1================================================================================ 2WHAT IS Flash-Friendly File System (F2FS)? 3================================================================================ 4 5NAND flash memory-based storage devices, such as SSD, eMMC, and SD cards, have 6been equipped on a variety systems ranging from mobile to server systems. Since 7they are known to have different characteristics from the conventional rotating 8disks, a file system, an upper layer to the storage device, should adapt to the 9changes from the sketch in the design level. 10 11F2FS is a file system exploiting NAND flash memory-based storage devices, which 12is based on Log-structured File System (LFS). The design has been focused on 13addressing the fundamental issues in LFS, which are snowball effect of wandering 14tree and high cleaning overhead. 15 16Since a NAND flash memory-based storage device shows different characteristic 17according to its internal geometry or flash memory management scheme, namely FTL, 18F2FS and its tools support various parameters not only for configuring on-disk 19layout, but also for selecting allocation and cleaning algorithms. 20 21The following git tree provides the file system formatting tool (mkfs.f2fs), 22a consistency checking tool (fsck.f2fs), and a debugging tool (dump.f2fs). 23>> git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs-tools.git 24 25For reporting bugs and sending patches, please use the following mailing list: 26>> linux-f2fs-devel@lists.sourceforge.net 27 28================================================================================ 29BACKGROUND AND DESIGN ISSUES 30================================================================================ 31 32Log-structured File System (LFS) 33-------------------------------- 34"A log-structured file system writes all modifications to disk sequentially in 35a log-like structure, thereby speeding up both file writing and crash recovery. 36The log is the only structure on disk; it contains indexing information so that 37files can be read back from the log efficiently. In order to maintain large free 38areas on disk for fast writing, we divide the log into segments and use a 39segment cleaner to compress the live information from heavily fragmented 40segments." from Rosenblum, M. and Ousterhout, J. K., 1992, "The design and 41implementation of a log-structured file system", ACM Trans. Computer Systems 4210, 1, 26–52. 43 44Wandering Tree Problem 45---------------------- 46In LFS, when a file data is updated and written to the end of log, its direct 47pointer block is updated due to the changed location. Then the indirect pointer 48block is also updated due to the direct pointer block update. In this manner, 49the upper index structures such as inode, inode map, and checkpoint block are 50also updated recursively. This problem is called as wandering tree problem [1], 51and in order to enhance the performance, it should eliminate or relax the update 52propagation as much as possible. 53 54[1] Bityutskiy, A. 2005. JFFS3 design issues. http://www.linux-mtd.infradead.org/ 55 56Cleaning Overhead 57----------------- 58Since LFS is based on out-of-place writes, it produces so many obsolete blocks 59scattered across the whole storage. In order to serve new empty log space, it 60needs to reclaim these obsolete blocks seamlessly to users. This job is called 61as a cleaning process. 62 63The process consists of three operations as follows. 641. A victim segment is selected through referencing segment usage table. 652. It loads parent index structures of all the data in the victim identified by 66 segment summary blocks. 673. It checks the cross-reference between the data and its parent index structure. 684. It moves valid data selectively. 69 70This cleaning job may cause unexpected long delays, so the most important goal 71is to hide the latencies to users. And also definitely, it should reduce the 72amount of valid data to be moved, and move them quickly as well. 73 74================================================================================ 75KEY FEATURES 76================================================================================ 77 78Flash Awareness 79--------------- 80- Enlarge the random write area for better performance, but provide the high 81 spatial locality 82- Align FS data structures to the operational units in FTL as best efforts 83 84Wandering Tree Problem 85---------------------- 86- Use a term, “node”, that represents inodes as well as various pointer blocks 87- Introduce Node Address Table (NAT) containing the locations of all the “node” 88 blocks; this will cut off the update propagation. 