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 reporting bugs and sending patches, please use the following mailing list: 29 30- linux-f2fs-devel@lists.sourceforge.net 31 32Background and Design issues 33============================ 34 35Log-structured File System (LFS) 36-------------------------------- 37"A log-structured file system writes all modifications to disk sequentially in 38a log-like structure, thereby speeding up both file writing and crash recovery. 39The log is the only structure on disk; it contains indexing information so that 40files can be read back from the log efficiently. In order to maintain large free 41areas on disk for fast writing, we divide the log into segments and use a 42segment cleaner to compress the live information from heavily fragmented 43segments." from Rosenblum, M. and Ousterhout, J. K., 1992, "The design and 44implementation of a log-structured file system", ACM Trans. Computer Systems 4510, 1, 26–52. 46 47Wandering Tree Problem 48---------------------- 49In LFS, when a file data is updated and written to the end of log, its direct 50pointer block is updated due to the changed location. Then the indirect pointer 51block is also updated due to the direct pointer block update. In this manner, 52the upper index structures such as inode, inode map, and checkpoint block are 53also updated recursively. This problem is called as wandering tree problem [1], 54and in order to enhance the performance, it should eliminate or relax the update 55propagation as much as possible. 56 57[1] Bityutskiy, A. 2005. JFFS3 design issues. http://www.linux-mtd.infradead.org/ 58 59Cleaning Overhead 60----------------- 61Since LFS is based on out-of-place writes, it produces so many obsolete blocks 62scattered across the whole storage. In order to serve new empty log space, it 63needs to reclaim these obsolete blocks seamlessly to users. This job is called 64as a cleaning process. 65 66The process consists of three operations as follows. 67 681. A victim segment is selected through referencing segment usage table. 692. It loads parent index structures of all the data in the victim identified by 70 segment summary blocks. 713. It checks the cross-reference between the data and its parent index structure. 724. It moves valid data selectively. 73 74This cleaning job may cause unexpected long delays, so the most important goal 75is to hide the latencies to users. And also definitely, it should reduce the 76amount of valid data to be moved, and move them quickly as well. 77 78Key Features 79============ 80 81Flash Awareness 82--------------- 83- Enlarge the random write area for better performance, but provide the high 84 spatial locality 85- Align FS data structures to the operational units in FTL as best efforts 86 87Wandering Tree Problem 88---------------------- 89- Use a term, “node”, that represents inodes as well as various pointer blocks 90- Introduce Node Address Table (NAT) containing the locations of all the “node” 91 blocks; this will cut off the update propagation. 92 93Cleaning Overhead 94----------------- 95- Support a background cleaning process 96- Support greedy and cost-benefit algorithms for victim selection policies 97- Support multi-head logs for static/dynamic hot and cold data separation 98- Introduce adaptive logging for efficient block allocation 99 100Mount Options 101============= 102 103 104====================== ============================================================ 105background_gc=%s Turn on/off cleaning operations, namely garbage 106 collection, triggered in background when I/O subsystem is 107 idle. If background_gc=on, it will turn on the garbage 108 collection and if background_gc=off, garbage collection 109 will be turned off. If background_gc=sync, it will turn 110 on synchronous garbage collection running in background. 111 Default value for this option is on. So garbage 112 collection is on by default. 113disable_roll_forward Disable the roll-forward recovery routine 114norecovery Disable the roll-forward recovery routine, mounted read- 115 only (i.e., -o ro,disable_roll_forward) 116discard/nodiscard Enable/disable real-time discard in f2fs, if discard is 117 enabled, f2fs will issue discard/TRIM commands when a 118 segment is cleaned. 119no_heap Disable heap-style segment allocation which finds free 120 segments for data from the beginning of main area, while 121 for node from the end of main area. 122nouser_xattr Disable Extended User Attributes. Note: xattr is enabled 123 by default if CONFIG_F2FS_FS_XATTR is selected. 124noacl Disable POSIX Access Control List. Note: acl is enabled 125 by default if CONFIG_F2FS_FS_POSIX_ACL is selected. 126active_logs=%u Support configuring the number of active logs. In the 127 current design, f2fs supports only 2, 4, and 6 logs. 128 Default number is 6. 129disable_ext_identify Disable the extension list configured by mkfs, so f2fs 130 does not aware of cold files such as media files. 131inline_xattr Enable the inline xattrs feature. 132noinline_xattr Disable the inline xattrs feature. 133inline_xattr_size=%u Support configuring inline xattr size, it depends on 134 flexible inline xattr feature. 