1 .\" This man page is Copyright (C) 1999 Andi Kleen <ak@muc.de>,
2 .\" Copyright (C) 2008-2014, Michael Kerrisk <mtk.manpages@gmail.com>,
3 .\" and Copyright (C) 2016, Heinrich Schuchardt <xypron.glpk@gmx.de>
5 .\" %%%LICENSE_START(VERBATIM_ONE_PARA)
6 .\" Permission is granted to distribute possibly modified copies
7 .\" of this page provided the header is included verbatim,
8 .\" and in case of nontrivial modification author and date
9 .\" of the modification is added to the header.
12 .\" Modified, 2003-12-02, Michael Kerrisk, <mtk.manpages@gmail.com>
13 .\" Modified, 2003-09-23, Adam Langley
14 .\" Modified, 2004-05-27, Michael Kerrisk, <mtk.manpages@gmail.com>
15 .\" Added SOCK_SEQPACKET
16 .\" 2008-05-27, mtk, Provide a clear description of the three types of
17 .\" address that can appear in the sockaddr_un structure: pathname,
18 .\" unnamed, and abstract.
20 .TH UNIX 7 2016-03-15 "Linux" "Linux Programmer's Manual"
22 unix \- sockets for local interprocess communication
24 .B #include <sys/socket.h>
26 .B #include <sys/un.h>
28 .IB unix_socket " = socket(AF_UNIX, type, 0);"
30 .IB error " = socketpair(AF_UNIX, type, 0, int *" sv ");"
36 socket family is used to communicate between processes on the same machine
38 Traditionally, UNIX domain sockets can be either unnamed,
39 or bound to a filesystem pathname (marked as being of type socket).
40 Linux also supports an abstract namespace which is independent of the
43 Valid socket types in the UNIX domain are:
45 for a stream-oriented socket;
47 for a datagram-oriented socket that preserves message boundaries
48 (as on most UNIX implementations, UNIX domain datagram
49 sockets are always reliable and don't reorder datagrams);
50 and (since Linux 2.6.4)
52 for a sequenced-packet socket that is connection-oriented,
53 preserves message boundaries,
54 and delivers messages in the order that they were sent.
56 UNIX domain sockets support passing file descriptors or process credentials
57 to other processes using ancillary data.
59 A UNIX domain socket address is represented in the following structure:
63 .\" #define UNIX_PATH_MAX 108
66 sa_family_t sun_family; /* AF_UNIX */
67 char sun_path[108]; /* pathname */
78 is 108 bytes in size; see also NOTES, below.
80 Various systems calls (for example,
88 Some other system calls (for example,
94 return an argument of this type.
96 Three types of address are distinguished in the
101 a UNIX domain socket can be bound to a null-terminated
102 filesystem pathname using
104 When the address of a pathname socket is returned
105 (by one of the system calls noted above),
108 offsetof(struct sockaddr_un, sun_path) + strlen(sun_path) + 1
112 contains the null-terminated pathname.
115 expression equates to the same value as
116 .IR sizeof(sa_family_t) ,
117 but some other implementations include other fields before
121 expression more portably describes the size of the address structure.)
123 For further details of pathname sockets, see below.
126 A stream socket that has not been bound to a pathname using
129 Likewise, the two sockets created by
132 When the address of an unnamed socket is returned,
134 .IR "sizeof(sa_family_t)" ,
137 should not be inspected.
138 .\" There is quite some variation across implementations: FreeBSD
139 .\" says the length is 16 bytes, HP-UX 11 says it's zero bytes.
142 an abstract socket address is distinguished (from a pathname socket)
145 is a null byte (\(aq\\0\(aq).
146 The socket's address in this namespace is given by the additional
149 that are covered by the specified length of the address structure.
150 (Null bytes in the name have no special significance.)
151 The name has no connection with filesystem pathnames.
152 When the address of an abstract socket is returned,
156 .IR "sizeof(sa_family_t)"
157 (i.e., greater than 2), and the name of the socket is contained in
159 .IR "(addrlen \- sizeof(sa_family_t))"
162 The abstract socket namespace is a nonportable Linux extension.
164 When binding a socket to a pathname, a few rules should be observed
165 for maximum portability and ease of coding:
169 should be null-terminated.
171 The length of the pathname, including the terminating null byte,
172 should not exceed the size of
177 argument that describes the enclosing
179 structure should have a value of at least:
182 offsetof(struct sockaddr_un, sun_path)+strlen(addr.sun_path)+1
188 .IR "sizeof(struct sockaddr_un)" .
