blob: 51afba17bbae89f433da7d150085df937c071c3c [file] [log] [blame]
Jaegeuk Kim98e4da82012-11-02 17:05:42 +09001================================================================================
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
Changman Leed51a7fb2013-07-04 17:12:47 +090021The following git tree provides the file system formatting tool (mkfs.f2fs),
22a consistency checking tool (fsck.f2fs), and a debugging tool (dump.f2fs).
Jaegeuk Kim5bb446a2012-11-27 14:36:14 +090023>> 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
Jaegeuk Kim98e4da82012-11-02 17:05:42 +090027
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
Namjae Jeon696c0182013-06-16 09:48:48 +0900101background_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 truned off.
106 Default value for this option is on. So garbage
107 collection is on by default.
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900108disable_roll_forward Disable the roll-forward recovery routine
109discard Issue discard/TRIM commands when a segment is cleaned.
110no_heap Disable heap-style segment allocation which finds free
111 segments for data from the beginning of main area, while
112 for node from the end of main area.
113nouser_xattr Disable Extended User Attributes. Note: xattr is enabled
114 by default if CONFIG_F2FS_FS_XATTR is selected.
115noacl Disable POSIX Access Control List. Note: acl is enabled
116 by default if CONFIG_F2FS_FS_POSIX_ACL is selected.
117active_logs=%u Support configuring the number of active logs. In the
118 current design, f2fs supports only 2, 4, and 6 logs.
119 Default number is 6.
120disable_ext_identify Disable the extension list configured by mkfs, so f2fs
121 does not aware of cold files such as media files.
Jaegeuk Kim66e960c2013-11-01 11:20:05 +0900122inline_xattr Enable the inline xattrs feature.
Huajun Lie4024e82013-11-10 23:13:21 +0800123inline_data Enable the inline data feature: New created small(<~3.4k)
124 files can be written into inode block.
Jaegeuk Kim6b4afdd2014-04-02 15:34:36 +0900125flush_merge Merge concurrent cache_flush commands as much as possible
126 to eliminate redundant command issues. If the underlying
127 device handles the cache_flush command relatively slowly,
128 recommend to enable this option.
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900129
130================================================================================
131DEBUGFS ENTRIES
132================================================================================
133
134/sys/kernel/debug/f2fs/ contains information about all the partitions mounted as
135f2fs. Each file shows the whole f2fs information.
136
137/sys/kernel/debug/f2fs/status includes:
138 - major file system information managed by f2fs currently
139 - average SIT information about whole segments
140 - current memory footprint consumed by f2fs.
141
142================================================================================
Namjae Jeonb59d0ba2013-08-04 23:09:40 +0900143SYSFS ENTRIES
144================================================================================
145
146Information about mounted f2f2 file systems can be found in
147/sys/fs/f2fs. Each mounted filesystem will have a directory in
148/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda).
149The files in each per-device directory are shown in table below.
150
151Files in /sys/fs/f2fs/<devname>
152(see also Documentation/ABI/testing/sysfs-fs-f2fs)
153..............................................................................
154 File Content
155
156 gc_max_sleep_time This tuning parameter controls the maximum sleep
157 time for the garbage collection thread. Time is
158 in milliseconds.
159
160 gc_min_sleep_time This tuning parameter controls the minimum sleep
161 time for the garbage collection thread. Time is
162 in milliseconds.
163
164 gc_no_gc_sleep_time This tuning parameter controls the default sleep
165 time for the garbage collection thread. Time is
166 in milliseconds.
167
Namjae Jeond2dc0952013-08-04 23:10:15 +0900168 gc_idle This parameter controls the selection of victim
169 policy for garbage collection. Setting gc_idle = 0
170 (default) will disable this option. Setting
171 gc_idle = 1 will select the Cost Benefit approach
172 & setting gc_idle = 2 will select the greedy aproach.
173
Jaegeuk Kimea91e9b2013-10-24 15:49:07 +0900174 reclaim_segments This parameter controls the number of prefree
175 segments to be reclaimed. If the number of prefree
Jaegeuk Kim58c41032014-03-19 14:17:21 +0900176 segments is larger than the number of segments
177 in the proportion to the percentage over total
178 volume size, f2fs tries to conduct checkpoint to
179 reclaim the prefree segments to free segments.
180 By default, 5% over total # of segments.
Jaegeuk Kimea91e9b2013-10-24 15:49:07 +0900181
Jaegeuk Kimba0697e2013-12-19 17:44:41 +0900182 max_small_discards This parameter controls the number of discard
183 commands that consist small blocks less than 2MB.
