1 .. SPDX-License-Identifier: GPL-2.0 1 .. SPDX-License-Identifier: GPL-2.0
2 2
3 ========================================== 3 ==========================================
4 WHAT IS Flash-Friendly File System (F2FS)? 4 WHAT IS Flash-Friendly File System (F2FS)?
5 ========================================== 5 ==========================================
6 6
7 NAND flash memory-based storage devices, such 7 NAND flash memory-based storage devices, such as SSD, eMMC, and SD cards, have
8 been equipped on a variety systems ranging fro 8 been equipped on a variety systems ranging from mobile to server systems. Since
9 they are known to have different characteristi 9 they are known to have different characteristics from the conventional rotating
10 disks, a file system, an upper layer to the st 10 disks, a file system, an upper layer to the storage device, should adapt to the
11 changes from the sketch in the design level. 11 changes from the sketch in the design level.
12 12
13 F2FS is a file system exploiting NAND flash me 13 F2FS is a file system exploiting NAND flash memory-based storage devices, which
14 is based on Log-structured File System (LFS). 14 is based on Log-structured File System (LFS). The design has been focused on
15 addressing the fundamental issues in LFS, whic 15 addressing the fundamental issues in LFS, which are snowball effect of wandering
16 tree and high cleaning overhead. 16 tree and high cleaning overhead.
17 17
18 Since a NAND flash memory-based storage device 18 Since a NAND flash memory-based storage device shows different characteristic
19 according to its internal geometry or flash me 19 according to its internal geometry or flash memory management scheme, namely FTL,
20 F2FS and its tools support various parameters 20 F2FS and its tools support various parameters not only for configuring on-disk
21 layout, but also for selecting allocation and 21 layout, but also for selecting allocation and cleaning algorithms.
22 22
23 The following git tree provides the file syste 23 The following git tree provides the file system formatting tool (mkfs.f2fs),
24 a consistency checking tool (fsck.f2fs), and a 24 a consistency checking tool (fsck.f2fs), and a debugging tool (dump.f2fs).
25 25
26 - git://git.kernel.org/pub/scm/linux/kernel/gi 26 - git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs-tools.git
27 27
28 For sending patches, please use the following !! 28 For reporting bugs and sending patches, please use the following mailing list:
29 29
30 - linux-f2fs-devel@lists.sourceforge.net 30 - linux-f2fs-devel@lists.sourceforge.net
31 31
32 For reporting bugs, please use the following f <<
33 <<
34 - https://bugzilla.kernel.org/enter_bug.cgi?pr <<
35 <<
36 Background and Design issues 32 Background and Design issues
37 ============================ 33 ============================
38 34
39 Log-structured File System (LFS) 35 Log-structured File System (LFS)
40 -------------------------------- 36 --------------------------------
41 "A log-structured file system writes all modif 37 "A log-structured file system writes all modifications to disk sequentially in
42 a log-like structure, thereby speeding up bot 38 a log-like structure, thereby speeding up both file writing and crash recovery.
43 The log is the only structure on disk; it cont 39 The log is the only structure on disk; it contains indexing information so that
44 files can be read back from the log efficientl 40 files can be read back from the log efficiently. In order to maintain large free
45 areas on disk for fast writing, we divide the 41 areas on disk for fast writing, we divide the log into segments and use a
46 segment cleaner to compress the live informati 42 segment cleaner to compress the live information from heavily fragmented
47 segments." from Rosenblum, M. and Ousterhout, 43 segments." from Rosenblum, M. and Ousterhout, J. K., 1992, "The design and
48 implementation of a log-structured file system 44 implementation of a log-structured file system", ACM Trans. Computer Systems
49 10, 1, 26–52. 45 10, 1, 26–52.
50 46
51 Wandering Tree Problem 47 Wandering Tree Problem
52 ---------------------- 48 ----------------------
53 In LFS, when a file data is updated and writte 49 In LFS, when a file data is updated and written to the end of log, its direct
54 pointer block is updated due to the changed lo 50 pointer block is updated due to the changed location. Then the indirect pointer
55 block is also updated due to the direct pointe 51 block is also updated due to the direct pointer block update. In this manner,
56 the upper index structures such as inode, inod 52 the upper index structures such as inode, inode map, and checkpoint block are
57 also updated recursively. This problem is call 53 also updated recursively. This problem is called as wandering tree problem [1],
58 and in order to enhance the performance, it sh 54 and in order to enhance the performance, it should eliminate or relax the update
59 propagation as much as possible. 55 propagation as much as possible.
60 56
61 [1] Bityutskiy, A. 2005. JFFS3 design issues. 57 [1] Bityutskiy, A. 2005. JFFS3 design issues. http://www.linux-mtd.infradead.org/
62 58
63 Cleaning Overhead 59 Cleaning Overhead
64 ----------------- 60 -----------------
65 Since LFS is based on out-of-place writes, it 61 Since LFS is based on out-of-place writes, it produces so many obsolete blocks
66 scattered across the whole storage. In order t 62 scattered across the whole storage. In order to serve new empty log space, it
67 needs to reclaim these obsolete blocks seamles 63 needs to reclaim these obsolete blocks seamlessly to users. This job is called
68 as a cleaning process. 64 as a cleaning process.
69 65
70 The process consists of three operations as fo 66 The process consists of three operations as follows.
71 67
72 1. A victim segment is selected through refere 68 1. A victim segment is selected through referencing segment usage table.
73 2. It loads parent index structures of all the 69 2. It loads parent index structures of all the data in the victim identified by
74 segment summary blocks. 70 segment summary blocks.
75 3. It checks the cross-reference between the d 71 3. It checks the cross-reference between the data and its parent index structure.
76 4. It moves valid data selectively. 72 4. It moves valid data selectively.
77 73
78 This cleaning job may cause unexpected long de 74 This cleaning job may cause unexpected long delays, so the most important goal
79 is to hide the latencies to users. And also de 75 is to hide the latencies to users. And also definitely, it should reduce the
80 amount of valid data to be moved, and move the 76 amount of valid data to be moved, and move them quickly as well.
81 77
82 Key Features 78 Key Features
83 ============ 79 ============
84 80
85 Flash Awareness 81 Flash Awareness
86 --------------- 82 ---------------
87 - Enlarge the random write area for better per 83 - Enlarge the random write area for better performance, but provide the high
88 spatial locality 84 spatial locality
89 - Align FS data structures to the operational 85 - Align FS data structures to the operational units in FTL as best efforts
90 86
91 Wandering Tree Problem 87 Wandering Tree Problem
92 ---------------------- 88 ----------------------
93 - Use a term, “node”, that represents inod 89 - Use a term, “node”, that represents inodes as well as various pointer blocks
94 - Introduce Node Address Table (NAT) containin 90 - Introduce Node Address Table (NAT) containing the locations of all the “node”
95 blocks; this will cut off the update propaga 91 blocks; this will cut off the update propagation.
96 92
97 Cleaning Overhead 93 Cleaning Overhead
98 ----------------- 94 -----------------
99 - Support a background cleaning process 95 - Support a background cleaning process
100 - Support greedy and cost-benefit algorithms f 96 - Support greedy and cost-benefit algorithms for victim selection policies
101 - Support multi-head logs for static/dynamic h 97 - Support multi-head logs for static/dynamic hot and cold data separation
102 - Introduce adaptive logging for efficient blo 98 - Introduce adaptive logging for efficient block allocation
103 99
104 Mount Options 100 Mount Options
105 ============= 101 =============
106 102
107 103
108 ======================== ===================== !! 104 ====================== ============================================================
109 background_gc=%s Turn on/off cleaning !! 105 background_gc=%s Turn on/off cleaning operations, namely garbage
110 collection, triggered !! 106 collection, triggered in background when I/O subsystem is
111 idle. If background_g !! 107 idle. If background_gc=on, it will turn on the garbage
112 collection and if bac !! 108 collection and if background_gc=off, garbage collection
113 will be turned off. I !! 109 will be turned off. If background_gc=sync, it will turn
114 on synchronous garbag !! 110 on synchronous garbage collection running in background.
