1 ================================================================================
2 WHAT IS Flash-Friendly File System (F2FS)?
3 ================================================================================
5 NAND flash memory-based storage devices, such as SSD, eMMC, and SD cards, have
6 been equipped on a variety systems ranging from mobile to server systems. Since
7 they are known to have different characteristics from the conventional rotating
8 disks, a file system, an upper layer to the storage device, should adapt to the
9 changes from the sketch in the design level.
11 F2FS is a file system exploiting NAND flash memory-based storage devices, which
12 is based on Log-structured File System (LFS). The design has been focused on
13 addressing the fundamental issues in LFS, which are snowball effect of wandering
14 tree and high cleaning overhead.
16 Since a NAND flash memory-based storage device shows different characteristic
17 according to its internal geometry or flash memory management scheme, namely FTL,
18 F2FS and its tools support various parameters not only for configuring on-disk
19 layout, but also for selecting allocation and cleaning algorithms.
21 The following git tree provides the file system formatting tool (mkfs.f2fs),
22 a 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
25 For reporting bugs and sending patches, please use the following mailing list:
26 >> linux-f2fs-devel@lists.sourceforge.net
28 ================================================================================
29 BACKGROUND AND DESIGN ISSUES
30 ================================================================================
32 Log-structured File System (LFS)
33 --------------------------------
34 "A log-structured file system writes all modifications to disk sequentially in
35 a log-like structure, thereby speeding up both file writing and crash recovery.
36 The log is the only structure on disk; it contains indexing information so that
37 files can be read back from the log efficiently. In order to maintain large free
38 areas on disk for fast writing, we divide the log into segments and use a
39 segment cleaner to compress the live information from heavily fragmented
40 segments." from Rosenblum, M. and Ousterhout, J. K., 1992, "The design and
41 implementation of a log-structured file system", ACM Trans. Computer Systems
44 Wandering Tree Problem
45 ----------------------
46 In LFS, when a file data is updated and written to the end of log, its direct
47 pointer block is updated due to the changed location. Then the indirect pointer
48 block is also updated due to the direct pointer block update. In this manner,
49 the upper index structures such as inode, inode map, and checkpoint block are
50 also updated recursively. This problem is called as wandering tree problem [1],
51 and in order to enhance the performance, it should eliminate or relax the update
52 propagation as much as possible.
54 [1] Bityutskiy, A. 2005. JFFS3 design issues. http://www.linux-mtd.infradead.org/
58 Since LFS is based on out-of-place writes, it produces so many obsolete blocks
59 scattered across the whole storage. In order to serve new empty log space, it
60 needs to reclaim these obsolete blocks seamlessly to users. This job is called
61 as a cleaning process.
63 The process consists of three operations as follows.
64 1. A victim segment is selected through referencing segment usage table.
65 2. It loads parent index structures of all the data in the victim identified by
66 segment summary blocks.
67 3. It checks the cross-reference between the data and its parent index structure.
68 4. It moves valid data selectively.
70 This cleaning job may cause unexpected long delays, so the most important goal
71 is to hide the latencies to users. And also definitely, it should reduce the
72 amount of valid data to be moved, and move them quickly as well.
74 ================================================================================
76 ================================================================================
80 - Enlarge the random write area for better performance, but provide the high
82 - Align FS data structures to the operational units in FTL as best efforts
84 Wandering 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.
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
97 ================================================================================
99 ================================================================================
101 background_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.
109 disable_roll_forward Disable the roll-forward recovery routine
110 norecovery Disable the roll-forward recovery routine, mounted read-
111 only (i.e., -o ro,disable_roll_forward)
112 discard Issue discard/TRIM commands when a segment is cleaned.
113 no_heap Disable heap-style segment allocation which finds free
114 segments for data from the beginning of main area, while
115 for node from the end of main area.
116 nouser_xattr Disable Extended User Attributes. Note: xattr is enabled
117 by default if CONFIG_F2FS_FS_XATTR is selected.
118 noacl Disable POSIX Access Control List. Note: acl is enabled
119 by default if CONFIG_F2FS_FS_POSIX_ACL is selected.
120 active_logs=%u Support configuring the number of active logs. In the
121 current design, f2fs supports only 2, 4, and 6 logs.
123 disable_ext_identify Disable the extension list configured by mkfs, so f2fs
124 does not aware of cold files such as media files.
125 inline_xattr Enable the inline xattrs feature.
126 inline_data Enable the inline data feature: New created small(<~3.4k)
127 files can be written into inode block.
128 inline_dentry Enable the inline dir feature: data in new created
129 directory entries can be written into inode block. The
130 space of inode block which is used to store inline
131 dentries is limited to ~3.4k.
