Debian Codegraph @

rfc1044.html

raw
1 <pre>Network Working Group K. Hardwick
2 Request for Comments: 1044 NSC
3 J. Lekashman
4 NASA-Ames GE
5 February 1988
6
7
8 <span class="h1">Internet Protocol on Network Systems HYPERchannel</span>
9 <span class="h1">Protocol Specification</span>
10
11 STATUS OF THIS MEMO
12
13 The intent of this document is to provide a complete discussion of
14 the protocols and techniques used to embed DoD standard Internet
15 Protocol datagrams (and its associated higher level protocols) on
16 Network Systems Corporation's HYPERchannel [<a href="#ref-1">1</a>] equipment.
17 Distribution of this memo is unlimited.
18
19 This document is intended for network planners and implementors who
20 are already familiar with the TCP/IP protocol suite and the
21 techniques used to carry TCP/IP traffic on common networks such as
22 the DDN or Ethernet. No great familiarity with NSC products is
23 assumed; an appendix is devoted to a review of NSC technologies and
24 protocols.
25
26 At the time of this first RFC edition, the contents of this document
27 has already been reviewed by about a dozen vendors and users active
28 in the use of TCP/IP on HYPERchannel media. Comments and suggestions
29 are still welcome (and implementable,) however.
30
31 Any comments or questions on this specification may be directed to:
32
33 Ken Hardwick
34 Director, Software Technology
35 Network Systems Corporation MS029
36 7600 Boone Avenue North
37 Brooklyn Park, MN 55428
38
39 Phone: (612) 424-1607
40
41 John Lekashman
42 Nasa Ames Research Center. NAS/GE
43 MS 258-6
44 Moffett Field, CA, 94035
45 [email protected]
46
47 Phone: (415) 694-4359
48
49
50
51
52 <span class="grey">Hardwick &amp; Lekashman [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
55
56
57 TABLE OF CONTENTS
58
59 Status of this memo . . . . . . . . . . . . . . . . . . . . . . <a href="#page-1">1</a>
60 Goals of this document . . . . . . . . . . . . . . . . . . . . <a href="#page-3">3</a>
61 Basic HYPERchannel network messages . . . . . . . . . . . . . . <a href="#page-4">4</a>
62 Basic (16-bit address) Message Proper header . . . . . . . . . <a href="#page-5">5</a>
63 TO addresses and open driver architecture . . . . . . . . . . <a href="#page-7">7</a>
64 Extended (32-bit address) Message Proper header . . . . . . . . <a href="#page-8">8</a>
65 Address Recognition and message forwarding . . . . . . . . . . <a href="#page-10">10</a>
66 32-bit message fields . . . . . . . . . . . . . . . . . . . . <a href="#page-12">12</a>
67 Broadcasting . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-14">14</a>
68
69 PROTOCOL SPECIFICATION . . . . . . . . . . . . . . . . . . . . <a href="#page-17">17</a>
70 Basic (16-bit) Message Encapsulation . . . . . . . . . . . . <a href="#page-18">18</a>
71 Compatibility with existing implementations . . . . . . . . . . <a href="#page-21">21</a>
72 Extended (32-bit) Message Encapsulation . . . . . . . . . . . <a href="#page-24">24</a>
73 Address Resolution Protocol . . . . . . . . . . . . . . . . . <a href="#page-27">27</a>
74 Maximum Transmission Unit . . . . . . . . . . . . . . . . . . . <a href="#page-31">31</a>
75
76 ADDRESS RESOLUTION . . . . . . . . . . . . . . . . . . . . . . <a href="#page-32">32</a>
77 Local Address Resolution . . . . . . . . . . . . . . . . . . . <a href="#page-33">33</a>
78 Configuration file format . . . . . . . . . . . . . . . . . . <a href="#page-34">34</a>
79 ARP servers . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-35">35</a>
80 Broadcast ARP . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-36">36</a>
81
82 <a href="#appendix-A">Appendix A</a>.
83 NSC Product Architecture and Addressing . . . . . . . . . . . . <a href="#page-38">38</a>
84
85 <a href="#appendix-B">Appendix B</a>.
86 Network Systems HYPERchannel protocols . . . . . . . . . . . . <a href="#page-42">42</a>
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108 <span class="grey">Hardwick &amp; Lekashman [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
111
112
113 GOALS OF THIS DOCUMENT
114
115 In this document, there are four major technical objectives:
116
117 1. To bless a "de facto" standard for IP on HYPERchannel that has
118 been implemented by Tektronix, Cray, NASA Ames, and others.
119 We are attempting to resolve some interoperability problems with
120 this standard so as to minimize the changes to existing IP on
121 HYPERchannel software. If any ambiguities remain in the de facto
122 standard, we wish to assist in their resolution.
123
124 2. To address larger networks, NSC's newer network products are
125 moving to a 32-bit address from the current 16-bit TO address.
126 This document would introduce the addressing extension to the
127 user community and specify how IP datagrams would work in the
128 new addressing mode.
129
130 3. To define an Address Resolution Protocol for HYPERchannel and
131 other NSC products. It is probably well known that current NSC
132 products do not support the broadcast modes that make ARP
133 particularly useful. However, many have expressed interest in
134 "ARP servers" at a known network address. These servers could
135 fade away as NSC products with broadcast capability come into
136 existence. Host drivers that can generate and recognize this
137 ARP protocol would be prepared to take advantage of it as the
138 pieces fall into place.
139
140 4. Part of this effort is to standardize the unofficial "message
141 type" field that reserves byte 8 of the HYPERchannel network
142 message. To permit better interoperability, NSC will initiate a
143 "network protocol registry" where any interested party may
144 obtain a unique value in byte 8 (or bytes 8 and 9) for their own
145 public, private, commercial or proprietary protocol. Lists of
146 assigned protocol type numbers and their "owners" will be
147 periodically published by NSC and would be available to
148 interested parties.
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164 <span class="grey">Hardwick &amp; Lekashman [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
167
168
169 BASIC HYPERCHANNEL NETWORK MESSAGES
170
171 Unlike most datagram delivery systems, the HYPERchannel network
172 message consists of two parts:
173
174 Message Proper
175 +--------------------+
176 | |
177 | |
178 | 10-64 bytes |
179 | |
180 | |
181 +--------------------+
182
183 Associated Data
184 +----------------------------------------------------+
185 | |
186 | |
187 | |
188 | |
189 | Unlimited length |
190 | |
191 | |
192 | |
193 | |
194 +----------------------------------------------------+
195
196 The first part is a message header that can be up to 64 bytes in
197 length. The first 10 bytes contain information required for the
198 delivery of the entire message, and the remainder can be used by
199 higher level protocols. The second part of the message, the
200 "Associated Data," can be optionally included with the message
201 proper. In most cases (transmission over HYPERchannel A trunks), the
202 length of the associated data is literally unlimited. Others (such
203 as HYPERchannel B or transmission within a local HYPERchannel A A400
204 adapter) limit the size of the Associated Data to 4K bytes. If the
205 information sent can be contained within the Message Proper, then the
206 Associated Data need not be sent.
207
208 HYPERchannel lower link protocols treat messages with and without
209 Associated Data quite differently; "Message only" transmissions are
210 sent using abbreviated protocols and can be queued in the receiving
211 network adapter, thus minimizing the elapsed time needed to send and
212 receive the messages. When associated data is provided, the
213 HYPERchannel A adapters free their logical resources towards driving
214 the host interface and coaxial trunks.
215
216
217
218
219
220 <span class="grey">Hardwick &amp; Lekashman [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
223
224
225 BASIC (16-BIT ADDRESS) MESSAGE PROPER HEADER
226
227 The first 10 bytes of the network Message Proper are examined by the
228 network adapters to control delivery of the network message. Its
229 format is as follows:
230
231 byte Message Proper
232 +------------------------------+-----------------------------+
233 0 | Trunks to Try | Message Flags |
234 | TO trunks | FROM trunks | |EXC|BST|A/D|
235 +--------------+---------------+-----------------+---+---+---+
236 2 | Access code |
237 | |
238 +------------------------------+-----------------------------+
239 4 | Physical addr of | | TO Port |
240 | destination adapter (TO) | | number |
241 +------------------------------+-----------------------------+
242 6 | Physical addr of source | |FROM port|
243 | adapter (FROM) | | number |
244 +------------------------------+-----------------------------+
245 8 | Message type |
246 | |
247 +------------------------------+-----------------------------+
248 10 | |
249 | Available for higher level protocols |
250 | |
251 | |
252 +------------------------------+-----------------------------+
253
254 TRUNKS TO TRY
255
256 Consists of two four bit masks indicating which of four possible
257 HYPERchannel A coaxial data trunks are to be used to transmit the
258 message and to return it. If a bit in the mask is ON, then the
259 adapter firmware will logically AND it with the mask of installed
260 trunk interfaces and use the result as a candidate list of
261 interfaces. Whenever one of the internal "frames" are sent to
262 communicate with the destination adapter, the transmission hardware
263 electronically selects the first non-busy trunk out of the list of
264 candidates. Thus, selection of a data trunk is best performed by the
265 adapter itself rather than by the host. "Dedicating" trunks to
266 specific applications only makes sense in very critical real time
267 applications such as streaming data directly from high speed
268 overrunnable peripherals.
269
270 A second Trunk mask is provided for the receiving adapter when it
271 sends frames back to the transmitter, as it is possible to build
272 "asymmetric" configurations of data trunks where trunk 1 on one box
273
274
275
276 <span class="grey">Hardwick &amp; Lekashman [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
279
280
281 is connected to the trunk 3 interface of a second. Such
282 configurations are strongly discouraged, but the addressing structure
283 supports it if needed.
284
285 The "trunks to try" field is only used by HYPERchannel A. To assure
286 maximum interoperability, a value of 0xFF should be placed in this
287 field to assure delivery over any technology. Other values should
288 only be used if the particular site hardware is so configured to not
289 be physically connected via those trunks.
290
291 MESSAGE FLAGS
292
293 Contains options in message delivery. In the basic type of message,
294 three bits are used:
295
296 ASSOCIATED DATA PRESENT (A/D) is ON if an Associated Data block
297 follows the Message Proper. 0 if only a message proper is present in
298 the network message. The value of this bit is enforced by the
299 network adapter firmware.
300
301 BURST MODE (BST) Enables a special mode for time critical transfers
302 where a single HYPERchannel A coaxial trunk is dedicated during
303 transmission of the network message. Not recommended for anything
304 that won't cause peripheral device overruns if data isn't delivered
305 once message transmission starts.