184 The candidates to be discarded are cached until
185 checkpoint is triggered, and issued during the
186 checkpoint. By default, it is disabled with 0.
187
Jaegeuk Kim216fbd62013-11-07 13:13:42 +0900188 ipu_policy This parameter controls the policy of in-place
189 updates in f2fs. There are five policies:
190 0: F2FS_IPU_FORCE, 1: F2FS_IPU_SSR,
191 2: F2FS_IPU_UTIL, 3: F2FS_IPU_SSR_UTIL,
192 4: F2FS_IPU_DISABLE.
193
194 min_ipu_util This parameter controls the threshold to trigger
195 in-place-updates. The number indicates percentage
196 of the filesystem utilization, and used by
197 F2FS_IPU_UTIL and F2FS_IPU_SSR_UTIL policies.
198
Jaegeuk Kim3bac3802014-01-09 21:00:06 +0900199 max_victim_search This parameter controls the number of trials to
200 find a victim segment when conducting SSR and
201 cleaning operations. The default value is 4096
202 which covers 8GB block address range.
203
Jaegeuk Kimab9fa662014-02-27 20:09:05 +0900204 dir_level This parameter controls the directory level to
205 support large directory. If a directory has a
206 number of files, it can reduce the file lookup
207 latency by increasing this dir_level value.
208 Otherwise, it needs to decrease this value to
209 reduce the space overhead. The default value is 0.
210
Jaegeuk Kimcdfc41c2014-03-19 13:31:37 +0900211 ram_thresh This parameter controls the memory footprint used
212 by free nids and cached nat entries. By default,
213 10 is set, which indicates 10 MB / 1 GB RAM.
214
Namjae Jeonb59d0ba2013-08-04 23:09:40 +0900215================================================================================
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900216USAGE
217================================================================================
218
2191. Download userland tools and compile them.
220
2212. Skip, if f2fs was compiled statically inside kernel.
222 Otherwise, insert the f2fs.ko module.
223 # insmod f2fs.ko
224
2253. Create a directory trying to mount
226 # mkdir /mnt/f2fs
227
2284. Format the block device, and then mount as f2fs
229 # mkfs.f2fs -l label /dev/block_device
230 # mount -t f2fs /dev/block_device /mnt/f2fs
231
Changman Leed51a7fb2013-07-04 17:12:47 +0900232mkfs.f2fs
233---------
234The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem,
235which builds a basic on-disk layout.
236
237The options consist of:
Changman Lee1571f842013-04-03 15:26:49 +0900238-l [label] : Give a volume label, up to 512 unicode name.
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900239-a [0 or 1] : Split start location of each area for heap-based allocation.
240 1 is set by default, which performs this.
241-o [int] : Set overprovision ratio in percent over volume size.
242 5 is set by default.
243-s [int] : Set the number of segments per section.
244 1 is set by default.
245-z [int] : Set the number of sections per zone.
246 1 is set by default.
247-e [str] : Set basic extension list. e.g. "mp3,gif,mov"
Changman Lee1571f842013-04-03 15:26:49 +0900248-t [0 or 1] : Disable discard command or not.
249 1 is set by default, which conducts discard.
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900250
Changman Leed51a7fb2013-07-04 17:12:47 +0900251fsck.f2fs
252---------
253The fsck.f2fs is a tool to check the consistency of an f2fs-formatted
254partition, which examines whether the filesystem metadata and user-made data
255are cross-referenced correctly or not.
256Note that, initial version of the tool does not fix any inconsistency.
257
258The options consist of:
259 -d debug level [default:0]
260
261dump.f2fs
262---------
263The dump.f2fs shows the information of specific inode and dumps SSA and SIT to
264file. Each file is dump_ssa and dump_sit.
265
266The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem.
267It shows on-disk inode information reconized by a given inode number, and is
268able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and
269./dump_sit respectively.
270
271The options consist of:
272 -d debug level [default:0]
273 -i inode no (hex)
274 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1]
275 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1]
276
277Examples:
278# dump.f2fs -i [ino] /dev/sdx
279# dump.f2fs -s 0~-1 /dev/sdx (SIT dump)
280# dump.f2fs -a 0~-1 /dev/sdx (SSA dump)
281
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900282================================================================================
283DESIGN
284================================================================================
285
286On-disk Layout
287--------------
288
289F2FS divides the whole volume into a number of segments, each of which is fixed
290to 2MB in size. A section is composed of consecutive segments, and a zone
291consists of a set of sections. By default, section and zone sizes are set to one
292segment size identically, but users can easily modify the sizes by mkfs.
293
294F2FS splits the entire volume into six areas, and all the areas except superblock
295consists of multiple segments as described below.
