1recv(2) System Calls Manual recv(2)
2
6 recv, recvfrom, recvmsg - receive a message from a socket
7
9 Standard C library (libc, -lc)
10
12 #include <sys/socket.h>
13 ssize_t recv(int sockfd, void buf[.len], size_t len,
15 int flags);
16 ssize_t recvfrom(int sockfd, void buf[restrict .len], size_t len,
17 int flags,
18 struct sockaddr *_Nullable restrict src_addr,
19 socklen_t *_Nullable restrict addrlen);
20 ssize_t recvmsg(int sockfd, struct msghdr *msg, int flags);
21
23 The recv(), recvfrom(), and recvmsg() calls are used to receive mes‐
24 sages from a socket. They may be used to receive data on both connec‐
25 tionless and connection-oriented sockets. This page first describes
26 common features of all three system calls, and then describes the dif‐
27 ferences between the calls.
28 The only difference between recv() and read(2) is the presence of
30 flags. With a zero flags argument, recv() is generally equivalent to
31 read(2) (but see NOTES). Also, the following call
32 recv(sockfd, buf, len, flags);
34 is equivalent to
36 recvfrom(sockfd, buf, len, flags, NULL, NULL);
38 All three calls return the length of the message on successful comple‐
40 tion. If a message is too long to fit in the supplied buffer, excess
41 bytes may be discarded depending on the type of socket the message is
42 received from.
43 If no messages are available at the socket, the receive calls wait for
45 a message to arrive, unless the socket is nonblocking (see fcntl(2)),
46 in which case the value -1 is returned and errno is set to EAGAIN or
47 EWOULDBLOCK. The receive calls normally return any data available, up
48 to the requested amount, rather than waiting for receipt of the full
49 amount requested.
50 An application can use select(2), poll(2), or epoll(7) to determine
52 when more data arrives on a socket.
53 The flags argument
55 The flags argument is formed by ORing one or more of the following val‐
56 ues:
57 MSG_CMSG_CLOEXEC (recvmsg() only; since Linux 2.6.23)
59 Set the close-on-exec flag for the file descriptor received via
60 a UNIX domain file descriptor using the SCM_RIGHTS operation
61 (described in unix(7)). This flag is useful for the same rea‐
62 sons as the O_CLOEXEC flag of open(2).
63 MSG_DONTWAIT (since Linux 2.2)
65 Enables nonblocking operation; if the operation would block, the
66 call fails with the error EAGAIN or EWOULDBLOCK. This provides
67 similar behavior to setting the O_NONBLOCK flag (via the fc‐
68 ntl(2) F_SETFL operation), but differs in that MSG_DONTWAIT is a
69 per-call option, whereas O_NONBLOCK is a setting on the open
70 file description (see open(2)), which will affect all threads in
71 the calling process and as well as other processes that hold
72 file descriptors referring to the same open file description.
73 MSG_ERRQUEUE (since Linux 2.2)
75 This flag specifies that queued errors should be received from
76 the socket error queue. The error is passed in an ancillary
77 message with a type dependent on the protocol (for IPv4
78 IP_RECVERR). The user should supply a buffer of sufficient
79 size. See cmsg(3) and ip(7) for more information. The payload
80 of the original packet that caused the error is passed as normal
81 data via msg_iovec. The original destination address of the
82 datagram that caused the error is supplied via msg_name.
83 The error is supplied in a sock_extended_err structure:
85 #define SO_EE_ORIGIN_NONE 0
87 #define SO_EE_ORIGIN_LOCAL 1
88 #define SO_EE_ORIGIN_ICMP 2
89 #define SO_EE_ORIGIN_ICMP6 3
90 struct sock_extended_err
92 {
93 uint32_t ee_errno; /* Error number */
94 uint8_t ee_origin; /* Where the error originated */
95 uint8_t ee_type; /* Type */
96 uint8_t ee_code; /* Code */
97 uint8_t ee_pad; /* Padding */
98 uint32_t ee_info; /* Additional information */
99 uint32_t ee_data; /* Other data */
100 /* More data may follow */
101 };
102 struct sockaddr *SO_EE_OFFENDER(struct sock_extended_err *);
104 ee_errno contains the errno number of the queued error. ee_ori‐
106 gin is the origin code of where the error originated. The other
107 fields are protocol-specific. The macro SO_EE_OFFENDER returns
108 a pointer to the address of the network object where the error
109 originated from given a pointer to the ancillary message. If
110 this address is not known, the sa_family member of the sockaddr
111 contains AF_UNSPEC and the other fields of the sockaddr are un‐
112 defined. The payload of the packet that caused the error is
113 passed as normal data.