190 There is some variation in how implementations handle UNIX domain
191 socket addresses that do not follow the above rules.
192 For example, some (but not all) implementations
193 .\" Linux does this, including for the case where the supplied path
195 append a null terminator if none is present in the supplied
198 When coding portable applications,
199 keep in mind that some implementations
203 as short as 92 bytes.
204 .\" Modern BSDs generally have 104, Tru64 and AIX have 104,
205 .\" Solaris and Irix have 108
212 return socket address structures.
213 When applied to UNIX domain sockets, the value-result
215 argument supplied to the call should be initialized as above.
216 Upon return, the argument is set to indicate the
218 size of the address structure.
219 The caller should check the value returned in this argument:
220 if the output value exceeds the input value,
221 then there is no guarantee that a null terminator is present in
225 For historical reasons, these socket options are specified with a
227 type even though they are
236 as the socket family.
239 Enables the receiving of the credentials of the sending process in an
241 When this option is set and the socket is not yet connected
242 a unique name in the abstract namespace will be generated automatically.
243 Expects an integer boolean flag.
250 .IR sizeof(sa_family_t) ,
251 .\" i.e., sizeof(short)
254 socket option was specified for a socket that was
255 not explicitly bound to an address,
256 then the socket is autobound to an abstract address.
257 The address consists of a null byte
258 followed by 5 bytes in the character set
260 Thus, there is a limit of 2^20 autobind addresses.
261 (From Linux 2.1.15, when the autobind feature was added,
262 8 bytes were used, and the limit was thus 2^32 autobind addresses.
263 The change to 5 bytes came in Linux 2.3.15.)
265 The following paragraphs describe domain-specific details and
266 unsupported features of the sockets API for UNIX domain sockets on Linux.
268 UNIX domain sockets do not support the transmission of
269 out-of-band data (the
279 flag is not supported by UNIX domain sockets.
287 is not supported by UNIX domain sockets.
291 socket option does have an effect for UNIX domain sockets, but the
294 For datagram sockets, the
296 value imposes an upper limit on the size of outgoing datagrams.
297 This limit is calculated as the doubled (see
299 option value less 32 bytes used for overhead.
300 .SS Ancillary messages
301 Ancillary data is sent and received using
305 For historical reasons the ancillary message types listed below
308 type even though they are
320 For more information see
324 Send or receive a set of open file descriptors from another process.
325 The data portion contains an integer array of the file descriptors.
326 The passed file descriptors behave as though they have been created with
330 Send or receive UNIX credentials.
331 This can be used for authentication.
332 The credentials are passed as a
335 Thus structure is defined in
342 pid_t pid; /* process ID of the sending process */
343 uid_t uid; /* user ID of the sending process */
344 gid_t gid; /* group ID of the sending process */
351 feature test macro must be defined (before including
353 header files) in order to obtain the definition
356 The credentials which the sender specifies are checked by the kernel.
357 A process with effective user ID 0 is allowed to specify values that do
359 The sender must specify its own process ID (unless it has the capability
361 its user ID, effective user ID, or saved set-user-ID (unless it has
363 and its group ID, effective group ID, or saved set-group-ID
370 option must be enabled on the socket.
374 calls return information in
376 The correct syntax is:
381 .IB error " = ioctl(" unix_socket ", " ioctl_type ", &" value ");"
391 socket the function returns the amount of queued unread data in the receive buffer.
392 The socket must not be in LISTEN state, otherwise an error
397 .IR <linux/sockios.h> .
398 .\" FIXME . http://sources.redhat.com/bugzilla/show_bug.cgi?id=12002,
399 .\" filed 2010-09-10, may cause SIOCINQ to be defined in glibc headers
401 you can use the synonymous
405 .\" SIOCOUTQ also has an effect for UNIX domain sockets, but not
406 .\" quite what userland might expect. It seems to return the number
407 .\" of bytes allocated for buffers containing pending output.
408 .\" That number is normally larger than the number of bytes of pending
409 .\" output. Since this info is, from userland's point of view, imprecise,
410 .\" and it may well change, probably best not to document this now.
414 the returned value is the same as
415 for Internet domain datagram socket;
421 The specified local address is already in use or the filesystem socket
422 object already exists.
425 The remote address specified by
427 was not a listening socket.
428 This error can also occur if the target pathname is not a socket.
431 Remote socket was unexpectedly closed.
434 User memory address was not valid.
437 Invalid argument passed.
438 A common cause is that the value
440 was not specified in the
442 field of passed addresses, or the socket was in an
443 invalid state for the applied operation.
