ext4.txt 8.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236
  1. Ext4 Filesystem
  2. ===============
  3. This is a development version of the ext4 filesystem, an advanced level
  4. of the ext3 filesystem which incorporates scalability and reliability
  5. enhancements for supporting large filesystems (64 bit) in keeping with
  6. increasing disk capacities and state-of-the-art feature requirements.
  7. Mailing list: linux-ext4@vger.kernel.org
  8. 1. Quick usage instructions:
  9. ===========================
  10. - Grab updated e2fsprogs from
  11. ftp://ftp.kernel.org/pub/linux/kernel/people/tytso/e2fsprogs-interim/
  12. This is a patchset on top of e2fsprogs-1.39, which can be found at
  13. ftp://ftp.kernel.org/pub/linux/kernel/people/tytso/e2fsprogs/
  14. - It's still mke2fs -j /dev/hda1
  15. - mount /dev/hda1 /wherever -t ext4dev
  16. - To enable extents,
  17. mount /dev/hda1 /wherever -t ext4dev -o extents
  18. - The filesystem is compatible with the ext3 driver until you add a file
  19. which has extents (ie: `mount -o extents', then create a file).
  20. NOTE: The "extents" mount flag is temporary. It will soon go away and
  21. extents will be enabled by the "-o extents" flag to mke2fs or tune2fs
  22. - When comparing performance with other filesystems, remember that
  23. ext3/4 by default offers higher data integrity guarantees than most. So
  24. when comparing with a metadata-only journalling filesystem, use `mount -o
  25. data=writeback'. And you might as well use `mount -o nobh' too along
  26. with it. Making the journal larger than the mke2fs default often helps
  27. performance with metadata-intensive workloads.
  28. 2. Features
  29. ===========
  30. 2.1 Currently available
  31. * ability to use filesystems > 16TB
  32. * extent format reduces metadata overhead (RAM, IO for access, transactions)
  33. * extent format more robust in face of on-disk corruption due to magics,
  34. * internal redunancy in tree
  35. 2.1 Previously available, soon to be enabled by default by "mkefs.ext4":
  36. * dir_index and resize inode will be on by default
  37. * large inodes will be used by default for fast EAs, nsec timestamps, etc
  38. 2.2 Candidate features for future inclusion
  39. There are several under discussion, whether they all make it in is
  40. partly a function of how much time everyone has to work on them:
  41. * improved file allocation (multi-block alloc, delayed alloc; basically done)
  42. * fix 32000 subdirectory limit (patch exists, needs some e2fsck work)
  43. * nsec timestamps for mtime, atime, ctime, create time (patch exists,
  44. needs some e2fsck work)
  45. * inode version field on disk (NFSv4, Lustre; prototype exists)
  46. * reduced mke2fs/e2fsck time via uninitialized groups (prototype exists)
  47. * journal checksumming for robustness, performance (prototype exists)
  48. * persistent file preallocation (e.g for streaming media, databases)
  49. Features like metadata checksumming have been discussed and planned for
  50. a bit but no patches exist yet so I'm not sure they're in the near-term
  51. roadmap.
  52. The big performance win will come with mballoc and delalloc. CFS has
  53. been using mballoc for a few years already with Lustre, and IBM + Bull
  54. did a lot of benchmarking on it. The reason it isn't in the first set of
  55. patches is partly a manageability issue, and partly because it doesn't
  56. directly affect the on-disk format (outside of much better allocation)
  57. so it isn't critical to get into the first round of changes. I believe
  58. Alex is working on a new set of patches right now.
  59. 3. Options
  60. ==========
  61. When mounting an ext4 filesystem, the following option are accepted:
  62. (*) == default
  63. extents ext4 will use extents to address file data. The
  64. file system will no longer be mountable by ext3.
  65. journal=update Update the ext4 file system's journal to the current
  66. format.
  67. journal=inum When a journal already exists, this option is ignored.
  68. Otherwise, it specifies the number of the inode which
  69. will represent the ext4 file system's journal file.
  70. journal_dev=devnum When the external journal device's major/minor numbers
  71. have changed, this option allows the user to specify
  72. the new journal location. The journal device is
  73. identified through its new major/minor numbers encoded
  74. in devnum.
  75. noload Don't load the journal on mounting.
  76. data=journal All data are committed into the journal prior to being
  77. written into the main file system.
  78. data=ordered (*) All data are forced directly out to the main file
  79. system prior to its metadata being committed to the
  80. journal.
