TW420961B - Ethernet switch and the switching method - Google Patents

Abstract

An Ethernet switch and the switching method which can selectively transmit or filter out network packets between a plurality of data ports. The Ethernet switch is composed of a plurality of network ports, the first and the second memory devices, the first and the second memory controllers, the switching device and the second memory management device. The network ports are used to receive or transmit network packets. The first memory device is used to store the source address of network packets and the relative information of these network ports. The second memory device is used to store the received network packets from the network ports. The first and the second memory controllers are connected to the first and the second memory devices respectively for controlling the writing and reading actions of the first and the second memory devices; further, the switching device connects to the network ports and the first memory controller so as to establish the relative information between the source addresses of each network packets and the network ports and establishing the relative information of the addresses of each network packets and the network ports according to the content of the first memory device so as to manage the content of the first memory device. The second memory management device connects the network ports and the second memory controller to manage the content of the second memory device. The common memory access control circuit is used to control the access of the common memory device.

Description

420961 五、發明說明（i) 本發明是有關於一種、網^路元件且特別是有關於一種 乙太網路交換器及其交換方法，其可以選擇性地在若干個 資教A F4身送或學赛網各封/包。 乙太網路是數據通訊市場中最受歡迎的區域網路 (Local Area Network， LAN)。早期，乙太網路的結構是 匯流排拓撲網路（Bus topology network)的 10BASE5。而 10BASE2則是10BASE5為降低成本的改良。雖然10BASE2及 10BASE5同樣屬於匯流排拓撲網路且操作於10MHz時脈頻 率’但由於10BASE2使用不同類型的同軸電纜（Coaxial cable)，故傳輸長度必需限定在200公尺以内。另外， 知-- 10BASE5及10BASE2網路的缺點還包括厂^裝成本極高、安 裝彈性較差、無法使用大樓既有的配線（Wiring system)…等等。而10BASE-T網路，其利用雙絞線以結構 星版―拓 - top qlIqjli network)，則提直以克服 10BASE2 及l〇BASE5：M」^JS。 由於區域網路的規模愈來愈大且區域網路間的數據傳 輸量愈來愈多，乙太區域網路的效能亦隨著連結區域網路 的節點數增加而降低。因此，為滿足大規模區域網路或高 速數據終端設備的要求，操作於高時脈頻率的乙上 I傈1運^lJL。高速乙太網路操作，l〇OMILz時脈頻率，主 要目的是改進10BASE乙太網路的效能。另外，高速乙太網 路亦定義有介質獨立界面（Media Independent I n t e r f a c e ’ ΜII)，用以使該網路節點輕易連接至雙絞線 介質或光纖介質所結構的網路，如：100BASE-TX，420961 V. Description of the invention (i) The present invention relates to a network circuit element, and in particular to an Ethernet switch and a switching method thereof. Or Xuesai each package / package. Ethernet is the most popular local area network (LAN) in the data communications market. In the early days, the structure of Ethernet was 10BASE5 of a bus topology network. 10BASE2 is an improvement of 10BASE5 to reduce costs. Although 10BASE2 and 10BASE5 also belong to the bus topology network and operate at 10MHz clock frequency ’, because 10BASE2 uses different types of coaxial cables, the transmission length must be limited to 200 meters. In addition, the shortcomings of the 10BASE5 and 10BASE2 networks include the extremely high installation cost, poor installation flexibility, the inability to use the building's existing wiring (Wiring system), and so on. The 10BASE-T network, which uses twisted-pair cables to structure the star version (top-top qlIqjli network), is straightened to overcome 10BASE2 and 10BASE5: M "^ JS. As the size of the local area network becomes larger and more data is transmitted between the local area networks, the performance of the Ethernet local area network decreases as the number of nodes connected to the local area network increases. Therefore, in order to meet the requirements of large-scale local area networks or high-speed data terminal equipment, it operates at high clock frequencies. High-speed Ethernet operation, 10OMILz clock frequency, the main purpose is to improve the performance of 10BASE Ethernet. In addition, high-speed Ethernet also defines a Media Independent Interface (MII) to allow the network node to easily connect to a network constructed with twisted-pair media or fiber optic media, such as: 100BASE-TX ,

