1WRITE(2) Linux Programmer's Manual WRITE(2)
2
6 write - write to a file descriptor
7
9 #include <unistd.h>
10 ssize_t write(int fd, const void *buf, size_t count);
12
14 write() writes up to count bytes from the buffer starting at buf to the
15 file referred to by the file descriptor fd.
16 The number of bytes written may be less than count if, for example,
18 there is insufficient space on the underlying physical medium, or the
19 RLIMIT_FSIZE resource limit is encountered (see setrlimit(2)), or the
20 call was interrupted by a signal handler after having written less than
21 count bytes. (See also pipe(7).)
22 For a seekable file (i.e., one to which lseek(2) may be applied, for
24 example, a regular file) writing takes place at the file offset, and
25 the file offset is incremented by the number of bytes actually written.
26 If the file was open(2)ed with O_APPEND, the file offset is first set
27 to the end of the file before writing. The adjustment of the file off‐
28 set and the write operation are performed as an atomic step.
29 POSIX requires that a read(2) that can be proved to occur after a
31 write() has returned will return the new data. Note that not all
32 filesystems are POSIX conforming.
33 According to POSIX.1, if count is greater than SSIZE_MAX, the result is
35 implementation-defined; see NOTES for the upper limit on Linux.
36
38 On success, the number of bytes written is returned. On error, -1 is
39 returned, and errno is set to indicate the cause of the error.
40 Note that a successful write() may transfer fewer than count bytes.
42 Such partial writes can occur for various reasons; for example, because
43 there was insufficient space on the disk device to write all of the re‐
44 quested bytes, or because a blocked write() to a socket, pipe, or simi‐
45 lar was interrupted by a signal handler after it had transferred some,
46 but before it had transferred all of the requested bytes. In the event
47 of a partial write, the caller can make another write() call to trans‐
48 fer the remaining bytes. The subsequent call will either transfer fur‐
49 ther bytes or may result in an error (e.g., if the disk is now full).
50 If count is zero and fd refers to a regular file, then write() may re‐
52 turn a failure status if one of the errors below is detected. If no
53 errors are detected, or error detection is not performed, 0 will be re‐
54 turned without causing any other effect. If count is zero and fd
55 refers to a file other than a regular file, the results are not speci‐
56 fied.
57
59 EAGAIN The file descriptor fd refers to a file other than a socket and
60 has been marked nonblocking (O_NONBLOCK), and the write would
61 block. See open(2) for further details on the O_NONBLOCK flag.
62 EAGAIN or EWOULDBLOCK
64 The file descriptor fd refers to a socket and has been marked
65 nonblocking (O_NONBLOCK), and the write would block.
66 POSIX.1-2001 allows either error to be returned for this case,
67 and does not require these constants to have the same value, so
68 a portable application should check for both possibilities.
69 EBADF fd is not a valid file descriptor or is not open for writing.
71 EDESTADDRREQ
73 fd refers to a datagram socket for which a peer address has not
74 been set using connect(2).
75 EDQUOT The user's quota of disk blocks on the filesystem containing the
77 file referred to by fd has been exhausted.
78 EFAULT buf is outside your accessible address space.
80 EFBIG An attempt was made to write a file that exceeds the implementa‐
82 tion-defined maximum file size or the process's file size limit,
83 or to write at a position past the maximum allowed offset.
84 EINTR The call was interrupted by a signal before any data was writ‐
86 ten; see signal(7).
87 EINVAL fd is attached to an object which is unsuitable for writing; or
89 the file was opened with the O_DIRECT flag, and either the ad‐
90 dress specified in buf, the value specified in count, or the
91 file offset is not suitably aligned.
92 EIO A low-level I/O error occurred while modifying the inode. This
94 error may relate to the write-back of data written by an earlier
95 write(), which may have been issued to a different file descrip‐
96 tor on the same file. Since Linux 4.13, errors from write-back
97 come with a promise that they may be reported by subsequent.
98 write() requests, and will be reported by a subsequent fsync(2)
99 (whether or not they were also reported by write()). An alter‐
100 nate cause of EIO on networked filesystems is when an advisory
101 lock had been taken out on the file descriptor and this lock has
102 been lost. See the Lost locks section of fcntl(2) for further
103 details.
104 ENOSPC The device containing the file referred to by fd has no room for
106 the data.
107 EPERM The operation was prevented by a file seal; see fcntl(2).
109 EPIPE fd is connected to a pipe or socket whose reading end is closed.
111 When this happens the writing process will also receive a SIG‐
112 PIPE signal. (Thus, the write return value is seen only if the
113 program catches, blocks or ignores this signal.)
114 Other errors may occur, depending on the object connected to fd.
116
118 SVr4, 4.3BSD, POSIX.1-2001.
119 Under SVr4 a write may be interrupted and return EINTR at any point,
121 not just before any data is written.
122
124 The types size_t and ssize_t are, respectively, unsigned and signed in‐
125 teger data types specified by POSIX.1.
126 A successful return from write() does not make any guarantee that data
128 has been committed to disk. On some filesystems, including NFS, it
129 does not even guarantee that space has successfully been reserved for
130 the data. In this case, some errors might be delayed until a future
131 write(), fsync(2), or even close(2). The only way to be sure is to
132 call fsync(2) after you are done writing all your data.
133 If a write() is interrupted by a signal handler before any bytes are
135 written, then the call fails with the error EINTR; if it is interrupted
136 after at least one byte has been written, the call succeeds, and re‐
137 turns the number of bytes written.
138 On Linux, write() (and similar system calls) will transfer at most
140 0x7ffff000 (2,147,479,552) bytes, returning the number of bytes actu‐
141 ally transferred. (This is true on both 32-bit and 64-bit systems.)
142 An error return value while performing write() using direct I/O does
144 not mean the entire write has failed. Partial data may be written and
145 the data at the file offset on which the write() was attempted should
146 be considered inconsistent.
147
149 According to POSIX.1-2008/SUSv4 Section XSI 2.9.7 ("Thread Interactions
150 with Regular File Operations"):
151 All of the following functions shall be atomic with respect to each
153 other in the effects specified in POSIX.1-2008 when they operate on
154 regular files or symbolic links: ...
155 Among the APIs subsequently listed are write() and writev(2). And
157 among the effects that should be atomic across threads (and processes)
158 are updates of the file offset. However, on Linux before version 3.14,
159 this was not the case: if two processes that share an open file de‐
160 scription (see open(2)) perform a write() (or writev(2)) at the same
161 time, then the I/O operations were not atomic with respect updating the
162 file offset, with the result that the blocks of data output by the two
163 processes might (incorrectly) overlap. This problem was fixed in Linux
164 3.14.
165
167 close(2), fcntl(2), fsync(2), ioctl(2), lseek(2), open(2), pwrite(2),
168 read(2), select(2), writev(2), fwrite(3)
169
171 This page is part of release 5.10 of the Linux man-pages project. A
172 description of the project, information about reporting bugs, and the
173 latest version of this page, can be found at
174 https://www.kernel.org/doc/man-pages/.
175Linux 2019-10-10 WRITE(2)