1read(2) System Calls Manual read(2)
2
6 read - read from a file descriptor
7
9 Standard C library (libc, -lc)
10
12 #include <unistd.h>
13 ssize_t read(int fd, void buf[.count], size_t count);
15
17 read() attempts to read up to count bytes from file descriptor fd into
18 the buffer starting at buf.
19 On files that support seeking, the read operation commences at the file
21 offset, and the file offset is incremented by the number of bytes read.
22 If the file offset is at or past the end of file, no bytes are read,
23 and read() returns zero.
24 If count is zero, read() may detect the errors described below. In the
26 absence of any errors, or if read() does not check for errors, a read()
27 with a count of 0 returns zero and has no other effects.
28 According to POSIX.1, if count is greater than SSIZE_MAX, the result is
30 implementation-defined; see NOTES for the upper limit on Linux.
31
33 On success, the number of bytes read is returned (zero indicates end of
34 file), and the file position is advanced by this number. It is not an
35 error if this number is smaller than the number of bytes requested;
36 this may happen for example because fewer bytes are actually available
37 right now (maybe because we were close to end-of-file, or because we
38 are reading from a pipe, or from a terminal), or because read() was in‐
39 terrupted by a signal. See also NOTES.
40 On error, -1 is returned, and errno is set to indicate the error. In
42 this case, it is left unspecified whether the file position (if any)
43 changes.
44
46 EAGAIN The file descriptor fd refers to a file other than a socket and
47 has been marked nonblocking (O_NONBLOCK), and the read would
48 block. See open(2) for further details on the O_NONBLOCK flag.
49 EAGAIN or EWOULDBLOCK
51 The file descriptor fd refers to a socket and has been marked
52 nonblocking (O_NONBLOCK), and the read would block.
53 POSIX.1-2001 allows either error to be returned for this case,
54 and does not require these constants to have the same value, so
55 a portable application should check for both possibilities.
56 EBADF fd is not a valid file descriptor or is not open for reading.
58 EFAULT buf is outside your accessible address space.
60 EINTR The call was interrupted by a signal before any data was read;
62 see signal(7).
63 EINVAL fd is attached to an object which is unsuitable for reading; or
65 the file was opened with the O_DIRECT flag, and either the ad‐
66 dress specified in buf, the value specified in count, or the
67 file offset is not suitably aligned.
68 EINVAL fd was created via a call to timerfd_create(2) and the wrong
70 size buffer was given to read(); see timerfd_create(2) for fur‐
71 ther information.
72 EIO I/O error. This will happen for example when the process is in
74 a background process group, tries to read from its controlling
75 terminal, and either it is ignoring or blocking SIGTTIN or its
76 process group is orphaned. It may also occur when there is a
77 low-level I/O error while reading from a disk or tape. A fur‐
78 ther possible cause of EIO on networked filesystems is when an
79 advisory lock had been taken out on the file descriptor and this
80 lock has been lost. See the Lost locks section of fcntl(2) for
81 further details.
82 EISDIR fd refers to a directory.
84 Other errors may occur, depending on the object connected to fd.
86
88 POSIX.1-2008.
89
91 SVr4, 4.3BSD, POSIX.1-2001.
92
94 On Linux, read() (and similar system calls) will transfer at most
95 0x7ffff000 (2,147,479,552) bytes, returning the number of bytes actu‐
96 ally transferred. (This is true on both 32-bit and 64-bit systems.)
97 On NFS filesystems, reading small amounts of data will update the time‐
99 stamp only the first time, subsequent calls may not do so. This is
100 caused by client side attribute caching, because most if not all NFS
101 clients leave st_atime (last file access time) updates to the server,
102 and client side reads satisfied from the client's cache will not cause
103 st_atime updates on the server as there are no server-side reads. UNIX
104 semantics can be obtained by disabling client-side attribute caching,
105 but in most situations this will substantially increase server load and
106 decrease performance.
107
109 According to POSIX.1-2008/SUSv4 Section XSI 2.9.7 ("Thread Interactions
110 with Regular File Operations"):
111 All of the following functions shall be atomic with respect to each
113 other in the effects specified in POSIX.1-2008 when they operate on
114 regular files or symbolic links: ...
115 Among the APIs subsequently listed are read() and readv(2). And among
117 the effects that should be atomic across threads (and processes) are
118 updates of the file offset. However, before Linux 3.14, this was not
119 the case: if two processes that share an open file description (see
120 open(2)) perform a read() (or readv(2)) at the same time, then the I/O
121 operations were not atomic with respect updating the file offset, with
122 the result that the reads in the two processes might (incorrectly)
123 overlap in the blocks of data that they obtained. This problem was
124 fixed in Linux 3.14.
125
127 close(2), fcntl(2), ioctl(2), lseek(2), open(2), pread(2), readdir(2),
128 readlink(2), readv(2), select(2), write(2), fread(3)
129Linux man-pages 6.04 2023-04-03 read(2)