第5頁 420961 五、發明說明（2) 100BASE-T4 或 100BASE-FX 網路。 基本上，10BASE5，10BASE2，10BASE-T， 10 0BASE-TX，100BASE-T4 或 100BASE-FX 均是屬於分享式網 路系統（Shared 。也就是說，整個網路 系統的頻寬會由連結網路系統的所有節點分享。丨〇Base網 路系統的頻寬為10MHz，100BASE網路系統的頻寬為 100MHz。因此，當連結分享式網路系統的節點數逐漸超過 臨界值’該網路系統的效能便會飽和。為克服這個問題， 、分割方一法丄及將整個網路系統分 成數個獨立的次網並」位於不恩韵碰撞^ ，。如_此，連結不不—會分享其他次網路 ^頻寬。舉例來說’若未分割的分享式1〇〇BASE；區域網路 系統連結有1 00個節點，網路滿載時各節點只能汾享^心 頻寬。相對地，若這個分享式1〇 0BASE區域網路系統分割 成兩個區段’則網路滿載時各節點可分享2MHz頻寬。應用 乙太網路協定的分割式區域網路系統稱為式乙太網 路，而網路系統在不同區段間的封包傳送則^乙太路交 換器切換。乙太網路交換器的主要目的是將封包切換於區 域網路系統的不同區段間，因此，由一區段至另一區段的 封包將被傳送，同區段内由一節點至另一節點的封包則不 會被傳送。 相關技術則揭露於美國專利527463 1，5491 694…及 558151。 第1圖係習知具有分散式記憶體結構的乙太網路交換Page 5 420961 V. Description of the invention (2) 100BASE-T4 or 100BASE-FX network. Basically, 10BASE5, 10BASE2, 10BASE-T, 10 0BASE-TX, 100BASE-T4, or 100BASE-FX are all shared network systems (Shared. That is, the bandwidth of the entire network system is determined by the connected network). All nodes of the system are shared. The bandwidth of the base network system is 10MHz, and the bandwidth of the 100BASE network system is 100MHz. Therefore, when the number of nodes connected to the shared network system gradually exceeds the threshold, the network system's The performance will be saturated. To overcome this problem, divide the method and divide the entire network system into several independent sub-networks and be located in the collision of ungraceful rhymes ^, such as _this, the link is not-will share other Secondary network ^ Bandwidth. For example, 'If the shared 100BASE is not divided; the local network system is connected to 100 nodes. When the network is fully loaded, each node can only enjoy the heartband bandwidth. Relatively If this shared 100BASE area network system is divided into two sections, then each node can share 2MHz bandwidth when the network is fully loaded. A divided area network system using the Ethernet protocol is called a type Ethernet Network, and the network system is in a different area Packet transmission between segments ^ Ethernet switch switching. The main purpose of an Ethernet switch is to switch packets between different sections of the local area network system. Therefore, from one section to another Packets will be transmitted. Packets from one node to another node in the same segment will not be transmitted. The related technology is disclosed in US patents 527463 1,5491 694 ... and 558151. Figure 1 is known to have distributed memory Ethernet Switching

