tjh.dev
Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1DCCP protocol
2============
3
4
5Contents
6========
7
8- Introduction
9- Missing features
10- Socket options
11- Notes
12
13Introduction
14============
15
16Datagram Congestion Control Protocol (DCCP) is an unreliable, connection
17based protocol designed to solve issues present in UDP and TCP particularly
18for real time and multimedia traffic.
19
20It has a base protocol and pluggable congestion control IDs (CCIDs).
21
22It is at proposed standard RFC status and the homepage for DCCP as a protocol
23is at:
24 http://www.read.cs.ucla.edu/dccp/
25
26Missing features
27================
28
29The DCCP implementation does not currently have all the features that are in
30the RFC.
31
32The known bugs are at:
33 http://linux-net.osdl.org/index.php/TODO#DCCP
34
35Socket options
36==============
37
38DCCP_SOCKOPT_SERVICE sets the service. The specification mandates use of
39service codes (RFC 4340, sec. 8.1.2); if this socket option is not set,
40the socket will fall back to 0 (which means that no meaningful service code
41is present). On active sockets this is set before connect(); specifying more
42than one code has no effect (all subsequent service codes are ignored). The
43case is different for passive sockets, where multiple service codes (up to 32)
44can be set before calling bind().
45
46DCCP_SOCKOPT_GET_CUR_MPS is read-only and retrieves the current maximum packet
47size (application payload size) in bytes, see RFC 4340, section 14.
48
49DCCP_SOCKOPT_SEND_CSCOV and DCCP_SOCKOPT_RECV_CSCOV are used for setting the
50partial checksum coverage (RFC 4340, sec. 9.2). The default is that checksums
51always cover the entire packet and that only fully covered application data is
52accepted by the receiver. Hence, when using this feature on the sender, it must
53be enabled at the receiver, too with suitable choice of CsCov.
54
55DCCP_SOCKOPT_SEND_CSCOV sets the sender checksum coverage. Values in the
56 range 0..15 are acceptable. The default setting is 0 (full coverage),
57 values between 1..15 indicate partial coverage.
58DCCP_SOCKOPT_RECV_CSCOV is for the receiver and has a different meaning: it
59 sets a threshold, where again values 0..15 are acceptable. The default
60 of 0 means that all packets with a partial coverage will be discarded.
61 Values in the range 1..15 indicate that packets with minimally such a
62 coverage value are also acceptable. The higher the number, the more
63 restrictive this setting (see [RFC 4340, sec. 9.2.1]). Partial coverage
64 settings are inherited to the child socket after accept().
65
66The following two options apply to CCID 3 exclusively and are getsockopt()-only.
67In either case, a TFRC info struct (defined in <linux/tfrc.h>) is returned.
68DCCP_SOCKOPT_CCID_RX_INFO
69 Returns a `struct tfrc_rx_info' in optval; the buffer for optval and
70 optlen must be set to at least sizeof(struct tfrc_rx_info).
71DCCP_SOCKOPT_CCID_TX_INFO
72 Returns a `struct tfrc_tx_info' in optval; the buffer for optval and
73 optlen must be set to at least sizeof(struct tfrc_tx_info).
74
75
76Sysctl variables
77================
78Several DCCP default parameters can be managed by the following sysctls
79(sysctl net.dccp.default or /proc/sys/net/dccp/default):
80
81request_retries
82 The number of active connection initiation retries (the number of
83 Requests minus one) before timing out. In addition, it also governs
84 the behaviour of the other, passive side: this variable also sets
85 the number of times DCCP repeats sending a Response when the initial
86 handshake does not progress from RESPOND to OPEN (i.e. when no Ack
87 is received after the initial Request). This value should be greater
88 than 0, suggested is less than 10. Analogue of tcp_syn_retries.
89
90retries1
91 How often a DCCP Response is retransmitted until the listening DCCP
92 side considers its connecting peer dead. Analogue of tcp_retries1.
93
94retries2
95 The number of times a general DCCP packet is retransmitted. This has
96 importance for retransmitted acknowledgments and feature negotiation,
97 data packets are never retransmitted. Analogue of tcp_retries2.
98
99send_ndp = 1
100 Whether or not to send NDP count options (sec. 7.7.2).
101
102send_ackvec = 1
103 Whether or not to send Ack Vector options (sec. 11.5).
104
105ack_ratio = 2
106 The default Ack Ratio (sec. 11.3) to use.
107
108tx_ccid = 2
109 Default CCID for the sender-receiver half-connection.
110
111rx_ccid = 2
112 Default CCID for the receiver-sender half-connection.
113
114seq_window = 100
115 The initial sequence window (sec. 7.5.2).
116
117tx_qlen = 5
118 The size of the transmit buffer in packets. A value of 0 corresponds
119 to an unbounded transmit buffer.
120
121sync_ratelimit = 125 ms
122 The timeout between subsequent DCCP-Sync packets sent in response to
123 sequence-invalid packets on the same socket (RFC 4340, 7.5.4). The unit
124 of this parameter is milliseconds; a value of 0 disables rate-limiting.
125
126Notes
127=====
128
129DCCP does not travel through NAT successfully at present on many boxes. This is
130because the checksum covers the pseudo-header as per TCP and UDP. Linux NAT
131support for DCCP has been added.