115 Default value for thi !! 111 Default value for this option is on. So garbage
116 collection is on by d !! 112 collection is on by default.
117 gc_merge When background_gc is !! 113 disable_roll_forward Disable the roll-forward recovery routine
118 let background GC thr !! 114 norecovery Disable the roll-forward recovery routine, mounted read-
119 it can eliminate the !! 115 only (i.e., -o ro,disable_roll_forward)
120 GC operation when GC !! 116 discard/nodiscard Enable/disable real-time discard in f2fs, if discard is
121 I/O and CPU resources !! 117 enabled, f2fs will issue discard/TRIM commands when a
122 nogc_merge Disable GC merge feat !! 118 segment is cleaned.
123 disable_roll_forward Disable the roll-forw !! 119 no_heap Disable heap-style segment allocation which finds free
124 norecovery Disable the roll-forw !! 120 segments for data from the beginning of main area, while
125 only (i.e., -o ro,dis !! 121 for node from the end of main area.
126 discard/nodiscard Enable/disable real-t !! 122 nouser_xattr Disable Extended User Attributes. Note: xattr is enabled
127 enabled, f2fs will is !! 123 by default if CONFIG_F2FS_FS_XATTR is selected.
128 segment is cleaned. !! 124 noacl Disable POSIX Access Control List. Note: acl is enabled
129 heap/no_heap Deprecated. !! 125 by default if CONFIG_F2FS_FS_POSIX_ACL is selected.
130 nouser_xattr Disable Extended User !! 126 active_logs=%u Support configuring the number of active logs. In the
131 by default if CONFIG_ !! 127 current design, f2fs supports only 2, 4, and 6 logs.
132 noacl Disable POSIX Access !! 128 Default number is 6.
133 by default if CONFIG_ !! 129 disable_ext_identify Disable the extension list configured by mkfs, so f2fs
134 active_logs=%u Support configuring t !! 130 does not aware of cold files such as media files.
135 current design, f2fs !! 131 inline_xattr Enable the inline xattrs feature.
136 Default number is 6. !! 132 noinline_xattr Disable the inline xattrs feature.
137 disable_ext_identify Disable the extension !! 133 inline_xattr_size=%u Support configuring inline xattr size, it depends on
138 is not aware of cold !! 134 flexible inline xattr feature.
139 inline_xattr Enable the inline xat !! 135 inline_data Enable the inline data feature: New created small(<~3.4k)
140 noinline_xattr Disable the inline xa !! 136 files can be written into inode block.
141 inline_xattr_size=%u Support configuring i !! 137 inline_dentry Enable the inline dir feature: data in new created
142 flexible inline xattr !! 138 directory entries can be written into inode block. The
143 inline_data Enable the inline dat !! 139 space of inode block which is used to store inline
144 files can be written !! 140 dentries is limited to ~3.4k.
145 inline_dentry Enable the inline dir !! 141 noinline_dentry Disable the inline dentry feature.
146 directory entries can !! 142 flush_merge Merge concurrent cache_flush commands as much as possible
147 space of inode block !! 143 to eliminate redundant command issues. If the underlying
148 dentries is limited t !! 144 device handles the cache_flush command relatively slowly,
149 noinline_dentry Disable the inline de !! 145 recommend to enable this option.
150 flush_merge Merge concurrent cach !! 146 nobarrier This option can be used if underlying storage guarantees
151 to eliminate redundan !! 147 its cached data should be written to the novolatile area.
152 device handles the ca !! 148 If this option is set, no cache_flush commands are issued
153 recommend to enable t !! 149 but f2fs still guarantees the write ordering of all the
154 nobarrier This option can be us !! 150 data writes.
155 its cached data shoul !! 151 fastboot This option is used when a system wants to reduce mount
156 If this option is set !! 152 time as much as possible, even though normal performance
157 but f2fs still guaran !! 153 can be sacrificed.
158 data writes. !! 154 extent_cache Enable an extent cache based on rb-tree, it can cache
159 barrier If this option is set !! 155 as many as extent which map between contiguous logical
160 issued. !! 156 address and physical address per inode, resulting in
161 fastboot This option is used w !! 157 increasing the cache hit ratio. Set by default.
162 time as much as possi !! 158 noextent_cache Disable an extent cache based on rb-tree explicitly, see
163 can be sacrificed. !! 159 the above extent_cache mount option.
164 extent_cache Enable an extent cach !! 160 noinline_data Disable the inline data feature, inline data feature is
165 as many as extent whi !! 161 enabled by default.
166 address and physical !! 162 data_flush Enable data flushing before checkpoint in order to
167 increasing the cache !! 163 persist data of regular and symlink.
168 noextent_cache Disable an extent cac !! 164 reserve_root=%d Support configuring reserved space which is used for
169 the above extent_cach !! 165 allocation from a privileged user with specified uid or
170 noinline_data Disable the inline da !! 166 gid, unit: 4KB, the default limit is 0.2% of user blocks.
171 enabled by default. !! 167 resuid=%d The user ID which may use the reserved blocks.
172 data_flush Enable data flushing !! 168 resgid=%d The group ID which may use the reserved blocks.
173 persist data of regul !! 169 fault_injection=%d Enable fault injection in all supported types with
174 reserve_root=%d Support configuring r !! 170 specified injection rate.
175 allocation from a pri !! 171 fault_type=%d Support configuring fault injection type, should be
176 gid, unit: 4KB, the d !! 172 enabled with fault_injection option, fault type value
177 resuid=%d The user ID which may !! 173 is shown below, it supports single or combined type.
178 resgid=%d The group ID which ma !! 174
179 fault_injection=%d Enable fault injectio !! 175 =================== ===========
180 specified injection r !! 176 Type_Name Type_Value
181 fault_type=%d Support configuring f !! 177 =================== ===========
182 enabled with fault_in !! 178 FAULT_KMALLOC 0x000000001
183 is shown below, it su !! 179 FAULT_KVMALLOC 0x000000002
184 !! 180 FAULT_PAGE_ALLOC 0x000000004
185 ===================== !! 181 FAULT_PAGE_GET 0x000000008
186 Type_Name !! 182 FAULT_ALLOC_BIO 0x000000010
187 ===================== !! 183 FAULT_ALLOC_NID 0x000000020
188 FAULT_KMALLOC !! 184 FAULT_ORPHAN 0x000000040
189 FAULT_KVMALLOC !! 185 FAULT_BLOCK 0x000000080
190 FAULT_PAGE_ALLOC !! 186 FAULT_DIR_DEPTH 0x000000100
191 FAULT_PAGE_GET !! 187 FAULT_EVICT_INODE 0x000000200
192 FAULT_ALLOC_BIO !! 188 FAULT_TRUNCATE 0x000000400
193 FAULT_ALLOC_NID !! 189 FAULT_READ_IO 0x000000800
194 FAULT_ORPHAN !! 190 FAULT_CHECKPOINT 0x000001000
195 FAULT_BLOCK !! 191 FAULT_DISCARD 0x000002000
196 FAULT_DIR_DEPTH !! 192 FAULT_WRITE_IO 0x000004000
197 FAULT_EVICT_INODE !! 193 =================== ===========
198 FAULT_TRUNCATE !! 194 mode=%s Control block allocation mode which supports "adaptive"
199 FAULT_READ_IO !! 195 and "lfs". In "lfs" mode, there should be no random
200 FAULT_CHECKPOINT !! 196 writes towards main area.
201 FAULT_DISCARD !! 197 io_bits=%u Set the bit size of write IO requests. It should be set
202 FAULT_WRITE_IO !! 198 with "mode=lfs".
203 FAULT_SLAB_ALLOC !! 199 usrquota Enable plain user disk quota accounting.
204 FAULT_DQUOT_INIT !! 200 grpquota Enable plain group disk quota accounting.
205 FAULT_LOCK_OP !! 201 prjquota Enable plain project quota accounting.
206 FAULT_BLKADDR_VALIDIT !! 202 usrjquota=<file> Appoint specified file and type during mount, so that quota
207 FAULT_BLKADDR_CONSIST !! 203 grpjquota=<file> information can be properly updated during recovery flow,
208 FAULT_NO_SEGMENT !! 204 prjjquota=<file> <quota file>: must be in root directory;
209 ===================== !! 205 jqfmt=<quota type> <quota type>: [vfsold,vfsv0,vfsv1].
