CA2048012A1 - Detection of duplicate alias addresses - Google Patents

Abstract

ABSTRACT

A duplicate address condition in a computer network may be detected by a station of the type for attachment to a computer communications network, the network capable of maintaining communications among a plurality of stations, the station having means for receiving a frame, the frame having a source address field and a frame control field; means for maintaining an individual address of the station; means for maintaining a source address list of address, the source address list not containing the individual address; means for determining that a contents of the source address field in the frame matches at least one address in the source address list; means for determining that the frame control field of the frame has a predetermined contents; and, means, responsive to the source address of the frame matching at least one address in the source address list and the frame control field of the frame having the predetermined value, for setting an indicator that a duplicate address condition exists. The predetermined contents of the frame control field further comprises an indication that the frame is a predetermined type of control frame. The predetermined type of claim may be a beacon frame, a claim frame, or some other MAC protocol frame. The list of alias addresses may be stored in a content addressable memory.
A cyclical redundancy check may be used for determining that bits from the source address field and bits from the frame control field are error free.

Description

3 -~ ~

F I ELD OF TH}~ I NVENT I ON

This invention relates generally to token controlled computer networks for communication among a plurality of computers, and more particularly to fault sequences related to duplicate ~tation addresses on the network.

BACKGROUND OF THE INVENTION

Stations make use of a unique station addre~s, referred to as "my address" to identify the station. (Quote marks are used herein to set off compound noun3 when their use helpq to clarify the meaning of a sentence, as in "my addres~". Alqo quote marks are used to set off other words when their use help~ to clarify the meaning of the sentence.) Stations also maintain a list of address which may contain my addregs for the purpose of comparing an address in the destination address field of a frame detected on the network. In the event that a station finds a match between the destination addres~ of a frame and an addres~ in the list of addresses, the station takes predetermined actions.

Also in most token controlled networks, the qtation transmitting a frame onto the network is required to remove, or "strip", the frame from the network after the frame circumnavigatas the network and returns to the originating ~tation. In some networks, stripping is accomplished by a station detecting a frame on the network, comparing the ~ource address SA

. . . - ' ~' ' ~

.
. .

2 ~ 2 of th~ frame with the my addres~ of the station, and ~tripping the frame in the event that the SA matches the my address. Thi~ form of stripping is called "source address stripping".

As the re~ult of a fault, a stat:ion may transmit a frame onto the ring where the source address SA field of the frame is the same a~ the my addreqs of another 3tation of the network. For example, a software or hardware fault in a station may cause the fault. Or, for a further example, a human error in entering a station address may have resulted in a duplicate address on the network. The fault of having a duplicate station address on a single local area network i~ referred to as the "duplicate addrass problem".

In some cases the duplicate address fault may have little or no effect on the network, where in other situations the fault condition may de~troy the operability of the network.

A ring network may have two states, a "ring operating state", and a "ring not oper~ting state". Whsn the ring i-- in the "ring not operating state", the ring may be in recovery and initialization proce~s. A typical token ring protocol may use Beacon, Claim, or other M~C (Medium Acces~ Control), frames for recovery and initialization.

In the event that two stations have the ~ame my addres~, or the event that one st:ation haq a list of addre~s and the list 2~

contain~ an address the qame aq the "my addresq" of anoth~r station on the network, then when the network is in the "ring not operating state" the network may fail to become operational.
Alternatively, after the ring has been in the "ring operating"
state, the ring may cea~e to operate after a period of successful operation.

More details of network failure modes resulting from duplicate address situations are discussed in the APPENDIX
herebelow. Also details of the Fiber Distributed Data Interface, FDDI, ring network along with duplicate addres~ failure modes ars discussed in the APPENDIX.

Also discussed in the Appendix is the ANSI/FDDI ~ing Management protocol3, referred to as RMT. RMT may detect duplicate "my addresseq" on a communications ring. RMT requires executing complex logic in order to accompli~h d~tection of duplicate "my addresseQ". Accordingly, RMT is too time consuming to execute in the event that a source address in a frame is compared with a li~t of addresses in the station. The difficulty level, such as the time required to compare a source addres~ in a fr~me with a list of addresses, or to put the logic for compari~on with a list into a silicon chip, increases approximately proportionally to the number of addresses in the list. For a list of rea~onable length the time to execute the RMT logic iQ too long for ordinary operation of the station.

2 ~ ~ 3 ~ :~ ,?, The detection of duplicate addre~ses during ring recovery and initilization has not been satisfactorily solved, and 90 rings may enter a series of oscillation~ and fail to become operational.

SU~RY OF THE INVENTION

The invention in its broad form residesl in method and a ~tation of the type for attachment to a computer communicationR network, the network capable of maintaining communications among a plurality of station~, comprises, means for receiving a frame by the qtation, the frame having a source addres~ field and a frama control field;
means for maintaining an individual addxe~ of the station; means for maintaining a source address list of address, the source address list not containing the individual address; mean for determining that a contentQ of the source addres~ field in the frame matches at leaQt one address in the source address li~t;
means for determining that the frame control field of the frame ha3 a predetermined contents; and, means, responsive to the source address of the the frame matching at least one address in the source address li~t and the frame control field of the frame having ths predetermined value, for setting an indicator that a duplicate addre~s condition exist~.

The predetermined contentQ of the frame control field further comprises an indication that the frame is a predetermined type of control frame,. The predetermined type of claim may be a beacon frame, a claim frame, or some other MAC protocol frame.

6~6~ 6 The li~3t oi~ alia3 addreqses may be stored in a content addreq~able memory.

A cyclical rPdundancy check may be u~ed for determining that bits from the ~ource addreqs fielcl and bits from the fram~
control field are error free.

Further, the apparatu~ may have a means for Qetting a first ~ignal to a first predetermined value in the event that the content~ of the ~ource addre~s field of the frame match a QOurCe addreq~ in the ~ourc~ addreqs li~t; a means or ~etting a Qecond ~ignal to a fir~t predetermin~d value in the event that the content~ of the frame control field ha~ the predetermined contentq; a meanq for Qetting a third ~ignal to a firqt predetermined value in the event that the source addresq ~ield and the frame control field pa~Q the cyclical redundancy check; and a mean~, re~pon~ive to the fir~t signal being at the fir~t predetermined value and the ~econd qignal being at the second predetermined value and the third ~ignal being at the third predeterminad value, for indicating that a duplicate addre~Q
condition existq.

The means respon~qive to the fir~t signal, the Yecond signal and the third ~ignal and for indicating that a duplicate addreQs condition exiqt~ may be an AND logic meanq. The AND logic meanQ may be an AND circuit.