89 90Cleaning Overhead 91----------------- 92- Support a background cleaning process 93- Support greedy and cost-benefit algorithms for victim selection policies 94- Support multi-head logs for static/dynamic hot and cold data separation 95- Introduce adaptive logging for efficient block allocation 96 97================================================================================ 98MOUNT OPTIONS 99================================================================================ 100 101background_gc=%s Turn on/off cleaning operations, namely garbage 102 collection, triggered in background when I/O subsystem is 103 idle. If background_gc=on, it will turn on the garbage 104 collection and if background_gc=off, garbage collection 105 will be turned off. If background_gc=sync, it will turn 106 on synchronous garbage collection running in background. 107 Default value for this option is on. So garbage 108 collection is on by default. 109disable_roll_forward Disable the roll-forward recovery routine 110norecovery Disable the roll-forward recovery routine, mounted read- 111 only (i.e., -o ro,disable_roll_forward) 112discard/nodiscard Enable/disable real-time discard in f2fs, if discard is 113 enabled, f2fs will issue discard/TRIM commands when a 114 segment is cleaned. 115no_heap Disable heap-style segment allocation which finds free 116 segments for data from the beginning of main area, while 117 for node from the end of main area. 118nouser_xattr Disable Extended User Attributes. Note: xattr is enabled 119 by default if CONFIG_F2FS_FS_XATTR is selected. 120noacl Disable POSIX Access Control List. Note: acl is enabled 121 by default if CONFIG_F2FS_FS_POSIX_ACL is selected. 122active_logs=%u Support configuring the number of active logs. In the 123 current design, f2fs supports only 2, 4, and 6 logs. 124 Default number is 6. 125disable_ext_identify Disable the extension list configured by mkfs, so f2fs 126 does not aware of cold files such as media files. 127inline_xattr Enable the inline xattrs feature. 128noinline_xattr Disable the inline xattrs feature. 129inline_xattr_size=%u Support configuring inline xattr size, it depends on 130 flexible inline xattr feature. 131inline_data Enable the inline data feature: New created small(<~3.4k) 132 files can be written into inode block. 133inline_dentry Enable the inline dir feature: data in new created 134 directory entries can be written into inode block. The 135 space of inode block which is used to store inline 136 dentries is limited to ~3.4k. 137noinline_dentry Disable the inline dentry feature. 138flush_merge Merge concurrent cache_flush commands as much as possible 139 to eliminate redundant command issues. If the underlying 140 device handles the cache_flush command relatively slowly, 141 recommend to enable this option. 142nobarrier This option can be used if underlying storage guarantees 143 its cached data should be written to the novolatile area. 144 If this option is set, no cache_flush commands are issued 145 but f2fs still guarantees the write ordering of all the 146 data writes. 147fastboot This option is used when a system wants to reduce mount 148 time as much as possible, even though normal performance 149 can be sacrificed. 150extent_cache Enable an extent cache based on rb-tree, it can cache 151 as many as extent which map between contiguous logical 152 address and physical address per inode, resulting in 153 increasing the cache hit ratio. Set by default. 154noextent_cache Disable an extent cache based on rb-tree explicitly, see 155 the above extent_cache mount option. 156noinline_data Disable the inline data feature, inline data feature is 157 enabled by default. 158data_flush Enable data flushing before checkpoint in order to 159 persist data of regular and symlink. 160reserve_root=%d Support configuring reserved space which is used for 161 allocation from a privileged user with specified uid or 162 gid, unit: 4KB, the default limit is 0.2% of user blocks. 163resuid=%d The user ID which may use the reserved blocks. 164resgid=%d The group ID which may use the reserved blocks. 