114 For local errors, no address is passed (this can be checked with
116 the cmsg_len member of the cmsghdr). For error receives, the
117 MSG_ERRQUEUE flag is set in the msghdr. After an error has been
118 passed, the pending socket error is regenerated based on the
119 next queued error and will be passed on the next socket opera‐
120 tion.
121 MSG_OOB
123 This flag requests receipt of out-of-band data that would not be
124 received in the normal data stream. Some protocols place expe‐
125 dited data at the head of the normal data queue, and thus this
126 flag cannot be used with such protocols.
127 MSG_PEEK
129 This flag causes the receive operation to return data from the
130 beginning of the receive queue without removing that data from
131 the queue. Thus, a subsequent receive call will return the same
132 data.
133 MSG_TRUNC (since Linux 2.2)
135 For raw (AF_PACKET), Internet datagram (since Linux
136 2.4.27/2.6.8), netlink (since Linux 2.6.22), and UNIX datagram
137 as well as sequenced-packet (since Linux 3.4) sockets: return
138 the real length of the packet or datagram, even when it was
139 longer than the passed buffer.
140 For use with Internet stream sockets, see tcp(7).
142 MSG_WAITALL (since Linux 2.2)
144 This flag requests that the operation block until the full re‐
145 quest is satisfied. However, the call may still return less
146 data than requested if a signal is caught, an error or discon‐
147 nect occurs, or the next data to be received is of a different
148 type than that returned. This flag has no effect for datagram
149 sockets.
150 recvfrom()
152 recvfrom() places the received message into the buffer buf. The caller
153 must specify the size of the buffer in len.
154 If src_addr is not NULL, and the underlying protocol provides the
156 source address of the message, that source address is placed in the
157 buffer pointed to by src_addr. In this case, addrlen is a value-result
158 argument. Before the call, it should be initialized to the size of the
159 buffer associated with src_addr. Upon return, addrlen is updated to
160 contain the actual size of the source address. The returned address is
161 truncated if the buffer provided is too small; in this case, addrlen
162 will return a value greater than was supplied to the call.
163 If the caller is not interested in the source address, src_addr and ad‐
165 drlen should be specified as NULL.
166 recv()
168 The recv() call is normally used only on a connected socket (see con‐
169 nect(2)). It is equivalent to the call:
170 recvfrom(fd, buf, len, flags, NULL, 0);
172 recvmsg()
174 The recvmsg() call uses a msghdr structure to minimize the number of
175 directly supplied arguments. This structure is defined as follows in
176 <sys/socket.h>:
177 struct msghdr {
179 void *msg_name; /* Optional address */
180 socklen_t msg_namelen; /* Size of address */
181 struct iovec *msg_iov; /* Scatter/gather array */
182 size_t msg_iovlen; /* # elements in msg_iov */
183 void *msg_control; /* Ancillary data, see below */
184 size_t msg_controllen; /* Ancillary data buffer len */
185 int msg_flags; /* Flags on received message */
186 };
187 The msg_name field points to a caller-allocated buffer that is used to
189 return the source address if the socket is unconnected. The caller
190 should set msg_namelen to the size of this buffer before this call;
191 upon return from a successful call, msg_namelen will contain the length
192 of the returned address. If the application does not need to know the
193 source address, msg_name can be specified as NULL.
194 The fields msg_iov and msg_iovlen describe scatter-gather locations, as
196 discussed in readv(2).
197 The field msg_control, which has length msg_controllen, points to a
199 buffer for other protocol control-related messages or miscellaneous an‐
200 cillary data. When recvmsg() is called, msg_controllen should contain
201 the length of the available buffer in msg_control; upon return from a
202 successful call it will contain the length of the control message se‐
203 quence.