C~ ~ ~ ?~

BRIEF DESCRIPTION OF THE DRAWINGS

Turning now to the drawing~, in which like numeralq represent like partY in the ~everal view9:

Fig. 1 3hows schematically the connection of two local area networks by a bridge.

Fig. 2 ~qhows a ~ource address list, SA li~t, in accordance with a preferred embodiment of the invention.

Fig. 3 3ho~s a destination addre3s liqt, DA list, in accordance with a referred embodiment of the invention.

Fig. 4A - Fig 4D show source addresq list~ and de tination addre~
lists for a bridge, in accordance with a preferred embodiment of the invention.

Fig. 5A and Fig. 5B show frame format in accordance with the FDDI
3tandard.

Fig. 6 shows the corraspondence between the OSI seven layer communications modal and a LAN communications model in accordance with a preferred embodiment of the invention.

Fig. 7 iQ a block diagram of an implementation of the invention in ~ !J '')i a qtation of a LAN.

Fig. 8 is a block diagram of an alias comparator u~ing a CAM, in accordance with a preferr~d embodiment of the invention.

Fig.9 iR a block diagram o~ an alternative ambodimant o~ the invention.

DETAILED DESCRIPTION

Referring now to Figure 1, FDDI ring communication~ syqtem 100 and FDDI ring communications ~ystem 120 are connacted together for exchange of data frames by bridge 120. 3tation~ Sl, S2, and SL are connacted into ring 100, and stations S3, S4, and SR ara connected into ring 110. Station SL i9 the ~tation of bridge 120 connected into ring 100, and station SR i~ the ~tation of bridge 120 connected into ring 110.

Each ~tation has an individual address, referred to as "my addressll, MA. Each station in~erts itQ unique "my addre~s", MA, into a frame as the ~ource address SA of the frame before transmitting the frame onto the ring.

The individual addre~s of a qtation should be unique. The problem diqclo~ed and ~olved by the preRent invention ari~es when more than one station ha~ the ~ame individual addre~s, the duplicate addre~s problem.

''~
- .
, . : . . ~ - . . . .
: . - .,'' ' - : ~ :, For example, the individual address, MA, of each ~tation correspondq to the notation of the FDDI ~pecification referred to hereinbelow in the ~ppendix a~ "my lonq addre~", Mh~. In a ring operating in accordance with the FDDI clpecification, each Qtation insert~ an individual addre~ into a frame a~ the ~ource addre~
SA of the frame before transmitting the! frame onto the ring. For example, the station may inqert itA inclividual "my lo~,g addres~"
into the ~ource addreqs field of the frame. As a further example, the station may contain a liQt of other station' 9 "my addreq~es"
and may in~ert an addre~ from the li~t into the source addreQq field of the frame.

For example, an individual address may be identified by having a zero aq the first bit. The individual addre3Y i~ to be contrasted with a group addre~s where a group addre~ may have a 1 as the fir~t bit.

The uqe of the term "station" in the pre~ent context a~3igns to the ~tation one individual address which should be unique. Tho~e Qkilled in the art of computer network~ u~e the term "Qtation" in ~ome contexts to include an apparatus having more than one Medium Access Control entity (MAC en~ity), and one individual addre~s, which should be unique, is asqigned to each MAC entity. In thi~ context the station may have several MAC
entitie~, and each M~C entity hacl it~ own individual addres~.
Those 3killed in the art of computer network~ will recognize that , .. ' .: - .

2 ~ J

the preqont terminology wharein a station has only one individual address which Qhould be unique is equivalent to the alternative wherein a station may have ~everal MAC entitie3 each with it~ own individual addres~, and the equivalence i~ crystallized by referring then to each MAC entity as a "station".

There are applications that may require the use of addresses which are diff~rent from the station's individual addreAQ. An "alias address" i8 an address different from he "individual address" but used in a ~imilar manner a3 the individual addre~s. For example, an application may require that an alias address be used as a source addres~ when transmitting frames. Also the "alias address" may be used to compare with the destination address field of a frame when receiving frames.

Referring now to Fig. 2 and Fig. 3, each ~tation may have one or more alias addre~seq. The alias addrs3ge~ are arranged into a source addresQ list, SA li~t 130 as shown in Fig. 2, and into a destination addreqs li~t, DA list 132 as ~hown in Fig 3.
The SA list 130 does not contain the station "individual addreqs".
In contrast, the DA list 132 usually contains the station individual address.

For functions requiring a source addre~s, a station u~es it~ SA liQt 130, and also separately uses its individual address.

And the Qtation usss itQ DA list 132 for functions requiring a destination address.

': . ' - , $ ~

The SA list 130 has at leaYt two purpo~e~, the fi~t purpo~e being that an addre~s from the list may be inserted into the source addres3 field of a frame before the frame i8 transmitted onto the ring, and the second purpoqe being that addresses from the SA list 130 are uqed for ~tripping frames from the ring by using the "source addre~ stripping algorithm.

For frame reception the 3tation may uqe the DA list 132.
The station then compares the address in the DA list 132 ~ith the de3tination addres3 field of a "detected" frame in order to make a deci~ion as to whethex or not the frame i~ to be "received".

A Medium Acce33 Control entity, or MAC entity, u3es control frames to control network operation. Control frame3 include Beacon frameq, Claim frames, and other MAC frames.
Control frame~ commonly use only the "individual addre3s" assigned to the MAC entity for compari30n with the source addresQ Eield of a frame. Thi~ "individual addreqs" i~ the "individual addres3"
aqsigned to the station using the present terminology of one "individual addresq" per "station". This individual addres~
should be unique. In contra~t, frameq which are used to qend and receive information for clients of the MAC 3ublayer may use the "individual addresq", or additionally may u~e an addreq~ in the station 3A li3t 130 Eor sending and stripping, or may use the ~tation DA li~t 132 Eor receiving.

: . ' - ' . ~. -- . :
. .

For example, for the qtation~ of bridge 120 the SA list and DA list are ~hown in Fig. 4~ - Fig. 4D. Station SL 122 has both an SA liQt 140 of alias addresseq shown in Fig 4A, and a DA li~t 142 of alia3 addreqses shown in Fig. 4,B. Al90, qtation SR 124 of bridge 120 has an SA list 150 of aliasl addresses shown in Fig. 4C, and a DA liQt 152 of alia~ addresses shown in Fig. 4D.