447 called on an already connected socket or a target address was
448 specified on a connected socket.
451 The pathname in the remote address specified to
459 Socket operation needs a target address, but the socket is not connected.
462 Stream operation called on non-stream oriented socket or tried to
463 use the out-of-band data option.
466 The sender passed invalid credentials in the
470 Remote socket was closed on a stream socket.
474 This can be avoided by passing the
482 Passed protocol is not
486 Remote socket does not match the local socket type
494 Other errors can be generated by the generic socket layer or
495 by the filesystem while generating a filesystem socket object.
496 See the appropriate manual pages for more information.
499 and the abstract namespace were introduced with Linux 2.2 and should not
500 be used in portable programs.
501 (Some BSD-derived systems also support credential passing,
502 but the implementation details differ.)
504 Binding to a socket with a filename creates a socket
505 in the filesystem that must be deleted by the caller when it is no
508 The usual UNIX close-behind semantics apply; the socket can be unlinked
509 at any time and will be finally removed from the filesystem when the last
510 reference to it is closed.
512 To pass file descriptors or credentials over a
515 to send or receive at least one byte of nonancillary data in the same
521 UNIX domain stream sockets do not support the notion of out-of-band data.
523 .SS Socket ownership and permissions
524 In the Linux implementation, sockets which are visible in the
525 filesystem honor the permissions of the directory they are in.
526 Creation of a new socket will fail if the process does not have write and
527 search (execute) permission on the directory the socket is created in.
530 connecting to a stream socket object requires write permission on that socket;
531 sending a datagram to a datagram socket likewise
532 requires write permission on that socket.
533 POSIX does not make any statement about the effect of the permissions
534 on a socket file, and on many systems (e.g., several BSD derivatives),
535 the socket permissions are ignored.
536 Portable programs should not rely on
537 this feature for security.
539 When creating a new socket, the owner and group of the socket file
540 are set according to the usual rules.
541 The socket file has all permissions enabled,
542 other than those that are turned off by the process
545 For a socket that is visible in the filesystem,
546 the owner, group, and permissions can be changed (using
550 .\" However, fchown() and fchmod() do not seem to have an effect
554 When binding a socket to an address,
555 Linux is one of the implementations that appends a null terminator
556 if none is supplied in
558 In most cases this is unproblematic:
559 when the socket address is retrieved,
560 it will be one byte longer than that supplied when the socket was bound.
561 However, there is one case where confusing behavior can result:
562 if 108 non-null bytes are supplied when a socket is bound,
563 then the addition of the null terminator takes the length of
565 .IR sizeof(sun_path) .
566 Consequently, when retrieving the socket address
569 .\" The behavior on Solaris is quite similar.
572 argument for the retrieving call is specified as
573 .IR "sizeof(struct sockaddr_un)" ,
574 then the returned address structure
576 have a null terminator in
579 In addition, some implementations
580 .\" i.e., traditional BSD
581 don't require a null terminator when binding a socket (the
583 argument is used to determine the length of
585 and when the socket address is retrieved on these implementations,
586 there is no null terminator in
589 Applications that retrieve socket addresses can (portably) code
590 to handle the possibility that there is no null terminator in
592 by respecting the fact that the number of valid bytes in the pathname is:
594 strnlen(addr.sun_path, addrlen \- offsetof(sockaddr_un, sun_path))
595 .\" The following patch to amend kernel behavior was rejected:
596 .\" http://thread.gmane.org/gmane.linux.kernel.api/2437
597 .\" Subject: [patch] Fix handling of overlength pathname in AF_UNIX sun_path
599 .\" And there was a related discussion in the Austin list:
600 .\" http://thread.gmane.org/gmane.comp.standards.posix.austin.general/5735
601 .\" Subject: Having a sun_path with no null terminator
604 .\" FIXME . Track http://austingroupbugs.net/view.php?id=561
606 Alternatively, an application can retrieve
607 the socket address by allocating a buffer of size
608 .I "sizeof(struct sockaddr_un)+1"
609 that is zeroed out before the retrieval.
610 The retrieving call can specify
613 .IR "sizeof(struct sockaddr_un)" ,
614 and the extra zero byte ensures that there will be
615 a null terminator for the string returned in
622 addrlen = sizeof(struct sockaddr_un);
623 addrp = malloc(addrlen + 1);
626 memset(addrp, 0, addrlen + 1);
628 if (getsockname(sfd, (struct sockaddr *) addrp, &addrlen)) == \-1)
631 printf("sun_path = %s\\n", ((struct sockaddr_un *) addrp)\->sun_path);
635 This sort of messiness can be avoided if it is guaranteed
636 that the applications that
638 pathname sockets follow the rules outlined above under
639 .IR "Pathname sockets" .