306
307 EXCEPTION (EXC) Indicates to some channel programmed host interfaces
308 that the message is "out of band" in some way and requires special
309 processing.
310
311 ACCESS CODE
312
313 A feature to permit adapters to share use of a cable yet still permit
314 an "access matrix" of which adapter boxes and physically talk to
315 which others. Not currently in use by anyone, support is being
316 discontinued.
317
318 TO ADDRESS
319
320 Consists of three parts. The high order 8-bits contains the physical
321 address of the network adapter box which is to receive the message.
322 The low order 8-bits are interpreted in different ways depending on
323 the nature of the receiving network adapter. If the receiving
324 adapter has different host "ports," then the low order bits of the TO
325 field are used to designate which interface is to receive the
326 message. On IBM data channels, the entire "logical" TO field is
327 interpreted as the subchannel on which the incoming data is to be
328 presented. Parts of the logical TO field that are not interpreted by
329
330
331
332 <span class="grey">Hardwick &amp; Lekashman [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
335
336
337 the network adapter are passed to the host for further
338 interpretation.
339
340 FROM ADDRESS
341
342 The FROM address is not physically used during the process of
343 transmitting a network message, but is passed through to the
344 receiving host so that a response can be returned to the point of
345 origin. In general, reversing the TO and FROM 16-bit address fields
346 and the TO and FROM trunk masks can reliably return a message to its
347 destination.
348
349 MESSAGE TYPE
350
351 The following two bytes are reserved for NSC. Users have been
352 encouraged to put a zero in byte 8 and anything at all in byte 9 so
353 as to not conflict with internal processing of messages by NSC
354 firmware. In the past, this field has been loosely defined as
355 carrying information of interest to NSC equipment carrying the
356 message and not as a formal protocol type field. For example, 0xFF00
357 in bytes 8 and 9 of the message will cause the receiving adapter to
358 "loop back" the message without delivering it to the attached host.
359
360 Concurrent with this document, it is NSC's intent to use both bytes 8
361 and 9 as a formal "protocol type" designator. Major protocols will
362 be assigned a unique value in byte 8 that will (among good citizens)
363 not duplicate a value generated by a different protocol. Minor
364 protocols will have 16-bit values assigned to them so that we won't
365 run out when 256 protocols turn up. Any interested party could
366 obtain a protocol number or numbers by application to NSC. In this
367 document, protocol types specific to IP protocols are assigned.
368
369 TO ADDRESSES AND OPEN DRIVER ARCHITECTURE
370
371 Since not all 16-bits of the TO address are used for the physical
372 delivery of the network message, the remainder are considered
373 "logical" in that their meaning is physically determined by host
374 computer software or (in cases such as the FIPS data channel) by
375 hardware in the host interface.
376
377 Since HYPERchannel is and will be used to support a large variety of
378 general and special purpose protocols, it is desirable that several
379 independent protocol servers be able to independently share the
380 HYPERchannel network interface. The implementation of many of NSC's
381 device drivers as well as those of other parties (such as Cray
382 Research) support this service. Each protocol server that wishes to
383 send or receive HYPERchannel network messages logically "connects" to
384 a HYPERchannel device driver by specifying the complete 16-bit TO
385
386
387
388 <span class="grey">Hardwick &amp; Lekashman [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
391
392
393 address it will "own" in the sense that any network message with that
394 TO address will be delivered to that protocol server.
395
396 The logical TO field serves a function similar to the TYPE byte in
397 the Ethernet 802.2 message header, but differs from it in that the
398 width of the logical TO field varies from host to host, and that no
399 values of the logical TO address are reserved for particular
400 protocols. On the other hand, it is possible to have several
401 "identical" protocols (such as two independent copies of IP with
402 different HYPERchannel addresses) sharing the same physical
403 HYPERchannel interface. This makes NSC's addressing approach
404 identical to the OSI concept that the protocol server to reach is
405 embedded within the address, rather than the IP notion of addressing
406 a "host" and identifying a server through a message type.
407
408 Since the HYPERchannel header also has a "message type" field, there
409 is some ambiguity concerning the respective roles of the message type
410 and logical TO fields:
411
412 o The logical TO field is always used to identify the protocol
413 server which will receive the message. Once a server has
414 specified the complete TO address for the messages it wishes to
415 receive, the message will not be delivered to a different
416 protocol server regardless of the contents of the message type
417 field.
418
419 o Although the "type" field cannot change the protocol server at
420 the final destination of the message, the type field can be used
421 by intermediate processes on the network to process the message
422 before it reaches the server destination. An obvious example is
423 the 0xFF00 message loopback type function, where network
424 processing to loop back the message results in nondelivery to
425 the TO address. In the future, intermediate nodes may process
426 "in transit" messages based on the message type only for
427 purposes such as security validation, aging of certain
428 datagrams, and network management.
429
430 EXTENDED (32-BIT ADDRESS) MESSAGE PROPER HEADER
431
432 In the original days of HYPERchannel, the limitation of 256 adapter
433 "boxes" that could be addressed in a network message was deemed
434 sufficient as 40 or so adapters was considered a "large" network. As
435 with the Ethernet, more recent networks have resulted in a need to
436 address larger networks. Although a few ad hoc modes have existed to
437 address larger HYPERchannel networks for some years, newer
438 technologies of HYPERchannel equipment have logically extended the
439 network message to support 32-bits of addressing, with 24 of those
440 bits to designate a physical network adapter.
441
442
443
444 <span class="grey">Hardwick &amp; Lekashman [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
447
448
449 This 32-bit header has been designed so that existing network
450 adapters are capable of sending and receiving these messages. Only
451 the network bridges need the intelligence to select messages
452 designated for them.
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500 <span class="grey">Hardwick &amp; Lekashman [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
503
504
505 +------------------------------+-----------------------------+
506 0 | Trunks to Try | Message Flags |
507 | TO trunks | FROM trunks |GNA|CRC| |SRC|EXC|BST|A/D|
508 +--------------+---------------+---+---+--+--+---+---+---+---+
509 2 | TO Domain # | TO Network # |
510 | | |
511 +------------------------------+-----------------------------+
512 4 |O| Physical addr of | | TO Port |
513 |N| destination adapter (TO) | | number |
514 +------------------------------+-----------------------------+
515 6 |O| Physical addr of source | |FROM port|
516 |N| adapter (FROM) | | number |
517 +------------------------------+-----------------------------+
518 8 | Message type |
519 | |
520 +------------------------------+-----------------------------+
521 10 | FROM Domain # | FROM Network # |
522 | | |
523 +------------------------------+-----------------------------+
524 12 | - reserved - | age count |
525 | | |
526 +------------------------------+-----------------------------+
527 14 | Next Header Offset | Header End Offset |
528 | (normally 16) | (normally 16) |
529 +------------------------------+-----------------------------+
530 16 | Start of user protocol |
531 | bytes 16 - 64 of message proper |
532 | |
533 +------------------------------+-----------------------------+
534
535 Associated Data
536 +-----------------------------------------------------------------+
537 | |
538 | As with basic format network messages |
539 | |
540 +-----------------------------------------------------------------+
541
542 ADDRESS RECOGNITION AND MESSAGE FORWARDING
543
544 With the 32-bit form of addressing, NSC is keeping with the premise
545 that the native HYPERchannel address bears a direct relation to the
546 position of the equipment in an extended HYPERchannel network.
547
548 Each collection of "locally" attached NSC network adapters that are
549 connected by coax or fiber optic cable (with the possible addition of
550 nonselective repeaters such as the ATRn series) is considered a
551 "network". Each network can have up to 256 directly addressable
552 adapters attached to it which can be reached by the basic format
553
554
555
556 <span class="grey">Hardwick &amp; Lekashman [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
559
560
561 network message.
562
563 Existing bridges or "link adapters" can be programmed to become
564 "selective repeaters" in that they can receive network messages
565 containing a subset of network addresses send them over the bridge
566 medium (if present) and reintroduce them on the other network. Such
567 interconnected local area networks are considered a single network
568 from an addressing point of view.
569
570 A large NSC network can have up to 64K networks which can be
571 complexly interconnected by network bridges and/or "backbone"
572 networks which distribute data between other networks. To simplify
573 the mechanics of message forwarding, the 16-bit network field is
574 divided into two eight quantities, a "network number" identifying
575 which network is to receive the message and a "domain number" which
576 specifies which network of networks is the recipient.
577
578 The bridge technology adapters which move messages between networks
579 have address recognition hardware which examines all the 24-bits in
580 bytes 2-5 of the network message header to determine if the bridge
581 should accept the message for forwarding. At any given instant of
582 time in the network, each bridge will have a list of networks and
583 domains that it should accept for forwarding to a network at the
584 other end of the bridge. Each Adapter (Including Newer Technology
585 host adapters) contains in address recognition hardware:
586
587 o domainmask -- a 256-bit mask of domain numbers that should be
588 accepted for forwarding (not local processing) by this adapter.
589 o MyDomain -- the value of the domain on which this host
590 adapter or bridge end is installed.
591 o NetworkMask -- a 256-bit mask of network numbers that should be
592 accepted for forwarding by this adapter.
593 o MyNetwork - the value of the network on which this host
594 adapter or bridge end is installed.
595 o AddressMask -- A 256-bit mask of the local network addresses
596 that should be accepted by the adapter.
597 o MyAddress -- the "base address" of the box, which must be
598 supplied in any message that is directed to control processes
599 within the adapter, such as a loopback message.
600
601 Address recognition takes place using the algorithm:
602
603 IF Domain IN DomainMask OR
604 IF (Domain = MyDomain AND Network IN NetworkMask) OR
605 IF (Domain = MyDomain AND Network = MyNetwork AND
606 Address IN AddressMask) THEN accept-message
607 ELSE ignore-message.
608
609
610
611
612 <span class="grey">Hardwick &amp; Lekashman [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
615
616
617 This algorithm means that an adapter's hardware address recognition
618 logic will accept any messages to the box itself, any secondary or
619 aliased local addresses owned by the adapter, and any message
620 directed to a remote network or domain that that particular adapter
621 is prepared to forward.
622
623
624 32-BIT MESSAGE FIELDS
625
626 TRUNK MASK
627
628 Is as in the basic network message. Messages that are to be
629 delivered outside the immediate network should have 0xFF in this byte
630 so that all possible trunks in intermediate networks should be tried.
631 Locally delivered 32-bit messages may still contain specially
632 tailored trunk masks to satisfy local delivery needs.