Station SL 122 of bridge 120 u~e3 itq destination addreqs li~t, DA li~t 142 ~hown in Fig. 4B, to receive frameQ originating in ring 100 and destined for a ~tation on ring 110. Station SR
124 of bridge 120 use~ its qource addresQ liQt, SA list 150, for stripping frames on ring 110 where the frame~ originated from ring 100. Accordingly, DA liQt 142 of qtation SL 122 haq entries of addre~se~ of stations S3, S4, and SR. ~l~o, station SR 124 has entries in SA list 150 shown in Fig. 4C of addres~es of Qtations S1, S2 and SL, 30 that SR can strip from ring 110 frames created by station3 S1, S2, and SL.

For example, entries which can occur in a destination address list, DA list 132, but not in the corresponding ~ource addres3 li3t, SA list 130, of a particular station include, for example, the "individual address" of th~ station and multicast addresses.

A multicast addreAs i3 an addre3s that cauqa~ a plurality of stations to receiva the frame. The qpecialized multica3t addre3q iQ not uQed for stripping purposes.

~ ~ i. 3 3 ~ ~

A second example of entries which may be included in a destination address list but not in a source addre~s li~t include addresses for a monitor station. In a monitor station, address of statioll~ to be monitored may be included in the destination addres~ list 80 that the monitor station will recei~e all frameq addressed to a particular station, or group of 3tation~, for the purpose of monitoring traffic on the network. Also a monitor station may use addresses in the ~ource address list to receive frames sent by a particular station or group of stations, for the purpose of monitoring traffic on the network.

Likewise, a message sent from a station on ring 110, say qtation S3, to station S1 i~ forwarded by station SR 124 of bridge i20, and the DA list 152 of station SR has addre~ses of stations Sl, S2, and SL aR entries. Also, station SL 122 mu~t strip from ring 100 those messages forwarded from ring 110, and so SA li~t 190 of station SL 1?2 has as entries addresses of stations S3, S4, and SR, as shown in Fig. 4A.

As in the above example, destination address list DA li~t 152 of station SR 124 has multica~t addresq as entries for broadcast of me~sages to ring 100 bv forwarding of bridge 120.
However the corresponding source address list SA list 150 of ~tation SR 124 does not contain entries for the multicast addre3~es because these addresses are not used ~or source addresses. Because t:hese addresse3 are not used a9 ~ource ~J) lJ ;3~

addresseq, they are not u~ed ~or ~qource address stripping. That is, the source address list of station SR 124, SA list 150, does not contain the multicast addresses contained in station SR
destination addresq list 152.

A variety of different types of station~ may employ alia~
addresses, either in a source address list or in a destination address list. Typical example3 of such stations include: bridges, monitor stations, stationq having more than one protocol ~tack, Qtation~ having variouQ addresse3 for network control function~, etc.

Two types of "protocol data units" are used by the FDDI MAC
layer, token~ aq shown in Fig. 5~, and frames as Qhown in Fig. 5B.
The token, as shown in Fig. 5A, haQ first a preamble PA of 16 or more 3ymbols, followed by a Qtarting delimiter SD of 2 Qymbol~, followed by a frame control FC sequence indicating that the "protocol data unit" is a token, and finally an ending dalimiter ED of 2 symbols. In the event that frame control field FC
indicates that the incoming "protocol data unit" iq not a token, then the "protocol data unit" is a frame as ~hown in Fig. 5B.

The frame control field FC is a string of bits, and for example, in a Qyqtem designed according to the FDDI protocol i~ 8 bits. One combination of the FC bits indicates that the "protocol data unit" is a beacon frame, another qet of bits indicates that it is a claim frame, another set of bits indicates that it is another MAC protocol frame, another ~et o~ bits indicates that it iY an information frame, etc.

When a station, in accordance with the invention, creates a beacon frame or a claim ~rame or other MAC protocol frame the station always insert~ the station' 3 "my addreqs" MA into the source addre~s field SA of the frame. That is, a .~tation never createQ a beacon frame or a claim fram~s or other MAC protocol frame with an addre~ taken from the source add~e~ list SA li~t 130.

Most ring protocolq have the requlrement that neither beacon frames nor claim frames or other MAC protocol frame are forwarded by any type of station from the ring where they are created. For example, The FDDI protocol has this requiremant.
The preferred embodiment of the invention has this requirement that neither beacon nor claim frames nor other MAC protocol frames are forwarded by any type of station from the ring where they were created.

Since no beacon frame or claim frame or other ~AC protocol frame is forwarded to another ring, there should be no match between the source address of a beacon frame or claim frame or other MAC protocol frame and any entry in a source address list SA
130. A match between the source address SA of a beacon frame or a claim frame or other MAC protocol frame and an entry in the source addreq~ t SA li3t 130 of a station indicate~ a duplicate address problem.

Stated differently, since: (1) the source address list, SA
list 130, of a station contains source addresses, both from other rings and from other various sources, for the purpoQe of stripping frames having source addresses matching those ~tored in the SA
list 130; t2) no station put~ an entry from it3 source address list SA 130 in a beacon or claim frame or other MAC protocol frame as the source address SA of the frame; (3) no beacon or claim frame or other MAC protocol frame ls forwarded from the ring in which it is created; therefore, the occurrence of a match between an entry in the source addreq3 list, SA li3t 130, and the qource addre~ of a beacon or claim frame or other MAC protocol frame ic a po~itive indication that a duplicate address axists in the ring.

For orientation purpose~, Fig. 6 is a block diagram comparing the architecture of a local area network architecture 160 with the Open Systems Interconnect, OSI, seven (7) layer communication3 model 161. The communications medium 162 either delivers bits to the physical layer 164 or receive~ bits ~or transmission from the phyqical layer. The physical layer 154 either delivers bit3 to the Medium Access Control, MAC, layer 165 or receives bits from the MAC layer 164. Also, the MAC layer either receive data from the logical link layer, LLC, 166 or delivers data to the LLC 166. The LLC 166 layer delivers data to higher client l~yers of the architecture, or receives information from the higher client layers for tran3mission.

~, ~ j? /~

In the OSI seven (7) level reference model 161, Layer 1, the phyqical layer 170 correqponds to the LAN phy~ical layer 164.
The OSI layer 2, the data link layer 172 correqpondq to the LAN
architecture 160 level3 deqignated aq MAC 165 and LLC 166. The higher levelQ 174, above level 2, corr~spond to the LAN client level~ 168.

For the present invention, the logical link layer, LLC, need not be further specified. ~or example, the layered model provides for modular design. And the higher laye~s are deqigned independently of the particular detailQ of the lower layer. Many higher protocol layers, ~uch a~ the LLC layer or higher, may be uqed with and benefit from the invention.

The ~tation management 176, SMT, entities interact with variouq levels of the LAN architecture 160, including the PHY 164, the MAC 165, the LLC 166, and the higher client layerc 168. The network management entitie~ perform serviceq and reque~t service3 a~ needed to maintain and manage the data llnk layer.