  81. data=writeback Data ordering is not preserved, data may be written
  82. into the main file system after its metadata has been
  83. committed to the journal.
  84. commit=nrsec (*) Ext4 can be told to sync all its data and metadata
  85. every 'nrsec' seconds. The default value is 5 seconds.
  86. This means that if you lose your power, you will lose
  87. as much as the latest 5 seconds of work (your
  88. filesystem will not be damaged though, thanks to the
  89. journaling). This default value (or any low value)
  90. will hurt performance, but it's good for data-safety.
  91. Setting it to 0 will have the same effect as leaving
  92. it at the default (5 seconds).
  93. Setting it to very large values will improve
  94. performance.
  95. barrier=1 This enables/disables barriers. barrier=0 disables
  96. it, barrier=1 enables it.
  97. orlov (*) This enables the new Orlov block allocator. It is
  98. enabled by default.
  99. oldalloc This disables the Orlov block allocator and enables
  100. the old block allocator. Orlov should have better
  101. performance - we'd like to get some feedback if it's
  102. the contrary for you.
  103. user_xattr Enables Extended User Attributes. Additionally, you
  104. need to have extended attribute support enabled in the
  105. kernel configuration (CONFIG_EXT4_FS_XATTR). See the
  106. attr(5) manual page and http://acl.bestbits.at/ to
  107. learn more about extended attributes.
  108. nouser_xattr Disables Extended User Attributes.
  109. acl Enables POSIX Access Control Lists support.
  110. Additionally, you need to have ACL support enabled in
  111. the kernel configuration (CONFIG_EXT4_FS_POSIX_ACL).
  112. See the acl(5) manual page and http://acl.bestbits.at/
  113. for more information.
  114. noacl This option disables POSIX Access Control List
  115. support.
  116. reservation
  117. noreservation
  118. bsddf (*) Make 'df' act like BSD.
  119. minixdf Make 'df' act like Minix.
  120. check=none Don't do extra checking of bitmaps on mount.
  121. nocheck
  122. debug Extra debugging information is sent to syslog.
  123. errors=remount-ro(*) Remount the filesystem read-only on an error.
  124. errors=continue Keep going on a filesystem error.
  125. errors=panic Panic and halt the machine if an error occurs.
  126. grpid Give objects the same group ID as their creator.
  127. bsdgroups
  128. nogrpid (*) New objects have the group ID of their creator.
  129. sysvgroups
  130. resgid=n The group ID which may use the reserved blocks.
  131. resuid=n The user ID which may use the reserved blocks.
  132. sb=n Use alternate superblock at this location.
  133. quota
  134. noquota
  135. grpquota
  136. usrquota
  137. bh (*) ext4 associates buffer heads to data pages to
  138. nobh (a) cache disk block mapping information
  139. (b) link pages into transaction to provide
  140. ordering guarantees.
  141. "bh" option forces use of buffer heads.
  142. "nobh" option tries to avoid associating buffer
  143. heads (supported only for "writeback" mode).
  144. Data Mode
  145. ---------
  146. There are 3 different data modes:
  147. * writeback mode
  148. In data=writeback mode, ext4 does not journal data at all. This mode provides
  149. a similar level of journaling as that of XFS, JFS, and ReiserFS in its default
  150. mode - metadata journaling. A crash+recovery can cause incorrect data to
  151. appear in files which were written shortly before the crash. This mode will
  152. typically provide the best ext4 performance.
  153. * ordered mode
  154. In data=ordered mode, ext4 only officially journals metadata, but it logically
  155. groups metadata and data blocks into a single unit called a transaction. When
  156. it's time to write the new metadata out to disk, the associated data blocks
  157. are written first. In general, this mode performs slightly slower than
  158. writeback but significantly faster than journal mode.
  159. * journal mode
  160. data=journal mode provides full data and metadata journaling. All new data is
  161. written to the journal first, and then to its final location.
  162. In the event of a crash, the journal can be replayed, bringing both data and
  163. metadata into a consistent state. This mode is the slowest except when data
  164. needs to be read from and written to disk at the same time where it
  165. outperforms all others modes.
  166. References
  167. ==========
  168. kernel source: <file:fs/ext4/>
  169. <file:fs/jbd2/>
  170. programs: http://e2fsprogs.sourceforge.net/
  171. http://ext2resize.sourceforge.net
  172. useful links: http://fedoraproject.org/wiki/ext3-devel
  173. http://www.bullopensource.org/ext4/