135inline_data Enable the inline data feature: New created small(<~3.4k) 136 files can be written into inode block. 137inline_dentry Enable the inline dir feature: data in new created 138 directory entries can be written into inode block. The 139 space of inode block which is used to store inline 140 dentries is limited to ~3.4k. 141noinline_dentry Disable the inline dentry feature. 142flush_merge Merge concurrent cache_flush commands as much as possible 143 to eliminate redundant command issues. If the underlying 144 device handles the cache_flush command relatively slowly, 145 recommend to enable this option. 146nobarrier This option can be used if underlying storage guarantees 147 its cached data should be written to the novolatile area. 148 If this option is set, no cache_flush commands are issued 149 but f2fs still guarantees the write ordering of all the 150 data writes. 151fastboot This option is used when a system wants to reduce mount 152 time as much as possible, even though normal performance 153 can be sacrificed. 154extent_cache Enable an extent cache based on rb-tree, it can cache 155 as many as extent which map between contiguous logical 156 address and physical address per inode, resulting in 157 increasing the cache hit ratio. Set by default. 158noextent_cache Disable an extent cache based on rb-tree explicitly, see 159 the above extent_cache mount option. 160noinline_data Disable the inline data feature, inline data feature is 161 enabled by default. 162data_flush Enable data flushing before checkpoint in order to 163 persist data of regular and symlink. 164reserve_root=%d Support configuring reserved space which is used for 165 allocation from a privileged user with specified uid or 166 gid, unit: 4KB, the default limit is 0.2% of user blocks. 167resuid=%d The user ID which may use the reserved blocks. 168resgid=%d The group ID which may use the reserved blocks. 169fault_injection=%d Enable fault injection in all supported types with 170 specified injection rate. 171fault_type=%d Support configuring fault injection type, should be 172 enabled with fault_injection option, fault type value 173 is shown below, it supports single or combined type. 174 175 =================== =========== 176 Type_Name Type_Value 177 =================== =========== 178 FAULT_KMALLOC 0x000000001 179 FAULT_KVMALLOC 0x000000002 180 FAULT_PAGE_ALLOC 0x000000004 181 FAULT_PAGE_GET 0x000000008 182 FAULT_ALLOC_BIO 0x000000010 183 FAULT_ALLOC_NID 0x000000020 184 FAULT_ORPHAN 0x000000040 185 FAULT_BLOCK 0x000000080 186 FAULT_DIR_DEPTH 0x000000100 187 FAULT_EVICT_INODE 0x000000200 188 FAULT_TRUNCATE 0x000000400 189 FAULT_READ_IO 0x000000800 190 FAULT_CHECKPOINT 0x000001000 191 FAULT_DISCARD 0x000002000 192 FAULT_WRITE_IO 0x000004000 193 =================== =========== 194mode=%s Control block allocation mode which supports "adaptive" 195 and "lfs". In "lfs" mode, there should be no random 196 writes towards main area. 197io_bits=%u Set the bit size of write IO requests. It should be set 198 with "mode=lfs". 199usrquota Enable plain user disk quota accounting. 200grpquota Enable plain group disk quota accounting. 201prjquota Enable plain project quota accounting. 202usrjquota=<file> Appoint specified file and type during mount, so that quota 203grpjquota=<file> information can be properly updated during recovery flow, 204prjjquota=<file> <quota file>: must be in root directory; 205jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1]. 206offusrjquota Turn off user journelled quota. 207offgrpjquota Turn off group journelled quota. 208offprjjquota Turn off project journelled quota. 209quota Enable plain user disk quota accounting. 210noquota Disable all plain disk quota option. 211whint_mode=%s Control which write hints are passed down to block 212 layer. This supports "off", "user-based", and 213 "fs-based". In "off" mode (default), f2fs does not pass 214 down hints. In "user-based" mode, f2fs tries to pass 215 down hints given by users. And in "fs-based" mode, f2fs 216 passes down hints with its policy. 217alloc_mode=%s Adjust block allocation policy, which supports "reuse" 218 and "default". 219fsync_mode=%s Control the policy of fsync. Currently supports "posix", 220 "strict", and "nobarrier". In "posix" mode, which is 221 default, fsync will follow POSIX semantics and does a 222 light operation to improve the filesystem performance. 223 In "strict" mode, fsync will be heavy and behaves in line 224 with xfs, ext4 and btrfs, where xfstest generic/342 will 225 pass, but the performance will regress. "nobarrier" is 226 based on "posix", but doesn't issue flush command for 227 non-atomic files likewise "nobarrier" mount option. 228test_dummy_encryption 229test_dummy_encryption=%s 230 Enable dummy encryption, which provides a fake fscrypt 231 context. The fake fscrypt context is used by xfstests. 232 The argument may be either "v1" or "v2", in order to 233 select the corresponding fscrypt policy version. 234checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable" 235 to reenable checkpointing. Is enabled by default. While 236 disabled, any unmounting or unexpected shutdowns will cause 237 the filesystem contents to appear as they did when the 238 filesystem was mounted with that option. 