132 flush_merge Merge concurrent cache_flush commands as much as possible
133 to eliminate redundant command issues. If the underlying
134 device handles the cache_flush command relatively slowly,
135 recommend to enable this option.
136 nobarrier This option can be used if underlying storage guarantees
137 its cached data should be written to the novolatile area.
138 If this option is set, no cache_flush commands are issued
139 but f2fs still guarantees the write ordering of all the
141 fastboot This option is used when a system wants to reduce mount
142 time as much as possible, even though normal performance
144 extent_cache Enable an extent cache based on rb-tree, it can cache
145 as many as extent which map between contiguous logical
146 address and physical address per inode, resulting in
147 increasing the cache hit ratio. Set by default.
148 noextent_cache Disable an extent cache based on rb-tree explicitly, see
149 the above extent_cache mount option.
150 noinline_data Disable the inline data feature, inline data feature is
152 data_flush Enable data flushing before checkpoint in order to
153 persist data of regular and symlink.
154 mode=%s Control block allocation mode which supports "adaptive"
155 and "lfs". In "lfs" mode, there should be no random
156 writes towards main area.
158 ================================================================================
160 ================================================================================
162 /sys/kernel/debug/f2fs/ contains information about all the partitions mounted as
163 f2fs. Each file shows the whole f2fs information.
165 /sys/kernel/debug/f2fs/status includes:
166 - major file system information managed by f2fs currently
167 - average SIT information about whole segments
168 - current memory footprint consumed by f2fs.
170 ================================================================================
172 ================================================================================
174 Information about mounted f2f2 file systems can be found in
175 /sys/fs/f2fs. Each mounted filesystem will have a directory in
176 /sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda).
177 The files in each per-device directory are shown in table below.
179 Files in /sys/fs/f2fs/<devname>
180 (see also Documentation/ABI/testing/sysfs-fs-f2fs)
181 ..............................................................................
184 gc_max_sleep_time This tuning parameter controls the maximum sleep
185 time for the garbage collection thread. Time is
188 gc_min_sleep_time This tuning parameter controls the minimum sleep
189 time for the garbage collection thread. Time is
192 gc_no_gc_sleep_time This tuning parameter controls the default sleep
193 time for the garbage collection thread. Time is
196 gc_idle This parameter controls the selection of victim
197 policy for garbage collection. Setting gc_idle = 0
198 (default) will disable this option. Setting
199 gc_idle = 1 will select the Cost Benefit approach
200 & setting gc_idle = 2 will select the greedy approach.
202 reclaim_segments This parameter controls the number of prefree
203 segments to be reclaimed. If the number of prefree
204 segments is larger than the number of segments
205 in the proportion to the percentage over total
206 volume size, f2fs tries to conduct checkpoint to
207 reclaim the prefree segments to free segments.
208 By default, 5% over total # of segments.
210 max_small_discards This parameter controls the number of discard
211 commands that consist small blocks less than 2MB.
212 The candidates to be discarded are cached until
213 checkpoint is triggered, and issued during the
214 checkpoint. By default, it is disabled with 0.
216 trim_sections This parameter controls the number of sections
217 to be trimmed out in batch mode when FITRIM
218 conducts. 32 sections is set by default.
220 ipu_policy This parameter controls the policy of in-place
221 updates in f2fs. There are five policies:
222 0x01: F2FS_IPU_FORCE, 0x02: F2FS_IPU_SSR,
223 0x04: F2FS_IPU_UTIL, 0x08: F2FS_IPU_SSR_UTIL,
224 0x10: F2FS_IPU_FSYNC.
226 min_ipu_util This parameter controls the threshold to trigger
227 in-place-updates. The number indicates percentage
228 of the filesystem utilization, and used by
229 F2FS_IPU_UTIL and F2FS_IPU_SSR_UTIL policies.
231 min_fsync_blocks This parameter controls the threshold to trigger
232 in-place-updates when F2FS_IPU_FSYNC mode is set.
233 The number indicates the number of dirty pages
234 when fsync needs to flush on its call path. If
235 the number is less than this value, it triggers
238 max_victim_search This parameter controls the number of trials to
239 find a victim segment when conducting SSR and
240 cleaning operations. The default value is 4096
241 which covers 8GB block address range.
243 dir_level This parameter controls the directory level to
244 support large directory. If a directory has a
245 number of files, it can reduce the file lookup
246 latency by increasing this dir_level value.
247 Otherwise, it needs to decrease this value to
248 reduce the space overhead. The default value is 0.
250 ram_thresh This parameter controls the memory footprint used
251 by free nids and cached nat entries. By default,
252 10 is set, which indicates 10 MB / 1 GB RAM.
254 ================================================================================
256 ================================================================================
258 1. Download userland tools and compile them.
260 2. Skip, if f2fs was compiled statically inside kernel.