633
634 MESSAGE FLAGS
635
636 The currently defined bits remain as before. Three new bits have
637 been defined since that time.
638
639 CRC (END-END MESSAGE INTEGRITY). Newer technology host adapters are
640 capable of generating a 32-bit CRC for the entire network message as
641 soon as it is received over the channel or bus interface from the
642 host. This 32-bit CRC is appended to the end of the associated data
643 block and is preserved through the entire delivery process until it
644 is checked by the host adapter that is the ultimate recipient of the
645 message, which removes it. This end to end integrity checking is
646 designed to provide a high degree of assurance that data has been
647 correctly moved through all intermediate LAN's, geographic links, and
648 internal adapter hardware and processes.
649
650 SRC (SOURCE FROM ADDRESS CORRECT). This bit is provided to take
651 advantage of the physical nature of the network address to optionally
652 verify that the 32-bit FROM address provided in the network message
653 is in fact the location that the message originated. If the bit is
654 not set by the transmitting host, no particular processing occurs on
655 the message. If the bit is set, then all intermediate adapters
656 involved in the delivery of the message have the privilege of turning
657 the bit off if the received message FROM address is not a TO address
658 that would be delivered to the originator if the message were going
659 the opposite direction.
660
661 If the message is received by a host computer with this bit still
662 set, then the FROM address is guaranteed correct in the sense that
663 returning a message with TO and FROM information reversed will result
664 in delivery of the message to the process that actually originated
665
666
667
668 <span class="grey">Hardwick &amp; Lekashman [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
671
672
673 it. By careful attention to the physical security of adapters and
674 intermediate links between networks, a high degree of security can be
675 built into systems that simply examine the FROM address of a message
676 to determine the legitimacy of its associated request.
677
678 GNA (GLOBAL NETWORK ADDRESSING). This bit ON indicates that 32-bit
679 addressing is present in the message. When this bit is on, bytes 2-3
680 (Domain and Network numbers) should also be nonzero.
681
682 TO ADDRESS
683
684 Four bytes contain the TO address, which is used to deliver the
685 network message as described in "Address Recognition and Message
686 Forwarding" on page 8. The "logical" part of the TO address is used
687 to designate a protocol server exactly as in the basic format network
688 message header.
689
690 The existing "address" field has its high order bit reserved as an
691 outnet bit for compatibility with existing A-series network adapter
692 equipment. Were it not for this bit, the A-series adapters would
693 attempt to accept messages that were "passing through" the local
694 network on their way elsewhere simply because the address field
695 matched while the the Domain and Network numbers (ignored by the A-
696 series adapters) were quite different.
697
698 This "outnet" bit is used in the following way:
699
700 o All network adapters (of any type) in an extended set of
701 networks containing A-Series adapters that will ever use 32-bit
702 addressing must have their addresses in the range 00-7F (hex.)
703
704 o If a message is to be sent to a destination on a nonlocal
705 network and domain on such an extended network, then the
706 high order bit of the address field is turned on.
707
708 o When the last bridge in the chain realizes that it is about to
709 forward the message to its final destination (the Domain and
710 Network numbers are local), then it turns the Outnet bit off.
711 This will result in local delivery to the destination adapter.
712
713 FROM ADDRESS
714
715 The FROM address follows the same logic as the TO address in that any
716 message can be returned to its source by reversing the FROM and TO
717 fields of the message. Since so many protocols examine byte 8 of the
718 message to determine its type, the FROM field has been split so that
719 the Domain and Network numbers extend into bytes 10-11.
720
721
722
723
724 <span class="grey">Hardwick &amp; Lekashman [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
727
728
729 MESSAGE TYPE
730
731 This field (informally defined in the past) has been extended to 16-
732 bits so that a unique value can be assigned to any present or future
733 protocol which is layer on HYPERchannel messages for either private
734 or public use.
735
736 AGE COUNT
737
738 This field serves the same purpose as the IP "time to live" in that
739 it prevents datagrams from endlessly circulating about in an
740 improperly configured network. Each time a 32-bit message passes
741 through a bridge, the Age Count is decremented by one. When the
742 result is zero, the message is discarded by the bridge.
743
744 NEXT HEADER OFFSET AND HEADER END OFFSET
745
746 These are used as fields to optionally provide "loose source
747 routing", where a list of 32-bit TO addresses can be provided by the
748 transmitter to explicitly determine the path of a message through the
749 network. If this feature is not used, both these fields would
750 contain the value 16 (decimal) to both indicate extra TO addresses
751 are absent and that the beginning of protocol data following the
752 HYPERchannel header is in byte 16.
753
754 Although it is conceivable that a HYPERchannel IP process could use
755 this source routing capability to direct messages to hosts or
756 gateways, this capability is not felt to be of sufficient value to IP
757 to build it into a HYPERchannel IP protocol.
758
759 In the future, all higher level protocols should be able to examine
760 Header End Offset to determine the start of the higher level protocol
761 information.
762
763 BROADCASTING
764
765 NSC message forwarding protocols use low level link protocols to
766 negotiate transmission of a message to its next destination on the
767 network. Furthermore, NSC network boxes often "fan out" so that
768 several hosts share the same network transmission equipment as in the
769 A400 adapter. Both these characteristics mean that providing a
770 genuine broadcast capability is not a trivial task, and in fact no
771 current implementations of NSC technology support a broadcast
772 capability.
773
774 The last several years have seen broadcast applications mature to the
775 point where they have virtually unquestioned utility on a local and
776 sometimes campuswide basis. Accordingly, new NSC technologies will
777
778
779
780 <span class="grey">Hardwick &amp; Lekashman [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
783
784
785 support a broadcast capability. Information on the use of this
786 capability is included here as it is essential to the discussion of
787 the Address Resolution Protocol later in this document.
788
789 Broadcast capability will be supported only with the extended (32-bit
790 address) message format. A broadcast message will have the following
791 general appearance:
792
793 byte Message Proper
794 +------------------------------+-----------------------------+
795 0 | Trunks to Try | Message Flags |
796 | TO trunks | FROM trunks |GNA|CRC| |SRC|EXC|BST|A/D|
797 +--------------+---------------+---+---+--+--+---+---+---+---+
798 2 | TO Domain Number | TO Network Number |
799 | or 0xFF | or 0xFF |
800 +------------------------------+-----------------------------+
801 4 | 0xFF | Broadcast channel number |
802 | | |
803 +------------------------------+-----------------------------+
804 6 |O| Physical addr of source | |FROM port|
805 |N| adapter (FROM) | | number |
806 +------------------------------+-----------------------------+
807 8 | Message type |
808 | |
809 +------------------------------+-----------------------------+
810 10 | FROM Domain Number | FROM Network Number |
811 | | |
812 +------------------------------+-----------------------------+
813 12 | - reserved - | age count |
814 | | |
815 +------------------------------+-----------------------------+
816 14 | Next Header Offset | Header End Offset |
817 | (normally 16) | (normally 16) |
818 +------------------------------+-----------------------------+
819 16 | Start of user protocol |
820 | bytes 16 - 64 of message proper |
821 | |
822 +------------------------------+-----------------------------+
823 Associated Data
824 +-----------------------------------------------------------------+
825 | |
826 | As with basic format network messages |
827 | Maximum associated data size 1K bytes. |
828 | |
829 +-----------------------------------------------------------------+
830
831
832
833
834
835
836 <span class="grey">Hardwick &amp; Lekashman [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
839
840
841 TRUNKS TO TRY AND MESSAGE FLAGS
842
843 These fields are defined just as with a normal 32-bit message. All
844 bits in the Message Flags field are valid with broadcast modes.
845
846 BROADCAST ADDRESS
847
848 For Domain, Network and Adapter Address fields, the value 0xFF is
849 reserved for use by the broadcast mechanism. A value of 0xFF in the
850 adapter address field indicates to the local network hardware that
851 this message is to be sent to all connected network equipment on the
852 individual network.
853
854 A value of 0xFF in the network or domain fields, respectively
855 indicates a request that the scope of the broadcast exceed the local
856 network. The bridging link adapters will receive the broadcast
857 message along with everyone else and will examine the "Broadcast
858 Channel" field and their internal switches to determine if the
859 message should be forwarded to other remote networks.
860
861 If the Network and Domain fields contain the local network and
862 domain, then the broadcast message will only be broadcast within the
863 local network. If a remote Network and Domain is specified, then the
864 message will be delivered as a single message to the remote network
865 and broadcast there.
866
867 BROADCAST CHANNEL
868
869 Since individual hosts and protocol servers generally are not
870 interested in all broadcast messages that float about the network, a
871 filtering mechanism is provided in the header and network adapter
872 equipment so that only proper classes of broadcast messages are
873 delivered to the end point.
874
875 Broadcast channel numbers in the range 00-0xFF will be assigned by
876 NSC much like the "message type" field. Host protocol servers
877 specify a specific TO address containing a channel number (such as
878 0xFF04) when they bind themselves to the HYPERchannel device driver.
879 The driver and the underlying equipment will deliver only broadcast
880 messages with the correct channel number to the protocol server. If
881 a protocol server wishes to receive several different broadcast
882 messages, it must bind itself to the driver several times with the
883 desired addresses.
884
885 Link adapters that are prepared to handle multinetwork broadcast
886 messages may be equipped with switches to determine which broadcast
887 channels will be propagated into the next network. Since
888 multinetwork broadcast is an arrangement that must be configured with
889
890
891
892 <span class="grey">Hardwick &amp; Lekashman [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
895
896
897 care, these switches are off by default.
898
899 FROM ADDRESS
900
901 The FROM address is constructed just as with a normal 32-bit network
902 message. The Source Address Correct bit is processed just as with a
903 normal message.
904
905 MESSAGE TYPE
906
907 Message type is defined as with normal messages. Presumably
908 broadcast applications will have unique message types that are not
909 generally found in normal messages.
910
911 AGE COUNT
912
913 Age count is vitally important in a multinetwork broadcast as "loops"
914 in the network can cause a great deal of activity until all the
915 progeny of the original broadcast message die out.
916
917 PROTOCOL SPECIFICATION
918
919 This section contains information on the technique used to
920 encapsulate IP datagrams on the HYPERchannel network message. It
921 contains three sections to describe three protocol packagings:
922
923 o The technique used to encapsulate IP datagrams on the basic
924 16-bit network message. This is a de facto standard that has
925 been in use for several years and is documented here to make it
926 official.
927
928 o The encapsulation technique for IP datagrams on 32 bit network
929 messages.