Fig. 7 iQ a block diagram ~howing the control flow for an implementation in a Qtation u~ing the invention to detact the pre~ence of a duplicate aliaq addre~s. Line 200 is the input line from the ring (not Qhown), and line 202 is the output line from the Qtation to the ring. A frame enter~ the station a~ a stream of bit~ on line 200. The ~tream of bits is converted by the physical layer, PHY 170, 16~ of Fig. 6, into an information stream. The information stream is conveniently proceQseS
internally by the qtation. The bit stream first enters the phy~ical protocol apparatus, box 210. The information ~tream leaves the physical protocol apparatus along line ~12, and the information stream is conducted to the MAC control apparatu~ 214 by line 212. The information stream is delivered to the logical link layer LLC 216 by line 218.

The invention is mocleled within the MAC sublayer and modifies the operation of certain MAC functionY related to processing of frames and source addresses. Figure 7 detailq the control flow required for the invention. The timing o~ the various control 3ignals is not discussed as there are many possible detailed designs. For in~tance, the events which cause various signals to occur and the actions which result are specified, but the circuitry for causing the ~ignals to coincide in time or act at the correct time is not detailed. Functions necessary for operation of a MAC for a to~en ring but which are independent of the invention are included in MAC control 214.
For FDDI MAC, the functions 214 include detection of a lost token, control of ring initialization, and transmi3sion of PDU's based on the token protocol. The processing of the DA within a frame is also included in 214. The DA proces~ing is modeled to operate based on a DA addreQs list Qeparate from SA addres3es. In practice, hardware fc>r storage of DA and SA addresses may sometim~s be shared without changing the spirit of the invention.

2 ~ 3 'L3~.~

A frame coming from logical link layer LLC 216 i~ conducted as an information stream along line 220 to the MAC control apparatus 214. From the MAC control apparatus 214 the information stream iq conducted by line 222 to ~UX 22~, and from M~X 224 the information stream i~ conductad by line 226 to the physical control apparatus 210. MnX 224 is a multiplexer having the function of either conducting the signal on line 234 to line 226, or conducting the signal on line 222 to line 226. In the event that a frame is being repeated onto the ring, MUX 224 conducts the signal from lins 234 to line 226. In the event that the station is tran3mitting a frame onto the ring, MUX 22~ conducts the qignal on line 222 to line 226. The MUX 224 is controllsd (control line not shown) by M~C control 214.

The physical control apparatus 210 converts the information stream from line 226 into a stream of bits, and the stream of bits is transmitted onto the ring (not shown in Fig. 7) by line 202.

An incoming information stream is automatically repeated on the ring by being conducted by line 230, first to repeat inhibit block 232, and if the repeat function is not inhibited by repeat inhibit block 232, the information stream is conducted by line 234 to MUX 224 for transmission onto the ring through line 202.

Variou~ functional blocks "liQten" to the incoming information stream on line 212, and these functional block~

. ' ' ~ ' .

. ., ' , ' ' .

% ~ ~ 8~ ~ ~ t~
checker 240 indicating tha~ the information ~tream on line 212 i~
without error is a quality requirement that reduceq fal~e indications of a duplicate addreqs problem.

Once line 270 has been aQserted to indicate that a duplicate alias source addre3~ has been found corrective action i9 taken. For example, the processor serving the MAC layer can be interrupted and ~oftware executed in order to identify the duplicate address. Once the duplicate address is identified, further corrective action may be taken. The corrective action may be to inform the network management, remove the duplicate addre~s from the alias address li~t of the station, and to possibly remove the two qtation~ station having the duplicate addre3s from the net~ork by use a a network management protocol, or other corrective action as appropriate.

Once the duplicate addre~ is identified by the aYqertion of line 270, the time period for the corrective action to take ef~ect may be critical to the rest of the ring. For example, until the corrective action takes effect the ring may experience the Beacon/Claim o~cillation as described in the Appendix. Once line 270 ha~ been a~serted to indicate that a duplicate alia~
qouce address has been found corrective action is taken. For example, the proces30r serving the MAC layer can be interrupted and software executed in order to identify the duplicate addre~.
Once the duplicate address is identified, further corrective action may be taken. The corrective action may be to inform the network management, remove the duplicate address from the alias address list of the station, and to po~sibly remove the two stations having the duplicate address from the network by use of a network management protocol, or other correcctive action as appropriate.

However, if the Ring_Op signal 274 is not asserted high, then AND
gate 272 disables ~tripping of frames *rom the ring, where the stripping is based on a match in the SA li~t compari~on by SA list comparator 246. Disabling stripping allows the station to rep~at Beacon, Claim, or other MAC protocol framee even when tha source address field of the frame matchee an address in the SA liqt of SA
list comparator 246.

Concurrently, while the3e frames are being repeated, the AND gate 260 detects the duplicate addre3s problem. Accordingly, AND gate 272 allows the ring to become operational when an addre~s in the SA list 246 is a duplicate address. Th~ use and function o~ AND gate 272 is independent of duplicate address detection by AND gate 260 and indication by line 270.

A convenient implementation of alias comparator 246 is a "content addreqsable memory", referred to by those ~killed in the art as a CAM. A content addressable memory has the property that an addreq~ is asserted on a set of input lines to the CAM, and if that address iq ~tored as a number in the CAM memory, an output on a match line is asqerted true.

, Referring to Fig. 8 and to Fig. 7, in using a CAM 300, in alias comparator 246, the match line 302 of the CAM is uqed as output line 266. The list of alias adclres~e~ i9 Rtored in the CAM
memory 304. The SA of the incoming f.rame is delivered from line 212 to alias comparator 246 where the source address SA iQ read by SA parser 306. SA parser 306 delivers the source address SA of the incoming frame to the CAM input lines 308. If the SA of the incoming frame is stored in the memory 304 of CAM 300, then the match line 302 is asserted true and the true signal appear~ on output line ~66, indicating that an alias addres~ matches the SA
of the incoming frame. In the event that FC parser 242 indicateQ
that the frame is a beacon frame or a claim frame or other MAC
frame and that CRC check 240 indicates that the information stream is good, th~n a duplicate alias addre~s problem exi~t~, and AND
circuit 260 asserts a true on its output line 270.

Use of the logic circuits with the source addresR list ~tored in a CAM allows the duplicate address test to be completed in only one memory cycle time of the CAM. This rapid test i~ a great improvement over running an execution of the RMT protocol for each address in the source address li~t of a station.

While the invention has been particularly shown and described with reference to the particular embodiments described hereinabove, it will be underqtood by those skilled in the art that various changes in form and detailQ may be made therein ~ .3 ~J
without departing from the spirit and scope of the invention.

