Go to main content

man pages section 3: Basic Library Functions

Exit Print View

Updated: Thursday, June 13, 2019
 
 

setsockopt (3C)

Name

getsockopt, setsockopt - get and set options on sockets

Synopsis

#include <sys/socket.h>

int getsockopt(int 
s, int level, 
int optname, void *restrict 
optval,
     socklen_t *restrict optlen);
int setsockopt(int
 s, int level
, int optname, 
const void *optval,
     socklen_t optlen);

Description

The getsockopt() and setsockopt() functions manipulate options associated with a socket. Options may exist at multiple protocol levels; they are always present at the uppermost “socket” level.

The level argument specifies the protocol level at which the option resides. To manipulate options at the socket level, specify the level argument as SOL_SOCKET. To manipulate options at the protocol level, supply the appropriate protocol number for the protocol controlling the option. For example, to indicate that an option will be interpreted by the TCP, set level to the protocol number of TCP, as defined in the netinet/in.h header, or as determined by using getprotobyname(3C). Some socket protocol families may also define additional levels, such as SOL_ROUTE. Only socket-level options are described here.

The parameters optval and optlen are used to access option values for setsockopt(). For getsockopt(), they identify a buffer in which the value(s) for the requested option(s) are to be returned. For getsockopt(), optlen is a value-result parameter, initially containing the size of the buffer pointed to by optval, and modified on return to indicate the actual size of the value returned. Use a 0 optval if no option value is to be supplied or returned.

The optname and any specified options are passed uninterpreted to the appropriate protocol module for interpretation. The include file <sys/socket.h> contains definitions for the socket-level options described below. Options at other protocol levels vary in format and name.

Most socket-level options take an int for optval. For setsockopt(), the optval parameter should be non-zero to enable a boolean option, or zero if the option is to be disabled. SO_LINGER uses a struct linger parameter that specifies the desired state of the option and the linger interval. struct linger is defined in <sys/socket.h>. struct linger contains the following members:

l_onoff

on = 1/off = 0

l_linger

linger time, in seconds

The following options are recognized at the socket level. Except as noted, each may be examined with getsockopt() and set with setsockopt().

SO_DEBUG

enable/disable recording of debugging information

SO_REUSEADDR

enable/disable local address reuse

SO_REUSEPORT

enable/disable local port reuse for TCP/SCTP/UDP socket

SO_KEEPALIVE

enable/disable keep connections alive

SO_DONTROUTE

enable/disable routing bypass for outgoing messages

SO_LINGER

linger on close if data is present

SO_BROADCAST

enable/disable permission to transmit broadcast messages

SO_OOBINLINE

enable/disable reception of out-of-band data in band

SO_SNDBUF

set buffer size for output

SO_RCVBUF

set buffer size for input

SO_SNDTIMEO

set the send timeout for socket

SO_RCVTIMEO

set the receive timeout for socket

SO_DGRAM_ERRIND

application wants delayed error

SO_TIMESTAMP

enable/disable reception of timestamp with datagrams

SO_LISTENQLIMIT

get the current value of listen queue limit (initially set using the backlog argument of listen(3C)). This option is available only with the getsockopt() function.

SO_EXCLBIND

enable/disable exclusive binding of the socket

SO_TYPE

get the type of the socket (get only)

SO_ERROR

get and clear error on the socket (get only)

In addition to SO_ERROR, the socket error is retrieved and cleared by any ioctl(2) invocation. In this case, the device driver detects socket error and passes it through ioctl() without change to the caller.

SO_MAC_EXEMPT

get or set mandatory access control on the socket. This option is available only when the system is configured with Trusted Extensions.

SO_ALLZONES

bypass zone boundaries (privileged).

SO_DOMAIN

get the domain used in the socket (get only)

SO_PROTOTYPE

for socket in domains PF_INET and PF_INET6, get the underlying protocol number used in the socket. For socket in domain PF_ROUTE, get the address family used in the socket.

SO_PASSIVE_CONNECT

modify connect(3C) to wait for connection request from a peer instead of initiating a connection request to it. It is applicable to TCP/SCTP PF_INET/PF_INET6 socket.

SO_FLOW_SLA

set per socket service level properties: priority and bandwidth limit.

It is applicable to TCP/UDP PF_INET/PF_INET6 sockets. Requires PRIV_SYS_FLOW_CONFIG privilege.

SO_FLOW_NAME

Make the socket a flow filter of the specified MAC flow.

It is applicable to TCP/UDP PF_INET/PF_INET6 sockets. Requires PRIV_SYS_FLOW_CONFIG privilege.