296
297 align with the zone size <-|
298 |-> align with the segment size
299 _________________________________________________________________________
Huajun Li9268cc32012-12-31 13:59:04 +0800300 | | | Segment | Node | Segment | |
301 | Superblock | Checkpoint | Info. | Address | Summary | Main |
302 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | |
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900303 |____________|_____2______|______N______|______N______|______N_____|__N___|
304 . .
305 . .
306 . .
307 ._________________________________________.
308 |_Segment_|_..._|_Segment_|_..._|_Segment_|
309 . .
310 ._________._________
311 |_section_|__...__|_
312 . .
313 .________.
314 |__zone__|
315
316- Superblock (SB)
317 : It is located at the beginning of the partition, and there exist two copies
318 to avoid file system crash. It contains basic partition information and some
319 default parameters of f2fs.
320
321- Checkpoint (CP)
322 : It contains file system information, bitmaps for valid NAT/SIT sets, orphan
323 inode lists, and summary entries of current active segments.
324
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900325- Segment Information Table (SIT)
326 : It contains segment information such as valid block count and bitmap for the
327 validity of all the blocks.
328
Huajun Li9268cc32012-12-31 13:59:04 +0800329- Node Address Table (NAT)
330 : It is composed of a block address table for all the node blocks stored in
331 Main area.
332
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900333- Segment Summary Area (SSA)
334 : It contains summary entries which contains the owner information of all the
335 data and node blocks stored in Main area.
336
337- Main Area
338 : It contains file and directory data including their indices.
339
340In order to avoid misalignment between file system and flash-based storage, F2FS
341aligns the start block address of CP with the segment size. Also, it aligns the
342start block address of Main area with the zone size by reserving some segments
343in SSA area.
344
345Reference the following survey for additional technical details.
346https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey
347
348File System Metadata Structure
349------------------------------
350
351F2FS adopts the checkpointing scheme to maintain file system consistency. At
352mount time, F2FS first tries to find the last valid checkpoint data by scanning
353CP area. In order to reduce the scanning time, F2FS uses only two copies of CP.
354One of them always indicates the last valid data, which is called as shadow copy
355mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism.
356
357For file system consistency, each CP points to which NAT and SIT copies are
358valid, as shown as below.
359
360 +--------+----------+---------+
Huajun Li9268cc32012-12-31 13:59:04 +0800361 | CP | SIT | NAT |
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900362 +--------+----------+---------+
363 . . . .
364 . . . .
365 . . . .
366 +-------+-------+--------+--------+--------+--------+
Huajun Li9268cc32012-12-31 13:59:04 +0800367 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 |
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900368 +-------+-------+--------+--------+--------+--------+
369 | ^ ^
370 | | |
371 `----------------------------------------'
372
373Index Structure
374---------------
375
376The key data structure to manage the data locations is a "node". Similar to
377traditional file structures, F2FS has three types of node: inode, direct node,
Huajun Lid08ab082012-12-05 16:45:32 +0800378indirect node. F2FS assigns 4KB to an inode block which contains 923 data block
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900379indices, two direct node pointers, two indirect node pointers, and one double
380indirect node pointer as described below. One direct node block contains 1018
381data blocks, and one indirect node block contains also 1018 node blocks. Thus,
382one inode block (i.e., a file) covers:
383
384 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB.
385
386 Inode block (4KB)
387 |- data (923)
388 |- direct node (2)
389 | `- data (1018)
390 |- indirect node (2)
391 | `- direct node (1018)
392 | `- data (1018)
393 `- double indirect node (1)
394 `- indirect node (1018)
395 `- direct node (1018)
396 `- data (1018)
397
398Note that, all the node blocks are mapped by NAT which means the location of
399each node is translated by the NAT table. In the consideration of the wandering
400tree problem, F2FS is able to cut off the propagation of node updates caused by
401leaf data writes.
402
403Directory Structure
404-------------------
405
406A directory entry occupies 11 bytes, which consists of the following attributes.
407
408- hash hash value of the file name
409- ino inode number
410- len the length of file name
411- type file type such as directory, symlink, etc
412
413A dentry block consists of 214 dentry slots and file names. Therein a bitmap is
414used to represent whether each dentry is valid or not. A dentry block occupies
4154KB with the following composition.
416
417 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) +
418 dentries(11 * 214 bytes) + file name (8 * 214 bytes)
419
420 [Bucket]
421 +--------------------------------+
422 |dentry block 1 | dentry block 2 |
423 +--------------------------------+
424 . .
425 . .
426 . [Dentry Block Structure: 4KB] .