42096 五、發明說明（3) '~ -- 器的電路方塊圖。 如第1圖所示，乙太網路交換器主要是由交換積體電 路100及若干埠積體電路UOa，110b ’ 110c所構成。交換 積體電路100用來控制網路封包在各埠間的傳送與交換 而埠積體電路110a，ll〇b，n〇c則分別内建有記憶體、° 120a，120b，120c，用來提供網路封包的儲存與交換。在 這種乙太網路交換器的架構下，當網路封包要從埠13〇&傳 送至埠130c時，封包必須先儲存於記憶璧12〇a，然後再由 記憶二體1。因此，這種Λ (網硌交; ，的工作效能並不理想。此外，為使這種乙太網路交^器 能夠連結各種區域網路，如：1〇BASE5，1〇BASE2，、 10BASE-T，lOOBASE-τχ，1〇〇base_T441〇〇base fx ’每個 璋積體電路亦必須在設計之初 造成記憶體的浪費。 ，鑑，此，本發明的主要目的就是·提供一種乙—太雙路 交換器及交換·方法，一AX y有效率 區段之間。 ·*·祕明ί另一個目的就是提供一種,ϋ網路交換器及 交換其-可以>將1 0BASE網路系統連結$, iji〇RA3&. 系統’藉以使舊網路系統順利延展 ^為達成上述及其他目的，本發明乃提供一種乙太網路 乂检，其可以選擇性地轉一送或^濾里^^包。這種乙太 網路父換器是由·複數個網路埠、第一及第二記憶體裝置、 第-及第二記憶體控制裝置、交換裝i、及第：記憶體管42096 V. Description of the invention (3) '~-The circuit block diagram of the device. As shown in Fig. 1, the Ethernet switch is mainly composed of a switching integrated circuit 100 and a plurality of port integrated circuits UOa, 110b '110c. The switch integrated circuit 100 is used to control the transmission and exchange of network packets between ports. The port integrated circuits 110a, 110b, and noc have built-in memory, ° 120a, 120b, and 120c, respectively. Provides storage and exchange of network packets. Under the structure of such an Ethernet switch, when a network packet is to be transmitted from port 13 & to port 130c, the packet must be stored in memory 璧 12a, and then by memory two. Therefore, the work efficiency of this Λ (network communication;) is not ideal. In addition, in order to enable this Ethernet interface to connect to various local networks, such as: 10BASE5, 10BASE2, 10BASE -T, 100BASE-τχ, 100base_T441〇base fx 'Each convolution body circuit must also cause memory waste at the beginning of design. Therefore, the main purpose of the present invention is to provide a B- Too two-way switch and switching method, between AX y efficient sections. * * Miming ί Another purpose is to provide a network switch and exchange it-you can> 10BASE network System link $, iji〇RA3 &. The system 'by which the old network system can be smoothly extended ^ In order to achieve the above and other purposes, the present invention provides an Ethernet inspection, which can be selectively forwarded or filtered ^^ 包. This Ethernet parent switch is composed of a plurality of network ports, first and second memory devices, first- and second memory control devices, exchange device i, and first: memory Body tube

IH 第7頁 420961 五.發明說明（4) 理裝置所構成。網路埠係用 記憶體裝置係用以網路封勺^傳送網路封包。第一 之關連訊息。第二記憶f置—I ^來源位址及該些網路埠 網路封包。第-及第二= = 儲存網路璋所接收之 镇_ 1險舻联要 〜控制裝置則分別連接至第一 及第一 s己憶體裝置’用以枯 ^ β ^ ^ 2 . 一控舅第—及第二記億體裝置之寫 入及讀取H交換裝f係連接網路埠及第：記^f 、制裝置’藉以建立各網路封包之來源位址與網路埠：關： 訊息、並依據第-記憶體裝置之内容建立各 位址與網路痒之闕連訊息，以fi£n隐體裝置之内的 如此’篮路封包可經由銘^屢^<綠經撰擇 .模組1、.學習模而迅速地送至正確的區$ 並且’由於路、f表老化模組可周期性地將過於老 化之索弓丨重置’因此路徑表之索引亦可維持在最新的狀 態。 另外’為連結不同規格的區域網路，資料埠可由介質 獨立界面組成。而路徑表的索引則具有來源位址、位址旗 標、資料埠號碼、老化旗^標及_立化指數J分別暫存網路封 包對應之序號、序號之選取模式、資料埠、老化狀態及老 化程度。 路徑表老化模組可由下數計時器組成，其初始值由略 徑表各索引的預定壽命決定，用以周期地將所有索引的老 化指數減1及將索引指數小於1的索引重置’藉以建立其他IH Page 7 420961 V. Description of the invention (4) The structure of the device. For network ports, the memory device is used for network packets ^ to send network packets. First related message. The second memory f is set to the source address and the network packets of the network ports. The first-and second = = storage network 璋 Received town _ 1 The risk is connected ~ The control device is connected to the first and the first memory device 'to dry ^ β ^ ^ 2. One control舅 The first and the second write and read devices of the billion-body device H exchange device f is connected to the network port and the first: record ^ f, the device 'to establish the source address and network port of each network packet: Off: messages, and based on the contents of the -memory device to establish a connection message between each address and the Internet, so that the 'basket packets within the fi £ n hidden device can be passed through Ming ^^^ Compose the module 1. The learning module is quickly sent to the correct area $ and 'Because the road and f table aging modules can periodically reset the over-aged cable bow, so the path table index can also Stay up to date. In addition, it is used to connect the LANs of different specifications. The data port can be composed of media-independent interfaces. The index of the path table has the source address, the address flag, the data port number, the aging flag ^, and the _Lihua Index J to temporarily store the serial number corresponding to the network packet, the serial number selection mode, the data port, and the aging status. And aging. The path table aging module can be composed of a countdown timer. Its initial value is determined by the predetermined life of each index in the approximate path table. It is used to periodically reduce the aging index of all indexes by 1 and reset the indexes with index indexes less than 1. Create other