- . , :. . . :
.
- ' '. . .:
.. . .
,, ,': ,, ' :

.
. . . ~ .

-. . .

2~ 3 ~

APPENDIX

Computer Networks A network of computer3 using a shared channel for exchange of frames between qtation~ of the network muqt have some mean~ for deciding which station has access to tranqmit a frame onto the medium. A ring network ha3 segments of media of the share,d channel connecting the 3tations into the ring.

One method to control acce3s to the media is to connect the ~tations in a logical ring, and to use a token pas~ing protocol to control access to transmit frames on the media. The media, the connection into a logical ring, and the station3 connected to the media are referred to as the "ring". An example of a well known token passing protocol is: to have a token, the token i~ held by a station, and the station holding the token i8 granted acce3~ to tran3mit on the ring, and after tranqmitting a last frame on the ring the station passes the token to the next station, and so the token pas~as around the ring. Also, it i3 col~mon to require a station tran3mitting onto the ring to remove, or "strip", from the ring the frame~ that it transmitted.

The FDDI comm~mication~ network uses source address stripping. The FDDI network is described in a number of documents produced by the Ameri.can National Standards In~titute (hereinafter ANSI), including: (1), "FDDI Station Management," ANSI/IEEE

.3 i 2 X3T9/90-078, X3T9.5/84-49, Rev. 6.2, May 18, 1990; (2), "~iber Distributed Data Interface (FDDI) - Token Ring Media Accesq Control (~C) ", ANSI/IEEE X3 .139-l9a7; (3), and the phy~ical layar specification i~ discussed in "~NSI/IEEE standard ANSI
X3.148-1988". The FDDI network is a ring topology and the right to transmit i9 controlled by a token. Capture of the token by a ~tation permitQ the station to transmit.

Source addres3 stripping used by the FDDI co~nunication~
network makes use of the source addre~3 field SA in a ~rame.
Stripping of frames from the ring proceed3 by having each station check all frames detected on the ring, and for each station to strip those frames having the source address field of the frame, SA, correspond to the individual address of the station.

Source addres~ ~tripping works well 90 long as each 3tation ha3 a unique addre3s. In the event that two station~ have the same address, the "duplicate addre3s problem", then source addre~Y
~tripping may lead to operational problems with the ring.

Station Addresses A key feature of Local Area NetworkQ, LAN, and Wide Area Networks, WAN, communication~ networks is that each qtation on the network has at least one unique addres3, and may have qeveral unique addre~qes. Problems arise in the network when an addre~s 3 .i ;~

which ~hould be unique i~ in fact duplicated in two or more stations. The addre3~ is normally uqed in a MAC layer, of the qeven layer ISO communications model, :in order to provide an address function of the station.

Various well known L~N communication~ networks are expected to work together through bridges, etc, and 90 must have unique station addres~es, and include: the I~SEE 802 family of protocols, including IEEE 802.3 ETHERNET alQo kno~qn a~ Carrier Sense Multiple Access/Colliqion Detect CSMA/CD, IEEE 802.4 Token Bus, and IEEE
802.5 Token Ring; the Fiber Distributed Data Interface FDDI token ring, etc.

Administration of the addresses a3~ignment i~ partly handled by the In~titute of Electrical and Electronic Engineers (hereinafter IEEE~, New York, New York. The addres3 qpace may, for example, be a 48 bit addresq 3pace. The addra3s ~pace may be broken into sub-addres~ spaces. For example, administration of the addre3ses i3 broken into two 47 bit addre3~ spaces, a Globally Unique address space and a Locally Administered addre~s 3pace.
Each addresY space haq 2**47 addresses. Other sub-addres~ qpace3 may be defined, for example, there may be an individual 3tation addre~s space and a group address spac~ wherein groupq of stations may be identified.

The IEEE-handle~ admini~tration of the Globally Unique address space. The IEEE assigns blocks of addre3~es to each ~J~I~ $'~J.

manufacturer of equipment. The manufacturer then iq reqpon3ible for as3igning a unique individual addres~ from its asqigned block of addresQe3 into each ~tation that iq manufactured. Each manufacturer then p~ovide~ a mean~ for the station to "know" its addresq whenever the 3tation i~ active on the network. And thi~
means usually provides an automatic aq~3ignment of the addre3s to the MAC layer of the qtation in order l_o minimize errorq in the 48 bit addreq3.

The station~ interconnected into networks may be made by many different manufacturerq. Conceptually, all qtation~ ever manufactured, by all different manufacturers, mu~t be capable of operating on one communication~ network. This concept is necessary because it iq never known which ~ubset of 3tations will be connected into a particular combination of local area network3 and wide area network3. The administration of the Globally Unique addre~ 3pace addresqeq by the IEEE a~3ign~ unique blocks to equipment manufacturer3. Equipment manufacturers insure that each new manufactured station will have a unique addres~ taken from that block qo that global interconnection is po~sible. Also the admini~tration of the Locally Administered address space by network u~ers, managers or owners i~ also suppo~ed to insure unique addresqe~.

The Locally Administered addre~ space iq managed by Pither users or addreQs adminiqtrators for variou~ network~, organizations, stc. Addres3es from the Locally Adminiqtered ~ J~J~

address space ar~ typically handled manually cmd qo are pron~ to human error.

A frame contains a de~tination addres~ field DA which identifies the intended receiver station of the frame, or receivers when a group de~tination addlress is specified. Also, the frame contains a sourcls addres3 field SA which identifieQ the transmitter station of the frame.

The source address SA and destination address DA of a ~rame are detected by a station, and are u3ed by the station for a variety of purposes. The most basic purpose of DA i~ for datecting frames addressed to the station. The source addresqes SA may be used ror other purpose~ such a3 source addreqs stripping. Additionally, a station may use a frame, including the source address of the frame in ring recovery and initialization protocols. Al30, a statioll may utili~e a plurality of addresse3 for a variety of purposes. ~xamples of the use of a plurality addresses include: separa1:e addres~es for different protocol stacks in a station; as a forwarding list in a bridge; and, by a 3tation for network management purposes, etc.

Addre~s Detection A content addressab`Le memory (CAM) may be used for rapid comparison the source addres~es in a frame with a list of addressas in a station. For example, a bridge station on a FDDI

ring network must forward frames addressed to a station on an interconnscted LAN, and quch a station may use a CAM to aid compariQon of a ~ource address, SA, in a frame with addre~ses maintained in the bridge qtation address li~t.

FDDI Normal Ring Recovery and Initilization A 3tation begins a ring recovery and initilization process as a re~ult of, for example, some ring disruption, timeout of a timer, etc.

