| blob | e9e750e59efc50c922122d22ba1ec7ca69775a54 |
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
42 10, 1, 26–52.
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/
56 Cleaning Overhead
57 -----------------
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 ================================================================================
75 KEY FEATURES
76 ================================================================================
78 Flash 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
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.
90 Cleaning 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
97 ================================================================================
98 MOUNT OPTIONS
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 truned off.
106 Default value for this option is on. So garbage
107 collection is on by default.
108 disable_roll_forward Disable the roll-forward recovery routine
109 norecovery Disable the roll-forward recovery routine, mounted read-
110 only (i.e., -o ro,disable_roll_forward)
111 discard Issue discard/TRIM commands when a segment is cleaned.
112 no_heap Disable heap-style segment allocation which finds free
113 segments for data from the beginning of main area, while
114 for node from the end of main area.
115 nouser_xattr Disable Extended User Attributes. Note: xattr is enabled
116 by default if CONFIG_F2FS_FS_XATTR is selected.
117 noacl Disable POSIX Access Control List. Note: acl is enabled
118 by default if CONFIG_F2FS_FS_POSIX_ACL is selected.
119 active_logs=%u Support configuring the number of active logs. In the
120 current design, f2fs supports only 2, 4, and 6 logs.
121 Default number is 6.
122 disable_ext_identify Disable the extension list configured by mkfs, so f2fs
123 does not aware of cold files such as media files.
124 inline_xattr Enable the inline xattrs feature.
125 inline_data Enable the inline data feature: New created small(<~3.4k)
126 files can be written into inode block.
127 inline_dentry Enable the inline dir feature: data in new created
128 directory entries can be written into inode block. The
129 space of inode block which is used to store inline
130 dentries is limited to ~3.4k.
131 flush_merge Merge concurrent cache_flush commands as much as possible
132 to eliminate redundant command issues. If the underlying
133 device handles the cache_flush command relatively slowly,
134 recommend to enable this option.
135 nobarrier This option can be used if underlying storage guarantees
136 its cached data should be written to the novolatile area.
137 If this option is set, no cache_flush commands are issued
138 but f2fs still guarantees the write ordering of all the
139 data writes.
140 fastboot This option is used when a system wants to reduce mount
141 time as much as possible, even though normal performance
142 can be sacrificed.
143 extent_cache Enable an extent cache based on rb-tree, it can cache
144 as many as extent which map between contiguous logical
145 address and physical address per inode, resulting in
146 increasing the cache hit ratio.
147 noinline_data Disable the inline data feature, inline data feature is
148 enabled by default.
150 ================================================================================
151 DEBUGFS ENTRIES
152 ================================================================================
154 /sys/kernel/debug/f2fs/ contains information about all the partitions mounted as
155 f2fs. Each file shows the whole f2fs information.
157 /sys/kernel/debug/f2fs/status includes:
158 - major file system information managed by f2fs currently
159 - average SIT information about whole segments
160 - current memory footprint consumed by f2fs.
162 ================================================================================
163 SYSFS ENTRIES
164 ================================================================================
166 Information about mounted f2f2 file systems can be found in
167 /sys/fs/f2fs. Each mounted filesystem will have a directory in
168 /sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda).
169 The files in each per-device directory are shown in table below.
171 Files in /sys/fs/f2fs/<devname>
172 (see also Documentation/ABI/testing/sysfs-fs-f2fs)
173 ..............................................................................
174 File Content
176 gc_max_sleep_time This tuning parameter controls the maximum sleep
177 time for the garbage collection thread. Time is
178 in milliseconds.
180 gc_min_sleep_time This tuning parameter controls the minimum sleep
181 time for the garbage collection thread. Time is
182 in milliseconds.
184 gc_no_gc_sleep_time This tuning parameter controls the default sleep
185 time for the garbage collection thread. Time is
186 in milliseconds.
188 gc_idle This parameter controls the selection of victim
189 policy for garbage collection. Setting gc_idle = 0
190 (default) will disable this option. Setting
191 gc_idle = 1 will select the Cost Benefit approach
192 & setting gc_idle = 2 will select the greedy aproach.
194 reclaim_segments This parameter controls the number of prefree
195 segments to be reclaimed. If the number of prefree
196 segments is larger than the number of segments
197 in the proportion to the percentage over total
198 volume size, f2fs tries to conduct checkpoint to
199 reclaim the prefree segments to free segments.