239 While mounting with checkpoint=disabled, the filesystem must 240 run garbage collection to ensure that all available space can 241 be used. If this takes too much time, the mount may return 242 EAGAIN. You may optionally add a value to indicate how much 243 of the disk you would be willing to temporarily give up to 244 avoid additional garbage collection. This can be given as a 245 number of blocks, or as a percent. For instance, mounting 246 with checkpoint=disable:100% would always succeed, but it may 247 hide up to all remaining free space. The actual space that 248 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable 249 This space is reclaimed once checkpoint=enable. 250compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo", 251 "lz4", "zstd" and "lzo-rle" algorithm. 252compress_log_size=%u Support configuring compress cluster size, the size will 253 be 4KB * (1 << %u), 16KB is minimum size, also it's 254 default size. 255compress_extension=%s Support adding specified extension, so that f2fs can enable 256 compression on those corresponding files, e.g. if all files 257 with '.ext' has high compression rate, we can set the '.ext' 258 on compression extension list and enable compression on 259 these file by default rather than to enable it via ioctl. 260 For other files, we can still enable compression via ioctl. 261====================== ============================================================ 262 263Debugfs Entries 264=============== 265 266/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as 267f2fs. Each file shows the whole f2fs information. 268 269/sys/kernel/debug/f2fs/status includes: 270 271 - major file system information managed by f2fs currently 272 - average SIT information about whole segments 273 - current memory footprint consumed by f2fs. 274 275Sysfs Entries 276============= 277 278Information about mounted f2fs file systems can be found in 279/sys/fs/f2fs. Each mounted filesystem will have a directory in 280/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda). 281The files in each per-device directory are shown in table below. 282 283Files in /sys/fs/f2fs/<devname> 284(see also Documentation/ABI/testing/sysfs-fs-f2fs) 285 286Usage 287===== 288 2891. Download userland tools and compile them. 290 2912. Skip, if f2fs was compiled statically inside kernel. 292 Otherwise, insert the f2fs.ko module:: 293 294 # insmod f2fs.ko 295 2963. Create a directory trying to mount:: 297 298 # mkdir /mnt/f2fs 299 3004. Format the block device, and then mount as f2fs:: 301 302 # mkfs.f2fs -l label /dev/block_device 303 # mount -t f2fs /dev/block_device /mnt/f2fs 304 305mkfs.f2fs 306--------- 307The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem, 308which builds a basic on-disk layout. 309 310The options consist of: 311 312=============== =========================================================== 313``-l [label]`` Give a volume label, up to 512 unicode name. 314``-a [0 or 1]`` Split start location of each area for heap-based allocation. 315 316 1 is set by default, which performs this. 317``-o [int]`` Set overprovision ratio in percent over volume size. 318 319 5 is set by default. 320``-s [int]`` Set the number of segments per section. 321 322 1 is set by default. 323``-z [int]`` Set the number of sections per zone. 324 325 1 is set by default. 326``-e [str]`` Set basic extension list. e.g. "mp3,gif,mov" 327``-t [0 or 1]`` Disable discard command or not. 328 329 1 is set by default, which conducts discard. 330=============== =========================================================== 331 332fsck.f2fs 333--------- 334The fsck.f2fs is a tool to check the consistency of an f2fs-formatted 335partition, which examines whether the filesystem metadata and user-made data 336are cross-referenced correctly or not. 337Note that, initial version of the tool does not fix any inconsistency. 338 339The options consist of:: 340 341 -d debug level [default:0] 342 343dump.f2fs 344--------- 345The dump.f2fs shows the information of specific inode and dumps SSA and SIT to 346file. Each file is dump_ssa and dump_sit. 347 348The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem. 349It shows on-disk inode information recognized by a given inode number, and is 350able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and 351./dump_sit respectively. 352 353The options consist of:: 354 355 -d debug level [default:0] 356 -i inode no (hex) 357 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1] 358 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1] 359 360Examples:: 361 362 # dump.f2fs -i [ino] /dev/sdx 363 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 364 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 365 366Design 367====== 368 369On-disk Layout 370-------------- 371 372F2FS divides the whole volume into a number of segments, each of which is fixed 373to 2MB in size. A section is composed of consecutive segments, and a zone 374consists of a set of sections. By default, section and zone sizes are set to one 375segment size identically, but users can easily modify the sizes by mkfs. 376 377F2FS splits the entire volume into six areas, and all the areas except superblock 378consists of multiple segments as described below:: 379 380 align with the zone size <-| 381 |-> align with the segment size 382 _________________________________________________________________________ 383 | | | Segment | Node | Segment | | 384 | Superblock | Checkpoint | Info. | Address | Summary | Main | 385 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | | 386 |____________|_____2______|______N______|______N______|______N_____|__N___| 387 . . 