1PATH_RESOLUTION(2) Linux Programmer's Manual PATH_RESOLUTION(2)
2
6 Unix/Linux path resolution - find the file referred to by a filename
7
9 Some Unix/Linux system calls have as parameter one or more filenames.
10 A filename (or pathname) is resolved as follows.
11 Step 1: Start of the resolution process
14 If the pathname starts with the '/' character, the starting lookup
15 directory is the root directory of the current process. (A process
16 inherits its root directory from its parent. Usually this will be the
17 root directory of the file hierarchy. A process may get a different
18 root directory by use of the chroot(2) system call. A process may get
19 an entirely private namespace in case it — or one of its ancestors —
20 was started by an invocation of the clone(2) system call that had the
21 CLONE_NEWNS flag set.) This handles the '/' part of the pathname.
22 If the pathname does not start with the '/' character, the starting
24 lookup directory of the resolution process is the current working
25 directory of the process. (This is also inherited from the parent. It
26 can be changed by use of the chdir(2) system call.)
27 Pathnames starting with a '/' character are called absolute pathnames.
29 Pathnames not starting with a '/' are called relative pathnames.
30 Step 2: Walk along the path
33 Set the current lookup directory to the starting lookup directory.
34 Now, for each non-final component of the pathname, where a component is
35 a substring delimited by '/' characters, this component is looked up in
36 the current lookup directory.
37 If the process does not have search permission on the current lookup
39 directory, an EACCES error is returned ("Permission denied").
40 If the component is not found, an ENOENT error is returned ("No such
42 file or directory").
43 If the component is found, but is neither a directory nor a symbolic
45 link, an ENOTDIR error is returned ("Not a directory").
46 If the component is found and is a directory, we set the current lookup
48 directory to that directory, and go to the next component.
49 If the component is found and is a symbolic link (symlink), we first
51 resolve this symbolic link (with the current lookup directory as start‐
52 ing lookup directory). Upon error, that error is returned. If the
53 result is not a directory, an ENOTDIR error is returned. If the reso‐
54 lution of the symlink is successful and returns a directory, we set the
55 current lookup directory to that directory, and go to the next compo‐
56 nent. Note that the resolution process here involves recursion. In
57 order to protect the kernel against stack overflow, and also to protect
58 against denial of service, there are limits on the maximum recursion
59 depth, and on the maximum number of symlinks followed. An ELOOP error
60 is returned when the maximum is exceeded ("Too many levels of symbolic
61 links").
62 Step 3: Find the final entry
65 The lookup of the final component of the pathname goes just like that
66 of all other components, as described in the previous step, with two
67 differences: (i) the final component need not be a directory (at least
68 as far as the path resolution process is concerned — it may have to be
69 a directory, or a non-directory, because of the requirements of the
70 specific system call), and (ii) it is not necessarily an error if the
71 component is not found — maybe we are just creating it. The details on
72 the treatment of the final entry are described in the manual pages of
73 the specific system calls.
74 . and ..
77 By convention, every directory has the entries "." and "..", which
78 refer to the directory itself and to its parent directory, respec‐
79 tively.
80 The path resolution process will assume that these entries have their
82 conventional meanings, regardless of whether they are actually present
83 in the physical filesystem.
84 One cannot walk down past the root: "/.." is the same as "/".
86 Mount points
89 After a "mount dev path" command, the pathname "path" refers to the
90 root of the filesystem hierarchy on the device "dev", and no longer to
91 whatever it referred to earlier.
92 One can walk out of a mounted filesystem: "path/.." refers to the par‐
94 ent directory of "path", outside of the filesystem hierarchy on "dev".
95 Trailing slashes
98 If a pathname ends in a '/', that forces resolution of the preceding
99 component as in Step 2: it has to exist and resolve to a directory.
100 Otherwise a trailing '/' is ignored. (Or, equivalently, a pathname
101 with a trailing '/' is equivalent to the pathname obtained by appending
102 '.' to it.)
103 Final symlink
106 If the last component of a pathname is a symbolic link, then it depends
107 on the system call whether the file referred to will be the symbolic
108 link or the result of path resolution on its contents. For example,
109 the system call lstat(2) will operate on the symlink, while stat(2)
110 operates on the file pointed to by the symlink.
111 Length limit
114 There is a maximum length for pathnames. If the pathname (or some
115 intermediate pathname obtained while resolving symbolic links) is too
116 long, an ENAMETOOLONG error is returned ("File name too long").
117 Empty pathname
120 In the original Unix, the empty pathname referred to the current direc‐
121 tory. Nowadays POSIX decrees that an empty pathname must not be
122 resolved successfully. Linux returns ENOENT in this case.
123 Permissions
126 The permission bits of a file consist of three groups of three bits,
127 cf. chmod(1) and stat(2). The first group of three is used when the
128 effective user ID of the current process equals the owner ID of the
129 file. The second group of three is used when the group ID of the file
130 either equals the effective group ID of the current process, or is one
131 of the supplementary group IDs of the current process (as set by set‐
132 groups(2)). When neither holds, the third group is used.
133 Of the three bits used, the first bit determines read permission, the
135 second write permission, and the last execute permission in case of
136 ordinary files, or search permission in case of directories.
137 Linux uses the fsuid instead of the effective user ID in permission
139 checks. Ordinarily the fsuid will equal the effective user ID, but the
140 fsuid can be changed by the system call setfsuid(2).
141 (Here "fsuid" stands for something like "file system user ID". The
143 concept was required for the implementation of a user space NFS server
144 at a time when processes could send a signal to a process with the same
145 effective user ID. It is obsolete now. Nobody should use setfsuid(2).)
146 Similarly, Linux uses the fsgid ("file system group ID") instead of the
148 effective group ID. See setfsgid(2).
149 Bypassing permission checks: superuser and capabilities
152 On a traditional Unix system, the superuser (root, user ID 0) is all-
153 powerful, and bypasses all permissions restrictions when accessing
154 files.
155 On Linux, superuser privileges are divided into capabilities (see capa‐
157 bilities(7)). Two capabilities are relevant for file permissions
158 checks: CAP_DAC_OVERRIDE and CAP_DAC_READ_SEARCH. (A process has these
159 capabilities if its fsuid is 0.)
160 The CAP_DAC_OVERRIDE capability overrides all permission checking, but
162 only grants execute permission when at least one of the file's three
163 execute permission bits is set.
164 The CAP_DAC_READ_SEARCH capability grants read and search permission on
166 directories, and read permission on ordinary files.
167
170 capabilities(7)
171Linux 2.6.7 2004-06-21 PATH_RESOLUTION(2)