1LOCKF(P) POSIX Programmer's Manual LOCKF(P)
2
6 lockf - record locking on files
7
9 #include <unistd.h>
10 int lockf(int fildes, int function, off_t size);
12
15 The lockf() function shall lock sections of a file with advisory-mode
16 locks. Calls to lockf() from other threads which attempt to lock the
17 locked file section shall either return an error value or block until
18 the section becomes unlocked. All the locks for a process are removed
19 when the process terminates. Record locking with lockf() shall be sup‐
20 ported for regular files and may be supported for other files.
21 The fildes argument is an open file descriptor. To establish a lock
23 with this function, the file descriptor shall be opened with write-only
24 permission (O_WRONLY) or with read/write permission (O_RDWR).
25 The function argument is a control value which specifies the action to
27 be taken. The permissible values for function are defined in <unistd.h>
28 as follows:
29 Function Description
31 F_ULOCK Unlock locked sections.
32 F_LOCK Lock a section for exclusive use.
33 F_TLOCK Test and lock a section for exclusive use.
34 F_TEST Test a section for locks by other processes.
35 F_TEST shall detect if a lock by another process is present on the
37 specified section.
38 F_LOCK and F_TLOCK shall both lock a section of a file if the section
40 is available.
41 F_ULOCK shall remove locks from a section of the file.
43 The size argument is the number of contiguous bytes to be locked or
45 unlocked. The section to be locked or unlocked starts at the current
46 offset in the file and extends forward for a positive size or backward
47 for a negative size (the preceding bytes up to but not including the
48 current offset). If size is 0, the section from the current offset
49 through the largest possible file offset shall be locked (that is, from
50 the current offset through the present or any future end-of-file). An
51 area need not be allocated to the file to be locked because locks may
52 exist past the end-of-file.
53 The sections locked with F_LOCK or F_TLOCK may, in whole or in part,
55 contain or be contained by a previously locked section for the same
56 process. When this occurs, or if adjacent locked sections would occur,
57 the sections shall be combined into a single locked section. If the
58 request would cause the number of locks to exceed a system-imposed
59 limit, the request shall fail.
60 F_LOCK and F_TLOCK requests differ only by the action taken if the sec‐
62 tion is not available. F_LOCK shall block the calling thread until the
63 section is available. F_TLOCK shall cause the function to fail if the
64 section is already locked by another process.
65 File locks shall be released on first close by the locking process of
67 any file descriptor for the file.
68 F_ULOCK requests may release (wholly or in part) one or more locked
70 sections controlled by the process. Locked sections shall be unlocked
71 starting at the current file offset through size bytes or to the end-
72 of-file if size is (off_t)0. When all of a locked section is not
73 released (that is, when the beginning or end of the area to be unlocked
74 falls within a locked section), the remaining portions of that section
75 shall remain locked by the process. Releasing the center portion of a
76 locked section shall cause the remaining locked beginning and end por‐
77 tions to become two separate locked sections. If the request would
78 cause the number of locks in the system to exceed a system-imposed
79 limit, the request shall fail.
80 A potential for deadlock occurs if the threads of a process controlling
82 a locked section are blocked by accessing another process' locked sec‐
83 tion. If the system detects that deadlock would occur, lockf() shall
84 fail with an [EDEADLK] error.
85 The interaction between fcntl() and lockf() locks is unspecified.
87 Blocking on a section shall be interrupted by any signal.
89 An F_ULOCK request in which size is non-zero and the offset of the last
91 byte of the requested section is the maximum value for an object of
92 type off_t, when the process has an existing lock in which size is 0
93 and which includes the last byte of the requested section, shall be
94 treated as a request to unlock from the start of the requested section
95 with a size equal to 0. Otherwise, an F_ULOCK request shall attempt to
96 unlock only the requested section.
97 Attempting to lock a section of a file that is associated with a
99 buffered stream produces unspecified results.
100
102 Upon successful completion, lockf() shall return 0. Otherwise, it shall
103 return -1, set errno to indicate an error, and existing locks shall not
104 be changed.
105
107 The lockf() function shall fail if:
108 EBADF The fildes argument is not a valid open file descriptor; or
110 function is F_LOCK or F_TLOCK and fildes is not a valid file
111 descriptor open for writing.
112 EACCES or EAGAIN
114 The function argument is F_TLOCK or F_TEST and the section is
116 already locked by another process.
117 EDEADLK
119 The function argument is F_LOCK and a deadlock is detected.
120 EINTR A signal was caught during execution of the function.
122 EINVAL The function argument is not one of F_LOCK, F_TLOCK, F_TEST, or
124 F_ULOCK; or size plus the current file offset is less than 0.
125 EOVERFLOW
127 The offset of the first, or if size is not 0 then the last, byte
128 in the requested section cannot be represented correctly in an
129 object of type off_t.
130 The lockf() function may fail if:
133 EAGAIN The function argument is F_LOCK or F_TLOCK and the file is
135 mapped with mmap().
136 EDEADLK or ENOLCK
138 The function argument is F_LOCK, F_TLOCK, or F_ULOCK, and the
140 request would cause the number of locks to exceed a system-
141 imposed limit.
142 EOPNOTSUPP or EINVAL
144 The implementation does not support the locking of files of the
146 type indicated by the fildes argument.
147 The following sections are informative.
150
152 Locking a Portion of a File
153 In the following example, a file named /home/cnd/mod1 is being modi‐
154 fied. Other processes that use locking are prevented from changing it
155 during this process. Only the first 10000 bytes are locked, and the
156 lock call fails if another process has any part of this area locked
157 already.
158 #include <fcntl.h>
161 #include <unistd.h>
162 int fildes;
165 int status;
166 ...
167 fildes = open("/home/cnd/mod1", O_RDWR);
168 status = lockf(fildes, F_TLOCK, (off_t)10000);
169
171 Record-locking should not be used in combination with the fopen(),
172 fread(), fwrite(), and other stdio functions. Instead, the more primi‐
173 tive, non-buffered functions (such as open()) should be used. Unex‐
174 pected results may occur in processes that do buffering in the user
175 address space. The process may later read/write data which is/was
176 locked. The stdio functions are the most common source of unexpected
177 buffering.
178 The alarm() function may be used to provide a timeout facility in
180 applications requiring it.
181
183 None.
184
186 None.
187
189 alarm() , chmod() , close() , creat() , fcntl() , fopen() , mmap() ,
190 open() , read() , write() , the Base Definitions volume of
191 IEEE Std 1003.1-2001, <unistd.h>
192
194 Portions of this text are reprinted and reproduced in electronic form
195 from IEEE Std 1003.1, 2003 Edition, Standard for Information Technology
196 -- Portable Operating System Interface (POSIX), The Open Group Base
197 Specifications Issue 6, Copyright (C) 2001-2003 by the Institute of
198 Electrical and Electronics Engineers, Inc and The Open Group. In the
199 event of any discrepancy between this version and the original IEEE and
200 The Open Group Standard, the original IEEE and The Open Group Standard
201 is the referee document. The original Standard can be obtained online
202 at http://www.opengroup.org/unix/online.html .
203IEEE/The Open Group 2003 LOCKF(P)