SO_NOSIGPIPE

Get and set whether this specific socket returns only EPIPE on write when the socket is disconnected or whether a SIGPIPE is send.

The SO_DEBUG option enables debugging in the underlying protocol modules. The SO_REUSEADDR or SO_REUSEPORT options indicate that the rules used in validating addresses and ports supplied in a bind() call should allow reuse of local addresses or ports. For more information, see the bind(3C) man page. The SO_REUSEPORT option must be set by all the sockets including the first socket, before calling bind() to enable reuse. All the calling processes must have the same effective UID for the bind() call to succeed. This option allows multiple listeners binding to the same local address or port pair. For TCP/SCTP socket, new incoming connection requests will be distributed across those listeners bound to the same local address or port pair. For UDP socket, incoming datagrams will be distributed across those bound sockets. The algorithm used to determine how to select a socket using this option to deliver an incoming TCP/SCTP connection request or an UDP datagram can be changed by modifying the ipadm protocol property, reuseport-lbalg. The current supported algorithms are:

  • Round Robin (rr)

  • Hashing on source IP address (src-ip)

  • Hashing on source IP address and source port (src-ip-port)

  • Hashing on source and destination IP address (src-dst-ip)

  • Hashing on all 4 tuples (src-dst-ip-ports)

The string in parenthesis is the value used in ipadm to represent the algorithm. For more information, see the ipadm(8) man page.

The SO_KEEPALIVE option enables the periodic transmission of messages on a connected socket. If the connected party fails to respond to these messages, the connection is considered broken and threads using the socket are notified using a SIGPIPE signal.

The SO_DONTROUTE option indicates that outgoing messages should bypass the standard routing facilities. Instead, messages are directed to the appropriate network interface according to the network portion of the destination address.

The SO_LINGER option controls the action taken when unsent messages are queued on a socket and a close(2) is performed. If the socket promises reliable delivery of data and SO_LINGER is set, the system will block the thread on the close() attempt until it is able to transmit the data or until it decides it is unable to deliver the information (a timeout period, termed the linger interval, is specified in the setsockopt() call when SO_LINGER is requested). If SO_LINGER is disabled and a close() is issued, the system will process the close() in a manner that allows the thread to continue as quickly as possible.

The option SO_BROADCAST requests permission to send broadcast datagrams on the socket. With protocols that support out-of-band data, the SO_OOBINLINE option requests that out-of-band data be placed in the normal data input queue as received; it will then be accessible with recv() or read() calls without the MSG_OOB flag.

The SO_SNDBUF and SO_RCVBUF options adjust the normal buffer sizes allocated for output and input buffers, respectively. The buffer size may be increased for high-volume connections or may be decreased to limit the possible backlog of incoming data. The maximum buffer size for UDP/TCP is determined by the value of the ipadm variable max_buf for that particular protocol. Use the ipadm(8) utility to determine the current default values. See the Solaris Tunable Parameters Reference Manual for information on setting the values of max_buf for either TCP, UDP or both. At present, lowering SO_RCVBUF on a TCP connection after it has been established has no effect.

The SO_SNDTIMEO and SO_RCVTIMEO options specify the timeout for socket I/O functions. Both options take a pointer to struct timeval as their argument. The default timeout value is zero, which means the socket I/O functions will never timeout. If the timeout expires and no data has been sent, the socket I/O functions will return -1 and set errno to EAGAIN or EWOULDBLOCK. If some data has already been transferred when timeout fires, the socket I/O functions will return the number of bytes transferred so far. The socket functions affected by this option are write(), send(), sendto(), sendmsg(), read(), recv(), recvfrom(), and recvmsg().

By default, delayed errors (such as ICMP port unreachable packets) are returned only for connected datagram sockets. The SO_DGRAM_ERRIND option makes it possible to receive errors for datagram sockets that are not connected. When this option is set, certain delayed errors received after completion of a sendto() or sendmsg() operation will cause a subsequent sendto() or sendmsg() operation using the same destination address (to parameter) to fail with the appropriate error. See send(3C).

If the SO_TIMESTAMP option is enabled on a SO_DGRAM or a SO_RAW socket, the recvmsg(3C) call will return a timestamp in the native data format, corresponding to when the datagram was received.

The SO_EXCLBIND option is used to enable or disable the exclusive binding of a socket. It overrides the use of the SO_REUSEADDR option to reuse an address on bind(3C). The actual semantics of the SO_EXCLBIND option depend on the underlying protocol. See tcp(4P) or udp(4P) for more information.