165fault_injection=%d Enable fault injection in all supported types with 166 specified injection rate. 167fault_type=%d Support configuring fault injection type, should be 168 enabled with fault_injection option, fault type value 169 is shown below, it supports single or combined type. 170 Type_Name Type_Value 171 FAULT_KMALLOC 0x000000001 172 FAULT_KVMALLOC 0x000000002 173 FAULT_PAGE_ALLOC 0x000000004 174 FAULT_PAGE_GET 0x000000008 175 FAULT_ALLOC_BIO 0x000000010 176 FAULT_ALLOC_NID 0x000000020 177 FAULT_ORPHAN 0x000000040 178 FAULT_BLOCK 0x000000080 179 FAULT_DIR_DEPTH 0x000000100 180 FAULT_EVICT_INODE 0x000000200 181 FAULT_TRUNCATE 0x000000400 182 FAULT_READ_IO 0x000000800 183 FAULT_CHECKPOINT 0x000001000 184 FAULT_DISCARD 0x000002000 185 FAULT_WRITE_IO 0x000004000 186mode=%s Control block allocation mode which supports "adaptive" 187 and "lfs". In "lfs" mode, there should be no random 188 writes towards main area. 189io_bits=%u Set the bit size of write IO requests. It should be set 190 with "mode=lfs". 191usrquota Enable plain user disk quota accounting. 192grpquota Enable plain group disk quota accounting. 193prjquota Enable plain project quota accounting. 194usrjquota=<file> Appoint specified file and type during mount, so that quota 195grpjquota=<file> information can be properly updated during recovery flow, 196prjjquota=<file> <quota file>: must be in root directory; 197jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1]. 198offusrjquota Turn off user journelled quota. 199offgrpjquota Turn off group journelled quota. 200offprjjquota Turn off project journelled quota. 201quota Enable plain user disk quota accounting. 202noquota Disable all plain disk quota option. 203whint_mode=%s Control which write hints are passed down to block 204 layer. This supports "off", "user-based", and 205 "fs-based". In "off" mode (default), f2fs does not pass 206 down hints. In "user-based" mode, f2fs tries to pass 207 down hints given by users. And in "fs-based" mode, f2fs 208 passes down hints with its policy. 209alloc_mode=%s Adjust block allocation policy, which supports "reuse" 210 and "default". 211fsync_mode=%s Control the policy of fsync. Currently supports "posix", 212 "strict", and "nobarrier". In "posix" mode, which is 213 default, fsync will follow POSIX semantics and does a 214 light operation to improve the filesystem performance. 215 In "strict" mode, fsync will be heavy and behaves in line 216 with xfs, ext4 and btrfs, where xfstest generic/342 will 217 pass, but the performance will regress. "nobarrier" is 218 based on "posix", but doesn't issue flush command for 219 non-atomic files likewise "nobarrier" mount option. 220test_dummy_encryption 221test_dummy_encryption=%s 222 Enable dummy encryption, which provides a fake fscrypt 223 context. The fake fscrypt context is used by xfstests. 224 The argument may be either "v1" or "v2", in order to 225 select the corresponding fscrypt policy version. 226checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable" 227 to reenable checkpointing. Is enabled by default. While 228 disabled, any unmounting or unexpected shutdowns will cause 229 the filesystem contents to appear as they did when the 230 filesystem was mounted with that option. 231 While mounting with checkpoint=disabled, the filesystem must 232 run garbage collection to ensure that all available space can 233 be used. If this takes too much time, the mount may return 234 EAGAIN. You may optionally add a value to indicate how much 235 of the disk you would be willing to temporarily give up to 236 avoid additional garbage collection. This can be given as a 237 number of blocks, or as a percent. For instance, mounting 238 with checkpoint=disable:100% would always succeed, but it may 239 hide up to all remaining free space. The actual space that 240 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable 241 This space is reclaimed once checkpoint=enable. 242compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo", 243 "lz4" and "zstd" algorithm. 244compress_log_size=%u Support configuring compress cluster size, the size will 245 be 4KB * (1 << %u), 16KB is minimum size, also it's 246 default size. 247compress_extension=%s Support adding specified extension, so that f2fs can enable 248 compression on those corresponding files, e.g. if all files 249 with '.ext' has high compression rate, we can set the '.ext' 250 on compression extension list and enable compression on 251 these file by default rather than to enable it via ioctl. 252 For other files, we can still enable compression via ioctl. 253 254================================================================================ 255DEBUGFS ENTRIES 256================================================================================ 257 258/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as 259f2fs. Each file shows the whole f2fs information. 