210 mode=%s Control block allocat !! 206 offusrjquota Turn off user journelled quota.
211 and "lfs". In "lfs" m !! 207 offgrpjquota Turn off group journelled quota.
212 writes towards main a !! 208 offprjjquota Turn off project journelled quota.
213 "fragment:segment" an !! 209 quota Enable plain user disk quota accounting.
214 These are developer o !! 210 noquota Disable all plain disk quota option.
215 fragmentation/after-G !! 211 whint_mode=%s Control which write hints are passed down to block
216 modes to understand f !! 212 layer. This supports "off", "user-based", and
217 and eventually get so !! 213 "fs-based". In "off" mode (default), f2fs does not pass
218 In "fragment:segment" !! 214 down hints. In "user-based" mode, f2fs tries to pass
219 position. With this, !! 215 down hints given by users. And in "fs-based" mode, f2fs
220 In "fragment:block", !! 216 passes down hints with its policy.
221 "max_fragment_chunk" !! 217 alloc_mode=%s Adjust block allocation policy, which supports "reuse"
222 We added some randomn !! 218 and "default".
223 it close to realistic !! 219 fsync_mode=%s Control the policy of fsync. Currently supports "posix",
224 1..<max_fragment_chun !! 220 "strict", and "nobarrier". In "posix" mode, which is
225 length of 1..<max_fra !! 221 default, fsync will follow POSIX semantics and does a
226 allocated blocks will !! 222 light operation to improve the filesystem performance.
227 Note that "fragment:b !! 223 In "strict" mode, fsync will be heavy and behaves in line
228 option for more rando !! 224 with xfs, ext4 and btrfs, where xfstest generic/342 will
229 Please, use these opt !! 225 pass, but the performance will regress. "nobarrier" is
230 recommend to re-forma !! 226 based on "posix", but doesn't issue flush command for
231 usrquota Enable plain user dis !! 227 non-atomic files likewise "nobarrier" mount option.
232 grpquota Enable plain group di !! 228 test_dummy_encryption Enable dummy encryption, which provides a fake fscrypt
233 prjquota Enable plain project !! 229 context. The fake fscrypt context is used by xfstests.
234 usrjquota=<file> Appoint specified fil !! 230 checkpoint=%s[:%u[%]] Set to "disable" to turn off checkpointing. Set to "enable"
235 grpjquota=<file> information can be pr !! 231 to reenable checkpointing. Is enabled by default. While
236 prjjquota=<file> <quota file>: must be !! 232 disabled, any unmounting or unexpected shutdowns will cause
237 jqfmt=<quota type> <quota type>: [vfsold !! 233 the filesystem contents to appear as they did when the
238 offusrjquota Turn off user journal !! 234 filesystem was mounted with that option.
239 offgrpjquota Turn off group journa !! 235 While mounting with checkpoint=disabled, the filesystem must
240 offprjjquota Turn off project jour !! 236 run garbage collection to ensure that all available space can
241 quota Enable plain user dis !! 237 be used. If this takes too much time, the mount may return
242 noquota Disable all plain dis !! 238 EAGAIN. You may optionally add a value to indicate how much
243 alloc_mode=%s Adjust block allocati !! 239 of the disk you would be willing to temporarily give up to
244 and "default". !! 240 avoid additional garbage collection. This can be given as a
245 fsync_mode=%s Control the policy of !! 241 number of blocks, or as a percent. For instance, mounting
246 "strict", and "nobarr !! 242 with checkpoint=disable:100% would always succeed, but it may
247 default, fsync will f !! 243 hide up to all remaining free space. The actual space that
248 light operation to im !! 244 would be unusable can be viewed at /sys/fs/f2fs/<disk>/unusable
249 In "strict" mode, fsy !! 245 This space is reclaimed once checkpoint=enable.
250 with xfs, ext4 and bt !! 246 compress_algorithm=%s Control compress algorithm, currently f2fs supports "lzo",
251 pass, but the perform !! 247 "lz4" and "zstd" algorithm.
252 based on "posix", but !! 248 compress_log_size=%u Support configuring compress cluster size, the size will
253 non-atomic files like !! 249 be 4KB * (1 << %u), 16KB is minimum size, also it's
254 test_dummy_encryption !! 250 default size.
255 test_dummy_encryption=%s !! 251 compress_extension=%s Support adding specified extension, so that f2fs can enable
256 Enable dummy encrypti !! 252 compression on those corresponding files, e.g. if all files
257 context. The fake fsc !! 253 with '.ext' has high compression rate, we can set the '.ext'
258 The argument may be e !! 254 on compression extension list and enable compression on
259 select the correspond !! 255 these file by default rather than to enable it via ioctl.
260 checkpoint=%s[:%u[%]] Set to "disable" to t !! 256 For other files, we can still enable compression via ioctl.
261 to reenable checkpoin !! 257 ====================== ============================================================
262 disabled, any unmount <<
263 the filesystem conten <<
264 filesystem was mounte <<
265 While mounting with c <<
266 run garbage collectio <<
267 be used. If this take <<
268 EAGAIN. You may optio <<
269 of the disk you would <<
270 avoid additional garb <<
271 number of blocks, or <<
272 with checkpoint=disab <<
273 hide up to all remain <<
274 would be unusable can <<
275 This space is reclaim <<
276 checkpoint_merge When checkpoint is en <<
277 daemon and make it to <<
278 much as possible to e <<
279 we can eliminate the <<
280 operation when the ch <<
281 a cgroup having low i <<
282 do better, we set the <<
283 to "3", to give one h <<
284 This is the same way <<
285 journaling thread of <<
286 nocheckpoint_merge Disable checkpoint me <<
287 compress_algorithm=%s Control compress algo <<
288 "lz4", "zstd" and "lz <<
289 compress_algorithm=%s:%d Control compress algo <<
290 "lz4" and "zstd" supp <<
291 algorithm level <<
292 lz4 3 - 16 <<
293 zstd 1 - 22 <<
294 compress_log_size=%u Support configuring c <<
295 be 4KB * (1 << %u). T <<
296 compress_extension=%s Support adding specif <<
297 compression on those <<
298 with '.ext' has high <<
299 on compression extens <<
300 these file by default <<
301 For other files, we c <<
302 Note that, there is o <<
303 can be set to enable <<
304 nocompress_extension=%s Support adding specif <<
305 compression on those <<
306 If you know exactly w <<
307 The same extension na <<
308 extension at the same <<
309 If the compress exten <<
310 nocompress extension <<
311 Don't allow use '*' t <<
312 After add nocompress_ <<
313 dir_flag < comp_exten <<
314 See more in compressi <<
315 <<
316 compress_chksum Support verifying chk <<
317 compress_mode=%s Control file compress <<
318 modes. In "fs" mode ( <<
319 on the compression en <<
320 the automaic compress <<
321 choosing the target f <<
322 compression/decompres <<
323 ioctls. <<
324 compress_cache Support to use addres <<
325 cache compressed bloc <<
326 random read. <<
327 inlinecrypt When possible, encryp <<
328 files using the blk-c <<
329 filesystem-layer encr <<
330 inline encryption har <<
331 unaffected. For more <<
332 Documentation/block/i <<
333 atgc Enable age-threshold <<
334 effectiveness and eff <<
335 discard_unit=%s Control discard unit, <<
336 and "section", issued <<
337 aligned to the unit, <<
338 so that small discard <<
339 For blkzoned device, <<
340 default, it is helpfu <<
341 reduce memory cost by <<
342 discard. <<
343 memory=%s Control memory mode. <<
344 "low" mode is introdu <<
345 Because of the nature <<
346 will try to save memo <<
347 "normal" mode is the <<
348 age_extent_cache Enable an age extent <<
349 data block update fre <<
350 order to provide bett <<
351 allocation. <<
352 errors=%s Specify f2fs behavior <<
353 "panic", "continue" a <<
354 panic immediately, co <<
355 the partition in read <<
356 mode. <<
357 ===================== <<
358 mode <<
359 ===================== <<
360 access ops <<
361 syscall errors <<
362 mount option <<
363 pending dir write <<
364 pending non-dir write <<
365 pending node write <<
366 pending meta write <<
367 ===================== <<
368 ======================== ===================== <<
369 258
370 Debugfs Entries 259 Debugfs Entries
371 =============== 260 ===============
372 261
373 /sys/kernel/debug/f2fs/ contains information a 262 /sys/kernel/debug/f2fs/ contains information about all the partitions mounted as
374 f2fs. Each file shows the whole f2fs informati 263 f2fs. Each file shows the whole f2fs information.