930
931 o The definition of an Address Resolution Protocol on
932 HYPERchannel.
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948 <span class="grey">Hardwick &amp; Lekashman [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
951
952
953 BASIC (16-BIT) MESSAGE ENCAPSULATION
954
955 Message Proper
956 +------------------------------+-----------------------------+
957 0 | Trunks to Try | Message Flags |
958 | TO trunks | FROM trunks |GNA|CRC| |SRC|EXC|BST|A/D|
959 +------------------------------+-----------------------------+
960 2 | Access code 0000 |
961 | (no longer supported) |
962 +------------------------------+-----------------------------+
963 4 | Physical addr of | Protocol server |Dest Port|
964 | destination adapter | logical address | number |
965 +------------------------------+-----------------------------+
966 6 | Physical addr of | Originating | Src Port|
967 | source adapter | server address | number |
968 +------------------------------+-----------------------------+
969 8 | IP on HYPERchannel | Offset to start of IP |
970 | type code 0x05 | header from message start |
971 +------------------------------+-----------------------------+
972 10 | IP type designator | Offset to start of IP |
973 | 0x34 | header from byte 12 |
974 +------------------------------+-----------------------------+
975 12 | Padding (variable length incl. zero bytes) |
976 | |
977 +------------------------------+-----------------------------+
978 Off | First (64-Offset) bytes of IP datagram |
979 | |
980 | |
981 | |
982 +------------------------------+-----------------------------+
983 Associated Data
984 +------------------------------+-----------------------------+
985 | |
986 | Remainder of IP datagram |
987 | |
988 | No associated data is present if IP |
989 | datagram fits in the Message Proper |
990 | |
991 +------------------------------+-----------------------------+
992
993 TRUNK MASK
994
995 From the vantage of an IP driver, any trunk mask is valid so long as
996 it results in successful delivery of the HYPERchannel network message
997 to its destination. There is no reason to check this field for
998 validity on reception of the message. Specification of the Trunk
999 Mask on output is a local affair that could be specified by the
1000 transmitting driver's address resolution tables.
1001
1002
1003
1004 <span class="grey">Hardwick &amp; Lekashman [Page 18]</span></pre>
1005 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
1006 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1007
1008
1009 MESSAGE FLAGS
1010
1011 No use is made of the Flags field (byte 1) other than to
1012 appropriately set the Associated Data bit. Burst Mode and the
1013 Exception bit should not be used with IP.
1014
1015 ACCESS CODE
1016
1017 Although some current implementations of IP on HYPERchannel support
1018 the access code, no one appears to be using it at the current time.
1019 Since this field is currently reserved for the use of 32-bit
1020 addresses, no value other than 0000 should be placed in this field.
1021
1022 TO ADDRESS
1023
1024 The TO field is generally obtained by a local IP driver through a
1025 table lookup algorithm where a 16-bit TO address is found that
1026 corresponds to the IP address of a local host or gateway. The high
1027 order bits of the TO address of course refer to the adapter number
1028 the adapter attached to the destination host.
1029
1030 The logical TO field should contain the protocol server address of
1031 the HYPERchannel IP driver for that host as determined by the host's
1032 system administrator. Many HYPERchannel TCP/IP drivers in the field
1033 today are not "open" in that any network message delivered to that
1034 host will be presumed to be an IP datagram regardless of the logical
1035 TO field; however any transmitting IP process should be capable of
1036 generating the entire 16-bit TO field in order to generate a message
1037 capable of reaching a destination IP process.
1038
1039 The process of determining which HYPERchannel address will receive an
1040 IP datagram based on its IP address is a major topic that is covered
1041 in "Address Resolution".
1042
1043 FROM ADDRESS
1044
1045 The FROM address is filled in with the address that the local driver
1046 expects to receive from the network, but no particular use is make of
1047 the FROM address.
1048
1049 MESSAGE TYPE
1050
1051 Network Systems requests that a value of 5 (decimal) be placed in
1052 this byte to uniquely indicate that the network message is being used
1053 to carry IP traffic. No other well-behaved protocol using
1054 HYPERchannel should duplicate this value of 5.
1055
1056
1057
1058
1059
1060 <span class="grey">Hardwick &amp; Lekashman [Page 19]</span></pre>
1061 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
1062 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1063
1064
1065 Many current implementations of IP on HYPERchannel place a zero or
1066 other values in this field simply because no value was reserved for
1067 IP usage. Transmitting versions of IP should always place a 5 in
1068 this field; receiving IP's should presume a delivered message to be
1069 an IP datagram until proven otherwise regardless of the contents of
1070 the Message Type field.
1071
1072 Developers should note that it is often convenient to permit
1073 reception of the value 0xFF00 in bytes 8 and 9 of the IP datagram.
1074 Transmitting a message with this value will cause it to be looped
1075 back at the destination adapter and returned to the protocol server
1076 designate in the FROM address. This permits the developer have host
1077 applications talk to others on the same host for purposes of network
1078 interface or other protocol debugging.
1079 IP HEADER OFFSET
1080
1081 Byte 9 contains the offset to the start of the IP header within the
1082 message proper, such that the Message Proper address plus the IP
1083 header offset generates the address of the first byte of the IP
1084 header (at least on byte addressable machines.)
1085
1086 This field is redundant with the offset field in byte 11, and is
1087 present for cosmetic compatibility with 32-bit implementations. On
1088 reception, the value in byte 11 should take precedence.
1089
1090 As part of the migration to larger HYPERchannel headers, this field
1091 will become significant with the 32-bit addressing format, as the
1092 length of the header is no longer 10 bytes and byte 11 is used for
1093 other purposes.
1094
1095 IP TYPE DESIGNATOR
1096
1097 Early implementations of IP drivers on HYPERchannel wanted to leave
1098 bytes 8 and 9 alone for NSC use and place a "message type" field in
1099 later in the message. A value of 0x34 had been selected by earlier
1100 developers for reasons that are now of only historical interest.
1101 Once again, implementations should generate this value on
1102 transmission, but not check it on input, assuming that an IP datagram
1103 is present in the message.
1104
1105 IP HEADER OFFSET
1106
1107 This value is used by a number of commercial implementations of IP on
1108 HYPERchannel to align the start of the IP header within the network
1109 message. This offset is relative to byte 12 of the network message
1110 so that a value of zero indicates that the IP header begins in byte
1111 12. This value should be both correctly generated on transmission,
1112 and always respected on input processing.
1113
1114
1115
1116 <span class="grey">Hardwick &amp; Lekashman [Page 20]</span></pre>
1117 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-21" ></span>
1118 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1119
1120
1121 The maximum permissible offset in this field is 52 indicating that
1122 the IP header begins at the start of the associated data block.
1123
1124 IP DATAGRAM CONTENTS
1125
1126 Beginning at the offset designated in byte 11, the IP datagram is
1127 treated as a contiguous block of data that flows from byte 63 of the
1128 message proper into the first byte of associated data, so that the
1129 entire message plus data is treated as a single contiguous block.
1130
1131 If the IP header is small enough to fit within the entire network
1132 message, then only the message proper is transmitted. The length of
1133 the message proper sent should always be 64 bytes, even if the IP
1134 datagram and HYPERchannel header do not occupy all 64 bytes of the
1135 message proper.
1136
1137 If the datagram flows over into the associated data, then both
1138 message and data are sent. Since a number of machines cannot send a
1139 length of data to the HYPERchannel that is an exact number of bytes
1140 (due to 16-64 bits on the channel bus,) the length of the associated
1141 data received should not be used as a guide to the length of the IP
1142 datagram -- this should be extracted from the IP header. A driver
1143 should verify, of course, that the associated data received is at
1144 least as long as is needed to hold the entire IP datagram.
1145
1146 COMPATIBILITY WITH EXISTING IMPLEMENTATIONS
1147
1148 The basic format described here is clearly a compromise between
1149 several implementations of IP on HYPERchannel. Not all existing
1150 implementations are interoperable with the standard described above.
1151 Currently there are two known "families" of IP HYPERchannel drivers
1152 in existence:
1153
1154 THE "CRAY-NASA AMES" PROTOCOL
1155
1156 This protocol is in the widest production use and has the largest
1157 number of supported drivers in existence. It is interoperable and
1158 identical with the standard described above with the sole exception
1159 that bytes 8 and 9 are set to zero by these drivers. As these bytes
1160 are ignored by most implementations of this driver, they have been
1161 assigned values to formalize the use of the message type field and to
1162 make it consistent with the 32-bit protocol.
1163
1164 THE "TEKTRONIX-BERKELEY" PROTOCOL
1165
1166 This protocol was historically the first IP on HYPERchannel
1167 implementation developed (at Tektronix) and subsequently made its way
1168 to Berkeley and BSD UNIX. This protocol is not interoperable with
1169
1170
1171
1172 <span class="grey">Hardwick &amp; Lekashman [Page 21]</span></pre>
1173 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-22" ></span>
1174 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1175
1176
1177 the standard described above due to several distinct differences.
1178
1179 First, bytes 8 through 11 are always zero. The IP header always
1180 starts on byte 12. Comments in some of these drivers designate byte
1181 11 as an "IP header offset" field, but apparently this value is never
1182 processed.
1183
1184 The major difference (and the incompatibility) concerns the packaging
1185 of the IP datagram into the network message. Due to historical
1186 difficulties in the early 80's with the sending and receiving of very
1187 small blocks of associated data on VAXes, this protocol the takes a
1188 curious approach to the placement of the IP header and the headers of
1189 higher level protocols (such as TCP or UDP.)
1190
1191 o If the entire length of the IP datagram is 54 bytes or less,
1192 it is possible to fit the entire datagram and the HYPERchannel
1193 header in the 64 byte message proper. In this case, no
1194 associated data is sent; only a message proper is used to carry
1195 the data. The length of the message proper transmitted is the
1196 exact length needed to enclose the IP datagram; no padding bytes
1197 are sent at the end of the message.
1198
1199 o If the length of the IP header is greater than 54 bytes, then:
1200
1201 - All higher level protocol information (TCP/UDP header and
1202 their associated data fields) are placed in the associated
1203 data block, with the TCP/UDP header beginning at the start
1204 of the associated data block.
1205
1206 - On transmission, the length of the message proper
1207 transmitted is set to the length of the HYPERchannel header
1208 plus the IP header -- it is not padded out to 64 bytes.
1209 The length of the associated data sent should be sufficient
1210 to accommodate the TCP/UDP header and its data fields.