260 261/sys/kernel/debug/f2fs/status includes: 262 - major file system information managed by f2fs currently 263 - average SIT information about whole segments 264 - current memory footprint consumed by f2fs. 265 266================================================================================ 267SYSFS ENTRIES 268================================================================================ 269 270Information about mounted f2fs file systems can be found in 271/sys/fs/f2fs. Each mounted filesystem will have a directory in 272/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda). 273The files in each per-device directory are shown in table below. 274 275Files in /sys/fs/f2fs/<devname> 276(see also Documentation/ABI/testing/sysfs-fs-f2fs) 277 278================================================================================ 279USAGE 280================================================================================ 281 2821. Download userland tools and compile them. 283 2842. Skip, if f2fs was compiled statically inside kernel. 285 Otherwise, insert the f2fs.ko module. 286 # insmod f2fs.ko 287 2883. Create a directory trying to mount 289 # mkdir /mnt/f2fs 290 2914. Format the block device, and then mount as f2fs 292 # mkfs.f2fs -l label /dev/block_device 293 # mount -t f2fs /dev/block_device /mnt/f2fs 294 295mkfs.f2fs 296--------- 297The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem, 298which builds a basic on-disk layout. 299 300The options consist of: 301-l [label] : Give a volume label, up to 512 unicode name. 302-a [0 or 1] : Split start location of each area for heap-based allocation. 303 1 is set by default, which performs this. 304-o [int] : Set overprovision ratio in percent over volume size. 305 5 is set by default. 306-s [int] : Set the number of segments per section. 307 1 is set by default. 308-z [int] : Set the number of sections per zone. 309 1 is set by default. 310-e [str] : Set basic extension list. e.g. "mp3,gif,mov" 311-t [0 or 1] : Disable discard command or not. 312 1 is set by default, which conducts discard. 313 314fsck.f2fs 315--------- 316The fsck.f2fs is a tool to check the consistency of an f2fs-formatted 317partition, which examines whether the filesystem metadata and user-made data 318are cross-referenced correctly or not. 319Note that, initial version of the tool does not fix any inconsistency. 320 321The options consist of: 322 -d debug level [default:0] 323 324dump.f2fs 325--------- 326The dump.f2fs shows the information of specific inode and dumps SSA and SIT to 327file. Each file is dump_ssa and dump_sit. 328 329The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem. 330It shows on-disk inode information recognized by a given inode number, and is 331able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and 332./dump_sit respectively. 333 334The options consist of: 335 -d debug level [default:0] 336 -i inode no (hex) 337 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1] 338 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1] 339 340Examples: 341# dump.f2fs -i [ino] /dev/sdx 342# dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 343# dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 344 345================================================================================ 346DESIGN 347================================================================================ 348 349On-disk Layout 350-------------- 351 352F2FS divides the whole volume into a number of segments, each of which is fixed 353to 2MB in size. A section is composed of consecutive segments, and a zone 354consists of a set of sections. By default, section and zone sizes are set to one 355segment size identically, but users can easily modify the sizes by mkfs. 356 357F2FS splits the entire volume into six areas, and all the areas except superblock 358consists of multiple segments as described below. 359 360 align with the zone size <-| 361 |-> align with the segment size 362 _________________________________________________________________________ 363 | | | Segment | Node | Segment | | 364 | Superblock | Checkpoint | Info. | Address | Summary | Main | 365 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | | 366 |____________|_____2______|______N______|______N______|______N_____|__N___| 367 . . 368 . . 369 . . 370 ._________________________________________. 371 |_Segment_|_..._|_Segment_|_..._|_Segment_| 372 . . 373 ._________._________ 374 |_section_|__...__|_ 375 . . 376 .________. 377 |__zone__| 378 379- Superblock (SB) 380 : It is located at the beginning of the partition, and there exist two copies 381 to avoid file system crash. It contains basic partition information and some 382 default parameters of f2fs. 383 384- Checkpoint (CP) 385 : It contains file system information, bitmaps for valid NAT/SIT sets, orphan 386 inode lists, and summary entries of current active segments. 