies 6654KB with the following composition. 666 667:: 668 669 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) + 670 dentries(11 * 214 bytes) + file name (8 * 214 bytes) 671 672 [Bucket] 673 +--------------------------------+ 674 |dentry block 1 | dentry block 2 | 675 +--------------------------------+ 676 . . 677 . . 678 . [Dentry Block Structure: 4KB] . 679 +--------+----------+----------+------------+ 680 | bitmap | reserved | dentries | file names | 681 +--------+----------+----------+------------+ 682 [Dentry Block: 4KB] . . 683 . . 684 . . 685 +------+------+-----+------+ 686 | hash | ino | len | type | 687 +------+------+-----+------+ 688 [Dentry Structure: 11 bytes] 689 690F2FS implements multi-level hash tables for directory structure. Each level has 691a hash table with dedicated number of hash buckets as shown below. Note that 692"A(2B)" means a bucket includes 2 data blocks. 693 694:: 695 696 ---------------------- 697 A : bucket 698 B : block 699 N : MAX_DIR_HASH_DEPTH 700 ---------------------- 701 702 level #0 | A(2B) 703 | 704 level #1 | A(2B) - A(2B) 705 | 706 level #2 | A(2B) - A(2B) - A(2B) - A(2B) 707 . | . . . . 708 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) 709 . | . . . . 710 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) 711 712The number of blocks and buckets are determined by:: 713 714 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2, 715 # of blocks in level #n = | 716 `- 4, Otherwise 717 718 ,- 2^(n + dir_level), 719 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2, 720 # of buckets in level #n = | 721 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), 722 Otherwise 723 724When F2FS finds a file name in a directory, at first a hash value of the file 725name is calculated. Then, F2FS scans the hash table in level #0 to find the 726dentry consisting of the file name and its inode number. If not found, F2FS 727scans the next hash table in level #1. In this way, F2FS scans hash tables in 728each levels incrementally from 1 to N. In each level F2FS needs to scan only 729one bucket determined by the following equation, which shows O(log(# of files)) 730complexity:: 731 732 bucket number to scan in level #n = (hash value) % (# of buckets in level #n) 733 734In the case of file creation, F2FS finds empty consecutive slots that cover the 735file name. F2FS searches the empty slots in the hash tables of whole levels from 7361 to N in the same way as the lookup operation. 737 738The following figure shows an example of two cases holding children:: 739 740 --------------> Dir <-------------- 741 | | 742 child child 743 744 child - child [hole] - child 745 746 child - child - child [hole] - [hole] - child 747 748 Case 1: Case 2: 749 Number of children = 6, Number of children = 3, 750 File size = 7 File size = 7 751 752Default Block Allocation 753------------------------ 754 755At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node 756and Hot/Warm/Cold data. 757 758- Hot node contains direct node blocks of directories. 759- Warm node contains direct node blocks except hot node blocks. 760- Cold node contains indirect node blocks 761- Hot data contains dentry blocks 762- Warm data contains data blocks except hot and cold data blocks 763- Cold data contains multimedia data or migrated data blocks 764 765LFS has two schemes for free space management: threaded log and copy-and-compac- 766tion. The copy-and-compaction scheme which is known as cleaning, is well-suited 767for devices showing very good sequential write performance, since free segments 768are served all the time for writing new data. However, it suffers from cleaning 769overhead under high utilization. Contrarily, the threaded log scheme suffers 770from random writes, but no cleaning process is needed. F2FS adopts a hybrid 771scheme where the copy-and-compaction scheme is adopted by default, but the 772policy is dynamically changed to the threaded log scheme according to the file 773system status. 774 775In order to align F2FS with underlying flash-based storage, F2FS allocates a 776segment in a unit of section. F2FS expects that the section size would be the 777same as the unit size of garbage collection in FTL. Furthermore, with respect 778to the mapping granularity in FTL, F2FS allocates each section of the active 779logs from different zones as much as possible, since FTL can write the data in 780the active logs into one allocation unit according to its mapping granularity. 781 782Cleaning process 783---------------- 784 785F2FS does cleaning both on demand and in the background. On-demand cleaning is 786triggered when there are not enough free segments to serve VFS calls. Background 787cleaner is operated by a kernel thread, and triggers the cleaning job when the 788system is idle. 789 790F2FS supports two victim selection policies: greedy and cost-benefit algorithms. 