200 By default, 5% over total # of segments.
202 max_small_discards This parameter controls the number of discard
203 commands that consist small blocks less than 2MB.
204 The candidates to be discarded are cached until
205 checkpoint is triggered, and issued during the
206 checkpoint. By default, it is disabled with 0.
208 trim_sections This parameter controls the number of sections
209 to be trimmed out in batch mode when FITRIM
210 conducts. 32 sections is set by default.
212 ipu_policy This parameter controls the policy of in-place
213 updates in f2fs. There are five policies:
214 0x01: F2FS_IPU_FORCE, 0x02: F2FS_IPU_SSR,
215 0x04: F2FS_IPU_UTIL, 0x08: F2FS_IPU_SSR_UTIL,
216 0x10: F2FS_IPU_FSYNC.
218 min_ipu_util This parameter controls the threshold to trigger
219 in-place-updates. The number indicates percentage
220 of the filesystem utilization, and used by
221 F2FS_IPU_UTIL and F2FS_IPU_SSR_UTIL policies.
223 min_fsync_blocks This parameter controls the threshold to trigger
224 in-place-updates when F2FS_IPU_FSYNC mode is set.
225 The number indicates the number of dirty pages
226 when fsync needs to flush on its call path. If
227 the number is less than this value, it triggers
228 in-place-updates.
230 max_victim_search This parameter controls the number of trials to
231 find a victim segment when conducting SSR and
232 cleaning operations. The default value is 4096
233 which covers 8GB block address range.
235 dir_level This parameter controls the directory level to
236 support large directory. If a directory has a
237 number of files, it can reduce the file lookup
238 latency by increasing this dir_level value.
239 Otherwise, it needs to decrease this value to
240 reduce the space overhead. The default value is 0.
242 ram_thresh This parameter controls the memory footprint used
243 by free nids and cached nat entries. By default,
244 10 is set, which indicates 10 MB / 1 GB RAM.
246 ================================================================================
247 USAGE
248 ================================================================================
250 1. Download userland tools and compile them.
252 2. Skip, if f2fs was compiled statically inside kernel.
253 Otherwise, insert the f2fs.ko module.
254 # insmod f2fs.ko
256 3. Create a directory trying to mount
257 # mkdir /mnt/f2fs
259 4. Format the block device, and then mount as f2fs
260 # mkfs.f2fs -l label /dev/block_device
261 # mount -t f2fs /dev/block_device /mnt/f2fs
263 mkfs.f2fs
264 ---------
265 The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem,
266 which builds a basic on-disk layout.
268 The options consist of:
269 -l [label] : Give a volume label, up to 512 unicode name.
270 -a [0 or 1] : Split start location of each area for heap-based allocation.
271 1 is set by default, which performs this.
272 -o [int] : Set overprovision ratio in percent over volume size.
273 5 is set by default.
274 -s [int] : Set the number of segments per section.
275 1 is set by default.
276 -z [int] : Set the number of sections per zone.
277 1 is set by default.
278 -e [str] : Set basic extension list. e.g. "mp3,gif,mov"
279 -t [0 or 1] : Disable discard command or not.
280 1 is set by default, which conducts discard.
282 fsck.f2fs
283 ---------
284 The fsck.f2fs is a tool to check the consistency of an f2fs-formatted
285 partition, which examines whether the filesystem metadata and user-made data
286 are cross-referenced correctly or not.
287 Note that, initial version of the tool does not fix any inconsistency.
289 The options consist of:
290 -d debug level [default:0]
292 dump.f2fs
293 ---------
294 The dump.f2fs shows the information of specific inode and dumps SSA and SIT to
295 file. Each file is dump_ssa and dump_sit.
297 The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem.
298 It shows on-disk inode information reconized by a given inode number, and is
299 able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and
300 ./dump_sit respectively.
302 The options consist of:
303 -d debug level [default:0]
304 -i inode no (hex)
305 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1]
306 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1]
308 Examples:
309 # dump.f2fs -i [ino] /dev/sdx
310 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump)
311 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump)
313 ================================================================================
314 DESIGN
315 ================================================================================
317 On-disk Layout
318 --------------
320 F2FS divides the whole volume into a number of segments, each of which is fixed
321 to 2MB in size. A section is composed of consecutive segments, and a zone
322 consists of a set of sections. By default, section and zone sizes are set to one
323 segment size identically, but users can easily modify the sizes by mkfs.