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228 <span class="grey">Hardwick &amp; Lekashman [Page 22]</span></pre>
1229 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-23" ></span>
1230 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1231
1232
1233 WHICH PROTOCOL IS BEST?
1234
1235 In choosing which to follow, the "Cray-Ames" approach was taken for
1236 several reasons:
1237
1238 1. Cray Research has performed exemplary work in dealing with other
1239 vendors to provide IP on HYPERchannel from the Cray computers to
1240 other hosts. As a result, there are 4 or 5 vendor supported
1241 implementations of IP on HYPERchannel that use this approach.
1242
1243 2. The two part structure of the message proper has its uses when a
1244 machine wishes to make protocol decisions before staging the
1245 transfer of an immense block of associated data into memory.
1246 Many network coprocessors and intelligent I/O subsystems find it
1247 simpler to read in the entire network message before deciding
1248 what to do with it. Arbitrarily catenating the two components
1249 does this best and permits streaming of messages from future
1250 technology network adapters.
1251
1252 3. Some TCP users (mostly secure DoD sites) intend to load up IP
1253 datagrams with optional fields in the future. The
1254 Tektronix-Berkeley implementation has problems if the IP header
1255 length exceeds 54 bytes.
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284 <span class="grey">Hardwick &amp; Lekashman [Page 23]</span></pre>
1285 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-24" ></span>
1286 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1287
1288
1289 EXTENDED (32-BIT) MESSAGE ENCAPSULATION
1290
1291 Message Proper
1292 +------------------------------+-----------------------------+
1293 0 | Trunks to Try |1| Message Flags |
1294 | TO trunks | FROM trunks |GNA|CRC| |SRC|EXC|BST|A/D|
1295 +------------------------------+-----------------------------+
1296 2 | Destination Domain | Destination Network |
1297 | Number | Number |
1298 +------------------------------+-----------------------------+
1299 4 |O| Physical addr of | Protocol server |Dest Port|
1300 |N| destination adapter | logical address | number |
1301 +------------------------------+-----------------------------+
1302 6 |O| Physical addr of | Originating | Src Port|
1303 |N| source adapter | server address | number |
1304 +------------------------------+-----------------------------+
1305 8 | IP on HYPERchannel | Offset to start of IP |
1306 | type code 0x06 | datagram header |
1307 +------------------------------+-----------------------------+
1308 10 | Source Domain Number | Source Network Number |
1309 | | |
1310 +------------------------------+-----------------------------+
1311 12 | - reserved - | Age Count |
1312 +------------------------------+-----------------------------+
1313 14 | Next Header Offset | Header End Offset |
1314 | | (usually 16) |
1315 +------------------------------+-----------------------------+
1316 16 | Padding to IP header start (usually 0 bytes) |
1317 | |
1318 +------------------------------+-----------------------------+
1319 Off| Entire IP datagram if datagram length &lt;= (64-Offset) |
1320 | |
1321 | else first (64-Offset) bytes of IP datagram |
1322 +------------------------------+-----------------------------+
1323
1324 Associated Data
1325 +------------------------------+-----------------------------+
1326 | |
1327 | Remainder of IP datagram |
1328 | |
1329 | No associated data is present if IP |
1330 | datagram fits in the Message Proper |
1331 | |
1332 +------------------------------+-----------------------------+
1333
1334 TRUNK MASK
1335
1336 From the vantage of an IP driver, any trunk mask is valid so long as
1337
1338
1339
1340 <span class="grey">Hardwick &amp; Lekashman [Page 24]</span></pre>
1341 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-25" ></span>
1342 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1343
1344
1345 it results in successful delivery of the HYPERchannel network message
1346 to its destination. There is no reason to check this field for
1347 validity on reception of the message. Specification of the Trunk
1348 Mask on output is a local affair that can be specified by the
1349 transmitting driver's address resolution tables.
1350
1351 The use of 0xFF in this value is strongly encouraged for any message
1352 other than those using exotic trunk configurations on a single local
1353 network.
1354
1355 MESSAGE FLAGS
1356
1357 Several new bits have been defined here.
1358
1359 EXTENDED ADDRESSING. This bit should be set ON whenever a 32-bit
1360 address (Network and/or Domain numbers nonzero) is present in the
1361 message. It should always be OFF with the 16-bit message header. If
1362 this bit is improperly set, delivery of the message to the (apparent)
1363 destination is unlikely.
1364
1365 END-TO-END CRC. Some newer technology adapters are equipped to place
1366 a 32-bit CRC of the associated data at the end of the associated data
1367 block when this bit is on. Similarly equipped adapters will examine
1368 the trailing 32-bits of associated data (when the bit is on) to
1369 determine if the message contents have been corrupted at any stage of
1370 the transmission.
1371
1372 Transmitting device drivers should include the ability to set this
1373 bit on transmission as a configuration option similar to the specific
1374 HYPERchannel device interface used. The bit should be generated to
1375 be turned ON if the HYPERchannel IP driver is attached to an adapter
1376 equipped to generated CRC information -- it should be left OFF in all
1377 other circumstances.
1378
1379 If a message arrives at the host with the CRC bit still on, this
1380 indicates that the CRC information was placed at the end of
1381 associated data by the transmitting adapter and not removed by the
1382 receiving adapter; thus the associated data will be four bytes longer
1383 than otherwise expected. Since the IP datagram length is self
1384 contained in the network message, this should not impact IP drivers.
1385
1386 It is possible for host computers to both generate and check this CRC
1387 information to match the hardware assisted generation and checking
1388 logic in newer network adapters. Contact NSC if there are particular
1389 applications requiring exceptional data integrity that could benefit
1390 from host generation and checking.
1391
1392
1393
1394
1395
1396 <span class="grey">Hardwick &amp; Lekashman [Page 25]</span></pre>
1397 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-26" ></span>
1398 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1399
1400
1401 FROM ADDRESS CORRECT. This bit should be set by all transmitting IP
1402 drivers who have endeavored to provide a completely correct FROM
1403 address that properly reflects the adapter interface used. No action
1404 should be taken on this bit by the receiving IP driver at this time.
1405 Additional work needs to be done to determine the action an IP driver
1406 should take if it detects a real or imagined "security violation"
1407 should a message arrive with this bit absent.
1408
1409 TO ADDRESS
1410
1411 The TO address logically constitutes bytes 2-5 of the network
1412 message.
1413
1414 NETWORK AND DOMAIN NUMBERS. The Network and Domain numbers should
1415 both be nonzero when 32-bit addressing is used. If the message is
1416 local in nature, then the local Network and Domain numbers should be
1417 placed in this field.
1418
1419 ADAPTER ADDRESS. Contains the adapter address as in the basic
1420 message. The high order bit of this eight bit field (the "outnet"
1421 bit) should be set to zero if the destination network and domain are
1422 the same as the transmitting host's. The high order bit should be
1423 set to one if the destination host is not in the local network or
1424 domain.
1425
1426 LOGICAL TO AND SUBADDRESS. The logical TO field should contain the
1427 protocol server address of the HYPERchannel IP driver for that host
1428 as determined by the host's system administrator.
1429
1430 FROM ADDRESS
1431
1432 The FROM address is filled in with the address that the local driver
1433 expects to receive from the network, but no particular use is made of
1434 the FROM address.
1435
1436 MESSAGE TYPE
1437
1438 The value 6 must be placed in this byte to uniquely indicate that the
1439 network message is being used to carry IP traffic. No other well-
1440 behaved protocol using HYPERchannel should duplicate this value of 6.
1441
1442 Note that all IP drivers should be prepared to send and receive the
1443 basic format network messages using the 16-bit HYPERchannel
1444 addresses. The driver can distinguish an incoming network message by
1445 the value of byte 8 -- 32-bit messages will always have a 6 in byte
1446 8, while 16-bit messages should have a 5 here. For interoperability
1447 with older drivers, a value of 0 here should be treated as 16 address
1448 bit messages.
1449
1450
1451
1452 <span class="grey">Hardwick &amp; Lekashman [Page 26]</span></pre>
1453 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-27" ></span>
1454 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1455
1456
1457 IP HEADER OFFSET
1458
1459 Byte 9 contains the offset to the start of the IP header within the
1460 message proper, such that the Message Proper address plus the IP
1461 header offset generates the address of the first byte of the IP
1462 header (at least on byte addressable machines.)
1463
1464 Unlike the 16-bit header, receiving IP drivers should assume that
1465 this field contains a correct offset to the IP header and examine the
1466 information at that offset for conformance to an IP datagram header.
1467
1468 Valid offsets are in the range of 16 through 44 bytes, inclusive.
1469 The limitation of 44 bytes is imposed so that routing decisions on
1470 the vast majority of IP datagrams can be made by examining only the
1471 message proper, as the basic IP datagram will fit into the message
1472 proper if it begins at an offset of 44.
1473
1474 IP DATAGRAM CONTENTS
1475
1476 The message and data are treated as logically contiguous entities
1477 where the first byte of associated data immediately follows the 64th
1478 byte of the message proper.
1479
1480 If the entire IP datagram is less than or equal to (64-offset) bytes
1481 in length it will fit into the Message Proper. If so, only a message
1482 proper containing the HYPERchannel header and IP datagram is sent on
1483 the network.
1484
1485 If the IP datagram is greater than this length, the IP datagram
1486 spills over into the associated data. On transmission, a 64 byte
1487 message proper is sent followed by as many bytes of associated data
1488 as are needed to send the entire datagram.
1489
1490 On reception, the message proper can be read into the start of an IP
1491 input buffer and the associated data read into memory 64 bytes from
1492 the start of the message. If the received message is in fact a 32-
1493 bit address message, no "shuffling" of the message will be required
1494 to build a contiguous IP datagram -- it's right there at buffer+16.
1495
1496 ADDRESS RESOLUTION PROTOCOL
1497
1498 Address Resolution Protocol has achieved a great deal of success on
1499 the Ethernet as it permits a local IP network to configure itself
1500 simply by having each node know its own IP address. Those unfamiliar
1501 with the intent, protocol, and logic of the Address Resolution
1502 Protocol should refer to <a href="./rfc826">RFC-826</a>, "An Ethernet Address Resolution
1503 Protocol" [<a href="#ref-2" title="&quot;An Ethernet Address Resolution Protocol&quot;">2</a>].
1504
1505
1506
1507
1508 <span class="grey">Hardwick &amp; Lekashman [Page 27]</span></pre>
1509 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-28" ></span>
1510 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1511
1512
1513 A later section of this document describes four techniques where a
1514 target HYPERchannel address is to obtained given the destination's IP
1515 address. The protocol is defined in this section for completeness.