375 264
376 /sys/kernel/debug/f2fs/status includes: 265 /sys/kernel/debug/f2fs/status includes:
377 266
378 - major file system information managed by f2 267 - major file system information managed by f2fs currently
379 - average SIT information about whole segment 268 - average SIT information about whole segments
380 - current memory footprint consumed by f2fs. 269 - current memory footprint consumed by f2fs.
381 270
382 Sysfs Entries 271 Sysfs Entries
383 ============= 272 =============
384 273
385 Information about mounted f2fs file systems ca 274 Information about mounted f2fs file systems can be found in
386 /sys/fs/f2fs. Each mounted filesystem will ha 275 /sys/fs/f2fs. Each mounted filesystem will have a directory in
387 /sys/fs/f2fs based on its device name (i.e., / 276 /sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda).
388 The files in each per-device directory are sho 277 The files in each per-device directory are shown in table below.
389 278
390 Files in /sys/fs/f2fs/<devname> 279 Files in /sys/fs/f2fs/<devname>
391 (see also Documentation/ABI/testing/sysfs-fs-f 280 (see also Documentation/ABI/testing/sysfs-fs-f2fs)
392 281
393 Usage 282 Usage
394 ===== 283 =====
395 284
396 1. Download userland tools and compile them. 285 1. Download userland tools and compile them.
397 286
398 2. Skip, if f2fs was compiled statically insid 287 2. Skip, if f2fs was compiled statically inside kernel.
399 Otherwise, insert the f2fs.ko module:: 288 Otherwise, insert the f2fs.ko module::
400 289
401 # insmod f2fs.ko 290 # insmod f2fs.ko
402 291
403 3. Create a directory to use when mounting:: !! 292 3. Create a directory trying to mount::
404 293
405 # mkdir /mnt/f2fs 294 # mkdir /mnt/f2fs
406 295
407 4. Format the block device, and then mount as 296 4. Format the block device, and then mount as f2fs::
408 297
409 # mkfs.f2fs -l label /dev/block_device 298 # mkfs.f2fs -l label /dev/block_device
410 # mount -t f2fs /dev/block_device /mnt 299 # mount -t f2fs /dev/block_device /mnt/f2fs
411 300
412 mkfs.f2fs 301 mkfs.f2fs
413 --------- 302 ---------
414 The mkfs.f2fs is for the use of formatting a p 303 The mkfs.f2fs is for the use of formatting a partition as the f2fs filesystem,
415 which builds a basic on-disk layout. 304 which builds a basic on-disk layout.
416 305
417 The quick options consist of: !! 306 The options consist of:
418 307
419 =============== =========================== 308 =============== ===========================================================
420 ``-l [label]`` Give a volume label, up to 309 ``-l [label]`` Give a volume label, up to 512 unicode name.
421 ``-a [0 or 1]`` Split start location of eac 310 ``-a [0 or 1]`` Split start location of each area for heap-based allocation.
422 311
423 1 is set by default, which 312 1 is set by default, which performs this.
424 ``-o [int]`` Set overprovision ratio in 313 ``-o [int]`` Set overprovision ratio in percent over volume size.
425 314
426 5 is set by default. 315 5 is set by default.
427 ``-s [int]`` Set the number of segments 316 ``-s [int]`` Set the number of segments per section.
428 317
429 1 is set by default. 318 1 is set by default.
430 ``-z [int]`` Set the number of sections 319 ``-z [int]`` Set the number of sections per zone.
431 320
432 1 is set by default. 321 1 is set by default.
433 ``-e [str]`` Set basic extension list. e 322 ``-e [str]`` Set basic extension list. e.g. "mp3,gif,mov"
434 ``-t [0 or 1]`` Disable discard command or 323 ``-t [0 or 1]`` Disable discard command or not.
435 324
436 1 is set by default, which 325 1 is set by default, which conducts discard.
437 =============== =========================== 326 =============== ===========================================================
438 327
439 Note: please refer to the manpage of mkfs.f2fs <<
440 <<
441 fsck.f2fs 328 fsck.f2fs
442 --------- 329 ---------
443 The fsck.f2fs is a tool to check the consisten 330 The fsck.f2fs is a tool to check the consistency of an f2fs-formatted
444 partition, which examines whether the filesyst 331 partition, which examines whether the filesystem metadata and user-made data
445 are cross-referenced correctly or not. 332 are cross-referenced correctly or not.
446 Note that, initial version of the tool does no 333 Note that, initial version of the tool does not fix any inconsistency.
447 334
448 The quick options consist of:: !! 335 The options consist of::
449 336
450 -d debug level [default:0] 337 -d debug level [default:0]
451 338
452 Note: please refer to the manpage of fsck.f2fs <<
453 <<
454 dump.f2fs 339 dump.f2fs
455 --------- 340 ---------
456 The dump.f2fs shows the information of specifi 341 The dump.f2fs shows the information of specific inode and dumps SSA and SIT to
457 file. Each file is dump_ssa and dump_sit. 342 file. Each file is dump_ssa and dump_sit.
458 343
459 The dump.f2fs is used to debug on-disk data st 344 The dump.f2fs is used to debug on-disk data structures of the f2fs filesystem.
460 It shows on-disk inode information recognized 345 It shows on-disk inode information recognized by a given inode number, and is
461 able to dump all the SSA and SIT entries into 346 able to dump all the SSA and SIT entries into predefined files, ./dump_ssa and
462 ./dump_sit respectively. 347 ./dump_sit respectively.