427 +--------+----------+----------+------------+
428 | bitmap | reserved | dentries | file names |
429 +--------+----------+----------+------------+
430 [Dentry Block: 4KB] . .
431 . .
432 . .
433 +------+------+-----+------+
434 | hash | ino | len | type |
435 +------+------+-----+------+
436 [Dentry Structure: 11 bytes]
437
438F2FS implements multi-level hash tables for directory structure. Each level has
439a hash table with dedicated number of hash buckets as shown below. Note that
440"A(2B)" means a bucket includes 2 data blocks.
441
442----------------------
443A : bucket
444B : block
445N : MAX_DIR_HASH_DEPTH
446----------------------
447
448level #0 | A(2B)
449 |
450level #1 | A(2B) - A(2B)
451 |
452level #2 | A(2B) - A(2B) - A(2B) - A(2B)
453 . | . . . .
454level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B)
455 . | . . . .
456level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B)
457
458The number of blocks and buckets are determined by,
459
460 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2,
461 # of blocks in level #n = |
462 `- 4, Otherwise
463
Chao Yubfec07d2014-05-28 08:56:09 +0800464 ,- 2^(n + dir_level),
465 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2,
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900466 # of buckets in level #n = |
Chao Yubfec07d2014-05-28 08:56:09 +0800467 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1),
468 Otherwise
Jaegeuk Kim98e4da82012-11-02 17:05:42 +0900469
470When F2FS finds a file name in a directory, at first a hash value of the file
471name is calculated. Then, F2FS scans the hash table in level #0 to find the
472dentry consisting of the file name and its inode number. If not found, F2FS
473scans the next hash table in level #1. In this way, F2FS scans hash tables in
474each levels incrementally from 1 to N. In each levels F2FS needs to scan only
475one bucket determined by the following equation, which shows O(log(# of files))
476complexity.
477
478 bucket number to scan in level #n = (hash value) % (# of buckets in level #n)
479
480In the case of file creation, F2FS finds empty consecutive slots that cover the
481file name. F2FS searches the empty slots in the hash tables of whole levels from
4821 to N in the same way as the lookup operation.
483
484The following figure shows an example of two cases holding children.
485 --------------> Dir <--------------
486 | |
487 child child
488
489 child - child [hole] - child
490
491 child - child - child [hole] - [hole] - child
492
493 Case 1: Case 2:
494 Number of children = 6, Number of children = 3,
495 File size = 7 File size = 7
496
497Default Block Allocation
498------------------------
499
500At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node
501and Hot/Warm/Cold data.
502
503- Hot node contains direct node blocks of directories.
504- Warm node contains direct node blocks except hot node blocks.
505- Cold node contains indirect node blocks
506- Hot data contains dentry blocks
507- Warm data contains data blocks except hot and cold data blocks
508- Cold data contains multimedia data or migrated data blocks
509
510LFS has two schemes for free space management: threaded log and copy-and-compac-
511tion. The copy-and-compaction scheme which is known as cleaning, is well-suited
512for devices showing very good sequential write performance, since free segments
513are served all the time for writing new data. However, it suffers from cleaning
514overhead under high utilization. Contrarily, the threaded log scheme suffers
515from random writes, but no cleaning process is needed. F2FS adopts a hybrid
516scheme where the copy-and-compaction scheme is adopted by default, but the
517policy is dynamically changed to the threaded log scheme according to the file
518system status.
519
520In order to align F2FS with underlying flash-based storage, F2FS allocates a
521segment in a unit of section. F2FS expects that the section size would be the
522same as the unit size of garbage collection in FTL. Furthermore, with respect
523to the mapping granularity in FTL, F2FS allocates each section of the active
524logs from different zones as much as possible, since FTL can write the data in
525the active logs into one allocation unit according to its mapping granularity.
526
527Cleaning process
528----------------
529
530F2FS does cleaning both on demand and in the background. On-demand cleaning is
531triggered when there are not enough free segments to serve VFS calls. Background
532cleaner is operated by a kernel thread, and triggers the cleaning job when the
533system is idle.
534
535F2FS supports two victim selection policies: greedy and cost-benefit algorithms.
536In the greedy algorithm, F2FS selects a victim segment having the smallest number
537of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment
538according to the segment age and the number of valid blocks in order to address
539log block thrashing problem in the greedy algorithm. F2FS adopts the greedy
540algorithm for on-demand cleaner, while background cleaner adopts cost-benefit
541algorithm.
542
543In order to identify whether the data in the victim segment are valid or not,
544F2FS manages a bitmap. Each bit represents the validity of a block, and the
545bitmap is composed of a bit stream covering whole blocks in main area.