Socket%20option

1
Socket option

• Getsockopt ou setsockopt
• fcntl
• ioctl

2
Socket option

• include ltsys/socket.hgt
• int getsockopt(int sockfd, int level, int
  optname, void optval, socklen_t optlen)
• int setsockopt(int sockfd, int level, int
  optname, const void optval socklen_t optlen)
• Both return 0 if OK,1 on error

• Level code in the system to interpret the
  option
• Optval pointer to a variable from which the new
  value of the option is fetched by setsocopt or
  into which the current value of the option is
  stored

3
Options
4
Options
5
Socket options

• Options inherited by a connected TCP socket from
  the listening socket
• SO_DEBUG
• SO_DONTROUTE
• SO_KEEPALIVE
• SO_LINGER
• SO_OOBINLINE
• SO_RCVBUF
• SO_SNDBUF.

6
SO_BROADCAST

• Enables/disable the ability to send broadcast
  messages
• Only for networks that allow broadcast
• Code EACCESS returned

7
SO_DEBUG

• Valid for TCP
• Keeps track of all segment sent and received by
  TCP, kept on a circular buffer which can be
  examined by the kernel using the trpt function

8
SO_DONTROUTE

• Used to bypass the routing mechanism
• For UDP sockets, can set flag MSG_DONTROUTE when
  using send, sento, sendmsg
• Used by routing daemons to ignore the content of
  routing tables

9
SO_ERROR

• Used to retrieve error code
• Blocked process is notified via a select or a
  signal driven I/O

10
SO_KEEPALIVE

• If option is set TCP send automatically a
  keepalive message every two hours without
  exchanging data across the socket.
• What can happen
• TCP peer sends an ACK. The applications is not
  notified
• TCP peer sends na RST. Socket is closed and the
  error message ECONNRESET is returned.
• If no response is given, at most eight messages
  are sent, one every 75 seconds. After that an
  error code ETIMEDOUT is sent

11
SO_KEEPALIVE

• (cont.)
• If an ICMP message host unreachable is
  recieved, it is passed to the application
• The duration of the interval to send the
  keepalive message can be altered to all sockets
• Used in servers

12
SO_KEEPALIVE
13
SO_LINGER

• By default, close send data in buffer and returns
  immediately.
• So-linger allows change this way of operation

struct linger int l_onoff / 0off,
nonzeroon / int l_linger / linger time
(seconds) /

• l_onoff is zero, option is turned off close is
  called. Option is ignored
• l_onoff is nonzsero and l_linger zero buffer
  content is discarded and connection is
  discarded, an RST is sent to the peer, avoids
  time-wait

14
SO_LINGER

• l_onoff non zero, l_linger non zero process is
  put to sleep until all data in buffer is sent and
  acked or linger period expires
• The expiration of lingertime can be checked

15
SO_LINGER
16
SO_LINGER
17
SO_LINGER
18
SO_LINGER

• Application level ack
• Client
• char ack
• Write(sockfd, data, nbytes)
• n Read(sockfd, ack, 1)
• Server
• nbytes Read(sockfd, buff, sizeof(buff))
  Write(sockfd, "", 1)

19
SO_LINGER
20
SO_LINGER
21
SO_OOBINLINE

• Out-of-band data is put in the normal buffer queue

22
SO_RCVBUF e SO_SNDBUF

• Set value for the size of send buffer and receive
  buffer
• Default values
• TCP old versions 4096 bytes
• TCP current versions 8192 a 61440 bytes
• UDP send buffer 9000 bytes and UDP receive buffer
  40000

23
SO_RCVBUF e SO_SNDBUF

• The size of the buffer needs to be set before
  connect is called at the client and before listen
  is called at the server in the server A
  atribuição do tamanho do buffer deve ser feita no
  cliente antes de chamar connect e no servidor
  antes de chamar listen
• At least 3 MSS is recommended

24
Bandwidth delay product

• Bandwidth Delay product amount of byte in the
  pipe
• Transmission rate x RTT
• For instance T1 (1.536.000 bytes/sec) with an
  RTT of 60ms, bandwidth-delay product 11,520
  bytes
• If buffer is smaller than bandwidth-delay product
  TCP becomes the bottleneck
• Need larger buffers for high speed networks or
  networks with long propagation delays (satellite
  networks)

25
SO_RCVLOWAT e SO_SNDLOWAT

• Changes the low-water mark of the buffer
• Minimum amount of data required by select to
  consider data as either read or written

26
SO_REUSEADDR

• Allows server to perform bind on a port even if
  it is being used by another process (typically
  connected socket derived from a listening
  sockets)
• Allows multiple instances of the same server to
  be started on the same port as each instance
  binds a different local IP address

27
SO_REUSEADDR

• Allows a single process to bind the same port to
  multiple sockets, as long as each bind specifies
  a different local IP
• Allows completely duplicated bindings a bind of
  an IP address and port when the same IP address
  and port are already bound to another socket.
  Useful for multicast when the socket option
  SO_REUSEPORT is not available

28
SO_REUSEPORT

• SO_REUSEADDR is equivalent to SO_REUSEPORT when
  multicast is used

29
SO_TYPE e SO_USELOOPBACK

• SO_TYPE returns the socket type
• SO_USELOOPBACK receives copy of all transmitted

30
IPv4 Options

• Level IPROTO_IP
• IP_HDRINCL allows writing to IP header used
  in RAW_SOCKET (ICMP)
• What cannot be attributed
• Checksum
• IP identification field
• Output interface INADDR_V4
• Implementation dependent option

31
IPv4 Options

• IP_options allow setting of IP datagram fields
• IP_RECVDSTADDR allow to return a UDP packet as
  ancillary data (control) by using recvmsg
• IP_TOS allow setting field ToS

32
IPv4 Options

• IP_TTL allow writing and reading to filed Time
  to Live (TTL). Default 64

33
IPv6 Options

• LEVEL IPTPROTO_IPv6
• IPv6_ADDRFOMR allows converting an IPv4 socket
  to an IPv6 socket
• IPv6 checksum computes and store checksum for
  outgoing packets and verify checksum for incoming
  packets. Parameter indicates where checksum
  field is located

34
IPv6 Options

• IPv6_DSTOPTS specifies that any received IPv6
  destination option are to be returned as
  ancillary data by recvmsg
• IPv6_HOPLIMIT specifies that the received hop
  limit field to be returned as ancillary data by
  recvmsg
• IPv6_HOPOPTS specifies that the option
  hop-by-hop has to be returned as ancillary by
  recvmsg
• IPv6_NEXTHOP specifies the next hop a packet
  should go

35
IPv6 Options

• IPv6_PKTINFO specifies that destination address
  and interface are to be returned as ancillary
  data
• IPv6_RTHDR specifies that the routing fileds of
  the header should be returned as ancillary data
  by rcvmsg
• IPv6-UNICAST_HOPS set thehop limit

36
ICMPv6

• Level IPROTO_ICMPv6
• ICMP6_FILTER specifies which of the 256 ICMPv6
  messages can be sent on a podem RAWSOCKET

37
Opções TCP

• TCP_KEEPALIVE specifies the time interval in
  seconds that the connection should send keepalive
  messages
• TCP_MAXSEG allows setting and retrieving value
  of MSS
• TCP_NODELAY disable Nagles algorithm which
  prevents small packets to be sent. Small segments
  are sent in case there are pending acks

38
Opções TCP
39
Opções TCP
40
Opções TCP

• TCP_STDURG changes interpretation of the urgent
  pointer field