A station begin~ a ring recovery and initilization process by ~irQt entering a claim process by tranqmitting claim frame~.
In the event that the claim process fail~, the station begins beaconing. Upon succes~fu;l completion of beaconing, the ~tation again enters the claim process.

Ring recovery norma:Lly proceed~ as ~ollowq:

A. A ~tation which is beaconing continues to do so until:

1. it receiv~s anot:her station's beacon frame, at which time it repeats the other beacon;

2. it receives its own beacon, at which time it begins to emit claim frames.

.
~ .

.
.: , . ., . . . .

2~ .3 B. A station emits claim frames until:

1. it reCeiVeQ a beacon frame from another station, atwhich time it repeat3 the beacon;

2. it receives a claim frame from ~nother tation, and then it either continues to emit its own claim frame or b~gins repeating the other ~tation's claim frame, according to a ~et of rules;

C. a ~tation quits claiming when it:

1. times out and then begins beaconing;

2. receives a claim frame with precedence, according to the rules, and then begins repeating the claim frame with precedence;

3. receive3 its own claim frame, at which time the Qtation makes a token.

A simple ~et of rules for dstermining whether a station, upon receiving a claim frame, either copies the received claim frame or emits it's own claim frame is, for example, to determine precedence on the basiQ of the numerical value of the station address. For example, in token ring networks, it i~ a common . .
, : ' , ' : . :
-. , - - .
': ' .
.
.:, . ~ . . .

practice for the claim frame having the qource addre~3 SA field having the large3t value to have precadence. And 80 the receiving qtation copie~ the received claim frame if the SA field of the raceived frame is larger than the addre~s of the station, and if the S~ field of the received frame is smaller than the station address then the station emit~ its own claim frame. Thiq simple rule, leads in normal operation, to the 3ingle Rtation having the largest value of address to make the token.

The FDDI token ring uqeq an additional layer of rule~ in determining precedence by requiring that the qtation~ bid for a timer value. And in the event that there iq a tie in bidding for the timer value, the conflict i~ reqolved by the valua o* the station addre Q ~ .

Duplicate Addresses It is poQsible, and almo~t inevitable, that mi~takeq in addre~s assignment will be made a~ many tens of thousandq of 3tation3 are interconnected. Somewhere someone will enter a wrong addre~q~ into a station, and two ~tations will have the 3ame addreq~. For example, an address may be 48 bitq long, making (2**48) or 2.8*(10**14) posQible addresses. If a worker enter~q a 1 or a 0 ~rong in the 48 bit string in loading an addreqs, then two qtation~ may havs the qame addre~. And all approximately 2.8*(10**14) addreQ3es mu~t be capable of operating on one FDDI

:. .' ' , :' , s~ ~j. 2 network because it i~ unknown what 3ub~et will actually be connected into any particular network. A very sevare address quality control problem exists in the Locally AdminiQtered Addre~s asQignments, because manual entry of the bits of the address i~
common.

An even further complication ari~es because a qtation may utilize a li3t of addresses. A particular example i~ a bridge. A
bridge may, for example, join a firQt FDDI ring with a second FDDI
ring. The bridge iq Qpecific equipment having a first station on the fixst ring and a sacond station on the ~econd ring. The bridge must have a list of addre~Res listing the station~ to which the bridge forwards frame~. For example, the bridge first ~tation on the first ring must have a liqt of addresseQ of all ~tation~ on the Qecond ring qo that it forwards frames on the first ring onto the Qecond ring. LikewiQe, the second ~tation of the bridge on the second ring must have an addresq list li~ting all stations on the first ring 80 that it can forward frameQ from the second ring onto the first ring. A1QO~ a bridge may connect an FDDI ring with, for example, an ETHERNET network, or a token buQ network, or a IEEE 802.5 token ring network, and alqo mu~t buffer the3e incompatible protocolY.

Ring Disruption by Duplicate AddreQses Severe disrup1:ion of the FDDI communication3 network may ' .
- .
' . , ~ ,, , ~J~

occur when two station~ have the qame address, or a duplicate addres~ occur~ in an addre~s list The disruption occurs because each Qtation u~es the source addresR, SA, in a frame to take action. Actions taken by a station, baqed on the qource addre3s in a frame matching an addres3 in an address list includ~s, ~or example, qtripping of the frame, and ring recovery and initilization protocol steps.

A station, in doing 30urce addresq 3tripping, qtrips all frame~ having the ~ource addreqs, SA, of the frame matching any address in the station, including all addre~ses in the addreqs liqt of the station;

A pos~ible outcome i~ that a non-duplicate qtation may complete the beacon and claim proce~s.

In the ~eaconing ~tep each station te3ts the continuity of the ring by emitting a "Beacon" frame, and then stripping it3 "own" frame off the ring. The station identifie3 its "own" frame by reading the qource address SA in the frame header. And if two stations have the same addresQ then one will qtrip the other's Beacon framas from the ring, thereby interfering with the initialization proceQ3. Al~o, if a station ha~ an address list containing a duplicate address, and the list is used for ~ource addres3 stripping, it will strip the other station's beacon frame~
from the ring.

~ ~ lJ 3 ~

However, the l'Beaconl' proce~s may completa with two stations having duplicate addrs~ses by the two stations ~tripping each other's Beacon frame and not detectin~ the duplicate addre~q qituation .

After a qtation removes it's 'lown" Beacon frame, the qtation begin3 the "Claim" process in order to determine which station issues the ~irst token. The Claim process proceeds by each station i~quing a "Claim" frame and reading all other Claim frames from the other station~, a~ 3et forth her0inabove.

The ring can fail to recover and initialize as a result of a duplicate address.

An example of a ring failing to recover and initialize as a rasult of a duplicate addre~3s qituation is given as follows: two qtationq have the ~ame addreQ~; one, the first qtation, of the stations begins beaconing; the second ~tation receives its "own"
beacon frame from the first station, and begin~ claiming; the first ~tation receives the claim frame from the second station, ignores the claim frame ~ince it is beaconing, and continueq to smit beacon frames; the secc)nd station again receiveq its "own"
beacon frame from the first station, continue~ to claim, and the network i~ stuck and no token can be made. This condition i3 called "Claim Beacon Deadlock". Claim beacon deadlock can also occur if one ~tation has the address of another station in an address list.

The rules for claim frame precedence for the FDDI/ANSI
Qtandard ring are more complex than the above example, and are as follows. Each Claim frama has two operative parts, a firqt part called "Target Token Rotation Time Bid" (hereinafter TTRT_Bid), and a qecond part con3i3ting of the station address. The TTRT_Bid iq a value given to the MAC layer by higher layer~, and iQ a value that the higher layerq have determined would be a u3eful value for that particular station con;sidering its intended application.