1516
1517 Message Proper
1518 +------------------------------+-----------------------------+
1519 0 | Trunks to Try |1| Message Flags |
1520 | TO trunks | FROM trunks |GNA|CRC| |SRC|EXC|BST|A/D|
1521 +------------------------------+-----------------------------+
1522 2 | Server Domain or | Server Network or |
1523 | 0xFF | 0xFF |
1524 +------------------------------+-----------------------------+
1525 4 | Server Adapter Address or | Server logical addr/port or |
1526 | 0xFF | 07 |
1527 +------------------------------+-----------------------------+
1528 6 |O| Physical addr of | Originating | Src Port|
1529 |N| source adapter | server address | number |
1530 +------------------------------+-----------------------------+
1531 8 | NSC ARP type code |
1532 | 07 | 00 |
1533 +------------------------------+-----------------------------+
1534 10 | Source Domain | Source Network |
1535 +------------------------------+-----------------------------+
1536 12 | - reserved - | Age Count |
1537 +------------------------------+-----------------------------+
1538 14 | Next Header Offset | Header End Offset |
1539 | (usually 16) | (usually 16) |
1540 +------------------------------+-----------------------------+
1541 16 | Padding to start of IP info (usually 0 bytes) |
1542 +------------------------------+-----------------------------+
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564 <span class="grey">Hardwick &amp; Lekashman [Page 28]</span></pre>
1565 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-29" ></span>
1566 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1567
1568
1569 +------------------------------+-----------------------------+
1570 Off | ARP hardware address type for HYPERchannel |
1571 | 8 |
1572 +------------------------------+-----------------------------+
1573 +2 | HYPERchannel protocol type |
1574 | 06 00 |
1575 +------------------------------+-----------------------------+
1576 +4 | HYPERchannel address length | IP address length |
1577 | 6 | 4 |
1578 +------------------------------+-----------------------------+
1579 +6 | ARP opcode (request or reply) |
1580 +------------------------------+-----------------------------+
1581 +8 | Domain | Network |
1582 +- Sender's 32-bit HYPERchannel address -+
1583 +10 | Adapter address | Logical addr/port |
1584 +------------------------------+-----------------------------+
1585 +12 | Source's MTU size |
1586 +------------------------------+-----------------------------+
1587 +14 | | |
1588 +- Sender's 32-bit IP address -+
1589 +16 | |
1590 +------------------------------+-----------------------------+
1591 +18 | Domain | Network |
1592 +- Destination's 32-bit HYPERchannel address -+
1593 +20 | (to be determined on request) |
1594 | Adapter address | Logical addr/port |
1595 +------------------------------+-----------------------------+
1596 +22 | Destination's MTU size |
1597 | (to be determined on request) |
1598 +------------------------------+-----------------------------+
1599 +24 | | |
1600 +- Destination's 32-bit IP address -+
1601 +26 | |
1602 +------------------------------+-----------------------------+
1603
1604 Layout of the fields of this ARP message is a fairly straightforward
1605 exercise given the standards of ARP and the 32-bit message header. A
1606 few fields are worth remarking upon:
1607
1608 TO ADDRESS
1609
1610 The TO address of an ARP message will be one of two classes of
1611 address. A "normal" address indicates that the message is an ARP
1612 response, or that it is an ARP request directed at an ARP server at a
1613 well known address on the local network. For those HYPERchannel
1614 networks which are equipped to broadcast, a value of 0xFFFFFF07 in
1615 the TO address will (by convention) be picked up only by those
1616 protocol servers prepared to interpret and respond to ARP messages.
1617
1618
1619
1620 <span class="grey">Hardwick &amp; Lekashman [Page 29]</span></pre>
1621 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-30" ></span>
1622 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1623
1624
1625 The issue of which address to use in an ARP request is discussed in
1626 the Address Resolution section.
1627
1628 FROM ADDRESS
1629
1630 Must be the correct FROM address of the user protocol server issuing
1631 an ARP request. The Source Correct bit in the Message Flags byte
1632 should be set by this requesting server, as some ARP servers may
1633 someday choose to issue ARP information on an "need to know" basis in
1634 secure environments. With an ARP response, the FROM address will
1635 contain the "normal" HYPERchannel address of the protocol server
1636 responding to the ARP address, even if that server was reached via
1637 broadcast mechanisms.
1638
1639 ARP responses are returned to the party specified in the FROM address
1640 specified in the message header, rather than the address in the
1641 "Source HYPERchannel Address" field within the body of the ARP
1642 message.
1643
1644 MESSAGE TYPE
1645
1646 The 16-bit value 0x0700 is reserved for the exclusive use of ARP.
1647 Unlike IP messages, no provision is made for the ARP message to begin
1648 at an arbitrary offset within the message proper, so the value in
1649 byte 9 is an extension of the message type.
1650
1651 HEADER END OFFSET
1652
1653 ARP uses the 32-bit addressing convention that byte 15 contains the
1654 offset to the start of user protocol (and hence the end of user
1655 protocol information). Note that this is not a substitute for the IP
1656 offset fields, as this field also serves as the end of HYPERchannel
1657 header information -- future NSC message processing code may well
1658 take exception to "garbage" between the actual header end and the
1659 start of user data.
1660
1661 HYPERCHANNEL HARDWARE TYPE CODE
1662
1663 This 16-bit number is assigned a formal ARP hardware type of 8.
1664
1665 HYPERCHANNEL PROTOCOL TYPE
1666
1667 On the Ethernet, this field is used to distinguish IP from all other
1668 protocols that may require address resolution. To be logically
1669 consistent, this field is identical to bytes 8 and 9 0x0600 in a 32-
1670 bit address HYPERchannel message carrying an IP datagram.
1671
1672
1673
1674
1675
1676 <span class="grey">Hardwick &amp; Lekashman [Page 30]</span></pre>
1677 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-31" ></span>
1678 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1679
1680
1681 HYPERCHANNEL ADDRESS LENGTH
1682
1683 This contains the value 6, a sufficient number of bytes to
1684 accommodate the four byte HYPERchannel address and 2 bytes to
1685 indicate the largest IP datagram size that source and destination can
1686 handle.
1687
1688 SOURCE AND DESTINATION HYPERCHANNEL ADDRESS
1689
1690 This field contains the Domain, Network, and Adapter/port address of
1691 source and destination, respectively. A value of 0000 in the Domain
1692 and Network fields has special significance as this is interpreted as
1693 a request to send and receive 16-bit HYPERchannel headers rather than
1694 32-bit headers. If 32-bit headers are to be used within a single
1695 HYPERchannel network, then the local domain and network numbers may
1696 be specified.
1697
1698 MAXIMUM TRANSMISSION UNIT
1699
1700 HYPERchannel LAN technology is such that messages of unlimited length
1701 may be sent between hosts. Since host throughput on a network is
1702 generally limited by the rate the network equipment can be
1703 functioned, larger transmission sizes result in higher bulk transfer
1704 performance. Since not every host will be able to handle the maximum
1705 size IP datagram, a more flexible means of MTU (maximum transmission
1706 unit) size negotiation than simply wiring the same value into every
1707 network host is needed. With this field, each host declares the
1708 maximum IP datagram size (not the associated data block size) it is
1709 prepared to receive. Transmitting IP drivers should be prepared to
1710 send the minimum of the source and destination IP sizes negotiated at
1711 ARP time.
1712
1713 The MTU size sent refers to the maximum size of IP header + data. It
1714 does not include the length of the HYPERchannel Hardware header or
1715 any offset between the header and the start of the IP datagram.
1716 Since it is the option of the transmitting hosts to use an offset of
1717 up to 44 bytes a receiving host must in any event be prepared to
1718 receive a 64 byte Message Proper and an Associated Data block of
1719 MTU-20 (that is 64 - 44, or the length of the basic IP header).
1720
1721 An example of a typical 16-bit packet is:
1722
1723 12 bytes hardware header.
1724 12 bytes offset.
1725 40 bytes IP/TCP header.
1726 4096 bytes of data.
1727 This gives an MTU of 4136.
1728
1729
1730
1731
1732 <span class="grey">Hardwick &amp; Lekashman [Page 31]</span></pre>
1733 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-32" ></span>
1734 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1735
1736
1737 An example of a typical 32-bit packet is:
1738
1739 16 bytes hardware header.
1740 8 bytes offset.
1741 40 bytes IP/TCP header.
1742 4096 bytes of associated data,
1743 This also gives an MTU of 4136.
1744
1745 The offset values are chosen so that the typical packet causes user
1746 data to be page aligned at the start of the associated data area.
1747 This is an implementation decision, which can certainly be modified
1748 as required.
1749
1750 The maximum maximum transmission unit is 65536, the current largest
1751 size IP datagram. In order to allow this value to fit into a 16-bit
1752 field, the offset length is not included in the MTU. This MTU size
1753 is not a requirement that a local host be equipped to send or receive
1754 datagrams of that size; it simply indicates the maximum capacity of
1755 the receiving host.
1756
1757 A note on trunk masks:
1758
1759 There is no field for specifying trunk masks. This is intentional,
1760 as new NSC hardware will contain trunk reachability information,
1761 eliminating the need for the host to maintain hardware configuration
1762 layouts. All HYPERchannel messages generated as a result of an ARP
1763 response should use 0xFF in the trunk mask.
1764
1765 ADDRESS RESOLUTION
1766
1767 This section describes techniques used by an IP driver to determine
1768 the HYPERchannel address and header that a message should contain
1769 given an IP datagram containing an IP address. It describes
1770 techniques that are local to specific hosts (and hence can be
1771 modified without regard to the activities or techniques of other
1772 hosts) as well as techniques to use the Address Resolution Protocol
1773 on existing HYPERchannel equipment to better manage IP addresses.
1774
1775 It also discusses the migration of name resolution on one of four
1776 steps.
1777
1778 1. Truncation of the IP address to form a HYPERchannel address.
1779
1780 2. Local resolution of HYPERchannel addresses through configuration
1781 files.
1782
1783 3. Centralized resolution of HYPERchannel addresses through an "ARP
1784 server" driven by a configuration file.
1785
1786
1787
1788 <span class="grey">Hardwick &amp; Lekashman [Page 32]</span></pre>
1789 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-33" ></span>
1790 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1791
1792
1793 4. Distributed resolution of HYPERchannel addresses using a "real"
1794 address Resolution Protocol on future HYPERchannel media
1795 supporting a broadcast mode.