641 The following code demonstrates the use of sequenced-packet
642 sockets for local interprocess communication.
643 It consists of two programs.
644 The server program waits for a connection from the client program.
645 The client sends each of its command-line arguments in separate messages.
646 The server treats the incoming messages as integers and adds them up.
647 The client sends the command string "END".
648 The server sends back a message containing the sum of the client's integers.
649 The client prints the sum and exits.
650 The server waits for the next client to connect.
651 To stop the server, the client is called with the command-line argument "DOWN".
653 The following output was recorded while running the server in the background
654 and repeatedly executing the client.
655 Execution of the server program ends when it receives the "DOWN" command.
663 $ \fB./client 11 \-5\fP
665 $ \fB./client DOWN\fP
677 #define SOCKET_NAME "/tmp/9Lq7BNBnBycd6nxy.socket"
678 #define BUFFER_SIZE 12
687 #include <sys/socket.h>
690 #include "connection.h"
693 main(int argc, char *argv[])
695 struct sockaddr_un name;
698 int connection_socket;
701 char buffer[BUFFER_SIZE];
704 * In case the program exited inadvertently on the last run,
710 /* Create local socket. */
712 connection_socket = socket(AF_UNIX, SOCK_SEQPACKET, 0);
713 if (connection_socket == \-1) {
719 * For portability clear the whole structure, since some
720 * implementations have additional (nonstandard) fields in
724 memset(&name, 0, sizeof(struct sockaddr_un));
726 /* Bind socket to socket name. */
728 name.sun_family = AF_UNIX;
729 strncpy(name.sun_path, SOCKET_NAME, sizeof(name.sun_path) \- 1);
731 ret = bind(connection_socket, (const struct sockaddr *) &name,
732 sizeof(struct sockaddr_un));
739 * Prepare for accepting connections. The backlog size is set
740 * to 20. So while one request is being processed other requests
744 ret = listen(connection_socket, 20);
750 /* This is the main loop for handling connections. */
754 /* Wait for incoming connection. */
756 data_socket = accept(connection_socket, NULL, NULL);
757 if (data_socket == \-1) {
765 /* Wait for next data packet. */
767 ret = read(data_socket, buffer, BUFFER_SIZE);
773 /* Ensure buffer is 0\-terminated. */
775 buffer[BUFFER_SIZE \- 1] = 0;
777 /* Handle commands. */
779 if (!strncmp(buffer, "DOWN", BUFFER_SIZE)) {
784 if (!strncmp(buffer, "END", BUFFER_SIZE)) {
788 /* Add received summand. */
790 result += atoi(buffer);
795 sprintf(buffer, "%d", result);
796 ret = write(data_socket, buffer, BUFFER_SIZE);
807 /* Quit on DOWN command. */
814 close(connection_socket);
816 /* Unlink the socket. */
831 #include <sys/socket.h>
834 #include "connection.h"
837 main(int argc, char *argv[])
839 struct sockaddr_un addr;
843 char buffer[BUFFER_SIZE];
845 /* Create local socket. */
847 data_socket = socket(AF_UNIX, SOCK_SEQPACKET, 0);
848 if (data_socket == \-1) {
854 * For portability clear the whole structure, since some
855 * implementations have additional (nonstandard) fields in
859 memset(&addr, 0, sizeof(struct sockaddr_un));
861 /* Connect socket to socket address */
863 addr.sun_family = AF_UNIX;
864 strncpy(addr.sun_path, SOCKET_NAME, sizeof(addr.sun_path) \- 1);
866 ret = connect (data_socket, (const struct sockaddr *) &addr,
867 sizeof(struct sockaddr_un));
869 fprintf(stderr, "The server is down.\\n");
873 /* Send arguments. */
875 for (i = 1; i < argc; ++i) {
876 ret = write(data_socket, argv[i], strlen(argv[i]) + 1);
883 /* Request result. */
885 strcpy (buffer, "END");
886 ret = write(data_socket, buffer, strlen(buffer) + 1);
892 /* Receive result. */
894 ret = read(data_socket, buffer, BUFFER_SIZE);
900 /* Ensure buffer is 0\-terminated. */
902 buffer[BUFFER_SIZE \- 1] = 0;
904 printf("Result = %s\\n", buffer);
914 For an example of the use of
924 .BR capabilities (7),