261 Otherwise, insert the f2fs.ko module.
264 3. Create a directory trying to mount
267 4. Format the block device, and then mount as f2fs
268 # mkfs.f2fs -l label /dev/block_device
269 # mount -t f2fs /dev/block_device /mnt/f2fs
273 The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem,
274 which builds a basic on-disk layout.
276 The options consist of:
277 -l [label] : Give a volume label, up to 512 unicode name.
278 -a [0 or 1] : Split start location of each area for heap-based allocation.
279 1 is set by default, which performs this.
280 -o [int] : Set overprovision ratio in percent over volume size.
282 -s [int] : Set the number of segments per section.
284 -z [int] : Set the number of sections per zone.
286 -e [str] : Set basic extension list. e.g. "mp3,gif,mov"
287 -t [0 or 1] : Disable discard command or not.
288 1 is set by default, which conducts discard.
292 The fsck.f2fs is a tool to check the consistency of an f2fs-formatted
293 partition, which examines whether the filesystem metadata and user-made data
294 are cross-referenced correctly or not.
295 Note that, initial version of the tool does not fix any inconsistency.
297 The options consist of:
298 -d debug level [default:0]
302 The dump.f2fs shows the information of specific inode and dumps SSA and SIT to
303 file. Each file is dump_ssa and dump_sit.
305 The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem.
306 It shows on-disk inode information recognized by a given inode number, and is
307 able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and
308 ./dump_sit respectively.
310 The options consist of:
311 -d debug level [default:0]
313 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1]
314 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1]
317 # dump.f2fs -i [ino] /dev/sdx
318 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump)
319 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump)
321 ================================================================================
323 ================================================================================
328 F2FS divides the whole volume into a number of segments, each of which is fixed
329 to 2MB in size. A section is composed of consecutive segments, and a zone
330 consists of a set of sections. By default, section and zone sizes are set to one
331 segment size identically, but users can easily modify the sizes by mkfs.
333 F2FS splits the entire volume into six areas, and all the areas except superblock
334 consists of multiple segments as described below.
336 align with the zone size <-|
337 |-> align with the segment size
338 _________________________________________________________________________
339 | | | Segment | Node | Segment | |
340 | Superblock | Checkpoint | Info. | Address | Summary | Main |
341 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | |
342 |____________|_____2______|______N______|______N______|______N_____|__N___|
346 ._________________________________________.
347 |_Segment_|_..._|_Segment_|_..._|_Segment_|
356 : It is located at the beginning of the partition, and there exist two copies
357 to avoid file system crash. It contains basic partition information and some
358 default parameters of f2fs.
361 : It contains file system information, bitmaps for valid NAT/SIT sets, orphan
362 inode lists, and summary entries of current active segments.
364 - Segment Information Table (SIT)
365 : It contains segment information such as valid block count and bitmap for the
366 validity of all the blocks.
368 - Node Address Table (NAT)
369 : It is composed of a block address table for all the node blocks stored in
372 - Segment Summary Area (SSA)
373 : It contains summary entries which contains the owner information of all the
374 data and node blocks stored in Main area.
377 : It contains file and directory data including their indices.
379 In order to avoid misalignment between file system and flash-based storage, F2FS
380 aligns the start block address of CP with the segment size. Also, it aligns the
381 start block address of Main area with the zone size by reserving some segments
384 Reference the following survey for additional technical details.
385 https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey
387 File System Metadata Structure
388 ------------------------------
390 F2FS adopts the checkpointing scheme to maintain file system consistency. At
391 mount time, F2FS first tries to find the last valid checkpoint data by scanning
392 CP area. In order to reduce the scanning time, F2FS uses only two copies of CP.
393 One of them always indicates the last valid data, which is called as shadow copy
394 mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism.
396 For file system consistency, each CP points to which NAT and SIT copies are
397 valid, as shown as below.
399 +--------+----------+---------+
401 +--------+----------+---------+
405 +-------+-------+--------+--------+--------+--------+
406 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 |
407 +-------+-------+--------+--------+--------+--------+
410 `----------------------------------------'
415 The key data structure to manage the data locations is a "node". Similar to
416 traditional file structures, F2FS has three types of node: inode, direct node,
417 indirect node. F2FS assigns 4KB to an inode block which contains 923 data block
418 indices, two direct node pointers, two indirect node pointers, and one double
419 indirect node pointer as described below. One direct node block contains 1018
420 data blocks, and one indirect node block contains also 1018 node blocks. Thus,
421 one inode block (i.e., a file) covers:
423 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB.