463 348
464 The options consist of:: 349 The options consist of::
465 350
466 -d debug level [default:0] 351 -d debug level [default:0]
467 -i inode no (hex) 352 -i inode no (hex)
468 -s [SIT dump segno from #1~#2 (decimal), for 353 -s [SIT dump segno from #1~#2 (decimal), for all 0~-1]
469 -a [SSA dump segno from #1~#2 (decimal), for 354 -a [SSA dump segno from #1~#2 (decimal), for all 0~-1]
470 355
471 Examples:: 356 Examples::
472 357
473 # dump.f2fs -i [ino] /dev/sdx 358 # dump.f2fs -i [ino] /dev/sdx
474 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump) 359 # dump.f2fs -s 0~-1 /dev/sdx (SIT dump)
475 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump) 360 # dump.f2fs -a 0~-1 /dev/sdx (SSA dump)
476 361
477 Note: please refer to the manpage of dump.f2fs <<
478 <<
479 sload.f2fs <<
480 ---------- <<
481 The sload.f2fs gives a way to insert files and <<
482 image. This tool is useful when building f2fs <<
483 <<
484 Note: please refer to the manpage of sload.f2f <<
485 <<
486 resize.f2fs <<
487 ----------- <<
488 The resize.f2fs lets a user resize the f2fs-fo <<
489 all the files and directories stored in the im <<
490 <<
491 Note: please refer to the manpage of resize.f2 <<
492 <<
493 defrag.f2fs <<
494 ----------- <<
495 The defrag.f2fs can be used to defragment scat <<
496 filesystem metadata across the disk. This can <<
497 more free consecutive space. <<
498 <<
499 Note: please refer to the manpage of defrag.f2 <<
500 <<
501 f2fs_io <<
502 ------- <<
503 The f2fs_io is a simple tool to issue various <<
504 f2fs-specific ones, which is very useful for Q <<
505 <<
506 Note: please refer to the manpage of f2fs_io(8 <<
507 <<
508 Design 362 Design
509 ====== 363 ======
510 364
511 On-disk Layout 365 On-disk Layout
512 -------------- 366 --------------
513 367
514 F2FS divides the whole volume into a number of 368 F2FS divides the whole volume into a number of segments, each of which is fixed
515 to 2MB in size. A section is composed of conse 369 to 2MB in size. A section is composed of consecutive segments, and a zone
516 consists of a set of sections. By default, sec 370 consists of a set of sections. By default, section and zone sizes are set to one
517 segment size identically, but users can easily 371 segment size identically, but users can easily modify the sizes by mkfs.
518 372
519 F2FS splits the entire volume into six areas, 373 F2FS splits the entire volume into six areas, and all the areas except superblock
520 consist of multiple segments as described belo !! 374 consists of multiple segments as described below::
521 375
522 al 376 align with the zone size <-|
523 |-> align with the segment si 377 |-> align with the segment size
524 _________________________________________ 378 _________________________________________________________________________
525 | | | Segment | 379 | | | Segment | Node | Segment | |
526 | Superblock | Checkpoint | Info. | 380 | Superblock | Checkpoint | Info. | Address | Summary | Main |
527 | (SB) | (CP) | Table (SIT) | 381 | (SB) | (CP) | Table (SIT) | Table (NAT) | Area (SSA) | |
528 |____________|_____2______|______N______|_ 382 |____________|_____2______|______N______|______N______|______N_____|__N___|
529 383 . .
530 384 . .
531 385 . .
532 ._________ 386 ._________________________________________.
533 |_Segment_ 387 |_Segment_|_..._|_Segment_|_..._|_Segment_|
534 . 388 . .
535 ._________ 389 ._________._________
536 |_section_ 390 |_section_|__...__|_
537 . 391 . .
538 .________. 392 .________.
539 |__zone__| 393 |__zone__|
540 394
541 - Superblock (SB) 395 - Superblock (SB)
542 It is located at the beginning of the parti 396 It is located at the beginning of the partition, and there exist two copies
543 to avoid file system crash. It contains bas 397 to avoid file system crash. It contains basic partition information and some
544 default parameters of f2fs. 398 default parameters of f2fs.
545 399
546 - Checkpoint (CP) 400 - Checkpoint (CP)
547 It contains file system information, bitmap 401 It contains file system information, bitmaps for valid NAT/SIT sets, orphan
548 inode lists, and summary entries of current 402 inode lists, and summary entries of current active segments.
549 403
550 - Segment Information Table (SIT) 404 - Segment Information Table (SIT)
551 It contains segment information such as val 405 It contains segment information such as valid block count and bitmap for the
552 validity of all the blocks. 406 validity of all the blocks.
553 407
554 - Node Address Table (NAT) 408 - Node Address Table (NAT)
555 It is composed of a block address table for 409 It is composed of a block address table for all the node blocks stored in
556 Main area. 410 Main area.
557 411
558 - Segment Summary Area (SSA) 412 - Segment Summary Area (SSA)
559 It contains summary entries which contains 413 It contains summary entries which contains the owner information of all the
560 data and node blocks stored in Main area. 414 data and node blocks stored in Main area.
561 415
562 - Main Area 416 - Main Area
563 It contains file and directory data includi 417 It contains file and directory data including their indices.
564 418
565 In order to avoid misalignment between file sy 419 In order to avoid misalignment between file system and flash-based storage, F2FS
566 aligns the start block address of CP with the 420 aligns the start block address of CP with the segment size. Also, it aligns the
567 start block address of Main area with the zone 421 start block address of Main area with the zone size by reserving some segments
568 in SSA area. 422 in SSA area.
569 423
570 Reference the following survey for additional 424 Reference the following survey for additional technical details.
571 https://wiki.linaro.org/WorkingGroups/Kernel/P 425 https://wiki.linaro.org/WorkingGroups/Kernel/Projects/FlashCardSurvey
572 426
573 File System Metadata Structure 427 File System Metadata Structure
574 ------------------------------ 428 ------------------------------
575 429
576 F2FS adopts the checkpointing scheme to mainta 430 F2FS adopts the checkpointing scheme to maintain file system consistency. At
577 mount time, F2FS first tries to find the last 431 mount time, F2FS first tries to find the last valid checkpoint data by scanning
578 CP area. In order to reduce the scanning time, 432 CP area. In order to reduce the scanning time, F2FS uses only two copies of CP.
579 One of them always indicates the last valid da 433 One of them always indicates the last valid data, which is called as shadow copy
580 mechanism. In addition to CP, NAT and SIT also 434 mechanism. In addition to CP, NAT and SIT also adopt the shadow copy mechanism.
581 435
582 For file system consistency, each CP points to 436 For file system consistency, each CP points to which NAT and SIT copies are
583 valid, as shown as below:: 437 valid, as shown as below::
584 438
585 +--------+----------+---------+ 439 +--------+----------+---------+
586 | CP | SIT | NAT | 440 | CP | SIT | NAT |
587 +--------+----------+---------+ 441 +--------+----------+---------+
588 . . . . 442 . . . .
589 . . . . 443 . . . .
590 . . . 444 . . . .
591 +-------+-------+--------+--------+--------+ 445 +-------+-------+--------+--------+--------+--------+
592 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | 446 | CP #0 | CP #1 | SIT #0 | SIT #1 | NAT #0 | NAT #1 |
593 +-------+-------+--------+--------+--------+ 447 +-------+-------+--------+--------+--------+--------+
594 | ^ 448 | ^ ^
595 | | 449 | | |
596 `---------------------------------------- 450 `----------------------------------------'
597 451
598 Index Structure 452 Index Structure
599 --------------- 453 ---------------
600 454
601 The key data structure to manage the data loca 455 The key data structure to manage the data locations is a "node". Similar to
602 traditional file structures, F2FS has three ty 456 traditional file structures, F2FS has three types of node: inode, direct node,
603 indirect node. F2FS assigns 4KB to an inode bl 457 indirect node. F2FS assigns 4KB to an inode block which contains 923 data block
604 indices, two direct node pointers, two indirec 458 indices, two direct node pointers, two indirect node pointers, and one double
605 indirect node pointer as described below. One 459 indirect node pointer as described below. One direct node block contains 1018
606 data blocks, and one indirect node block conta 460 data blocks, and one indirect node block contains also 1018 node blocks. Thus,
607 one inode block (i.e., a file) covers:: 461 one inode block (i.e., a file) covers::
608 462
609 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 10 463 4KB * (923 + 2 * 1018 + 2 * 1018 * 1018 + 1018 * 1018 * 1018) := 3.94TB.