The SO_TYPE and SO_ERROR options are used only with getsockopt(). The SO_TYPE option returns the type of the socket, for example, SOCK_STREAM. It is useful for servers that inherit sockets on startup. The SO_ERROR option returns any pending error on the socket and clears the error status. It may be used to check for asynchronous errors on connected datagram sockets or for other asynchronous errors.

The SO_MAC_EXEMPT option is used to toggle socket behavior with unlabeled peers. A socket that has this option enabled can communicate with an unlabeled peer if it is in the global zone or has a label that dominates the default label of the peer. Otherwise, the socket must have a label that is equal to the default label of the unlabeled peer. Calling setsockopt() with this option returns an EACCES error if the process lacks the NET_MAC_AWARE privilege or if the socket is bound. The SO_MAC_EXEMPT option is available only when the system is configured with Trusted Extensions.

The SO_ALLZONES option can be used to bypass zone boundaries between shared-IP zones. Normally, the system prevents a socket from being bound to an address that is not assigned to the current zone. It also prevents a socket that is bound to a wildcard address from receiving traffic for other zones. However, some daemons which run in the global zone might need to send and receive traffic using addresses that belong to other shared-IP zones. If set before a socket is bound, SO_ALLZONES causes the socket to ignore zone boundaries between shared-IP zones and permits the socket to be bound to any address assigned to the shared-IP zones. If the socket is bound to a wildcard address, it receives traffic intended for all shared-IP zones and behaves as if an equivalent socket were bound in each active shared-IP zone. Applications that use the SO_ALLZONES option to initiate connections or send datagram traffic should specify the source address for outbound traffic by binding to a specific address. There is no effect from setting this option in an exclusive-IP zone. Setting this option requires the sys_net_config privilege. See zones(7).

The SO_PASSIVE_CONNECT option can be used to modify connect() semantics for TCP and SCTP socket. After this option is set, calling connect() on the socket will not initiate a connection setup sequence. Instead, the transport end point is in listen state waiting for a connection request from the remote peer specified in connect(). After the expected connection is established, connect() returns.

SO_FLOW_SLA socket option can be used to set per socket service level properties for the socket. The properties, priority and bandwidth limit are supported in SOCK_FLOW_PROP_VERSION1. These properties are effective for both outbound and inbound packets.

Only TCP and UDP PF_INET/PF_INET6 sockets are supported at this time.

The socket option takes a pointer to a sock_flow_props_t structure as the value:


typedef struct sock_flow_props_s {
        int             sfp_version;
        uint32_t        sfp_mask;
        int             sfp_priority;   /* flow priority */
        uint64_t        sfp_maxbw;      /* bandwidth limit in bps */      
        int             sfp_status; 	/* flow create status for getsockopt */
     } sock_flow_props_t;

     #define SOCK_FLOW_PROP_VERSION1  1

     /* bit mask values for sfp_mask */
     #define SFP_MAXBW       0x00000001      /* BW Limit */
     #define SFP_PRIORITY    0x00000008      /* Priority */

     /* possible values for sfp_priority */
     #define SFP_PRIO_NORMAL 1
     #define SFP_PRIO_HIGH   2

A value of 0 for sfp_maxbw drops all packets for the socket.

The status of the flow creation can be obtained by a getsockopt(). The sfp_status field indicates the status. A value of 0 means a flow is successfully created. In case of an error, this field has the errno to indicate the error. This field is not set during setsockopt() and is ignored.

Usage

Example 1 Examples of Usage
 	a. at the time of socket(3c) call
     sock_flow_props_t sprop;
     ...
     sock = socket(AF_INET, SOCK_DGRAM, 0);
     sprop.sfp_version = SOCK_FLOW_PROP_VERSION1;
     sprop.sfp_mask = SFP_PRIORITY;
     sprop.sfp_priority = SFP_PRIO_HIGH;
     setsockopt(sock, SOL_SOCKET, SO_FLOW_SLA, &sprop, sizeof
		(sprop));
     ...

     b. at the time of the accept(3c) call
     sock_flow_props_t sprop;
     ...
     newsock = accept(listen_sock, (struct sockaddr *)&from,
		&fromlen);
     // construct sprop
     setsockopt(newsock, SOL_SOCKET, SO_FLOW_SLA, &sprop, sizeof
		(sprop));
     ...

A socket application using this socket option, causes the system to create a system flow. flowadm(8) can be used to observe these flows. These system generated flows have the prefix <id>.sys.sock.

The pfiles command prints the SO_FLOW_SLA socket option along with the other socket options.

