1PIPE(2) Linux Programmer's Manual PIPE(2)
2
3
4
6 pipe, pipe2 - create pipe
7
9 #include <unistd.h>
10
11 /* On Alpha, IA-64, MIPS, SuperH, and SPARC/SPARC64; see NOTES */
12 struct fd_pair {
13 long fd[2];
14 };
15 struct fd_pair pipe();
16
17 /* On all other architectures */
18 int pipe(int pipefd[2]);
19
20 #define _GNU_SOURCE /* See feature_test_macros(7) */
21 #include <fcntl.h> /* Obtain O_* constant definitions */
22 #include <unistd.h>
23
24 int pipe2(int pipefd[2], int flags);
25
27 pipe() creates a pipe, a unidirectional data channel that can be used
28 for interprocess communication. The array pipefd is used to return two
29 file descriptors referring to the ends of the pipe. pipefd[0] refers
30 to the read end of the pipe. pipefd[1] refers to the write end of the
31 pipe. Data written to the write end of the pipe is buffered by the
32 kernel until it is read from the read end of the pipe. For further
33 details, see pipe(7).
34
35 If flags is 0, then pipe2() is the same as pipe(). The following val‐
36 ues can be bitwise ORed in flags to obtain different behavior:
37
38 O_CLOEXEC
39 Set the close-on-exec (FD_CLOEXEC) flag on the two new file
40 descriptors. See the description of the same flag in open(2)
41 for reasons why this may be useful.
42
43 O_DIRECT (since Linux 3.4)
44 Create a pipe that performs I/O in "packet" mode. Each write(2)
45 to the pipe is dealt with as a separate packet, and read(2)s
46 from the pipe will read one packet at a time. Note the follow‐
47 ing points:
48
49 * Writes of greater than PIPE_BUF bytes (see pipe(7)) will be
50 split into multiple packets. The constant PIPE_BUF is
51 defined in <limits.h>.
52
53 * If a read(2) specifies a buffer size that is smaller than the
54 next packet, then the requested number of bytes are read, and
55 the excess bytes in the packet are discarded. Specifying a
56 buffer size of PIPE_BUF will be sufficient to read the
57 largest possible packets (see the previous point).
58
59 * Zero-length packets are not supported. (A read(2) that spec‐
60 ifies a buffer size of zero is a no-op, and returns 0.)
61
62 Older kernels that do not support this flag will indicate this
63 via an EINVAL error.
64
65 Since Linux 4.5, it is possible to change the O_DIRECT setting
66 of a pipe file descriptor using fcntl(2).
67
68 O_NONBLOCK
69 Set the O_NONBLOCK file status flag on the open file descrip‐
70 tions referred to by the new file descriptors. Using this flag
71 saves extra calls to fcntl(2) to achieve the same result.
72
74 On success, zero is returned. On error, -1 is returned, errno is set
75 appropriately, and pipefd is left unchanged.
76
77 On Linux (and other systems), pipe() does not modify pipefd on failure.
78 A requirement standardizing this behavior was added in POSIX.1-2016.
79 The Linux-specific pipe2() system call likewise does not modify pipefd
80 on failure.
81
83 EFAULT pipefd is not valid.
84
85 EINVAL (pipe2()) Invalid value in flags.
86
87 EMFILE The per-process limit on the number of open file descriptors has
88 been reached.
89
90 ENFILE The system-wide limit on the total number of open files has been
91 reached.
92
93 ENFILE The user hard limit on memory that can be allocated for pipes
94 has been reached and the caller is not privileged; see pipe(7).
95
97 pipe2() was added to Linux in version 2.6.27; glibc support is avail‐
98 able starting with version 2.9.
99
101 The SystemV ABI on some architectures allows the use of more than one
102 register for returning multiple values; several architectures (namely,
103 Alpha, IA-64, MIPS, SuperH, and SPARC/SPARC64) (ab)use this feature in
104 order to implement the pipe() system call in a functional manner: the
105 call doesn't take any arguments and returns a pair of file descriptors
106 as the return value on success. The glibc pipe() wrapper function
107 transparently deals with this. See syscall(2) for information regard‐
108 ing registers used for storing second file descriptor.
109
111 pipe(): POSIX.1-2001, POSIX.1-2008.
112
113 pipe2() is Linux-specific.
114
116 The following program creates a pipe, and then fork(2)s to create a
117 child process; the child inherits a duplicate set of file descriptors
118 that refer to the same pipe. After the fork(2), each process closes
119 the file descriptors that it doesn't need for the pipe (see pipe(7)).
120 The parent then writes the string contained in the program's command-
121 line argument to the pipe, and the child reads this string a byte at a
122 time from the pipe and echoes it on standard output.
123
124 Program source
125 #include <sys/types.h>
126 #include <sys/wait.h>
127 #include <stdio.h>
128 #include <stdlib.h>
129 #include <unistd.h>
130 #include <string.h>
131
132 int
133 main(int argc, char *argv[])
134 {
135 int pipefd[2];
136 pid_t cpid;
137 char buf;
138
139 if (argc != 2) {
140 fprintf(stderr, "Usage: %s <string>\n", argv[0]);
141 exit(EXIT_FAILURE);
142 }
143
144 if (pipe(pipefd) == -1) {
145 perror("pipe");
146 exit(EXIT_FAILURE);
147 }
148
149 cpid = fork();
150 if (cpid == -1) {
151 perror("fork");
152 exit(EXIT_FAILURE);
153 }
154
155 if (cpid == 0) { /* Child reads from pipe */
156 close(pipefd[1]); /* Close unused write end */
157
158 while (read(pipefd[0], &buf, 1) > 0)
159 write(STDOUT_FILENO, &buf, 1);
160
161 write(STDOUT_FILENO, "\n", 1);
162 close(pipefd[0]);
163 _exit(EXIT_SUCCESS);
164
165 } else { /* Parent writes argv[1] to pipe */
166 close(pipefd[0]); /* Close unused read end */
167 write(pipefd[1], argv[1], strlen(argv[1]));
168 close(pipefd[1]); /* Reader will see EOF */
169 wait(NULL); /* Wait for child */
170 exit(EXIT_SUCCESS);
171 }
172 }
173
175 fork(2), read(2), socketpair(2), splice(2), tee(2), vmsplice(2),
176 write(2), popen(3), pipe(7)
177
179 This page is part of release 5.04 of the Linux man-pages project. A
180 description of the project, information about reporting bugs, and the
181 latest version of this page, can be found at
182 https://www.kernel.org/doc/man-pages/.
183
184
185
186Linux 2019-03-06 PIPE(2)