387 388- Segment Information Table (SIT) 389 : It contains segment information such as valid block count and bitmap for the 390 validity of all the blocks. 391 392- Node Address Table (NAT) 393 : It is composed of a block address table for all the node blocks stored in 394 Main area. 395 396- Segment Summary Area (SSA) 397 : It contains summary entries which contains the owner information of all the 398 data and node blocks stored in Main area. 399 400- Main Area 401 : It contains file and directory data including their indices. 402 403In order to avoid misalignment between file system and flash-based storage, F2FS 404aligns the start block address of CP with the segment size. Also, it aligns the 405start block address of Main area with the zone size by reserving some segments 406in SSA area. 407 408Reference the following survey for additional technical details. 409https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey 410 411File System Metadata Structure 412------------------------------ 413 414F2FS adopts the checkpointing scheme to maintain file system consistency. At 415mount time, F2FS first tries to find the last valid checkpoint data by scanning 416CP area. In order to reduce the scanning time, F2FS uses only two copies of CP. 417One of them always indicates the last valid data, which is called as shadow copy 418mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism. 419 420For file system consistency, each CP points to which NAT and SIT copies are 421valid, as shown as below. 422 423 +--------+----------+---------+ 424 | CP | SIT | NAT | 425 +--------+----------+---------+ 426 . . . . 427 . . . . 428 . . . . 429 +-------+-------+--------+--------+--------+--------+ 430 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 | 431 +-------+-------+--------+--------+--------+--------+ 432 | ^ ^ 433 | | | 434 `----------------------------------------' 435 436Index Structure 437--------------- 438 439The key data structure to manage the data locations is a "node". Similar to 440traditional file structures, F2FS has three types of node: inode, direct node, 441indirect node. F2FS assigns 4KB to an inode block which contains 923 data block 442indices, two direct node pointers, two indirect node pointers, and one double 443indirect node pointer as described below. One direct node block contains 1018 444data blocks, and one indirect node block contains also 1018 node blocks. Thus, 445one inode block (i.e., a file) covers: 446 447 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB. 448 449 Inode block (4KB) 450 |- data (923) 451 |- direct node (2) 452 | `- data (1018) 453 |- indirect node (2) 454 | `- direct node (1018) 455 | `- data (1018) 456 `- double indirect node (1) 457 `- indirect node (1018) 458 `- direct node (1018) 459 `- data (1018) 460 461Note that, all the node blocks are mapped by NAT which means the location of 462each node is translated by the NAT table. In the consideration of the wandering 463tree problem, F2FS is able to cut off the propagation of node updates caused by 464leaf data writes. 465 466Directory Structure 467------------------- 468 469A directory entry occupies 11 bytes, which consists of the following attributes. 470 471- hash hash value of the file name 472- ino inode number 473- len the length of file name 474- type file type such as directory, symlink, etc 475 476A dentry block consists of 214 dentry slots and file names. Therein a bitmap is 477used to represent whether each dentry is valid or not. A dentry block occupies 4784KB with the following composition. 479 480 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) + 481 dentries(11 * 214 bytes) + file name (8 * 214 bytes) 482 483 [Bucket] 484 +--------------------------------+ 485 |dentry block 1 | dentry block 2 | 486 +--------------------------------+ 487 . . 488 . . 489 . [Dentry Block Structure: 4KB] . 490 +--------+----------+----------+------------+ 491 | bitmap | reserved | dentries | file names | 492 +--------+----------+----------+------------+ 493 [Dentry Block: 4KB] . . 494 . . 495 . . 496 +------+------+-----+------+ 497 | hash | ino | len | type | 498 +------+------+-----+------+ 499 [Dentry Structure: 11 bytes] 500 501F2FS implements multi-level hash tables for directory structure. Each level has 502a hash table with dedicated number of hash buckets as shown below. Note that 503"A(2B)" means a bucket includes 2 data blocks. 504 505---------------------- 506A : bucket 507B : block 508N : MAX_DIR_HASH_DEPTH 509---------------------- 510 511level #0 | A(2B) 512 | 513level #1 | A(2B) - A(2B) 514 | 515level #2 | A(2B) - A(2B) - A(2B) - A(2B) 516 . | . . . . 