Creation of a system flow requires the link name and the 5-tuple in case of a connected socket and the local 3-tuple in case of a listener/bound socket. The following restrictions exist for the cases where a link name or a value in the 5-tuple/3-tuple can not be obtained given a socket. Some of these restrictions may be relaxed in future.

  • Sockets using IP loopback addresses (/dev/lo0) are not supported.

  • Only connected UDP is supported. This means the socket call, connect(), must be made for the option to be enforced.

  • INADDR_ANY IP address is not supported. So, the socket call, bind(), must pass an IP address for the option to be enforced.

Note that a flow is not created if the above conditions are not met. It is not an error (i.e. bind/connect would not fail) if a flow is not created. getsockopt() should be used to determine if a flow is created.

SO_FLOW_NAME socket option makes a socket a member of the specified MAC flow. Multiple sockets can belong to the same flow. Each socket maps to a flow filter. The flow is created if it does not already exist. If the flow exists, it is assumed that the flow name is known and the application got it out of band.

The following is an example code using SO_FLOW_NAME:

char flow[MAXUSERFLOWNAMELEN];
(void) strlcpy(flow, "tenantA", MAXUSERFLOWNAMELEN);
/* Assign a socket to the flow */rv = setsockopt(sock1, SOL_SOCKET, SO_FLOW_NAME, flow,
strlen(flow));
rv = setsockopt(sock1, SOL_SOCKET, SO_FLOW_NAME, flow,     
     strlen(flow));
rv = setsockopt(sock2, SOL_SOCKET, SO_FLOW_NAME, flow,
     strlen(flow));
...

An application can use either SO_FLOW_SLA or SO_FLOW_NAME option exclusively. They can also be used together with the following rules.

For a given socket, this sequence of setsockopt() calls is allowed:

setsockopt(s, SO_FLOW_NAME, ...)
setsockopt(s, SO_FLOW_SLA, ...)

In this case, the SLA on the flow is modified. SO_FLOW_SLA does not create any system flow though the socket already belongs to a flow as indicated by SO_FLOW_NAME.

For a given socket, this sequence is not allowed:

setsockopt(s, SO_FLOW_SLA, ...);
setsockopt(s, SO_FLOW_NAME, ...);

We do not allow this because we already create a system flow <int>.sys.sock for the first call. The second call fails in this case.

getsockopt() can be called with either of the socket options at any time.

Return Values

If successful, getsockopt() and setsockopt() return 0. Otherwise, the functions return −1 and set errno to indicate the error.

Errors

The getsockopt() and setsockopt() calls succeed unless:

EBADF

The argument s is not a valid file descriptor.

EACCES

Permission denied.

EADDRINUSE

Address already joined for IP_ADD_MEMBERSHIP.

EADDRNOTAVAIL

Bad interface address for IP_ADD_MEMBERSHIP and IP_DROP_MEMBERSHIP.

EHOSTUNREACH

Invalid address for IP_MULTICAST_IF.

EINVAL

Invalid length for IP_OPTIONS.

Not a multicast address for IP_ADD_MEMBERSHIP and IP_DROP_MEMBERSHIP.

The specified option is invalid at the specified socket level, or the socket has been shut down.

ENOBUFS

SO_SNDBUF or SO_RCVBUF exceeds a system limit.

ENOENT

Address not joined for IP_DROP_MEMBERSHIP.

ENOMEM

There was insufficient memory available for the operation to complete.

ENOPROTOOPT

The option is unknown at the level indicated.

ENOSR

There were insufficient STREAMS resources available for the operation to complete.

ENOTSOCK

The argument s is not a socket.

EPERM

No permissions or sufficient privileges.

ENOTCONN

Flow creation status for getsockopt() for SO_FLOW_SLA. No flow is created yet as the application did not do a connect or bind.

EOPNOTSUPP

Flow creation status for getsockopt() for SO_FLOW_SLA. Flow creation is not supported for this socket.

EALREADY

A flow with identical attributes exists.

EINPROGRESS

A flow is being created.

Attributes

See attributes(7) for descriptions of the following attributes:

ATTRIBUTE TYPE
ATTRIBUTE VALUE
Interface Stability
Committed
MT-Level
Async-Signal-Safe
Standard

See Also

close(2), ioctl(2), read(2), write(2), bind(3C), connect(3C), getprotobyname(3C), recv(3C), recvmsg(3C), send(3C), socket(3C), socket.h(3HEAD), tcp(4P), udp(4P), attributes(7), zones(7), flowadm(8), ipadm(8)

Oracle Solaris 11.4 Tunable Parameters Reference Manual