^20961 五、發明說明（5) 網路封包之索弓丨 學習模組可根據網路封 取模式得到若干索引，當來源位址及位址旗標之選 索引對應之老化指數係設：：包對應其甲-索弓|時，該 未對應其中任一索引時， =值，反之，當該網路封包 建立該封包對應之新索引。老化指數較小的索引便清除以 路徑模組可根據該網路 之選取模式得到若干索？丨，i i之目的位址及該位址旗標 時，該網路封包係由該索引網路封包對應其中一索引 另外，本發明亦提供貝科卑傅送。 換方法選擇太網路交換器網路封包交 方泛^:提供該網路。^ 錄該些料幻原位址與該此Μ成之路徑*，用以 路徑學習裝置，以因應各封;網路埠之關連訊息。提供一 容，建立路徑表中之一新索來源位址及路徑表之内 以及，提供一路徑選擇裝置，丨或更新路徑表中之一索引。 路徑表之内容，決定各封包以因應各封包目的位址及該 為了使本發明前述之目目的位址與網路埠之關連。 乃列舉較佳實施例，並 沐特徵與優點得更易明瞭， 下。 。所附圖式，進-步予以說明如 第1圖係習知具有分散式 器的電路方塊圖； s憶體結構的乙太網路交換 第2圖係本發明乙太網路 供器的電路方塊圖； 420961 五、發明說明（6) 第3圖係本發明乙太網路交換器中記憶體裝置使用配 置的示意圖； 第4圖係本發明乙太網路交換器之路徑表中每個索引 的格式圖； 第5圖係本發明乙太網路交換器之路徑表起始化模組 的狀態流程圖； 第6圖係本發明乙太網路交換器之路徑表老化模組的 狀態流程圖； 第7圖係本發明乙太網路交換器之目的位址及來源位 址的貢料流向圖， 第8圖係本發明乙太網路交換器之學習模組的狀態流 程圖；以及 第9圖係本發明乙太網路交換器之路徑選擇模組的狀 態流程圖。 符號說明 100交換積體電路； 110a ' 1 10b、110c璋積體電路； 120a、120b、120c 記憶體； 130a、130b、130c 埠； 1 0分享式記憶體裝置； 1 2 a封包緩衝器； 14 0路徑表存取控制電路； 1 5 0路徑表； 2 0分享式記憶體存取控制電路；^ 20961 V. Description of the invention (5) The bow of the network packet 丨 The learning module can obtain several indexes according to the network blocking mode. When the source index and the selected index of the address flag correspond to the aging index :: When a packet corresponds to its A-cord ||, when it does not correspond to any of the indexes, the value = value, otherwise, when the network packet establishes a new index corresponding to the packet. The index with a small aging index is cleared. How many routes can the path module obtain according to the selection mode of the network?丨, when the destination address of i i and the address flag, the network packet is corresponding to one of the indexes by the index network packet. In addition, the present invention also provides a Becopier. To choose another method, select the Ethernet switch network packet delivery method. ^: Provide the network. ^ Record the original address of the material and the path * formed by the M * to be used for the path learning device to respond to the connection information of each network port. Provide content to create a new cable source address in the routing table and within the routing table, and provide a routing device, or update an index in the routing table. The contents of the route table determine each packet to correspond to the destination address of each packet and the connection between the destination address and the network port for the purpose of the present invention. The preferred embodiments are listed, and the features and advantages are more easily understood. . The drawings are further explained as shown in FIG. 1. FIG. 1 is a conventional block diagram of a circuit with a decentralized device. FIG. Block diagram; 420961 V. Description of the invention (6) Figure 3 is a schematic diagram of the memory device usage configuration in the Ethernet switch of the present invention; Figure 4 is each of the path table of the Ethernet switch of the present invention Index format diagram; Figure 5 is a state flowchart of the path table initialization module of the Ethernet switch of the present invention; Figure 6 is a state chart of the path table aging module of the Ethernet switch of the present invention Flow chart; Figure 7 is a flow diagram of the destination address and source address of the Ethernet switch of the present invention, and Figure 8 is a state flowchart of the learning module of the Ethernet switch of the present invention; And FIG. 9 is a state flowchart of the path selection module of the Ethernet switch of the present invention. Symbol description 100 exchange integrated circuit; 110a '1 10b, 110c integrated circuit; 120a, 120b, 120c memory; 130a, 130b, 130c port; 10 shared memory device; 1 2a packet buffer; 14 0 path table access control circuit; 150 path table; 20 shared memory access control circuit;