610 464
611 Inode block (4KB) 465 Inode block (4KB)
612 |- data (923) 466 |- data (923)
613 |- direct node (2) 467 |- direct node (2)
614 | `- data (1018) 468 | `- data (1018)
615 |- indirect node (2) 469 |- indirect node (2)
616 | `- direct node (1018) 470 | `- direct node (1018)
617 | `- data (1018) 471 | `- data (1018)
618 `- double indirect node (1) 472 `- double indirect node (1)
619 `- indirect node (101 473 `- indirect node (1018)
620 `- direc 474 `- direct node (1018)
621 475 `- data (1018)
622 476
623 Note that all the node blocks are mapped by NA !! 477 Note that, all the node blocks are mapped by NAT which means the location of
624 each node is translated by the NAT table. In t 478 each node is translated by the NAT table. In the consideration of the wandering
625 tree problem, F2FS is able to cut off the prop 479 tree problem, F2FS is able to cut off the propagation of node updates caused by
626 leaf data writes. 480 leaf data writes.
627 481
628 Directory Structure 482 Directory Structure
629 ------------------- 483 -------------------
630 484
631 A directory entry occupies 11 bytes, which con 485 A directory entry occupies 11 bytes, which consists of the following attributes.
632 486
633 - hash hash value of the file name 487 - hash hash value of the file name
634 - ino inode number 488 - ino inode number
635 - len the length of file name 489 - len the length of file name
636 - type file type such as directory, s 490 - type file type such as directory, symlink, etc
637 491
638 A dentry block consists of 214 dentry slots an 492 A dentry block consists of 214 dentry slots and file names. Therein a bitmap is
639 used to represent whether each dentry is valid 493 used to represent whether each dentry is valid or not. A dentry block occupies
640 4KB with the following composition. 494 4KB with the following composition.
641 495
642 :: 496 ::
643 497
644 Dentry Block(4 K) = bitmap (27 bytes) + rese 498 Dentry Block(4 K) = bitmap (27 bytes) + reserved (3 bytes) +
645 dentries(11 * 214 bytes) 499 dentries(11 * 214 bytes) + file name (8 * 214 bytes)
646 500
647 [Bucket] 501 [Bucket]
648 +-------------------------------- 502 +--------------------------------+
649 |dentry block 1 | dentry block 2 503 |dentry block 1 | dentry block 2 |
650 +-------------------------------- 504 +--------------------------------+
651 . . 505 . .
652 . . 506 . .
653 . [Dentry Block Structure: 4KB] 507 . [Dentry Block Structure: 4KB] .
654 +--------+----------+----------+------------ 508 +--------+----------+----------+------------+
655 | bitmap | reserved | dentries | file names 509 | bitmap | reserved | dentries | file names |
656 +--------+----------+----------+------------ 510 +--------+----------+----------+------------+
657 [Dentry Block: 4KB] . . 511 [Dentry Block: 4KB] . .
658 . . 512 . .
659 . . 513 . .
660 +------+------+-----+------+ 514 +------+------+-----+------+
661 | hash | ino | len | type | 515 | hash | ino | len | type |
662 +------+------+-----+------+ 516 +------+------+-----+------+
663 [Dentry Structure: 11 bytes] 517 [Dentry Structure: 11 bytes]
664 518
665 F2FS implements multi-level hash tables for di 519 F2FS implements multi-level hash tables for directory structure. Each level has
666 a hash table with dedicated number of hash buc 520 a hash table with dedicated number of hash buckets as shown below. Note that
667 "A(2B)" means a bucket includes 2 data blocks. 521 "A(2B)" means a bucket includes 2 data blocks.
668 522
669 :: 523 ::
670 524
671 ---------------------- 525 ----------------------
672 A : bucket 526 A : bucket
673 B : block 527 B : block
674 N : MAX_DIR_HASH_DEPTH 528 N : MAX_DIR_HASH_DEPTH
675 ---------------------- 529 ----------------------
676 530
677 level #0 | A(2B) 531 level #0 | A(2B)
678 | 532 |
679 level #1 | A(2B) - A(2B) 533 level #1 | A(2B) - A(2B)
680 | 534 |
681 level #2 | A(2B) - A(2B) - A(2B) - A(2B) 535 level #2 | A(2B) - A(2B) - A(2B) - A(2B)
682 . | . . . . 536 . | . . . .
683 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) 537 level #N/2 | A(2B) - A(2B) - A(2B) - A(2B) - A(2B) - ... - A(2B)
684 . | . . . . 538 . | . . . .
685 level #N | A(4B) - A(4B) - A(4B) - A(4B) 539 level #N | A(4B) - A(4B) - A(4B) - A(4B) - A(4B) - ... - A(4B)
686 540
687 The number of blocks and buckets are determine 541 The number of blocks and buckets are determined by::
688 542
689 ,- 2, if n < MAX_D 543 ,- 2, if n < MAX_DIR_HASH_DEPTH / 2,
690 # of blocks in level #n = | 544 # of blocks in level #n = |
691 `- 4, Otherwise 545 `- 4, Otherwise
692 546
693 ,- 2^(n + dir_lev 547 ,- 2^(n + dir_level),
694 | if n + d 548 | if n + dir_level < MAX_DIR_HASH_DEPTH / 2,
695 # of buckets in level #n = | 549 # of buckets in level #n = |
696 `- 2^((MAX_DIR_HA 550 `- 2^((MAX_DIR_HASH_DEPTH / 2) - 1),
697 Otherwis 551 Otherwise
698 552
699 When F2FS finds a file name in a directory, at 553 When F2FS finds a file name in a directory, at first a hash value of the file
700 name is calculated. Then, F2FS scans the hash 554 name is calculated. Then, F2FS scans the hash table in level #0 to find the
701 dentry consisting of the file name and its ino 555 dentry consisting of the file name and its inode number. If not found, F2FS
702 scans the next hash table in level #1. In this 556 scans the next hash table in level #1. In this way, F2FS scans hash tables in
703 each levels incrementally from 1 to N. In each !! 557 each levels incrementally from 1 to N. In each levels F2FS needs to scan only
704 one bucket determined by the following equatio 558 one bucket determined by the following equation, which shows O(log(# of files))
705 complexity:: 559 complexity::
706 560
707 bucket number to scan in level #n = (hash va 561 bucket number to scan in level #n = (hash value) % (# of buckets in level #n)
708 562
709 In the case of file creation, F2FS finds empty 563 In the case of file creation, F2FS finds empty consecutive slots that cover the
710 file name. F2FS searches the empty slots in th 564 file name. F2FS searches the empty slots in the hash tables of whole levels from
711 1 to N in the same way as the lookup operation 565 1 to N in the same way as the lookup operation.
712 566
713 The following figure shows an example of two c 567 The following figure shows an example of two cases holding children::
714 568
715 --------------> Dir <-------------- 569 --------------> Dir <--------------
716 | | 570 | |
717 child child 571 child child
718 572
719 child - child [hole] - 573 child - child [hole] - child
720 574
721 child - child - child [hole] - 575 child - child - child [hole] - [hole] - child
722 576
723 Case 1: Case 2: 577 Case 1: Case 2:
724 Number of children = 6, Number of 578 Number of children = 6, Number of children = 3,
725 File size = 7 File size 579 File size = 7 File size = 7
726 580
727 Default Block Allocation 581 Default Block Allocation
728 ------------------------ 582 ------------------------
729 583
730 At runtime, F2FS manages six active logs insid 584 At runtime, F2FS manages six active logs inside "Main" area: Hot/Warm/Cold node
731 and Hot/Warm/Cold data. 585 and Hot/Warm/Cold data.