388 . . 389 . . 390 ._________________________________________. 391 |_Segment_|_..._|_Segment_|_..._|_Segment_| 392 . . 393 ._________._________ 394 |_section_|__...__|_ 395 . . 396 .________. 397 |__zone__| 398 399- Superblock (SB) 400 It is located at the beginning of the partition, and there exist two copies 401 to avoid file system crash. It contains basic partition information and some 402 default parameters of f2fs. 403 404- Checkpoint (CP) 405 It contains file system information, bitmaps for valid NAT/SIT sets, orphan 406 inode lists, and summary entries of current active segments. 407 408- Segment Information Table (SIT) 409 It contains segment information such as valid block count and bitmap for the 410 validity of all the blocks. 411 412- Node Address Table (NAT) 413 It is composed of a block address table for all the node blocks stored in 414 Main area. 415 416- Segment Summary Area (SSA) 417 It contains summary entries which contains the owner information of all the 418 data and node blocks stored in Main area. 419 420- Main Area 421 It contains file and directory data including their indices. 422 423In order to avoid misalignment between file system and flash-based storage, F2FS 424aligns the start block address of CP with the segment size. Also, it aligns the 425start block address of Main area with the zone size by reserving some segments 426in SSA area. 427 428Reference the following survey for additional technical details. 429https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey 430 431File System Metadata Structure 432------------------------------ 433 434F2FS adopts the checkpointing scheme to maintain file system consistency. At 435mount time, F2FS first tries to find the last valid checkpoint data by scanning 436CP area. In order to reduce the scanning time, F2FS uses only two copies of CP. 437One of them always indicates the last valid data, which is called as shadow copy 438mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism. 439 440For file system consistency, each CP points to which NAT and SIT copies are 441valid, as shown as below:: 442 443 +--------+----------+---------+ 444 | CP | SIT | NAT | 445 +--------+----------+---------+ 446 . . . . 447 . . . . 448 . . . . 449 +-------+-------+--------+--------+--------+--------+ 450 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 | 451 +-------+-------+--------+--------+--------+--------+ 452 | ^ ^ 453 | | | 454 `----------------------------------------' 455 456Index Structure 457--------------- 458 459The key data structure to manage the data locations is a "node". Similar to 460traditional file structures, F2FS has three types of node: inode, direct node, 461indirect node. F2FS assigns 4KB to an inode block which contains 923 data block 462indices, two direct node pointers, two indirect node pointers, and one double 463indirect node pointer as described below. One direct node block contains 1018 464data blocks, and one indirect node block contains also 1018 node blocks. Thus, 465one inode block (i.e., a file) covers:: 466 467 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB. 468 469 Inode block (4KB) 470 |- data (923) 471 |- direct node (2) 472 | `- data (1018) 473 |- indirect node (2) 474 | `- direct node (1018) 475 | `- data (1018) 476 `- double indirect node (1) 477 `- indirect node (1018) 478 `- direct node (1018) 479 `- data (1018) 480 481Note that, all the node blocks are mapped by NAT which means the location of 482each node is translated by the NAT table. In the consideration of the wandering 483tree problem, F2FS is able to cut off the propagation of node updates caused by 484leaf data writes. 485 486Directory Structure 487------------------- 488 489A directory entry occupies 11 bytes, which consists of the following attributes. 490 491- hash hash value of the file name 492- ino inode number 493- len the length of file name 494- type file type such as directory, symlink, etc 495 496A dentry block consists of 214 dentry slots and file names. Therein a bitmap is 497used to represent whether each dentry is valid or not. A dentry block occupies 4984KB with the following composition. 499 500:: 501 502 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) + 503 dentries(11 * 214 bytes) + file name (8 * 214 bytes) 504 505 [Bucket] 506 +--------------------------------+ 507 |dentry block 1 | dentry block 2 | 508 +--------------------------------+ 509 . . 510 . . 511 . [Dentry Block Structure: 4KB] . 