第10頁 420981 五、發明說明（7) 21仲裁模組； 2 3資料多工模組； 2 5定址模組； 2 7指令模組； 3 0資料交換電路； 3 1路徑表起始化模組； 33仲裁模組； 3 5老化模組； 3 7路徑選擇模組； 3 9學習模組； 40a、40b網路埠； 41a、41b介質獨立界面； 42a、42b記錄單元； 44a、44b介質存取控制單元； 46a、46b接收直接記憶體存取單元； 48a、48b傳送直接記憶體存取單元； 5 0緩衝器管理單元； 5 2緩衝表； 60發光二極體顯示器； 7 0外部控制界面； 8 0 a、8 0 b接收直接記憶體存取； 8 2 a、8 2 b位址閂鎖電路； 84a目的位址（DA)路徑仲裁器84a ; 86a目的位址（DA)資料多工器；Page 10 420981 V. Description of the invention (7) 21 arbitration module; 2 3 data multiplexing module; 2 5 addressing module; 2 7 instruction module; 3 0 data exchange circuit; 3 1 path table initialization module Group; 33 arbitration module; 3 5 aging module; 3 7 path selection module; 3 9 learning module; 40a, 40b network port; 41a, 41b medium independent interface; 42a, 42b recording unit; 44a, 44b medium Access control unit; 46a, 46b receive direct memory access unit; 48a, 48b transmit direct memory access unit; 50 buffer management unit; 52 buffer table; 60 light emitting diode display; 70 external control Interface; 8 a, 80 b receive direct memory access; 8 2 a, 8 2 b address latch circuit; 84a destination address (DA) path arbiter 84a; 86a destination address (DA) data Worker