In arbitrating for the riyht to is3ue a fir3t token, the station bidding the smallest: value of TTRT_ Bid winY the right to iq~ue the fir~t token, and all ~tationq have their Token Rotation Timerq (hereinafter TRT) ~et; to the winning TTRT_Bid valua. The Target Token Rotation Timer value TRT iq the maximum length of time that a Qtation may hold a token during tran3miqqion.

In the event that two ~tation3 bid the ~ame value of TTRT_Bid, the bidding iQ decided on the ba3i~ of the length of the addreq3 and then on the baQis of the value of the addre3~.
Winning of the arbitration for the right to iQ~ue the first token i3 determined by the following three ruleq:

1. The bid with the lowest Target Token Rotation Time (TTRT_Bid) ha~ prececlence (i.e. the numerically loweqt bid value for TTRT.

~,fi~ iJ

2. Given equal TTRT Bid values in bidding, the bid with the large3t address has precedence, (that is, a station with a 48 bit addreQ~ win~ over a 3tation with a 16 bit addresq);

3. Given equal bid valueq for TTRT_Bid and e~ual addre~s length~, the bid with the highest address has precedence (i.e., tha numerically highest addres~).

The commonly used bidding protocol in the claim proce~ iq on the ba~is of the value of the station addre~. Although in FDDI the ~tation~ bid for TTRT Bid in order to determine which station will iqsue the first token, in many other token controlled communication~ networks the value of TRT i~ fixed and the bidding for the ~tation to create the firqt token i~ baqad entirely on the value of the ~tation address.

The duplicate addre~ problem in the FDDI network iQ
particularly qevere when the winning bid in the Claim process i9 by two ~tation~ having numerically identical addrasseq. If the two duplicataQ have bid different valueq for TTRT, ona duplicate3 will have the winning bid value of TTRT. Each ctation ~trip~ the other'Q Claim frameR. The station with the winning TTRT will never receive itQ Claim frameQ containing the winning bid TTRT
value, and ~o a token will never be isqued. Thu~, the FDDI Ring will never become operational.

2 ~

A duplicate address situatiorl causes further problems if the two duplicates hava the ~ame bid value for TTRT, aq both stations will "win" the Claiming procesq. Accordingly, both stations may start tranqmitting "Clalm" frames. The FDDI ring may repeatadly initiali7e rather than enter qtable operation.

In addition to the duplicate address problem being introduced by two stationQ on the ring having tha qams addr2s3, a station may maintain a list of addreqses, aQ described hereinabove. Accordingly, if an address in the address list iq numerically the ~amo a~ any address of a station in the FDDI
network, or in an list in any ~econd station on the FDDI network, then the FDDI network may fail to initialize to ~table operation as described above. With the interconnection of computer networks having addresses of, for example 48 bits in length, tens of thousands of stations may be connected together. And if any two of these stations has the same address, then the FDDI network may fail to function.

Solutions to Duplicate Address Problem Four conditions that are indicative of the existence of a duplicate address problem are:

1. A fir3t station receiving its "own" Beacon while Claiming for longer than the maximum delay of the FDDI ring. This condition indicates that another duplicate i9 Beaconing and the first ~tation i~ Claiming.

2 . A first station receiving it's own Claim frame while Beaconing for longer than maximum delay of the FDDI ring. Thi~
condition indicateq that another duplicate i9 Claiming while tha station is Beaconing.

3. A firQt station Receiving it' 3 own Claim frame while not Claiming. This condition indicates that both duplicate~ have the same winning bid of Target Token Rotation Time (TTRT) and address combination. The f~rst station haq stopped Claiming and isQued a token, while the other duplicate station continues to Claim.

4. A first station rec~iving valid Claim Frames with it' 3 "own" address and different bid value for TTRT. This condition indicates duplicates with different TTRT values are both Claiming.

These four conditionq in the duplicate address problem have been addressed in the publication "FDDI Ring Management", K.B.
Ocheltree and R.F. Montalvo, at the IEEE 14'th conference on Local Computer Networks, October 1989 in Minneapolis, Minnesota; and in the document "Reqearch Report, FDDI Duplicate Address Problems", by K.B. Ocheltree, IBM Research Division, Yorktown Height , New York, ~no publication date).

These four conditions are included in "FDDI Station 2 ~

Management SMT" Rev. 6.2 mentioned hereinabove as a standard Rlng Management Technique (hereinafter RMT). An examination of these conditions shows that they require a protocol to carry out the teqt steps in seacon or Claim frame passing through a ~tation, where the protocol i~ performed by Qoftware or speciality hardware.

A further point iq that RMT has no cyclical redundancy check, CRC. And so to prevent data errors from being mistaken for duplicate addresses, RMT examineq the source addre~s field SA of qeveral succe~sive frames. If the SA field of ~everal Ruccessive frames agree, then the protocol assumeQ that the SA fiald has b~en correctly read, and the protocol can then make decisions about duplicate addresses. That is, in this example, the RMT protocol can only reqpond to steady Ytate ring condition~ becauqe of the absence of cyclical redundancy check in the RMT protocol.

The implementation of ~MT, as set out in the ANSI Qtandards for FDDI referenced hereinabove, involves comparing many received Claim and Beacon frames to the frameq ~ourced by this MAC. Three steps are necessary befora the conditions may be applied. It will be readily apparent how to perform these qtepq, to a person skilled in the art of the FDDI Qpecifications as hereinabove referanced. First, the state of the MAC transmitter is aQcertained, including the SA of a tran~mitted frame, information contained in transmitted frames, and whether MAC is ~ending claim frames, beacon frames, or repeating frames from another Qtation.

% ~ 2 Use of the logic circuits with the source address li~t stored in a CAM allows the duplicate address test to be completed in only one memory cycle time of the CAM. This rapid test is a great improvement over running an execution of the RMT protocol for each address in the source address list of a station.

The function of AND gate 272 i~ to inhibit ~tripping of beacon, claim, or other MAC protocol frames in the event that the source addresQ field of the frame matches an addre~s in the source addreQs list, SA list, o the station. The use of AND gate 272 is one embodiment of the principle of inhibiting stripping when the source address field of a beacon frame, claim frame, or other MAC con~rol protocol frame matches an address in the station SA list.
Those _killed in the art will recognize that there .
' ,: ' ' ' . ' .

2 ~ l K 8 ~ ~ ~

Second, several of the received frameq are ~tor~d in memory. ~

special mode of operation i~ briefly enabled to receive Claim or Beacon frameq, as Claim or Beacon frameq are not normally received. Third, the state of ~C is verified to have been constant ~or the 8ample of frama~. If 90, the four detection rondition3 are applied to ths MAC transmitter state and received frame contentq. The SA in each received frame is compared to the list of addre~s~q u~s~d (and therefore stripped) by thi3 station.