325 F2FS splits the entire volume into six areas, and all the areas except superblock
326 consists of multiple segments as described below.
328 align with the zone size <-|
329 |-> align with the segment size
330 _________________________________________________________________________
331 | | | Segment | Node | Segment | |
332 | Superblock | Checkpoint | Info. | Address | Summary | Main |
333 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | |
334 |____________|_____2______|______N______|______N______|______N_____|__N___|
335 . .
336 . .
337 . .
338 ._________________________________________.
339 |_Segment_|_..._|_Segment_|_..._|_Segment_|
340 . .
341 ._________._________
342 |_section_|__...__|_
343 . .
344 .________.
345 |__zone__|
347 - Superblock (SB)
348 : It is located at the beginning of the partition, and there exist two copies
349 to avoid file system crash. It contains basic partition information and some
350 default parameters of f2fs.
352 - Checkpoint (CP)
353 : It contains file system information, bitmaps for valid NAT/SIT sets, orphan
354 inode lists, and summary entries of current active segments.
356 - Segment Information Table (SIT)
357 : It contains segment information such as valid block count and bitmap for the
358 validity of all the blocks.
360 - Node Address Table (NAT)
361 : It is composed of a block address table for all the node blocks stored in
362 Main area.
364 - Segment Summary Area (SSA)
365 : It contains summary entries which contains the owner information of all the
366 data and node blocks stored in Main area.
368 - Main Area
369 : It contains file and directory data including their indices.
371 In order to avoid misalignment between file system and flash-based storage, F2FS
372 aligns the start block address of CP with the segment size. Also, it aligns the
373 start block address of Main area with the zone size by reserving some segments
374 in SSA area.
376 Reference the following survey for additional technical details.
377 https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey
379 File System Metadata Structure
380 ------------------------------
382 F2FS adopts the checkpointing scheme to maintain file system consistency. At
383 mount time, F2FS first tries to find the last valid checkpoint data by scanning
384 CP area. In order to reduce the scanning time, F2FS uses only two copies of CP.
385 One of them always indicates the last valid data, which is called as shadow copy
386 mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism.
388 For file system consistency, each CP points to which NAT and SIT copies are
389 valid, as shown as below.
391 +--------+----------+---------+
392 | CP | SIT | NAT |
393 +--------+----------+---------+
394 . . . .
395 . . . .
396 . . . .
397 +-------+-------+--------+--------+--------+--------+
398 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 |
399 +-------+-------+--------+--------+--------+--------+
400 | ^ ^
401 | | |
402 `----------------------------------------'
404 Index Structure
405 ---------------
407 The key data structure to manage the data locations is a "node". Similar to
408 traditional file structures, F2FS has three types of node: inode, direct node,
409 indirect node. F2FS assigns 4KB to an inode block which contains 923 data block
410 indices, two direct node pointers, two indirect node pointers, and one double
411 indirect node pointer as described below. One direct node block contains 1018
412 data blocks, and one indirect node block contains also 1018 node blocks. Thus,
413 one inode block (i.e., a file) covers:
415 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB.
417 Inode block (4KB)
418 |- data (923)
419 |- direct node (2)
420 | `- data (1018)
421 |- indirect node (2)
422 | `- direct node (1018)
423 | `- data (1018)
424 `- double indirect node (1)
425 `- indirect node (1018)
426 `- direct node (1018)
427 `- data (1018)
429 Note that, all the node blocks are mapped by NAT which means the location of
430 each node is translated by the NAT table. In the consideration of the wandering
431 tree problem, F2FS is able to cut off the propagation of node updates caused by
432 leaf data writes.
434 Directory Structure
435 -------------------
437 A directory entry occupies 11 bytes, which consists of the following attributes.
439 - hash hash value of the file name
440 - ino inode number
441 - len the length of file name
442 - type file type such as directory, symlink, etc
444 A dentry block consists of 214 dentry slots and file names. Therein a bitmap is
445 used to represent whether each dentry is valid or not. A dentry block occupies
446 4KB with the following composition.
448 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) +
449 dentries(11 * 214 bytes) + file name (8 * 214 bytes)
451 [Bucket]
452 +--------------------------------+
453 |dentry block 1 | dentry block 2 |
454 +--------------------------------+
455 . .
456 . .
457 . [Dentry Block Structure: 4KB] .
458 +--------+----------+----------+------------+
459 | bitmap | reserved | dentries | file names |
460 +--------+----------+----------+------------+
461 [Dentry Block: 4KB] . .