420961420961

8 4 b來源位址（s A )路徑仲裁器 86b來源位址（sa)資料多工器 9 〇記憶體存取仲裁器； 9 2記憶體存取指令控制； 9 4記憶體位址多工器9 4 ;以及 96記憶體資料多工器96。 實施例 ’複^ί 發明之太網路交換器係 憶―體控剋1覃、.』換裝置第-及第、·^ 係用以財子網路封= 包。第-記憶體展置 息β第二記憶體裝置係用以 該些網路璋之關連訊 包。第一及第二記憶俨奸岳丨f存網路埠所接收之網路封 記憶體裝置1以；“：則分別連接至第-及第二 取。另外，交換裝置传支姑第二記憶體裝置之寫入及讀 置，藉以建立各網路封Ϊ 車及第-記憶體控制裝 息、並依據第一記憶體=來源，址與網路埠之關連訊 址與網路埠之關連訊t ,、，内各建立各網路封包目的位 第二記憶體管理裝w ^ ^管理第—記憶體裝置之内容。 以管理第二記憶體裝置之=路淳及第二記憶體控制裝置 路則控制分享式呓情 谷。分享式記憶體存取控制電 以下，便 請參考第2圖，此A 士\；兒明本發明之實施例。 馬本發明乙太網路交換器的電路方 420961 五、發明說明（9) 塊圖。 在第2圖中，路交接於兩區域網路 間，具有：分享式記憶體裝置10，分享式記憶體存取控制 電路20，資料交換電路30，緩衝器管理單元5〇，緩衝表 52，發光二極體顯不器60，外部控制界面7〇及兩個資料埠 40a、4 0b。資料蟑40a、40b分別具有：介質獨立界面 41a、41b，記錄單元42a、42b，介質存取控制單元44a、 44b，接收直接記憶體存取單元46a、46b，傳送直接記憶 體存取單元48a、48b ’路徑表150 ’及路徑表存取控制^ 路 140。 乙太網路交換器的各部分說明如下。 [分享式記憶體裝置1 〇 ] 如第2圖所示，埠個記憶體裝置10主要用以作為封包 緩衝器12a(Packet buffer)。第3圖係記憶體裝置的示 意圖。在這個實施例中，記憶體裝置丨〇的大小為25 6K個字 元組（Word)，以1.5Κ位元組為單位、分成341個緩衝單 元’用以作為封包緩衝器12a並接收兩資料埠4〇a、40b的 網路封包存取。 [路徑表1 5 0 ] 第4圖係路徑表丨5〇内每個索引的格式。每個索引在路 表150中佔一個的^大小_。j中第1個字中具有12位 元的辨別碼（ID pattern) ’2位元的老化旗標（ageflg)，1 仇元的資料埠號碼（pno)及1位元的位址旗標（fig)。第2個 字為廉商辨識碼之第11第16仗元。第3個字為廠商辨識碼8 4 b source address (s A) path arbiter 86 b source address (sa) data multiplexer 9 memory access arbiter; 9 2 memory access command control; 9 4 memory address multiplexer 9 4; and 96 memory data multiplexer 96. Example ‘replicate’ The Ethernet network switch system of the invention is recalled—the body is controlled by 1 Qin,. ”The first and second devices of the device are used to encapsulate the network packets. The first-memory spreading information β second-memory device is used for related network packets. The first and second memories are stored on the network memory device 1 received by the network port; ": are connected to the first and second fetches respectively. In addition, the exchange device transmits the second memory. The writing and reading of the external device, so as to establish each network enclosure and the first-memory control device, and according to the first memory = source, the connection between the address and the network port and the connection between the address and the network port The second memory management device w ^ ^ manages the contents of the first memory device. To manage the second memory device = Lu Chun and the second memory control device. The road controls the shared memory valley. The shared memory access control is as follows, please refer to Figure 2, this A \; the embodiment of the present invention. The circuit of the Ethernet switch of the present invention. Fang 420961 Fifth, the description of the invention (9) Block diagram. In the second diagram, the road is connected between two local area networks, and includes: shared memory device 10, shared memory access control circuit 20, and data exchange circuit 30. , Buffer management unit 50, buffer table 52, light emitting diode display 60, The external control interface 70 and two data ports 40a and 40b. The data cocks 40a and 40b respectively have a media independent interface 41a and 41b, a recording unit 42a and 42b, and a media access control unit 44a and 44b, which receive direct memory storage The fetching units 46a and 46b transmit the direct memory access units 48a and 48b 'path table 150' and the path table access control ^ path 140. Each part of the Ethernet switch is described below. [Shared memory device 1 〇] As shown in FIG. 2, the memory device 10 is mainly used as a packet buffer 12 a (Packet buffer). FIG. 3 is a schematic diagram of the memory device. In this embodiment, the memory device 丨 〇 It has a size of 25 6K characters (Word) and is divided into 341 buffer units in units of 1.5K bytes. It is used as the packet buffer 12a and receives the network packet access of the two data ports 40a and 40b. [Path table 1 50] Figure 4 shows the format of each index in the path table 丨 50. Each index occupies one of the size ^ in the road table 150. The first word in j has 12 bits. ID pattern '2-bit ageflg, 1 hate Data port number (pno) and one-bit address flag (fig). The second word is inexpensive supplier identification code 11 of 16 battle yuan. The first three words for the manufacturers identification code