1796
1797 IP ADDRESS TRUNCATION
1798
1799 A number of IP on HYPERchannel implementations support modes where
1800 the HYPERchannel address is generated by placing the low order 16-
1801 bits of the IP address in the TO address of the message proper. This
1802 more or less treats a set of HYPERchannel boxes addressable through
1803 16-bit HYPERchannel addresses as a Class B IP network.
1804
1805 This approach certainly offers simplicity: IP addresses are simply
1806 chosen to match HYPERchannel addresses and no IP address
1807 "configuration files" need be kept. Although this approach works in
1808 an environment where the HYPERchannel completely constitutes a Class
1809 B network, or where connection to a larger IP network is not a
1810 concern, its long term use is discouraged for several reasons:
1811
1812 o It simply will not work with any Class C address (the physical
1813 TO address is not controllable) or a Class A address (where host
1814 addresses are generally carefully administered.) In addition,
1815 it will not support subnetworks. It is quite incompatible with
1816 32-bit HYPERchannel addresses.
1817
1818 o By decoupling the IP and HYPERchannel addresses through more
1819 complex address resolution, the characters of the two addresses
1820 allow greater site flexibility: the IP address becomes
1821 "logical" in character so that an address can move about a site
1822 with the user or host; the HYPERchannel address maintains its
1823 physical character so that a HYPERchannel address carefully
1824 identifies the physical location of the source and destination
1825 within the extended HYPERchannel network.
1826
1827 LOCAL ADDRESS RESOLUTION
1828
1829 The current state of address resolution art with IP on HYPERchannel
1830 works as follows: given an arbitrary IP address, the IP HYPERchannel
1831 driver looks up an entry with that address in a (generally hashed)
1832 table. If found, the table entry contains the first 6 bytes of the
1833 HYPERchannel header that is used to send the IP datagram to the next
1834 IP node on the internet. Since implementations such as the 4.3BSD
1835 UNIX IP are clever enough to provide its lower level drivers with the
1836 IP address of the next gateway as well as the destination address on
1837 the internet (assuming the message is not delivered locally on the
1838 HYPERchannel,) the number of entries in this table is more or less
1839 manageable, as it must only contain the IP hosts and gateway
1840 addresses that are directly accessible on the HYPERchannel.
1841
1842
1843
1844 <span class="grey">Hardwick &amp; Lekashman [Page 33]</span></pre>
1845 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-34" ></span>
1846 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1847
1848
1849 CONFIGURATION FILE FORMAT
1850
1851 So long as this technique of address resolution is used, the
1852 techniques used are exclusively local to the host in the sense that
1853 the techniques used to generate and store the information in the
1854 table are irrelevant to other hosts.
1855
1856 Shown here is a typical file format. This file should probably be
1857 program generated from a database, as asymmetric trunk configurations
1858 and multiply homed hosts can cause differences in physical routing
1859 and trunk usage. This format is documented here to illustrate what
1860 sort of information must be kept at the link layer.
1861
1862 The file consists of source lines each with the form:
1863
1864 &lt;type&gt; &lt;hostname&gt; &lt;trunks/flags&gt; &lt;domain/net&gt; &lt;addr&gt; &lt;MTU&gt;
1865
1866 an example:
1867
1868 &lt;type&gt; &lt;hostname&gt; &lt;t/f&gt; &lt;dom/net&gt; &lt;addr&gt; &lt;MTU&gt;
1869 # Random front end
1870 host hyper.nsco.com FF88 0103 3702 4148
1871 # because we want to show the 4 byte format
1872 host 192.12.102.1 FF00 0000 2203 1024
1873 # Small packets, interactive traffic.
1874 host cray-b.nas.nasa.gov FF88 0103 4401 4148
1875 # The other interface, for big packets.
1876 ahost cray-b.nas.nasa.gov FF88 0103 4501 32768
1877 # A loopback interface, (What else)
1878 loop loop37.nsco.com FF00 0000 3700 4148
1879 # And of course an example of arp service.
1880 arpserver hcgate.nsco.com FF88 0103 7F07
1881
1882 Comments may begin with either # or ;.
1883 Case is not significant in any field.
1884
1885 &lt;type&gt; indicates the type of entity to be defined.
1886 Currently defined types are "host," "ahost", "loop," "address,"
1887 and "arpserver".
1888
1889 host This token indicates an IP host. The following field is
1890 expected to be a name that can be resolved to an IP
1891 address.
1892
1893 ahost This field indicates an additional network interface to
1894 the same host. This may be used for performance
1895 enhancements.
1896
1897
1898
1899
1900 <span class="grey">Hardwick &amp; Lekashman [Page 34]</span></pre>
1901 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-35" ></span>
1902 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1903
1904
1905 loop Sets a flag in the entry for that host so that 0xFF00 is
1906 placed in bytes 8 and 9 of the message. This will cause
1907 the IP datagram to be directed towards the specified host
1908 (which must still be a valid host name) and looped back
1909 within the remote adapter. This facility serves both as a
1910 debugging aid and as a crude probe of the availability of
1911 the remote network adapter.
1912
1913 arpserver This indicates an address to use for directing ARP
1914 requests to the network. If several arpserver addresses
1915 are specified, they will be tried in turn until a response
1916 is received (or we run out of servers.) An arpserver with
1917 the appropriate broadcast address of FFFF FF07 would
1918 cause an ARP broadcast to take place when broadcasting
1919 becomes available. Broadcast and specific addresses may
1920 be used in combination.
1921
1922 &lt;hostname&gt; This field is the logical name of the destination. For a
1923 host it is the logical name to be given to the local naming service
1924 to determine the associated IP address. This field may contain four
1925 decimal numbers separated by dots, in which case it is assumed to be
1926 the explicit IP address.
1927
1928 &lt;trunks/flags&gt; This field is the value to be placed in bytes 0 and 1
1929 of the message header when sending to this host. The associated data
1930 bit need not be supplied as the driver must control it. All other
1931 bits are sent as provided. This field is a hexidecimal number.
1932
1933 &lt;domain/net&gt; This field is the value to be placed in the Domain and
1934 Network number field of the message. A value of 0000 in this field
1935 indicates that the destination should be reached by constructing a
1936 16-bit HYPERchannel header, rather than a 32-bit header.
1937
1938 &lt;address&gt; This field is the value to be placed in the 16-bit TO field
1939 to reach &lt;hostname&gt;. This field is a hexidecimal number.
1940
1941 &lt;MTU&gt; This field contains the largest size IP datagram that the
1942 destination host is prepared to receive. This field is a decimal
1943 number. This field is optional. If not present, a value of 4148 is
1944 assumed. See the earlier discussion on Maximum Transmission Unit for
1945 more detail.
1946
1947 ARP SERVERS
1948
1949 The primary problem with local host address resolution is that
1950 changes or additions to hosts on the local net must be replicated to
1951 every HYPERchannel host in that network. While this is manageable
1952 for up to half a dozen hosts, it becomes quite unmanageable for
1953
1954
1955
1956 <span class="grey">Hardwick &amp; Lekashman [Page 35]</span></pre>
1957 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-36" ></span>
1958 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
1959
1960
1961 larger networks. An approach that can be implemented using existing
1962 HYPERchannel technology is to have a server on the HYPERchannel
1963 network provide the HYPERchannel destination address that is
1964 associated with an IP address.
1965
1966 Although this is strictly a point-to-point request/response dialogue
1967 between two network nodes, the Address Resolution Protocol which was
1968 originally designed for Ethernet (but thoughtfully constructed to
1969 work with any pair of link and network addresses) performs an
1970 excellent job.
1971
1972 ARP servers can be reached simply by placing the address of the
1973 server in the 32-bit TO address of the network message. ARP servers
1974 only "listen" to messages that arrive on their well known normal
1975 address; they do not respond to ARP broadcast messages. Properly
1976 equipped IP drivers should respond to the broadcast messages when
1977 they appear.
1978
1979 If an ARP server receives a message containing an IP address it does
1980 not know how to resolve, it ignores the message so that another ARP
1981 server might be addressed at the source's next attempt.
1982
1983 If the address is resolvable, it places the known HYPERchannel
1984 address and MTU size in the response and returns it to the location
1985 in the HYPERchannel header FROM address.
1986
1987 Unlike a broadcast ARP, the ARP server will be required to service
1988 two requests when two hosts that are initially unknown to one another
1989 attempt to get in touch. Since the destination did not receive the
1990 ARP request, it must contact the ARP server when its higher level
1991 protocols first generate a datagram to respond to the the source's
1992 first IP datagram to go through to the destination.
1993
1994 The source configuration file described in the previous section was
1995 explicitly designed so that it could be sufficient as a data base for
1996 an ARP server as well as an individual host.
1997
1998 BROADCAST ARP
1999
2000 When a local HYPERchannel network contains a broadcast capability,
2001 any IP driver wishing to perform HYPERchannel address resolution may
2002 be configured to emit the ARP message on a broadcast instead of a
2003 well known address. IP drivers on other hosts are presumed to know
2004 if their local HYPERchannel interface can send broadcast messages; if
2005 so, they arrange to "listen" on the FF07 broadcast TO address for
2006 ARP.
2007
2008 Processing of a received ARP broadcast message is otherwise identical
2009
2010
2011
2012 <span class="grey">Hardwick &amp; Lekashman [Page 36]</span></pre>
2013 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-37" ></span>
2014 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
2015
2016
2017 to <a href="./rfc826">RFC-826</a>:
2018
2019 o Messages are responded to if and only if the destination IP
2020 driver is authoritative for the designated IP address.
2021
2022 o Whenever an ARP message is processed, the IP driver takes the
2023 source HYPERchannel address and MTU size and adds it to its
2024 address resolution tables. Thus the driver is equipped to
2025 turn around the IP datagram that arrives from the destination
2026 host when contact is made.
2027
2028 Each IP driver may have address resolutions that are set through a
2029 static routing table (the configuration file specified above). If
2030 ARP information arrives that contradicts a static entry (as opposed
2031 to previously set dynamic ARP information) then the ARP information
2032 should be ignored. This decision is made on the premise that the
2033 only useful purpose of static routing in a broadcast ARP environment
2034 is to add authentication, as it's easy to lie with ARP.
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068 <span class="grey">Hardwick &amp; Lekashman [Page 37]</span></pre>
2069 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-38" ></span>
2070 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
2071
2072
2073 APPENDIX A. NSC PRODUCT ARCHITECTURE AND ADDRESSING
2074
2075 This section is intended to be a concise review of the state of the
2076 art in NSC networks and the techniques they provide for the delivery
2077 of messages. Those who are thoroughly familiar with HYPERchannel may
2078 wish to only skim this section; however, there is material on new
2079 technologies and addressing formats that are not yet generally known
2080 to most of NSC's customers.