430 | `- direct node (1018)
432 `- double indirect node (1)
433 `- indirect node (1018)
434 `- direct node (1018)
437 Note that, all the node blocks are mapped by NAT which means the location of
438 each node is translated by the NAT table. In the consideration of the wandering
439 tree problem, F2FS is able to cut off the propagation of node updates caused by
445 A directory entry occupies 11 bytes, which consists of the following attributes.
447 - hash hash value of the file name
449 - len the length of file name
450 - type file type such as directory, symlink, etc
452 A dentry block consists of 214 dentry slots and file names. Therein a bitmap is
453 used to represent whether each dentry is valid or not. A dentry block occupies
454 4KB with the following composition.
456 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) +
457 dentries(11 * 214 bytes) + file name (8 * 214 bytes)
460 +--------------------------------+
461 |dentry block 1 | dentry block 2 |
462 +--------------------------------+
465 . [Dentry Block Structure: 4KB] .
466 +--------+----------+----------+------------+
467 | bitmap | reserved | dentries | file names |
468 +--------+----------+----------+------------+
469 [Dentry Block: 4KB] . .
472 +------+------+-----+------+
473 | hash | ino | len | type |
474 +------+------+-----+------+
475 [Dentry Structure: 11 bytes]
477 F2FS implements multi-level hash tables for directory structure. Each level has
478 a hash table with dedicated number of hash buckets as shown below. Note that
479 "A(2B)" means a bucket includes 2 data blocks.
481 ----------------------
484 N : MAX_DIR_HASH_DEPTH
485 ----------------------
489 level #1 | A(2B) - A(2B)
491 level #2 | A(2B) - A(2B) - A(2B) - A(2B)
493 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B)
495 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B)
497 The number of blocks and buckets are determined by,
499 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2,
500 # of blocks in level #n = |
503 ,- 2^(n + dir_level),
504 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2,
505 # of buckets in level #n = |
506 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1),
509 When F2FS finds a file name in a directory, at first a hash value of the file
510 name is calculated. Then, F2FS scans the hash table in level #0 to find the
511 dentry consisting of the file name and its inode number. If not found, F2FS
512 scans the next hash table in level #1. In this way, F2FS scans hash tables in
513 each levels incrementally from 1 to N. In each levels F2FS needs to scan only
514 one bucket determined by the following equation, which shows O(log(# of files))
517 bucket number to scan in level #n = (hash value) % (# of buckets in level #n)
519 In the case of file creation, F2FS finds empty consecutive slots that cover the
520 file name. F2FS searches the empty slots in the hash tables of whole levels from
521 1 to N in the same way as the lookup operation.
523 The following figure shows an example of two cases holding children.
524 --------------> Dir <--------------
528 child - child [hole] - child
530 child - child - child [hole] - [hole] - child
533 Number of children = 6, Number of children = 3,
534 File size = 7 File size = 7
536 Default Block Allocation
537 ------------------------
539 At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node
540 and Hot/Warm/Cold data.
542 - Hot node contains direct node blocks of directories.
543 - Warm node contains direct node blocks except hot node blocks.
544 - Cold node contains indirect node blocks
545 - Hot data contains dentry blocks
546 - Warm data contains data blocks except hot and cold data blocks
547 - Cold data contains multimedia data or migrated data blocks
549 LFS has two schemes for free space management: threaded log and copy-and-compac-
550 tion. The copy-and-compaction scheme which is known as cleaning, is well-suited
551 for devices showing very good sequential write performance, since free segments
552 are served all the time for writing new data. However, it suffers from cleaning
553 overhead under high utilization. Contrarily, the threaded log scheme suffers
554 from random writes, but no cleaning process is needed. F2FS adopts a hybrid
555 scheme where the copy-and-compaction scheme is adopted by default, but the
556 policy is dynamically changed to the threaded log scheme according to the file
559 In order to align F2FS with underlying flash-based storage, F2FS allocates a
560 segment in a unit of section. F2FS expects that the section size would be the
561 same as the unit size of garbage collection in FTL. Furthermore, with respect
562 to the mapping granularity in FTL, F2FS allocates each section of the active
563 logs from different zones as much as possible, since FTL can write the data in
564 the active logs into one allocation unit according to its mapping granularity.
569 F2FS does cleaning both on demand and in the background. On-demand cleaning is
570 triggered when there are not enough free segments to serve VFS calls. Background
571 cleaner is operated by a kernel thread, and triggers the cleaning job when the
574 F2FS supports two victim selection policies: greedy and cost-benefit algorithms.
575 In the greedy algorithm, F2FS selects a victim segment having the smallest number
576 of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment
577 according to the segment age and the number of valid blocks in order to address
578 log block thrashing problem in the greedy algorithm. F2FS adopts the greedy
579 algorithm for on-demand cleaner, while background cleaner adopts cost-benefit
582 In order to identify whether the data in the victim segment are valid or not,
583 F2FS manages a bitmap. Each bit represents the validity of a block, and the
584 bitmap is composed of a bit stream covering whole blocks in main area.