第13頁 五、發明說明（10) 之第17至第24位元及8位元的老化指數（age incjex)。第4 個字則為保留字（reserved)。 [分享式記憶體存取控制電路2 〇 ] 如第2圖所示’分享式記憶體存取控制電路2〇主要具 有仲裁模組（Arbi tat ing)21、資料多工模組（Da ta Muxing)23、定址模組（Addressing)25、指令模-組 (Command) 27，用以控制分享式記憶體裝置1〇的存取。 [路徑表存取控制電路140] 如第2圖所示’路徑表存取控制電路係負責路徑表内 資料的讀出與寫入動作之控制。 [資料交換電路30] 如第2圖所示’資料交換電路30主要具有五個模組： 路徑表起始化模組（Routing Table Ini tiat i〇n )31、仲裁 rbi_taU_ng_H,3、路徑表老 —TaW-e rising) 35、路徑選擇模組（Routing )37及學習模組 (Learning)39 ’其分別說明如下。 路徑表起始化模組31λ 路徑表起始化模組可利用兩種方法完成起始化動作： 一種是利用硬體電源進行重置’另一種則是利用軟體程式 進行重置。硬體重置一般是將電源先關後開，或將晶片@ 重置端電rf立先降後升。而軟體重置則是由軟體程式控制外 取控制^界面70而完成。 舉例來說’路徑表起始化模組3 1係將路禋表丨5 〇中所 有索引的老化旗標及老化指數清除為OOOOH。Page 13 V. Description of Invention (10) The 17th to 24th bits and 8-bit aging index (age incjex). The fourth word is reserved. [Shared memory access control circuit 2 〇] As shown in FIG. 2 'Shared memory access control circuit 2 〇 mainly has an arbitration module (Arbi tat ing) 21, a data multiplexing module (Da ta Muxing ) 23. Addressing module 25 and Command module 27 are used to control the access of the shared memory device 10. [Path table access control circuit 140] As shown in FIG. 2, the 'path table access control circuit is responsible for controlling the reading and writing of data in the path table. [Data exchange circuit 30] As shown in FIG. 2 'the data exchange circuit 30 mainly has five modules: a routing table initialization module (Routing Table Ini tiat i〇n) 31, an arbitration rbi_taU_ng_H, 3, the path table is old —TaW-e rising) 35, Routing module 37 and Learning module 39 'are explained below. Path table initialization module 31λ The path table initialization module can use two methods to complete the initialization action: one is to reset using a hardware power supply, and the other is to reset using a software program. Hardware reset is usually to turn off the power first and then turn on, or to reset the chip @ RESET 端 电 rf first and then rise. The software reset is completed by the software program controlling the external control interface 70. For example, the 'path table initiation module 31' clears the aging flags and aging indexes of all indexes in the routing table 501 as OOOOH.

420961 五、發明說明（11) 另外，路徑表起始化模組3 1的中斷則可由軟體程式控 制外部控制界面7 0完成。 圖0是路徑表起始化模組31的狀態流程圖。如圖 中所示’路徑表起始化模組3 1具有五種狀態：間狀態 (、位址重置狀態（Ad(jj^g t)、位—妓一索引清除狀H (E n t y C 1 xJ+.、最後索引檢查狀態(Las t Chk )及位址向下計數 狀敷·4Αγ4·4ϊ^1ι〇—。 當機器被迫中斷或路徑氣起始化模組3動作正常結 束時，路徑表起始化模组31係維持在簡一置狀態；而當軟體 控制外部控制界面70下達起始指令時，路徑表起始化模組 3 1則離開閒置狀態並進入位址重置狀態。 體方式 將路徑 而 生路徑 在 是否為 態，否 而 向下計 (世:被模 位址重處_史、無論路徑表起始化模組3丨是以硬 重置或以軟體方式啟動，路徑表起始化模組31首先 表1 5 0的具最1位址的一個索引著除。 位址索引清除狀態巾’路徑表起始化模組31則會產 表的清除信號以重置老化旗標及老化指數。 最後索引檢查狀態中，當索引清除後 路徑表15°的最後-個索弓卜若是則回到閒一置狀 則便進入位址向下計數狀態。 置狀 位址向下計數狀態中，蹊麻 數-單位，用以繼續下一以=位址計數器係 組33 ·、 個索引的起始化動作。 時，仲裁同時輸入交換電㈣420961 V. Description of the invention (11) In addition, the interrupt of the path table initialization module 31 can be controlled by a software program to control the external control interface 70. FIG. 0 is a state flowchart of the path table initialization module 31. As shown in the figure, the path table initiation module 3 1 has five states: an interim state (, an address reset state (Ad (jj ^ gt), a bit-prostitutes index clear state H (E nty C 1 xJ +., the last index check status (Las t Chk), and the address is counted down. 4Αγ4 · 4ϊ ^ 1ι〇— When the machine is forced to interrupt or the path gas initialization module 3 operation ends normally, the path table The initialization module 31 is maintained in a simple state; when the software controls the external control interface 70 to issue an initialization command, the path table initialization module 31 leaves the idle state and enters the address reset state. The path is generated from the path. Whether the path is in the state or not (counter: the location of the module is replaced by _ history, whether the path table initialization module 3 丨 is hard reset or started by software, the path The table initialization module 31 firstly deletes an index with the most significant address in Table 1 50. The address index clears the status towel. The path table initialization module 31 generates a table clear signal to reset the aging. Flag and aging index. In the final index check state, when the index is cleared, the last 15 feet of the path table. If it returns to the idle state, it will enter the address down counting state. In the state of address down counting, the ramie number-unit is used to continue to the next address = 33 Initialization of the index. At the same time, the arbitration inputs the exchange voltage at the same time.