If a match i9 found, the transmitter state iq u~ed to determino if th~ frame waq transmitted by thiq station. A duplicate address problem iQ detected when the SA of the received frame matches the liqt of addresR but was not recently tran~mitted. A single quch frame iq ignored aq it may have been created by errors on the media, but many such frames indicate a duplicate addresq condition ha~ been detected by tha RMT proceQ~. The above proce~s mu~t be repeated often enough to meet the reaction time specified for RMT, roughly 80 milliqeconds.

Theqe four conditionq are difficult to implement in a station having an addres~ list containing thousand3 of addresses.

One difficulty iq that the protocol, software or hardwars, muqt make the required tests with all addresses in the addre~s list on each qource address, in a Beacon or Claim frame received by aach qtation. Also in the example in which RMT re3pond3 only to ring steady state conditions, the response i~ inconvsniently delayed.

~uYu -u ~

are many equivalent methods ~or inhi~iting stripping when a beacon frame, claim ~rame, o~ other MAC
protocol frame has an address match with the SA list of the station without departing from the spirit and s c ope o f the invent i on .

For example a second embodiment of the invention disables the content addre~sable memory, C~M, when a C~M is used to gtore the station SA
list. In the event that the station SA li~t is maintained in a CAM, as shown in Fig. 8, then the CAM may be disabled and so prevent the output from SA li~t comparator 246 from being asserted. Many C~M chips have a "disable" pin, and So disable pin 310 ~hown in Fig. 8 may be uQed to practice the present invention.

A~ a further example, a third embodiment of the principle of inhibiting stripping when the fram~
having a match be~ween its source address field and the SA li t of the station i~ a beacon frame, or claim frame, or other MAC protocol frame is show in Fig. 9. The output of FC parser line 264 connects to input 269 of AND gate 272. The signal on line 264, and therefore input 269 of AND gate 272, i~

.

~ h asserted high in the event that FC parser 242 detects a beacon frame, a claim frame, or another MAC protocol frame. The detection of the presence of a beacon frame, a claim frame, or another MAC
protocol frame is equivalent to the Ring_Op signal on line 274 of Fig. 7 indicating that the ring is not operational, and in some embodiment:s of the invention the output of FC parser 242 rnay be u~ed to e~tablish the value of the Ring Op signal on line 274. Line 269 provide~ the signal to prevent stripping of beacon frames, claim frames, or other MAC Control frames.

While the invention ha~ been particularly shown and described with reference to the particular embodiments described hereinabove, it will be understood by those ~killed in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention.

While the invention has been particularly shown and described with reference to the particular embodiments described hereinabove, it will be understood by those skilled ~n the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention.

Claims (13)

What is claimed is:

1. A station of the type for attachment to a computer communications network, said network being capable of maintaining communications among a plurality of stations, comprising:

means for receiving a frame by said station, said frame having a source address field and a frame control field;

means for maintaining an individual address of said station;

means for maintaining a source address list of address, said source address list not containing said individual address;

means for determining that contents of said source address field in said frame match at least one address in said source address list;

means for determining that said frame control field of said frame has predetermined contents; and, means, responsive to said source address of said frame matching at least one address in said source address list and said frame control field of said frame having said predetermined value, for setting an indicator that a duplicate address condition exists.

2. The apparatus as in Claim 1 wherein said predetermined contents of said frame control field further comprise an indication that said frame is a predetermined type of control frame.

3. The apparatus as in Claim 1 wherein said predetermined contents of said frame control field further comprise an indication that said frame is a beacon frame.

4. The apparatus as in Claim 1 wherein said predetermined contents of said frame control field further comprise an indication that said frame is a claim frame.

5. The apparatus as in Claim 1 wherein said means for maintaining a source address list further comprises:

means for storing said list of alias addresses in a content addressable memory.

6. The apparatus as in Claim 1 further comprising:

means for determining that bits from said source address field and bits from said frame control field pass a cyclical redundancy check, CRC; and, means, responsive to said bits passing said cyclical redundancy check, for indicating the detection of a duplicate address condition.

7. The apparatus as in Claim 1 further comprising:

means for setting a first signal to a first predetermined value in the event that said contents of said source address field of said frame match a source address in said source address list;

means for setting a second signal to a first predetermined value in the event that said contents of said frame control field has said predetermined contents;

means for setting a third signal to a first predetermined value in the event that said source address field and said frame control field pass said cyclical redundancy check;

means, responsive to said first signal being at said first predetermined value and said second signal being at said second predetermined value and said third signal being at said third predetermined value, for indicating that a duplicate address condition exists.

8. The apparatus as in Claim 7 wherein said means responsive to said first signal, said second signal and said third signal and for indicating that a duplicate address condition exists is an AND
logic means.

9. The apparatus as in claim 8 wherein said AND logic means is an AND circuit.

10. The apparatus as in Claim 8 wherein said communications system is a fiber distributed data interface.

11. A method for detecting a duplicate address condition in a station of the type for attachment to a computer communications network, said network being capable of maintaining communications among a plurality of stations, comprising:

receiving a frame by said station, said frame having a source address field and a frame control field;

maintaining an individual address of said station;

maintaining a source address list of address, said source address list not containing said individual address;

determining that contents of said source address field in said frame match at least one address in said source address list;

determining that said frame control field of said frame has predetermined contents; and, setting an indicator that a duplicate address condition exists in response to said source address of said frame matching at least one address in said source address list and said frame control field of said frame having said predetermined value.

12. A method as in Claim 11 wherein the step of setting an indicator comprises:

setting a first signal, in response to said content addressable memory, to a first predetermined value in the event that said contents of said source address field of said frame match a source address in said source address list;

setting a second signal to a first predetermined value in the event that said contents of said frame control field indicate that said frame is a beacon frame, or is a claim frame, or is another MAC protocol frame;

setting a third signal to a first predetermined value in the event that said source address field and said frame control field pass a cyclical redundancy check;

indicating by an AND circuit that a duplicate address condition exists, said AND circuit responsive to said first signal being at said first predetermined value and said second signal being at said second predetermined value and said third signal being at said third predetermined value.

13. A method as in claim 11 including the steps of:

determining whether the network is in an operational state or is in an initialization state;

disabling stripping frames from said network in response to said contents of said source field of said frame matching at least one address in said source address list and in response to said network being in said initialization state, whereby beacon, claim, and other MAC protocol frames are not stripped based on source address match with said source address list.