204 The messages are of the form:
206 struct cmsghdr {
208 size_t cmsg_len; /* Data byte count, including header
209 (type is socklen_t in POSIX) */
210 int cmsg_level; /* Originating protocol */
211 int cmsg_type; /* Protocol-specific type */
212 /* followed by
213 unsigned char cmsg_data[]; */
214 };
215 Ancillary data should be accessed only by the macros defined in
217 cmsg(3).
218 As an example, Linux uses this ancillary data mechanism to pass ex‐
220 tended errors, IP options, or file descriptors over UNIX domain sock‐
221 ets. For further information on the use of ancillary data in various
222 socket domains, see unix(7) and ip(7).
223 The msg_flags field in the msghdr is set on return of recvmsg(). It
225 can contain several flags:
226 MSG_EOR
228 indicates end-of-record; the data returned completed a record
229 (generally used with sockets of type SOCK_SEQPACKET).
230 MSG_TRUNC
232 indicates that the trailing portion of a datagram was discarded
233 because the datagram was larger than the buffer supplied.
234 MSG_CTRUNC
236 indicates that some control data was discarded due to lack of
237 space in the buffer for ancillary data.
238 MSG_OOB
240 is returned to indicate that expedited or out-of-band data was
241 received.
242 MSG_ERRQUEUE
244 indicates that no data was received but an extended error from
245 the socket error queue.
246 MSG_CMSG_CLOEXEC (since Linux 2.6.23)
248 indicates that MSG_CMSG_CLOEXEC was specified in the flags argu‐
249 ment of recvmsg().
250
252 These calls return the number of bytes received, or -1 if an error oc‐
253 curred. In the event of an error, errno is set to indicate the error.
254 When a stream socket peer has performed an orderly shutdown, the return
256 value will be 0 (the traditional "end-of-file" return).
257 Datagram sockets in various domains (e.g., the UNIX and Internet do‐
259 mains) permit zero-length datagrams. When such a datagram is received,
260 the return value is 0.
261 The value 0 may also be returned if the requested number of bytes to
263 receive from a stream socket was 0.
264
266 These are some standard errors generated by the socket layer. Addi‐
267 tional errors may be generated and returned from the underlying proto‐
268 col modules; see their manual pages.
269 EAGAIN or EWOULDBLOCK
271 The socket is marked nonblocking and the receive operation would
272 block, or a receive timeout had been set and the timeout expired
273 before data was received. POSIX.1 allows either error to be re‐
274 turned for this case, and does not require these constants to
275 have the same value, so a portable application should check for
276 both possibilities.
277 EBADF The argument sockfd is an invalid file descriptor.
279 ECONNREFUSED
281 A remote host refused to allow the network connection (typically
282 because it is not running the requested service).
283 EFAULT The receive buffer pointer(s) point outside the process's ad‐
285 dress space.
286 EINTR The receive was interrupted by delivery of a signal before any
288 data was available; see signal(7).
289 EINVAL Invalid argument passed.
291 ENOMEM Could not allocate memory for recvmsg().
293 ENOTCONN
295 The socket is associated with a connection-oriented protocol and
296 has not been connected (see connect(2) and accept(2)).
297 ENOTSOCK
299 The file descriptor sockfd does not refer to a socket.
300
302 According to POSIX.1, the msg_controllen field of the msghdr structure
303 should be typed as socklen_t, and the msg_iovlen field should be typed
304 as int, but glibc currently types both as size_t.
305
307 POSIX.1-2008.
308
310 POSIX.1-2001, 4.4BSD (first appeared in 4.2BSD).
311 POSIX.1 describes only the MSG_OOB, MSG_PEEK, and MSG_WAITALL flags.
313
315 If a zero-length datagram is pending, read(2) and recv() with a flags
316 argument of zero provide different behavior. In this circumstance,
317 read(2) has no effect (the datagram remains pending), while recv() con‐
318 sumes the pending datagram.
319 See recvmmsg(2) for information about a Linux-specific system call that
321 can be used to receive multiple datagrams in a single call.
322
324 An example of the use of recvfrom() is shown in getaddrinfo(3).
325
327 fcntl(2), getsockopt(2), read(2), recvmmsg(2), select(2), shutdown(2),
328 socket(2), cmsg(3), sockatmark(3), ip(7), ipv6(7), socket(7), tcp(7),
329 udp(7), unix(7)
330Linux man-pages 6.05 2023-07-18 recv(2)