791In the greedy algorithm, F2FS selects a victim segment having the smallest number 792of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment 793according to the segment age and the number of valid blocks in order to address 794log block thrashing problem in the greedy algorithm. F2FS adopts the greedy 795algorithm for on-demand cleaner, while background cleaner adopts cost-benefit 796algorithm. 797 798In order to identify whether the data in the victim segment are valid or not, 799F2FS manages a bitmap. Each bit represents the validity of a block, and the 800bitmap is composed of a bit stream covering whole blocks in main area. 801 802Write-hint Policy 803----------------- 804 805F2FS sets the whint all the time with the below policy. 806 807===================== ======================== =================== 808User F2FS Block 809===================== ======================== =================== 810N/A META WRITE_LIFE_NONE|REQ_META 811N/A HOT_NODE WRITE_LIFE_NONE 812N/A WARM_NODE WRITE_LIFE_MEDIUM 813N/A COLD_NODE WRITE_LIFE_LONG 814ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME 815extension list " " 816 817-- buffered io 818N/A COLD_DATA WRITE_LIFE_EXTREME 819N/A HOT_DATA WRITE_LIFE_SHORT 820N/A WARM_DATA WRITE_LIFE_NOT_SET 821 822-- direct io 823WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 824WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 825WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 826WRITE_LIFE_NONE " WRITE_LIFE_NONE 827WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM 828WRITE_LIFE_LONG " WRITE_LIFE_LONG 829===================== ======================== =================== 830 831Fallocate(2) Policy 832------------------- 833 834The default policy follows the below POSIX rule. 835 836Allocating disk space 837 The default operation (i.e., mode is zero) of fallocate() allocates 838 the disk space within the range specified by offset and len. The 839 file size (as reported by stat(2)) will be changed if offset+len is 840 greater than the file size. Any subregion within the range specified 841 by offset and len that did not contain data before the call will be 842 initialized to zero. This default behavior closely resembles the 843 behavior of the posix_fallocate(3) library function, and is intended 844 as a method of optimally implementing that function. 845 846However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to 847fallocate(fd, DEFAULT_MODE), it allocates on-disk block addresses having 848zero or random data, which is useful to the below scenario where: 849 850 1. create(fd) 851 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 852 3. fallocate(fd, 0, 0, size) 853 4. address = fibmap(fd, offset) 854 5. open(blkdev) 855 6. write(blkdev, address) 856 857Compression implementation 858-------------------------- 859 860- New term named cluster is defined as basic unit of compression, file can 861 be divided into multiple clusters logically. One cluster includes 4 << n 862 (n >= 0) logical pages, compression size is also cluster size, each of 863 cluster can be compressed or not. 864 865- In cluster metadata layout, one special block address is used to indicate 866 a cluster is a compressed one or normal one; for compressed cluster, following 867 metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs 868 stores data including compress header and compressed data. 869 870- In order to eliminate write amplification during overwrite, F2FS only 871 support compression on write-once file, data can be compressed only when 872 all logical blocks in cluster contain valid data and compress ratio of 873 cluster data is lower than specified threshold. 874 875- To enable compression on regular inode, there are four ways: 876 877 * chattr +c file 878 * chattr +c dir; touch dir/file 879 * mount w/ -o compress_extension=ext; touch file.ext 880 * mount w/ -o compress_extension=*; touch any_file 881 882- To disable compression on regular inode, there are two ways: 883 884 * chattr -c file 885 * mount w/ -o nocompress_extension=ext; touch file.ext 886 887- Priority in between FS_COMPR_FL, FS_NOCOMP_FS, extensions: 888 889 * compress_extension=so; nocompress_extension=zip; chattr +c dir; touch 890 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so and baz.txt 891 should be compresse, bar.zip should be non-compressed. chattr +c dir/bar.zip 892 can enable compress on bar.zip. 893 * compress_extension=so; nocompress_extension=zip; chattr -c dir; touch 894 dir/foo.so; touch dir/bar.zip; touch dir/baz.txt; then foo.so should be 895 compresse, bar.zip and baz.txt should be non-compressed. 896 chattr+c dir/bar.zip; chattr+c dir/baz.txt; can enable compress on bar.zip 897 and baz.txt. 898 899- At this point, compression feature doesn't expose compressed space to user 900 directly in order to guarantee potential data updates later to the space. 