732 586
733 - Hot node contains direct node blocks of 587 - Hot node contains direct node blocks of directories.
734 - Warm node contains direct node blocks ex 588 - Warm node contains direct node blocks except hot node blocks.
735 - Cold node contains indirect node blocks 589 - Cold node contains indirect node blocks
736 - Hot data contains dentry blocks 590 - Hot data contains dentry blocks
737 - Warm data contains data blocks except ho 591 - Warm data contains data blocks except hot and cold data blocks
738 - Cold data contains multimedia data or mi 592 - Cold data contains multimedia data or migrated data blocks
739 593
740 LFS has two schemes for free space management: 594 LFS has two schemes for free space management: threaded log and copy-and-compac-
741 tion. The copy-and-compaction scheme which is 595 tion. The copy-and-compaction scheme which is known as cleaning, is well-suited
742 for devices showing very good sequential write 596 for devices showing very good sequential write performance, since free segments
743 are served all the time for writing new data. 597 are served all the time for writing new data. However, it suffers from cleaning
744 overhead under high utilization. Contrarily, t 598 overhead under high utilization. Contrarily, the threaded log scheme suffers
745 from random writes, but no cleaning process is 599 from random writes, but no cleaning process is needed. F2FS adopts a hybrid
746 scheme where the copy-and-compaction scheme is 600 scheme where the copy-and-compaction scheme is adopted by default, but the
747 policy is dynamically changed to the threaded 601 policy is dynamically changed to the threaded log scheme according to the file
748 system status. 602 system status.
749 603
750 In order to align F2FS with underlying flash-b 604 In order to align F2FS with underlying flash-based storage, F2FS allocates a
751 segment in a unit of section. F2FS expects tha 605 segment in a unit of section. F2FS expects that the section size would be the
752 same as the unit size of garbage collection in 606 same as the unit size of garbage collection in FTL. Furthermore, with respect
753 to the mapping granularity in FTL, F2FS alloca 607 to the mapping granularity in FTL, F2FS allocates each section of the active
754 logs from different zones as much as possible, 608 logs from different zones as much as possible, since FTL can write the data in
755 the active logs into one allocation unit accor 609 the active logs into one allocation unit according to its mapping granularity.
756 610
757 Cleaning process 611 Cleaning process
758 ---------------- 612 ----------------
759 613
760 F2FS does cleaning both on demand and in the b 614 F2FS does cleaning both on demand and in the background. On-demand cleaning is
761 triggered when there are not enough free segme 615 triggered when there are not enough free segments to serve VFS calls. Background
762 cleaner is operated by a kernel thread, and tr 616 cleaner is operated by a kernel thread, and triggers the cleaning job when the
763 system is idle. 617 system is idle.
764 618
765 F2FS supports two victim selection policies: g 619 F2FS supports two victim selection policies: greedy and cost-benefit algorithms.
766 In the greedy algorithm, F2FS selects a victim 620 In the greedy algorithm, F2FS selects a victim segment having the smallest number
767 of valid blocks. In the cost-benefit algorithm 621 of valid blocks. In the cost-benefit algorithm, F2FS selects a victim segment
768 according to the segment age and the number of 622 according to the segment age and the number of valid blocks in order to address
769 log block thrashing problem in the greedy algo 623 log block thrashing problem in the greedy algorithm. F2FS adopts the greedy
770 algorithm for on-demand cleaner, while backgro 624 algorithm for on-demand cleaner, while background cleaner adopts cost-benefit
771 algorithm. 625 algorithm.
772 626
773 In order to identify whether the data in the v 627 In order to identify whether the data in the victim segment are valid or not,
774 F2FS manages a bitmap. Each bit represents the 628 F2FS manages a bitmap. Each bit represents the validity of a block, and the
775 bitmap is composed of a bit stream covering wh 629 bitmap is composed of a bit stream covering whole blocks in main area.
776 630
777 Write-hint Policy 631 Write-hint Policy
778 ----------------- 632 -----------------
779 633
780 F2FS sets the whint all the time with the belo !! 634 1) whint_mode=off. F2FS only passes down WRITE_LIFE_NOT_SET.
>> 635
>> 636 2) whint_mode=user-based. F2FS tries to pass down hints given by
>> 637 users.
781 638
782 ===================== ======================== 639 ===================== ======================== ===================
783 User F2FS 640 User F2FS Block
784 ===================== ======================== 641 ===================== ======================== ===================
785 N/A META !! 642 META WRITE_LIFE_NOT_SET
786 N/A HOT_NODE !! 643 HOT_NODE "
787 N/A WARM_NODE !! 644 WARM_NODE "
788 N/A COLD_NODE !! 645 COLD_NODE "
789 ioctl(COLD) COLD_DATA 646 ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME
790 extension list " 647 extension list " "
791 648
792 -- buffered io 649 -- buffered io
793 N/A COLD_DATA !! 650 WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
794 N/A HOT_DATA !! 651 WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
795 N/A WARM_DATA !! 652 WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET
>> 653 WRITE_LIFE_NONE " "
>> 654 WRITE_LIFE_MEDIUM " "
>> 655 WRITE_LIFE_LONG " "
>> 656
>> 657 -- direct io
>> 658 WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
>> 659 WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
>> 660 WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET
>> 661 WRITE_LIFE_NONE " WRITE_LIFE_NONE
>> 662 WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM
>> 663 WRITE_LIFE_LONG " WRITE_LIFE_LONG
>> 664 ===================== ======================== ===================
>> 665
>> 666 3) whint_mode=fs-based. F2FS passes down hints with its policy.
>> 667
>> 668 ===================== ======================== ===================
>> 669 User F2FS Block
>> 670 ===================== ======================== ===================
>> 671 META WRITE_LIFE_MEDIUM;
>> 672 HOT_NODE WRITE_LIFE_NOT_SET
>> 673 WARM_NODE "
>> 674 COLD_NODE WRITE_LIFE_NONE
>> 675 ioctl(COLD) COLD_DATA WRITE_LIFE_EXTREME
>> 676 extension list " "
>> 677
>> 678 -- buffered io
>> 679 WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
>> 680 WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
>> 681 WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_LONG
>> 682 WRITE_LIFE_NONE " "
>> 683 WRITE_LIFE_MEDIUM " "
>> 684 WRITE_LIFE_LONG " "
796 685
797 -- direct io 686 -- direct io
798 WRITE_LIFE_EXTREME COLD_DATA 687 WRITE_LIFE_EXTREME COLD_DATA WRITE_LIFE_EXTREME
799 WRITE_LIFE_SHORT HOT_DATA 688 WRITE_LIFE_SHORT HOT_DATA WRITE_LIFE_SHORT
800 WRITE_LIFE_NOT_SET WARM_DATA 689 WRITE_LIFE_NOT_SET WARM_DATA WRITE_LIFE_NOT_SET
801 WRITE_LIFE_NONE " 690 WRITE_LIFE_NONE " WRITE_LIFE_NONE
802 WRITE_LIFE_MEDIUM " 691 WRITE_LIFE_MEDIUM " WRITE_LIFE_MEDIUM
803 WRITE_LIFE_LONG " 692 WRITE_LIFE_LONG " WRITE_LIFE_LONG
804 ===================== ======================== 693 ===================== ======================== ===================
805 694
806 Fallocate(2) Policy 695 Fallocate(2) Policy
807 ------------------- 696 -------------------
808 697
809 The default policy follows the below POSIX rul !! 698 The default policy follows the below posix rule.
810 699
811 Allocating disk space 700 Allocating disk space
812 The default operation (i.e., mode is zero) 701 The default operation (i.e., mode is zero) of fallocate() allocates
813 the disk space within the range specified 702 the disk space within the range specified by offset and len. The
814 file size (as reported by stat(2)) will be 703 file size (as reported by stat(2)) will be changed if offset+len is
815 greater than the file size. Any subregion 704 greater than the file size. Any subregion within the range specified
816 by offset and len that did not contain dat 705 by offset and len that did not contain data before the call will be
817 initialized to zero. This default behavio 706 initialized to zero. This default behavior closely resembles the
818 behavior of the posix_fallocate(3) library 707 behavior of the posix_fallocate(3) library function, and is intended
819 as a method of optimally implementing that 708 as a method of optimally implementing that function.