517level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B) 518 . | . . . . 519level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B) 520 521The number of blocks and buckets are determined by, 522 523 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2, 524 # of blocks in level #n = | 525 `- 4, Otherwise 526 527 ,- 2^(n + dir_level), 528 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2, 529 # of buckets in level #n = | 530 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1), 531 Otherwise 532 533When F2FS finds a file name in a directory, at first a hash value of the file 534name is calculated. Then, F2FS scans the hash table in level #0 to find the 535dentry consisting of the file name and its inode number. If not found, F2FS 536scans the next hash table in level #1. In this way, F2FS scans hash tables in 537each levels incrementally from 1 to N. In each levels F2FS needs to scan only 538one bucket determined by the following equation, which shows O(log(# of files)) 539complexity. 540 541 bucket number to scan in level #n = (hash value) % (# of buckets in level #n) 542 543In the case of file creation, F2FS finds empty consecutive slots that cover the 544file name. F2FS searches the empty slots in the hash tables of whole levels from 5451 to N in the same way as the lookup operation. 546 547The following figure shows an example of two cases holding children. 548 --------------> Dir <-------------- 549 | | 550 child child 551 552 child - child [hole] - child 553 554 child - child - child [hole] - [hole] - child 555 556 Case 1: Case 2: 557 Number of children = 6, Number of children = 3, 558 File size = 7 File size = 7 559 560Default Block Allocation 561------------------------ 562 563At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node 564and Hot/Warm/Cold data. 565 566- Hot node contains direct node blocks of directories. 567- Warm node contains direct node blocks except hot node blocks. 568- Cold node contains indirect node blocks 569- Hot data contains dentry blocks 570- Warm data contains data blocks except hot and cold data blocks 571- Cold data contains multimedia data or migrated data blocks 572 573LFS has two schemes for free space management: threaded log and copy-and-compac- 574tion. The copy-and-compaction scheme which is known as cleaning, is well-suited 575for devices showing very good sequential write performance, since free segments 576are served all the time for writing new data. However, it suffers from cleaning 577overhead under high utilization. Contrarily, the threaded log scheme suffers 578from random writes, but no cleaning process is needed. F2FS adopts a hybrid 579scheme where the copy-and-compaction scheme is adopted by default, but the 580policy is dynamically changed to the threaded log scheme according to the file 581system status. 582 583In order to align F2FS with underlying flash-based storage, F2FS allocates a 584segment in a unit of section. F2FS expects that the section size would be the 585same as the unit size of garbage collection in FTL. Furthermore, with respect 586to the mapping granularity in FTL, F2FS allocates each section of the active 587logs from different zones as much as possible, since FTL can write the data in 588the active logs into one allocation unit according to its mapping granularity. 589 590Cleaning process 591---------------- 592 593F2FS does cleaning both on demand and in the background. On-demand cleaning is 594triggered when there are not enough free segments to serve VFS calls. Background 595cleaner is operated by a kernel thread, and triggers the cleaning job when the 596system is idle. 597 598F2FS supports two victim selection policies: greedy and cost-benefit algorithms. 599In the greedy algorithm, F2FS selects a victim segment having the smallest number 600of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment 601according to the segment age and the number of valid blocks in order to address 602log block thrashing problem in the greedy algorithm. F2FS adopts the greedy 603algorithm for on-demand cleaner, while background cleaner adopts cost-benefit 604algorithm. 605 606In order to identify whether the data in the victim segment are valid or not, 607F2FS manages a bitmap. Each bit represents the validity of a block, and the 608bitmap is composed of a bit stream covering whole blocks in main area. 609 610Write-hint Policy 611----------------- 612 6131) whint_mode=off. F2FS only passes down WRITE_LIFE_NOT_SET. 614 6152) whint_mode=user-based. F2FS tries to pass down hints given by 616users. 