462 . .
463 . .
464 +------+------+-----+------+
465 | hash | ino | len | type |
466 +------+------+-----+------+
467 [Dentry Structure: 11 bytes]
469 F2FS implements multi-level hash tables for directory structure. Each level has
470 a hash table with dedicated number of hash buckets as shown below. Note that
471 "A(2B)" means a bucket includes 2 data blocks.
473 ----------------------
474 A : bucket
475 B : block
476 N : MAX_DIR_HASH_DEPTH
477 ----------------------
479 level #0 | A(2B)
480 |
481 level #1 | A(2B) - A(2B)
482 |
483 level #2 | A(2B) - A(2B) - A(2B) - A(2B)
484 . | . . . .
485 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B)
486 . | . . . .
487 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B)
489 The number of blocks and buckets are determined by,
491 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2,
492 # of blocks in level #n = |
493 `- 4, Otherwise
495 ,- 2^(n + dir_level),
496 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2,
497 # of buckets in level #n = |
498 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1),
499 Otherwise
501 When F2FS finds a file name in a directory, at first a hash value of the file
502 name is calculated. Then, F2FS scans the hash table in level #0 to find the
503 dentry consisting of the file name and its inode number. If not found, F2FS
504 scans the next hash table in level #1. In this way, F2FS scans hash tables in
505 each levels incrementally from 1 to N. In each levels F2FS needs to scan only
506 one bucket determined by the following equation, which shows O(log(# of files))
507 complexity.
509 bucket number to scan in level #n = (hash value) % (# of buckets in level #n)
511 In the case of file creation, F2FS finds empty consecutive slots that cover the
512 file name. F2FS searches the empty slots in the hash tables of whole levels from
513 1 to N in the same way as the lookup operation.
515 The following figure shows an example of two cases holding children.
516 --------------> Dir <--------------
517 | |
518 child child
520 child - child [hole] - child
522 child - child - child [hole] - [hole] - child
524 Case 1: Case 2:
525 Number of children = 6, Number of children = 3,
526 File size = 7 File size = 7
528 Default Block Allocation
529 ------------------------
531 At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node
532 and Hot/Warm/Cold data.
534 - Hot node contains direct node blocks of directories.
535 - Warm node contains direct node blocks except hot node blocks.
536 - Cold node contains indirect node blocks
537 - Hot data contains dentry blocks
538 - Warm data contains data blocks except hot and cold data blocks
539 - Cold data contains multimedia data or migrated data blocks
541 LFS has two schemes for free space management: threaded log and copy-and-compac-
542 tion. The copy-and-compaction scheme which is known as cleaning, is well-suited
543 for devices showing very good sequential write performance, since free segments
544 are served all the time for writing new data. However, it suffers from cleaning
545 overhead under high utilization. Contrarily, the threaded log scheme suffers
546 from random writes, but no cleaning process is needed. F2FS adopts a hybrid
547 scheme where the copy-and-compaction scheme is adopted by default, but the
548 policy is dynamically changed to the threaded log scheme according to the file
549 system status.
551 In order to align F2FS with underlying flash-based storage, F2FS allocates a
552 segment in a unit of section. F2FS expects that the section size would be the
553 same as the unit size of garbage collection in FTL. Furthermore, with respect
554 to the mapping granularity in FTL, F2FS allocates each section of the active
555 logs from different zones as much as possible, since FTL can write the data in
556 the active logs into one allocation unit according to its mapping granularity.
558 Cleaning process
559 ----------------
561 F2FS does cleaning both on demand and in the background. On-demand cleaning is
562 triggered when there are not enough free segments to serve VFS calls. Background
563 cleaner is operated by a kernel thread, and triggers the cleaning job when the
564 system is idle.
566 F2FS supports two victim selection policies: greedy and cost-benefit algorithms.
567 In the greedy algorithm, F2FS selects a victim segment having the smallest number
568 of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment
569 according to the segment age and the number of valid blocks in order to address
570 log block thrashing problem in the greedy algorithm. F2FS adopts the greedy
571 algorithm for on-demand cleaner, while background cleaner adopts cost-benefit
572 algorithm.
574 In order to identify whether the data in the victim segment are valid or not,
575 F2FS manages a bitmap. Each bit represents the validity of a block, and the
576 bitmap is composed of a bit stream covering whole blocks in main area.