820 709
821 However, once F2FS receives ioctl(fd, F2FS_IOC 710 However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to
822 fallocate(fd, DEFAULT_MODE), it allocates on-d !! 711 fallocate(fd, DEFAULT_MODE), it allocates on-disk blocks addressess having
823 zero or random data, which is useful to the be 712 zero or random data, which is useful to the below scenario where:
824 713
825 1. create(fd) 714 1. create(fd)
826 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE) 715 2. ioctl(fd, F2FS_IOC_SET_PIN_FILE)
827 3. fallocate(fd, 0, 0, size) 716 3. fallocate(fd, 0, 0, size)
828 4. address = fibmap(fd, offset) 717 4. address = fibmap(fd, offset)
829 5. open(blkdev) 718 5. open(blkdev)
830 6. write(blkdev, address) 719 6. write(blkdev, address)
831 720
832 Compression implementation 721 Compression implementation
833 -------------------------- 722 --------------------------
834 723
835 - New term named cluster is defined as basic u 724 - New term named cluster is defined as basic unit of compression, file can
836 be divided into multiple clusters logically. 725 be divided into multiple clusters logically. One cluster includes 4 << n
837 (n >= 0) logical pages, compression size is 726 (n >= 0) logical pages, compression size is also cluster size, each of
838 cluster can be compressed or not. 727 cluster can be compressed or not.
839 728
840 - In cluster metadata layout, one special bloc 729 - In cluster metadata layout, one special block address is used to indicate
841 a cluster is a compressed one or normal one; !! 730 cluster is compressed one or normal one, for compressed cluster, following
842 metadata maps cluster to [1, 4 << n - 1] phy 731 metadata maps cluster to [1, 4 << n - 1] physical blocks, in where f2fs
843 stores data including compress header and co 732 stores data including compress header and compressed data.
844 733
845 - In order to eliminate write amplification du 734 - In order to eliminate write amplification during overwrite, F2FS only
846 support compression on write-once file, data 735 support compression on write-once file, data can be compressed only when
847 all logical blocks in cluster contain valid !! 736 all logical blocks in file are valid and cluster compress ratio is lower
848 cluster data is lower than specified thresho !! 737 than specified threshold.
849 738
850 - To enable compression on regular inode, ther !! 739 - To enable compression on regular inode, there are three ways:
851 740
852 * chattr +c file 741 * chattr +c file
853 * chattr +c dir; touch dir/file 742 * chattr +c dir; touch dir/file
854 * mount w/ -o compress_extension=ext; touch 743 * mount w/ -o compress_extension=ext; touch file.ext
855 * mount w/ -o compress_extension=*; touch an <<
856 <<
857 - To disable compression on regular inode, the <<
858 <<
859 * chattr -c file <<
860 * mount w/ -o nocompress_extension=ext; touc <<
861 <<
862 - Priority in between FS_COMPR_FL, FS_NOCOMP_F <<
863 <<
864 * compress_extension=so; nocompress_extensio <<
865 dir/foo.so; touch dir/bar.zip; touch dir/b <<
866 should be compresse, bar.zip should be non <<
867 can enable compress on bar.zip. <<
868 * compress_extension=so; nocompress_extensio <<
869 dir/foo.so; touch dir/bar.zip; touch dir/b <<
870 compresse, bar.zip and baz.txt should be n <<
871 chattr+c dir/bar.zip; chattr+c dir/baz.txt <<
872 and baz.txt. <<
873 <<
874 - At this point, compression feature doesn't e <<
875 directly in order to guarantee potential dat <<
876 Instead, the main goal is to reduce data wri <<
877 possible, resulting in extending disk life t <<
878 congestion. Alternatively, we've added ioctl <<
879 interface to reclaim compressed space and sh <<
880 special flag to the inode. Once the compress <<
881 will block writing data to the file until ei <<
882 reserved via ioctl(F2FS_IOC_RESERVE_COMPRESS <<
883 truncated to zero. <<
884 744
885 Compress metadata layout:: 745 Compress metadata layout::
886 746
887 [Dnode Structu 747 [Dnode Structure]
888 +----------------------------- 748 +-----------------------------------------------+
889 | cluster 1 | cluster 2 | .... 749 | cluster 1 | cluster 2 | ......... | cluster N |
890 +----------------------------- 750 +-----------------------------------------------+
891 . . 751 . . . .
892 . . !! 752 . . . .
893 . Compressed Cluster . 753 . Compressed Cluster . . Normal Cluster .
894 +----------+---------+---------+---------+ 754 +----------+---------+---------+---------+ +---------+---------+---------+---------+
895 |compr flag| block 1 | block 2 | block 3 | 755 |compr flag| block 1 | block 2 | block 3 | | block 1 | block 2 | block 3 | block 4 |
896 +----------+---------+---------+---------+ 756 +----------+---------+---------+---------+ +---------+---------+---------+---------+
897 . . !! 757 . .
898 . 758 . .
899 . 759 . .
900 +-------------+-------------+--------- 760 +-------------+-------------+----------+----------------------------+
901 | data length | data chksum | reserved 761 | data length | data chksum | reserved | compressed data |
902 +-------------+-------------+--------- 762 +-------------+-------------+----------+----------------------------+
903 <<
904 Compression mode <<
905 -------------------------- <<
906 <<
907 f2fs supports "fs" and "user" compression mode <<
908 With this option, f2fs provides a choice to se <<
909 compression enabled files (refer to "Compressi <<
910 enable compression on a regular inode). <<
911 <<
912 1) compress_mode=fs <<
913 This is the default option. f2fs does automati <<
914 compression enabled files. <<
915 <<
916 2) compress_mode=user <<
917 This disables the automatic compression and gi <<
918 target file and the timing. The user can do ma <<
919 compression enabled files using F2FS_IOC_DECOM <<
920 ioctls like the below. <<
921 <<
922 To decompress a file, <<
923 <<
924 fd = open(filename, O_WRONLY, 0); <<
925 ret = ioctl(fd, F2FS_IOC_DECOMPRESS_FILE); <<
926 <<
927 To compress a file, <<
928 <<
929 fd = open(filename, O_WRONLY, 0); <<
930 ret = ioctl(fd, F2FS_IOC_COMPRESS_FILE); <<
931 <<
932 NVMe Zoned Namespace devices <<
933 ---------------------------- <<
934 <<
935 - ZNS defines a per-zone capacity which can be <<
936 zone-size. Zone-capacity is the number of us <<
937 F2FS checks if zone-capacity is less than zo <<
938 segment which starts after the zone-capacity <<
939 the free segment bitmap at initial mount tim <<
940 as permanently used so they are not allocate <<
941 consequently are not needed to be garbage co <<
942 zone-capacity is not aligned to default segm <<
943 can start before the zone-capacity and span <<
944 Such spanning segments are also considered a <<
945 past the zone-capacity are considered unusab <<
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.