1 .\" Copyright (C) 2011 by Andi Kleen <andi@firstfloor.org>
2 .\" and Copyright (c) 2011 by Michael Kerrisk <mtk.manpages@gmail.com>
4 .\" %%%LICENSE_START(VERBATIM)
5 .\" Permission is granted to make and distribute verbatim copies of this
6 .\" manual provided the copyright notice and this permission notice are
7 .\" preserved on all copies.
9 .\" Permission is granted to copy and distribute modified versions of this
10 .\" manual under the conditions for verbatim copying, provided that the
11 .\" entire resulting derived work is distributed under the terms of a
12 .\" permission notice identical to this one.
14 .\" Since the Linux kernel and libraries are constantly changing, this
15 .\" manual page may be incorrect or out-of-date. The author(s) assume no
16 .\" responsibility for errors or omissions, or for damages resulting from
17 .\" the use of the information contained herein. The author(s) may not
18 .\" have taken the same level of care in the production of this manual,
19 .\" which is licensed free of charge, as they might when working
22 .\" Formatted or processed versions of this manual, if unaccompanied by
23 .\" the source, must acknowledge the copyright and authors of this work.
26 .\" Syscall added in following commit
27 .\" commit a2e2725541fad72416326798c2d7fa4dafb7d337
28 .\" Author: Arnaldo Carvalho de Melo <acme@redhat.com>
29 .\" Date: Mon Oct 12 23:40:10 2009 -0700
31 .TH RECVMMSG 2 2019-03-06 "Linux" "Linux Programmer's Manual"
33 recvmmsg \- receive multiple messages on a socket
36 .BR "#define _GNU_SOURCE" " /* See feature_test_macros(7) */"
37 .BI "#include <sys/socket.h>"
39 .BI "int recvmmsg(int " sockfd ", struct mmsghdr *" msgvec \
40 ", unsigned int " vlen ","
41 .BI " int " flags ", struct timespec *" timeout ");"
46 system call is an extension of
48 that allows the caller to receive multiple messages from a socket
49 using a single system call.
50 (This has performance benefits for some applications.)
51 A further extension over
53 is support for a timeout on the receive operation.
57 argument is the file descriptor of the socket to receive data from.
61 argument is a pointer to an array of
64 The size of this array is specified in
69 structure is defined in
76 struct msghdr msg_hdr; /* Message header */
77 unsigned int msg_len; /* Number of received bytes for header */
86 structure, as described in
90 field is the number of bytes returned for the message in the entry.
91 This field has the same value as the return value of a single
97 argument contains flags ORed together.
98 The flags are the same as documented for
100 with the following addition:
102 .BR MSG_WAITFORONE " (since Linux 2.6.34)"
105 after the first message has been received.
112 .BR clock_gettime (2))
113 defining a timeout (seconds plus nanoseconds) for the receive operation
114 .RI ( "but see BUGS!" ).
115 (This interval will be rounded up to the system clock granularity,
116 and kernel scheduling delays mean that the blocking interval
117 may overrun by a small amount.)
120 is NULL, then the operation blocks indefinitely.
126 messages have been received
127 or until the timeout expires.
128 A nonblocking call reads as many messages as are available
129 (up to the limit specified by
131 and returns immediately.
135 successive elements of
137 are updated to contain information about each received message:
139 contains the size of the received message;
142 are updated as described in
144 The return value of the call indicates the number of elements of
146 that have been updated.
150 returns the number of messages received in
152 on error, \-1 is returned, and
154 is set to indicate the error.
158 In addition, the following error can occur:
168 system call was added in Linux 2.6.33.
169 Support in glibc was added in version 2.12.
176 argument does not work as intended.
177 .\" FIXME . https://bugzilla.kernel.org/show_bug.cgi?id=75371
178 .\" http://thread.gmane.org/gmane.linux.man/5677
179 The timeout is checked only after the receipt of each datagram,
182 datagrams are received before the timeout expires,
183 but then no further datagrams are received, the call will block forever.
185 If an error occurs after at least one message has been received,
186 the call succeeds, and returns the number of messages received.
187 The error code is expected to be returned on a subsequent call to
189 In the current implementation, however, the error code can be overwritten
190 in the meantime by an unrelated network event on a socket,
191 for example an incoming ICMP packet.
194 The following program uses
196 to receive multiple messages on a socket and stores
197 them in multiple buffers.
198 The call returns if all buffers are filled or if the
199 timeout specified has expired.
201 The following snippet periodically generates UDP datagrams
202 containing a random number:
206 .RB "$" " while true; do echo $RANDOM > /dev/udp/127.0.0.1/1234; "
207 .B " sleep 0.25; done"
211 These datagrams are read by the example application, which
212 can give the following output:
229 #include <netinet/ip.h>
233 #include <sys/socket.h>
241 int sockfd, retval, i;
242 struct sockaddr_in addr;
243 struct mmsghdr msgs[VLEN];
244 struct iovec iovecs[VLEN];
245 char bufs[VLEN][BUFSIZE+1];
246 struct timespec timeout;
248 sockfd = socket(AF_INET, SOCK_DGRAM, 0);
254 addr.sin_family = AF_INET;
255 addr.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
256 addr.sin_port = htons(1234);
257 if (bind(sockfd, (struct sockaddr *) &addr, sizeof(addr)) == \-1) {
262 memset(msgs, 0, sizeof(msgs));
263 for (i = 0; i < VLEN; i++) {
264 iovecs[i].iov_base = bufs[i];
265 iovecs[i].iov_len = BUFSIZE;
266 msgs[i].msg_hdr.msg_iov = &iovecs[i];
267 msgs[i].msg_hdr.msg_iovlen = 1;
270 timeout.tv_sec = TIMEOUT;
273 retval = recvmmsg(sockfd, msgs, VLEN, 0, &timeout);
275 perror("recvmmsg()");
279 printf("%d messages received\en", retval);
280 for (i = 0; i < retval; i++) {
281 bufs[i][msgs[i].msg_len] = 0;
282 printf("%d %s", i+1, bufs[i]);
288 .BR clock_gettime (2),