617 618User F2FS Block 619---- ---- ----- 620 META WRITE_LIFE_NOT_SET 621 HOT_NODE " 622 WARM_NODE " 623 COLD_NODE " 624*ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME 625*extension list " " 626 627-- buffered io 628WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 629WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 630WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 631WRITE_LIFE_NONE " " 632WRITE_LIFE_MEDIUM " " 633WRITE_LIFE_LONG " " 634 635-- direct io 636WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 637WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 638WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 639WRITE_LIFE_NONE " WRITE_LIFE_NONE 640WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM 641WRITE_LIFE_LONG " WRITE_LIFE_LONG 642 6433) whint_mode=fs-based. F2FS passes down hints with its policy. 644 645User F2FS Block 646---- ---- ----- 647 META WRITE_LIFE_MEDIUM; 648 HOT_NODE WRITE_LIFE_NOT_SET 649 WARM_NODE " 650 COLD_NODE WRITE_LIFE_NONE 651ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME 652extension list " " 653 654-- buffered io 655WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 656WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 657WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_LONG 658WRITE_LIFE_NONE " " 659WRITE_LIFE_MEDIUM " " 660WRITE_LIFE_LONG " " 661 662-- direct io 663WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME 664WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT 665WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET 666WRITE_LIFE_NONE " WRITE_LIFE_NONE 667WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM 668WRITE_LIFE_LONG " WRITE_LIFE_LONG 669 670Fallocate(2) Policy 671------------------- 672 673The default policy follows the below posix rule. 674 675Allocating disk space 676 The default operation (i.e., mode is zero) of fallocate() allocates 677 the disk space within the range specified by offset and len. The 678 file size (as reported by stat(2)) will be changed if offset+len is 679 greater than the file size. Any subregion within the range specified 680 by offset and len that did not contain data before the call will be 681 initialized to zero. This default behavior closely resembles the 682 behavior of the posix_fallocate(3) library function, and is intended 683 as a method of optimally implementing that function. 684 685However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to 686fallocate(fd, DEFAULT_MODE), it allocates on-disk blocks addressess having 687zero or random data, which is useful to the below scenario where: 688 1. create(fd) 689 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 690 3. fallocate(fd, 0, 0, size) 691 4. address = fibmap(fd, offset) 692 5. open(blkdev) 693 6. write(blkdev, address) 694 695Compression implementation 696-------------------------- 697 698- New term named cluster is defined as basic unit of compression, file can 699be divided into multiple clusters logically. One cluster includes 4 << n 700(n >= 0) logical pages, compression size is also cluster size, each of 701cluster can be compressed or not. 702 703- In cluster metadata layout, one special block address is used to indicate 704cluster is compressed one or normal one, for compressed cluster, following 705metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs 706stores data including compress header and compressed data. 707 708- In order to eliminate write amplification during overwrite, F2FS only 709support compression on write-once file, data can be compressed only when 710all logical blocks in file are valid and cluster compress ratio is lower 711than specified threshold. 712 713- To enable compression on regular inode, there are three ways: 714* chattr +c file 715* chattr +c dir; touch dir/file 716* mount w/ -o compress_extension=ext; touch file.ext 717 718Compress metadata layout: 719 [Dnode Structure] 720 +-----------------------------------------------+ 721 | cluster 1 | cluster 2 | ......... | cluster N | 722 +-----------------------------------------------+ 723 . . . . 724 . . . . 725 . Compressed Cluster . . Normal Cluster . 726+----------+---------+---------+---------+ +---------+---------+---------+---------+ 727|compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 | 728+----------+---------+---------+---------+ +---------+---------+---------+---------+ 729 . . 730 . . 731 . . 732 +-------------+-------------+----------+----------------------------+ 733 | data length | data chksum | reserved | compressed data | 734 +-------------+-------------+----------+----------------------------+ 735