| 1 |
<pre>Network Working Group J. Zweig |
| 2 |
Request for Comments: 1145 UIUC |
| 3 |
C. Partridge |
| 4 |
BBN |
| 5 |
February 1990 |
| 6 |
|
| 7 |
|
| 8 |
<span class="h1">TCP Alternate Checksum Options</span> |
| 9 |
|
| 10 |
Status of This Memo |
| 11 |
|
| 12 |
This memo suggests a pair of TCP options to allow use of alternate |
| 13 |
data checksum algorithms in the TCP header. The use of these options |
| 14 |
is experimental, and not recommended for production use. |
| 15 |
|
| 16 |
Distribution of this memo is unlimited. |
| 17 |
|
| 18 |
Introduction |
| 19 |
|
| 20 |
Some members of the networking community have expressed interest in |
| 21 |
using checksum-algorithms with different error detection and |
| 22 |
correction properties than the standard TCP checksum. The option |
| 23 |
described in this memo provides a mechanism to negotiate the use of |
| 24 |
an alternate checksum at connection-establishment time, as well as a |
| 25 |
mechanism to carry additional checksum information for algorithms |
| 26 |
that utilize checksums that are longer than 16 bits. |
| 27 |
|
| 28 |
Definition of the Options |
| 29 |
|
| 30 |
The TCP Alternate Checksum Request Option may be sent in a SYN |
| 31 |
segment by a TCP to indicate that the TCP is prepared to both |
| 32 |
generate and receive checksums based on an alternate algorithm. |
| 33 |
During communication, the alternate checksum replaces the regular TCP |
| 34 |
checksum in the checksum field of the TCP header. Should the |
| 35 |
alternate checksum require more than 2 octets to transmit, the |
| 36 |
checksum may either be moved into a TCP Alternate Checksum Data |
| 37 |
Option and the checksum field of the TCP header be sent as 0, or the |
| 38 |
data may be split between the header field and the option. Alternate |
| 39 |
checksums are computed over the same data as the regular TCP checksum |
| 40 |
(see TCP Alternate Checksum Data Option discussion below). |
| 41 |
|
| 42 |
TCP Alternate Checksum Request Option |
| 43 |
|
| 44 |
The format of the TCP Alternate Checksum Request Option is: |
| 45 |
|
| 46 |
+----------+----------+----------+ |
| 47 |
| Kind=14 | Length=3 | chksum | |
| 48 |
+----------+----------+----------+ |
| 49 |
|
| 50 |
|
| 51 |
|
| 52 |
<span class="grey">Zweig & Partridge [Page 1]</span></pre> |
| 53 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span> |
| 54 |
<span class="grey"><a href="./rfc1145">RFC 1145</a> TCP Alternate Checksum Options February 1990</span> |
| 55 |
|
| 56 |
|
| 57 |
Here chksum is a number identifying the type of checksum to be used. |
| 58 |
|
| 59 |
This option is not copied into each fragment, and appears only in the |
| 60 |
first fragment. |
| 61 |
|
| 62 |
The currently defined values of chksum are: |
| 63 |
|
| 64 |
0 -- TCP checksum |
| 65 |
1 -- 8-bit Fletcher's algorithm (see <a href="#appendix-I">Appendix I</a>) |
| 66 |
2 -- 16-bit Fletcher's algorithm (see <a href="#appendix-I">Appendix I</a>I) |
| 67 |
|
| 68 |
Note that the 8-bit Fletcher algorithm gives a 16-bit checksum and |
| 69 |
the 16-bit algorithm gives a 32-bit checksum. |
| 70 |
|
| 71 |
Alternate checksum negotiation proceeds as follows: |
| 72 |
|
| 73 |
A SYN segment used to originate a connection may contain the |
| 74 |
Alternate Checksum Request Option, which specifies an alternate |
| 75 |
checksum-calculation algorithm to be used for the connection. The |
| 76 |
acknowledging SYN-ACK segment may also carry the option. |
| 77 |
|
| 78 |
If both SYN segments carry the Alternate Checksum Request option, |
| 79 |
and both specify the same algorithm, that algorithm must be used |
| 80 |
for the remainder of the connection. Otherwise, the standard TCP |
| 81 |
checksum algorithm must be used for the entire connection. Thus, |
| 82 |
for example, if one TCP specifies type 1 checksums, and the other |
| 83 |
specifies type 2 checksums, then they will use type 0 (the regular |
| 84 |
TCP checksum). Note that in practice, one TCP will typically be |
| 85 |
responding to the other's SYN, and thus either accepting or |
| 86 |
rejecting the proposed alternate checksum algorithm. |
| 87 |
|
| 88 |
Any segment with the SYN bit set must always use the standard TCP |
| 89 |
checksum algorithm. Thus the SYN segment will always be |
| 90 |
