1ATTR(5) File Formats Manual ATTR(5)
2
6 attr - Extended attributes
7
9 Extended attributes are name:value pairs associated permanently with
10 files and directories, similar to the environment strings associated
11 with a process. An attribute may be defined or undefined. If it is
12 defined, its value may be empty or non-empty.
13 Extended attributes are extensions to the normal attributes which are
15 associated with all inodes in the system (i.e. the stat(2) data). They
16 are often used to provide additional functionality to a filesystem -
17 for example, additional security features such as Access Control Lists
18 (ACLs) may be implemented using extended attributes.
19 Users with search access to a file or directory may retrieve a list of
21 attribute names defined for that file or directory.
22 Extended attributes are accessed as atomic objects. Reading retrieves
24 the whole value of an attribute and stores it in a buffer. Writing
25 replaces any previous value with the new value.
26 Space consumed for extended attributes is counted towards the disk quo‐
28 tas of the file owner and file group.
29 Currently, support for extended attributes is implemented on Linux by
31 the ext2, ext3 and XFS filesystem patches, which can be downloaded from
32 http://acl.bestbits.at/ and http://oss.sgi.com/projects/xfs/ respec‐
33 tively.
34
36 Attribute names are zero-terminated strings. The attribute name is
37 always specified in the fully qualified namespace.attribute form, eg.
38 user.mime_type, trusted.md5sum, system.posix_acl_access, or secu‐
39 rity.selinux.
40 The namespace mechanism is used to define different classes of extended
42 attributes. These different classes exist for several reasons, e.g.
43 the permissions and capabilities required for manipulating extended
44 attributes of one namespace may differ to another.
45 Currently the security, system, trusted, and user extended attribute
47 classes are defined as described below. Additional classes may be added
48 in the future.
49 Extended security attributes
51 The security attribute namespace is used by kernel security modules,
52 such as Security Enhanced Linux. Read and write access permissions to
53 security attributes depend on the policy implemented for each security
54 attribute by the security module. When no security module is loaded,
55 all processes have read access to extended security attributes, and
56 write access is limited to processes that have the CAP_SYS_ADMIN capa‐
57 bility.
58 Extended system attributes
60 Extended system attributes are used by the kernel to store system
61 objects such as Access Control Lists and Capabilities. Read and write
62 access permissions to system attributes depend on the policy imple‐
63 mented for each system attribute implemented by filesystems in the ker‐
64 nel.
65 Trusted extended attributes
67 Trusted extended attributes are visible and accessible only to pro‐
68 cesses that have the CAP_SYS_ADMIN capability (the super user usually
69 has this capability). Attributes in this class are used to implement
70 mechanisms in user space (i.e., outside the kernel) which keep informa‐
71 tion in extended attributes to which ordinary processes should not have
72 access.
73 Extended user attributes
75 Extended user attributes may be assigned to files and directories for
76 storing arbitrary additional information such as the mime type, charac‐
77 ter set or encoding of a file. The access permissions for user
78 attributes are defined by the file permission bits.
79 The file permission bits of regular files and directories are inter‐
81 preted differently from the file permission bits of special files and
82 symbolic links. For regular files and directories the file permission
83 bits define access to the file's contents, while for device special
84 files they define access to the device described by the special file.
85 The file permissions of symbolic links are not used in access checks.
86 These differences would allow users to consume filesystem resources in
87 a way not controllable by disk quotas for group or world writable spe‐
88 cial files and directories.
89 For this reason, extended user attributes are only allowed for regular
91 files and directories, and access to extended user attributes is
92 restricted to the owner and to users with appropriate capabilities for
93 directories with the sticky bit set (see the chmod(1) manual page for
94 an explanation of Sticky Directories).
95
97 The kernel and the filesystem may place limits on the maximum number
98 and size of extended attributes that can be associated with a file.
99 In the current ext2 and ext3 filesystem implementations, all extended
101 attributes must fit on a single filesystem block (1024, 2048 or 4096
102 bytes, depending on the block size specified when the filesystem was
103 created). This limit may be removed in a future version.
104 In the XFS filesystem implementation, there is no practical limit on
106 the number of extended attributes associated with a file, and the algo‐
107 rithms used to store extended attribute information on disk are scal‐
108 able (stored either inline in the inode, as an extent, or in a B+
109 tree).
110
112 Since the filesystems on which extended attributes are stored might
113 also be used on architectures with a different byte order and machine
114 word size, care should be taken to store attribute values in an archi‐
115 tecture independent format.
116
118 Andreas Gruenbacher, <a.gruenbacher@bestbits.at> and the SGI XFS devel‐
119 opment team, <linux-xfs@oss.sgi.com>.
120
122 getfattr(1), setfattr(1).
123 ATTR(5)