512 +--------+----------+----------+------------+ 513 | bitmap | reserved | dentries | file names | 514 +--------+----------+----------+------------+ 515 [Dentry Block: 4KB] . . 516 . . 517 . . 518 +------+------+-----+------+ 519 | hash | ino | len | type | 520 +------+------+-----+------+ 521 [Dentry Structure: 11 bytes] 522 523F2FS implements multi-level hash tables for directory structure. Each level has 524a hash table with dedicated number of hash buckets as shown below. Note that 525"A(2B)" means a bucket includes 2 data blocks. 526 527:: 528 529 ---------------------- 530 A : bucket 531 B : block 532 N : MAX_DIR_HASH_DEPTH 533 ---------------------- 534 535 level #0 | A(2B) 536 | 537 level #1 | A(2B) - A(2B) 538 | 539 level #2 | A(2B) - A(2B) - A(2B) - A(2B) 540 . | . . . . 541 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) 542 . | . . . . 543 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) 544 545The number of blocks and buckets are determined by:: 546 547 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2, 548 # of blocks in level #n = | 549 `- 4, Otherwise 550 551 ,- 2^(n + dir_level), 552 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2, 553 # of buckets in level #n = | 554 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), 555 Otherwise 556 557When F2FS finds a file name in a directory, at first a hash value of the file 558name is calculated. Then, F2FS scans the hash table in level #0 to find the 559dentry consisting of the file name and its inode number. If not found, F2FS 560scans the next hash table in level #1. In this way, F2FS scans hash tables in 561each levels incrementally from 1 to N. In each levels F2FS needs to scan only 562one bucket determined by the following equation, which shows O(log(# of files)) 563complexity:: 564 565 bucket number to scan in level #n = (hash value) % (# of buckets in level #n) 566 567In the case of file creation, F2FS finds empty consecutive slots that cover the 568file name. F2FS searches the empty slots in the hash tables of whole levels from 5691 to N in the same way as the lookup operation. 570 571The following figure shows an example of two cases holding children:: 572 573 --------------> Dir <-------------- 574 | | 575 child child 576 577 child - child [hole] - child 578 579 child - child - child [hole] - [hole] - child 580 581 Case 1: Case 2: 582 Number of children = 6, Number of children = 3, 583 File size = 7 File size = 7 584 585Default Block Allocation 586------------------------ 587 588At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node 589and Hot/Warm/Cold data. 590 591- Hot node contains direct node blocks of directories. 592- Warm node contains direct node blocks except hot node blocks. 593- Cold node contains indirect node blocks 594- Hot data contains dentry blocks 595- Warm data contains data blocks except hot and cold data blocks 596- Cold data contains multimedia data or migrated data blocks 597 598LFS has two schemes for free space management: threaded log and copy-and-compac- 599tion. The copy-and-compaction scheme which is known as cleaning, is well-suited 600for devices showing very good sequential write performance, since free segments 601are served all the time for writing new data. However, it suffers from cleaning 602overhead under high utilization. Contrarily, the threaded log scheme suffers 603from random writes, but no cleaning process is needed. F2FS adopts a hybrid 604scheme where the copy-and-compaction scheme is adopted by default, but the 605policy is dynamically changed to the threaded log scheme according to the file 606system status. 607 608In order to align F2FS with underlying flash-based storage, F2FS allocates a 609segment in a unit of section. F2FS expects that the section size would be the 610same as the unit size of garbage collection in FTL. Furthermore, with respect 611to the mapping granularity in FTL, F2FS allocates each section of the active 612logs from different zones as much as possible, since FTL can write the data in 613the active logs into one allocation unit according to its mapping granularity. 614 615Cleaning process 616---------------- 617 618F2FS does cleaning both on demand and in the background. On-demand cleaning is 619triggered when there are not enough free segments to serve VFS calls. Background 620cleaner is operated by a kernel thread, and triggers the cleaning job when the 621system is idle. 622 623F2FS supports two victim selection policies: greedy and cost-benefit algorithms. 624In the greedy algorithm, F2FS selects a victim segment having the smallest number 625of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment 626according to the segment age and the number of valid blocks in order to address 627log block thrashing problem in the greedy algorithm. F2FS adopts the greedy 628algorithm for on-demand cleaner, while background cleaner adopts cost-benefit 629algorithm. 630 631In order to identify whether the data in the victim segment are valid or not, 632F2FS manages a bitmap. Each bit represents the validity of a block, and the 633bitmap is composed of a bit stream covering whole blocks in main area. 634 635Write-hint Policy 636----------------- 637 6381) whint_mode=off. F2FS only passes down WRITE_LIFE_NOT_SET. 639 6402) whint_mode=user-based. F2FS tries to pass down hints given by 641users. 