understood by the receiving TCP. The alternate checksum must not |
| 91 |
be used until the first non-SYN segment. In addition, because RST |
| 92 |
segments may also be received or sent without complete state |
| 93 |
information, any segment with the RST bit set must use the |
| 94 |
standard TCP checksum. |
| 95 |
|
| 96 |
The option may not be sent in any segment that does not have the |
| 97 |
SYN bit set. |
| 98 |
|
| 99 |
An implementation of TCP which does not support the option should |
| 100 |
silently ignore it (as <a href="./rfc1122">RFC 1122</a> requires). Ignoring the option |
| 101 |
will force any TCP attempting to use an alternate checksum to use |
| 102 |
the standard TCP checksum algorithm, thus ensuring |
| 103 |
interoperability. |
| 104 |
|
| 105 |
|
| 106 |
|
| 107 |
|
| 108 |
<span class="grey">Zweig & Partridge [Page 2]</span></pre> |
| 109 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span> |
| 110 |
<span class="grey"><a href="./rfc1145">RFC 1145</a> TCP Alternate Checksum Options February 1990</span> |
| 111 |
|
| 112 |
|
| 113 |
TCP Alternate Checksum Data Option |
| 114 |
|
| 115 |
The format of the TCP Alternate Checksum Data Option is: |
| 116 |
|
| 117 |
+---------+---------+---------+ +---------+ |
| 118 |
| Kind=15 |Length=N | data | ... | data | |
| 119 |
+---------+---------+---------+ +---------+ |
| 120 |
|
| 121 |
This field is used only when the alternate checksum that is |
| 122 |
negotiated is longer than 16 bits. These checksums will not fit in |
| 123 |
the checksum field of the TCP header and thus at least part of them |
| 124 |
must be put in an option. Whether the checksum is split between the |
| 125 |
checksum field in the TCP header and the option or the entire |
| 126 |
checksum is placed in the option is determined on a checksum by |
| 127 |
checksum basis. |
| 128 |
|
| 129 |
The length of this option will depend on the choice of alternate |
| 130 |
checksum algorithm for this connection. |
| 131 |
|
| 132 |
This option is not copied into each fragment, and appears only in the |
| 133 |
first fragment. |
| 134 |
|
| 135 |
While computing the alternate checksum, the TCP checksum field and |
| 136 |
the data portion TCP Alternate Checksum Data Option are replaced with |
| 137 |
zeros. |
| 138 |
|
| 139 |
An otherwise acceptable segment carrying this option on a connection |
| 140 |
using a 16-bit checksum algorithm, or carrying this option with an |
| 141 |
inappropriate number of data octets for the chosen alternate checksum |
| 142 |
algorithm is in error and must be discarded; a RST-segment must be |
| 143 |
generated, and the connection aborted. |
| 144 |
|
| 145 |
Note the requirement above that RST and SYN segments must always use |
| 146 |
the standard TCP checksum. |
| 147 |
|
| 148 |
APPENDIX I: The 8-bit Fletcher Checksum Algorithm |
| 149 |
|
| 150 |
The 8-bit Fletcher Checksum Algorithm is calculated over a sequence |
| 151 |
of data octets (call them D[1] through D[N]) by maintaining 2 |
| 152 |
unsigned 1's-complement 8-bit accumulators A and B whose contents are |
| 153 |
initially zero, and performing the following loop where i ranges from |
| 154 |
1 to N: |
| 155 |
|
| 156 |
A := A + D[i] |
| 157 |
B := B + A |
| 158 |
|
| 159 |
It can be shown that at the end of the loop A will contain the 8-bit |
| 160 |
1's complement sum of all octets in the datagram, and that B will |
| 161 |
|
| 162 |
|
| 163 |
|
| 164 |
<span class="grey">Zweig & Partridge [Page 3]</span></pre> |
| 165 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span> |
| 166 |
<span class="grey"><a href="./rfc1145">RFC 1145</a> TCP Alternate Checksum Options February 1990</span> |
| 167 |
|
| 168 |
|
| 169 |
contain (N)D[1] + (N-1)D[2] + ... + D[N]. |
| 170 |
|
| 171 |
The octets covered by this algorithm should be the same as those over |
| 172 |
which the standard TCP checksum calculation is performed, with the |
| 173 |
pseudoheader being D[1] through D[12] and the TCP header beginning at |
| 174 |
D[13]. Note that, for purposes of the checksum computation, the |
| 175 |
checksum field itself must be equal to zero. |
| 176 |
|
| 177 |
At the end of the loop, the A goes in the first byte of the TCP |
| 178 |
checksum and B goes in the second byte. |
| 179 |
|
| 180 |