901 Instead, the main goal is to reduce data writes to flash disk as much as 902 possible, resulting in extending disk life time as well as relaxing IO 903 congestion. Alternatively, we've added ioctl(F2FS_IOC_RELEASE_COMPRESS_BLOCKS) 904 interface to reclaim compressed space and show it to user after setting a 905 special flag to the inode. Once the compressed space is released, the flag 906 will block writing data to the file until either the compressed space is 907 reserved via ioctl(F2FS_IOC_RESERVE_COMPRESS_BLOCKS) or the file size is 908 truncated to zero. 909 910Compress metadata layout:: 911 912 [Dnode Structure] 913 +-----------------------------------------------+ 914 | cluster 1 | cluster 2 | ......... | cluster N | 915 +-----------------------------------------------+ 916 . . . . 917 . . . . 918 . Compressed Cluster . . Normal Cluster . 919 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 920 |compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 | 921 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 922 . . 923 . . 924 . . 925 +-------------+-------------+----------+----------------------------+ 926 | data length | data chksum | reserved | compressed data | 927 +-------------+-------------+----------+----------------------------+ 928 929Compression mode 930-------------------------- 931 932f2fs supports "fs" and "user" compression modes with "compression_mode" mount option. 933With this option, f2fs provides a choice to select the way how to compress the 934compression enabled files (refer to "Compression implementation" section for how to 935enable compression on a regular inode). 936 9371) compress_mode=fs 938This is the default option. f2fs does automatic compression in the writeback of the 939compression enabled files. 940 9412) compress_mode=user 942This disables the automatic compression and gives the user discretion of choosing the 943target file and the timing. The user can do manual compression/decompression on the 944compression enabled files using F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE 945ioctls like the below. 946 947To decompress a file, 948 949fd = open(filename, O_WRONLY, 0); 950ret = ioctl(fd, F2FS_IOC_DECOMPRESS_FILE); 951 952To compress a file, 953 954fd = open(filename, O_WRONLY, 0); 955ret = ioctl(fd, F2FS_IOC_COMPRESS_FILE); 956 957NVMe Zoned Namespace devices 958---------------------------- 959 960- ZNS defines a per-zone capacity which can be equal or less than the 961 zone-size. Zone-capacity is the number of usable blocks in the zone. 962 F2FS checks if zone-capacity is less than zone-size, if it is, then any 963 segment which starts after the zone-capacity is marked as not-free in 964 the free segment bitmap at initial mount time. These segments are marked 965 as permanently used so they are not allocated for writes and 966 consequently are not needed to be garbage collected. In case the 967 zone-capacity is not aligned to default segment size(2MB), then a segment 968 can start before the zone-capacity and span across zone-capacity boundary. 969 Such spanning segments are also considered as usable segments. All blocks 970 past the zone-capacity are considered unusable in these segments. 971 972Device aliasing feature 973----------------------- 974 975f2fs can utilize a special file called a "device aliasing file." This file allows 976the entire storage device to be mapped with a single, large extent, not using 977the usual f2fs node structures. This mapped area is pinned and primarily intended 978for holding the space. 979 980Essentially, this mechanism allows a portion of the f2fs area to be temporarily 981reserved and used by another filesystem or for different purposes. Once that 982external usage is complete, the device aliasing file can be deleted, releasing 983the reserved space back to F2FS for its own use. 984 985<use-case> 986 987# ls /dev/vd* 988/dev/vdb (32GB) /dev/vdc (32GB) 989# mkfs.ext4 /dev/vdc 990# mkfs.f2fs -c /dev/vdc@vdc.file /dev/vdb 991# mount /dev/vdb /mnt/f2fs 992# ls -l /mnt/f2fs 993vdc.file 994# df -h 995/dev/vdb 64G 33G 32G 52% /mnt/f2fs 996 997# mount -o loop /dev/vdc /mnt/ext4 998# df -h 999/dev/vdb 64G 33G 32G 52% /mnt/f2fs 1000/dev/loop7 32G 24K 30G 1% /mnt/ext4 1001# umount /mnt/ext4 1002 1003# f2fs_io getflags /mnt/f2fs/vdc.file 1004get a flag on /mnt/f2fs/vdc.file ret=0, flags=nocow(pinned),immutable 1005# f2fs_io setflags noimmutable /mnt/f2fs/vdc.file 1006get a flag on noimmutable ret=0, flags=800010 1007set a flag on /mnt/f2fs/vdc.file ret=0, flags=noimmutable 1008# rm /mnt/f2fs/vdc.file 1009# df -h 1010/dev/vdb 64G 753M 64G 2% /mnt/f2fs 1011 1012So, the key idea is, user can do any file operations on /dev/vdc, and 1013reclaim the space after the use, while the space is counted as /data. 1014That doesn't require modifying partition size and filesystem format. 1015