642 643===================== ======================== =================== 644User F2FS Block 645===================== ======================== =================== 646 META WRITE_LIFE_NOT_SET 647 HOT_NODE " 648 WARM_NODE " 649 COLD_NODE " 650ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME 651extension list " " 652 653-- buffered io 654WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 655WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 656WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 657WRITE_LIFE_NONE " " 658WRITE_LIFE_MEDIUM " " 659WRITE_LIFE_LONG " " 660 661-- direct io 662WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 663WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 664WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 665WRITE_LIFE_NONE " WRITE_LIFE_NONE 666WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM 667WRITE_LIFE_LONG " WRITE_LIFE_LONG 668===================== ======================== =================== 669 6703) whint_mode=fs-based. F2FS passes down hints with its policy. 671 672===================== ======================== =================== 673User F2FS Block 674===================== ======================== =================== 675 META WRITE_LIFE_MEDIUM; 676 HOT_NODE WRITE_LIFE_NOT_SET 677 WARM_NODE " 678 COLD_NODE WRITE_LIFE_NONE 679ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME 680extension list " " 681 682-- buffered io 683WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 684WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 685WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_LONG 686WRITE_LIFE_NONE " " 687WRITE_LIFE_MEDIUM " " 688WRITE_LIFE_LONG " " 689 690-- direct io 691WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 692WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 693WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 694WRITE_LIFE_NONE " WRITE_LIFE_NONE 695WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM 696WRITE_LIFE_LONG " WRITE_LIFE_LONG 697===================== ======================== =================== 698 699Fallocate(2) Policy 700------------------- 701 702The default policy follows the below posix rule. 703 704Allocating disk space 705 The default operation (i.e., mode is zero) of fallocate() allocates 706 the disk space within the range specified by offset and len. The 707 file size (as reported by stat(2)) will be changed if offset+len is 708 greater than the file size. Any subregion within the range specified 709 by offset and len that did not contain data before the call will be 710 initialized to zero. This default behavior closely resembles the 711 behavior of the posix_fallocate(3) library function, and is intended 712 as a method of optimally implementing that function. 713 714However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to 715fallocate(fd, DEFAULT_MODE), it allocates on-disk blocks addressess having 716zero or random data, which is useful to the below scenario where: 717 718 1. create(fd) 719 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 720 3. fallocate(fd, 0, 0, size) 721 4. address = fibmap(fd, offset) 722 5. open(blkdev) 723 6. write(blkdev, address) 724 725Compression implementation 726-------------------------- 727 728- New term named cluster is defined as basic unit of compression, file can 729 be divided into multiple clusters logically. One cluster includes 4 << n 730 (n >= 0) logical pages, compression size is also cluster size, each of 731 cluster can be compressed or not. 732 733- In cluster metadata layout, one special block address is used to indicate 734 cluster is compressed one or normal one, for compressed cluster, following 735 metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs 736 stores data including compress header and compressed data. 737 738- In order to eliminate write amplification during overwrite, F2FS only 739 support compression on write-once file, data can be compressed only when 740 all logical blocks in file are valid and cluster compress ratio is lower 741 than specified threshold. 742 743- To enable compression on regular inode, there are three ways: 744 745 * chattr +c file 746 * chattr +c dir; touch dir/file 747 * mount w/ -o compress_extension=ext; touch file.ext 748 749Compress metadata layout:: 750 751 [Dnode Structure] 752 +-----------------------------------------------+ 753 | cluster 1 | cluster 2 | ......... | cluster N | 754 +-----------------------------------------------+ 755 . . . . 756 . . . . 757 . Compressed Cluster . . Normal Cluster . 758 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 759 |compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 | 760 +----------+---------+---------+---------+ +---------+---------+---------+---------+ 761 . . 762 . . 763 . . 764 +-------------+-------------+----------+----------------------------+ 765 | data length | data chksum | reserved | compressed data | 766 +-------------+-------------+----------+----------------------------+ 767