Note that, unlike the OSI version of the Fletcher checksum, this |
| 181 |
checksum does not adjust the check bytes so that the receiver |
| 182 |
checksum is 0. |
| 183 |
|
| 184 |
There are a number of much faster algorithms for calculating the two |
| 185 |
octets of the 8-bit Fletcher checksum. For more information see |
| 186 |
[<a href="#ref-Sklower89" title=""Improving the Efficiency of the OSI Checksum Calculation"">Sklower89</a>], [<a href="#ref-Nakassis88" title=""Fletcher's Error Detection Algorithm: How to implement it efficiently and how to avoid the most common pitfalls"">Nakassis88</a>] and [<a href="#ref-Fletcher82" title=""An Arithmetic Checksum for Serial Transmissions"">Fletcher82</a>]. Naturally, any |
| 187 |
computation which computes the same number as would be calculated by |
| 188 |
the loop above may be used to calculate the checksum. One advantage |
| 189 |
of the Fletcher algorithms over the standard TCP checksum algorithm |
| 190 |
is the ability to detect the transposition of octets/words of any |
| 191 |
size within a datagram. |
| 192 |
|
| 193 |
APPENDIX II: The 16-bit Fletcher Checksum Algorithm |
| 194 |
|
| 195 |
The 16-bit Fletcher Checksum algorithm proceeds in precisely the same |
| 196 |
manner as the 8-bit checksum algorithm,, except that A, B and the |
| 197 |
D[i] are 16-bit quantities. It is necessary (as it is with the |
| 198 |
standard TCP checksum algorithm) to pad a datagram containing an odd |
| 199 |
number of octets with a zero octet. |
| 200 |
|
| 201 |
Result A should be placed in the TCP header checksum field and Result |
| 202 |
B should appear in an TCP Alternate Checksum Data option. This |
| 203 |
option must be present in every TCP header. The two bytes reserved |
| 204 |
for B should be set to zero during the calculation of the checksum. |
| 205 |
|
| 206 |
The checksum field of the TCP header shall contain the contents of A |
| 207 |
at the end of the loop. The TCP Alternate Checksum Data option must |
| 208 |
be present and contain the contents of B at the end of the loop. |
| 209 |
|
| 210 |
BIBLIOGRAPHY: |
| 211 |
|
| 212 |
[<a id="ref-BrBoPa89">BrBoPa89</a>] Braden, R., Borman, D., and C. Partridge, "Computing |
| 213 |
the Internet Checksum", ACM Computer Communication |
| 214 |
Review, Vol. 19, No. 2, pp. 86-101, April 1989. |
| 215 |
[Note that this includes Plummer, W. "IEN-45: TCP |
| 216 |
Checksum Function Design" (1978) as an appendix.] |
| 217 |
|
| 218 |
|
| 219 |
|
| 220 |
<span class="grey">Zweig & Partridge [Page 4]</span></pre> |
| 221 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span> |
| 222 |
<span class="grey"><a href="./rfc1145">RFC 1145</a> TCP Alternate Checksum Options February 1990</span> |
| 223 |
|
| 224 |
|
| 225 |
[<a id="ref-Fletcher82">Fletcher82</a>] Fletcher, J., "An Arithmetic Checksum for Serial |
| 226 |
Transmissions", IEEE Transactions on Communication, |
| 227 |
Vol. COM-30, No. 1, pp. 247-252, January 1982. |
| 228 |
|
| 229 |
[<a id="ref-Nakassis88">Nakassis88</a>] Nakassis, T., "Fletcher's Error Detection Algorithm: |
| 230 |
How to implement it efficiently and how to avoid the |
| 231 |
most common pitfalls", ACM Computer Communication |
| 232 |
Review, Vol. 18, No. 5, pp. 86-94, October 1988. |
| 233 |
|
| 234 |
[<a id="ref-Sklower89">Sklower89</a>] Sklower, K., "Improving the Efficiency of the OSI |
| 235 |
Checksum Calculation", ACM Computer Communication |
| 236 |
Review, Vol. 19, No. 5, pp. 32-43, October 1989. |
| 237 |
|
| 238 |
Security Considerations |
| 239 |
|
| 240 |
Security issues are not addressed in this memo. |
| 241 |
|
| 242 |
Authors' Addresses |
| 243 |
|
| 244 |
Johnny Zweig |
| 245 |
Digital Computer Lab |
| 246 |
University of Illinois (UIUC) |
| 247 |
1304 West Springfield Avenue |
| 248 |
CAMPUS MC 258 |
| 249 |
Urbana, IL 61801 |
| 250 |
|
| 251 |
Phone: (217) 333-7937 |
| 252 |
|
| 253 |
EMail: [email protected] |
| 254 |
|
| 255 |
|
| 256 |
Craig Partridge |
| 257 |
Bolt Beranek and Newman Inc. |
| 258 |
50 Moulton Street |
| 259 |
Cambridge, MA 02138 |
| 260 |
|
| 261 |
Phone: (617) 873-2459 |
| 262 |
|
| 263 |
EMail: [email protected] |
| 264 |
|
| 265 |
|
| 266 |
|
| 267 |
|
| 268 |
|
| 269 |
|
| 270 |
|
| 271 |
|
| 272 |
|
| 273 |
|
| 274 |
|
| 275 |
|
| 276 |
Zweig & Partridge [Page 5] |
| 277 |
</pre> |