2081
2082 NETWORK SYSTEMS HYPERCHANNEL TECHNOLOGIES
2083
2084 Network Systems manufactures several different network technologies
2085 that use very different media and link controls, but still provide a
2086 common host interface in both the protocol and hardware sense of the
2087 term. These four technologies are:
2088
2089 o HYPERchannel A -- A 50-megabit, baseband, CSMA with collision
2090 avoidance network using a coaxial cable bus. Individual
2091 HYPERchannel "network adapters" can control up to 4 of these
2092 coaxial cable "trunks," providing up to 200 megabits of
2093 capacity on a fully interconnected network. HYPERchannel A
2094 is NSC's earliest product and has been in production since
2095 1977. It is principally used to interconnect larger
2096 mainframe computers and high speed mainframe peripherals such
2097 as tape drives and laser printers.
2098
2099 o HYPERchannel B -- a 10-megabit, baseband, CSMA with collision
2100 avoidance network using a single coaxial cable bus. This
2101 technology is used for direct host to host communications under
2102 the name HYPERchannel B, and for terminal connections under the
2103 name HYPERbus. It is currently used for three major
2104 applications -- local networks of ASCII terminals, networks
2105 of IBM 3270 terminals, and host to host communications of
2106 smaller computers.
2107
2108 o DATAPIPE[3] -- a 275-megabit fiber optic "backbone" network
2109 that interconnects lower speed local area networks within a 20
2110 mile range, and to provide an ultra-high-performance network for
2111 the next generation of supercomputers and optical storage
2112 systems. A prototype version of DATApipe is currently under
2113 development at a customer site.
2114
2115 o Bridges and Network Distance Extensions -- NSC quickly
2116 discovered that its customers wanted very high speeds over
2117 geographic areas, not just within the range of several miles
2118 that is conceivable with a coaxial cable network. Starting
2119 in 1978, NSC began to build a series of "link adapters" that
2120 are integral bridges between local area networks. These link
2121
2122
2123
2124 <span class="grey">Hardwick &amp; Lekashman [Page 38]</span></pre>
2125 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-39" ></span>
2126 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
2127
2128
2129 adapters support common high speed communications media such
2130 as Telco T1 circuits, private microwave, high speed
2131 satellite links, and fiber optic point to point connections.
2132
2133 ATTACHMENT TO HOST COMPUTERS
2134
2135 Network Systems' high speed interfaces use the attachment techniques
2136 of the manufacturer's highest speed peripheral controllers in order
2137 to achieve burst transfer rates of tens of megabits per second.
2138 These attachment techniques fall into three categories:
2139
2140 "MAINFRAME" DATA CHANNEL ATTACHMENT
2141 +-----------+-------+ +------------+ | | | |
2142 | | | |HYPERchannel+--+ | | |
2143 | | +-------------------+ Network +--|-+ | |
2144 | Host | I/O +-------------------+ Adapter +--|-|-+ |
2145 | | | Standard host | +--|-|-|-+
2146 | Computer |Control| data channel +------------+ | | | |
2147 | | |
2148 | | |
2149 | | |
2150 | | |
2151 +-----------+-------+
2152
2153 The network adapter contains interface boards and firmware to be
2154 cabled to the manufacturer's data channel, such as would be done with
2155 a disk or tape controller. Mainframe network adapters do not emulate
2156 an existing manufacturer's device (such as a tape drive) but are
2157 supported by software which functions the channel and adapter to send
2158 and receive network messages.
2159
2160 Models of HYPERchannel adapters are available for essentially all
2161 large scale computers worldwide.
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180 <span class="grey">Hardwick &amp; Lekashman [Page 39]</span></pre>
2181 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-40" ></span>
2182 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
2183
2184
2185 MINICOMPUTER AND WORKSTATION ATTACHMENT
2186
2187 Since the network adapter contains lots of expensive, high speed
2188 logic, a different technique is used to provide attachment to
2189 minicomputers and workstations.
2190
2191 +-------------+ +---------------+ +--------------+
2192 | | | | | |
2193 | Minicomputer| | Supermini | | Workstation |
2194 | | | | | |
2195 +-----+-------+ +-------+-------+ +-------+------+
2196 | | DMA | | | DMA | | DMA | |
2197 | |control| | |control| |control| |
2198 +-----+---++--+ +-------+--++---+ +--++---+------+
2199 || || ||
2200 || || ||
2201 |+----------+ || +---------+|
2202 +----------+| || |+---------+
2203 || || ||
2204 +-++--+-----+--++-+--++-+
2205 | | | | |
2206 +-----+-----+-----+-----+
2207 | x400 |
2208 | Network Adapter |
2209 | |
2210 +-------+-+-+-+---------+
2211 | | | |
2212 ------------------|-|-|-+----------------
2213 ------------------|-|-+------------------
2214 ------------------|-+--------------------
2215 ------------------+----------------------
2216
2217 In this case, NSC provides a DMA controller designed for direct
2218 connection to that minicomputer's backplane bus. These DMA
2219 controllers accept functions and burst blocks of data from host
2220 memory to a channel cable that is connected to one of four ports on a
2221 "general purpose computer adapter." This adapter multiplexes
2222 transmissions to and from the HYPERchannel trunks from up to four
2223 attached processors.
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236 <span class="grey">Hardwick &amp; Lekashman [Page 40]</span></pre>
2237 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-41" ></span>
2238 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
2239
2240
2241 NETWORK COPROCESSORS
2242
2243 For about 10 different bus systems, Network systems provides a
2244 "smart" DMA controller containing onboard memory and a Motorola 68010
2245 protocol processor.
2246
2247 +------------+-----+---------------+-------+
2248 | | | Coprocessor | | +--------+
2249 | |Host | MC 68010 |Adapter+--------+ x400 |
2250 | HOST |DMA | 256K memory | DMA +--------+ Adapter|
2251 | | | | | +--------+
2252 | Memory +-----+---------------+-------+
2253 | |
2254 +------------+
2255
2256 This class of interface works through the network coprocessor's
2257 direct access to host memory. Network transmit and receive request
2258 packets are placed in a common "mailbox" area and extracted by the
2259 coprocessor. The coprocessor reads and writes system memory as
2260 required to service network requests in the proper order. The
2261 coprocessors currently provide a service to read or write network
2262 messages (called Driver service as it is more or less identical to
2263 HYPERchannel dumb DMA drivers) and a service for NETEX, which is
2264 NSC's OSI-like communications protocol.
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292 <span class="grey">Hardwick &amp; Lekashman [Page 41]</span></pre>
2293 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-42" ></span>
2294 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
2295
2296
2297 APPENDIX B. NETWORK SYSTEMS HYPERCHANNEL PROTOCOLS
2298
2299 The protocols implemented by NSC within its own boxes are designed
2300 for the needs of the different technologies. A compact summation of
2301 these protocols is:
2302
2303 HYPERchannel B HYPERchannel A DATApipe
2304 10 Mbits/second 50-200 Mbits/second 275 Mbits/second
2305 +----------------------+----------------------+---------------------+
2306 | |
2307 | HYPERchannel network message |
2308 | connectionless datagram protocol |
2309 | |
2310 +----------------------+----------------------+---------------------+
2311 | "HYPERchannel | | |
2312 | compatibility mode" | HYPERchannel A | DATApipe |
2313 | Virtual circuit | reservation and | acknowledgment |
2314 | estab. &amp; control | flow control | &amp; flow control |
2315 +----------------------+ protocol | protocol |
2316 | | | |
2317 | Virtual Circuits | | |
2318 | Flow Control | | |
2319 +----------------------+----------------------+---------------------+
2320 | CSMA / VT | CSMA / CA | |
2321 | frame (datagram) | frame (datagram) | TDMA packet delivery|
2322 | delivery and | delivery and | |
2323 | acknowledgment | acknowledgment | |
2324 +----------------------+----------------------+---------------------+
2325 | | | Fiber optics |
2326 | 75 ohm coax | 1-4 75 ohm coax | (various cable sizes|
2327 | cable | cables | and xmission modes)|
2328 +----------------------+----------------------+---------------------+
2329
2330 Without getting into great detail on these internal protocols, a few
2331 points are particularly interesting to system designers:
2332
2333 o All three technologies supply the same interface to the host
2334 computer or network coprocessor, a service to send and receive
2335 network messages that are datagrams from the host's vantage in
2336 that each contains sufficient information to deliver the message
2337 in and of itself. Since this datagram and its header fields are
2338 of paramount interest to the host implementor, it is discussed
2339 in detail below.
2340
2341 o All technologies use acknowledgments at a very low level to
2342 determine if packets have been successfully delivered. In
2343 addition to permitting a highly tuned contention mechanism for
2344 the coax medium, it also permits HYPERchannel A to balance the
2345
2346
2347
2348 <span class="grey">Hardwick &amp; Lekashman [Page 42]</span></pre>
2349 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-43" ></span>
2350 <span class="grey"><a href="./rfc1044">RFC 1044</a> IP on Network Systems HYPERchannel February 1988</span>
2351
2352
2353 load over several coax cables -- a feat that has proven very
2354 difficult on, for example, Ethernet.
2355
2356 o All boxes go to some lengths to assure that resources exist
2357 in the receiving box before actual transmission takes place.
2358 HYPERchannel B uses a virtual circuit that endures for several
2359 seconds of inactivity after one host first attempts to send a
2360 message to the other. Traffic over this "working virtual
2361 circuit" is flow controlled from source to destination and
2362 buffer resources are reserved for the path.
2363
2364 HYPERchannel A exchanges frames at very high rates to determine that
2365 the receiver is ready to receive data and to control its flow as data
2366 moves through the network.
2367
2368 DATApipe propagation time is relatively long compared to the time
2369 needed to send an internal packet of 2K-4K bytes. As a result,
2370 DATApipe controllers use a streamlined TP4-like transport protocol to
2371 assure delivery of frames between DATApipe boxes.
2372
2373 REFERENCES
2374
2375 [<a id="ref-1">1</a>] HYPERchannel is a trademark of Network Systems Corporation.
2376
2377 [<a id="ref-2">2</a>] Plummer, D., "An Ethernet Address Resolution Protocol",
2378 <a href="./rfc826">RFC-826</a>, Symbolics, September 1982.
2379
2380 [<a id="ref-3">3</a>] DATApipe is a registered trademark of Network Systems
2381 Corporation.
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404 Hardwick &amp; Lekashman [Page 43]
2405 </pre>