420961 發明說明（12) " * 表老化槿釦π 在本實施例中’路一备個索引的壽命均必須 限制，藉以避免過舊資料佔妹1徑表丨50空間。而路徑表 老化模組35則是周期性地老化路徑表丨5()中}索引的去化 jl.羞CAgeLo-dax)。 口 路徑表老化模組3 5的動作係由路徑表計時器的觸發信 號驅動。觸發信波寬JU_2ILiLa3。路徑表計時器的脈衝 產生頻率則由哥命時間以決定。舉例來說，若壽命時間設 定為1 ’ _路徑表1 50中在丄秒後老化去而若壽 命時間設定^_UL0_l則，路徑上1 5 0中各索引會在1 〇 〇秒後老 化去除。 對於25MHz的輸入時脈，5也元漣浊ff熬裴可用以n 781.5KHz的時脈。781· 5 ΚΗζ ά時脈則用以做為20位元向下 計數器的輸入時脈。为下的起始值斑壽命時間相 同’而當向下計數器下數至零，1登信號便會砉生。同 時，壽命周期的起始值又會重新載入，用以產生下次觸發 信號。另外’壽命周期的起始值亦可由軟體透過外部控制 .界面70進行|定。420961 Description of the invention (12) " * Table aging hibiscus π In this embodiment, the lifespan of an index must be limited, so as to avoid excessive old data occupying the space of the table. The path table aging module 35 is a periodic deaging of the {}} index in the path table 5 () (jl. CAgeLo-dax). The operation of the path table aging module 35 is driven by the trigger signal of the path table timer. Trigger signal width JU_2ILiLa3. The pulse generation frequency of the path table timer is determined by the brother's life time. For example, if the life time is set to 1 '__ path table 1 50, it will age after leap seconds, and if the life time is set to ^ _UL0_l, then each index on the path 150 will be aged and removed after 100 seconds. For the input clock of 25MHz, 5 yuan can also be used to clock 781.5KHz. 781 · 5 ΚΗζ ά clock is used as the input clock of the 20-bit down counter. For the same starting value, the spot life time is the same, and when the down counter counts down to zero, a 1-signal signal will be generated. At the same time, the start value of the life cycle is reloaded to generate the next trigger signal. In addition, the initial value of the 'life cycle' can also be determined by software through an external control interface 70.

每次產生觸發信號時，路徑表150中4096個索引的老 化指數便會下數一個單位。若i命時1設定為2，且所J ，引旨數設定為191(BFH」，』096個索拉典 幾^全部老化ϋ二除，如下式： 40ns X32 X2 X4096 xl91 = 2. 00278016 (秒） 第6圖係路徑表老化模組35的狀態流程圖。如圖中所Each time a trigger signal is generated, the aging index of the 4096 indexes in the path table 150 is counted down by one unit. If i is set to 2 and i is set to 1, and the number of quotes is set to 191 (BFH "," 096 Soradians ^ all aging and divided by two, as follows: 40ns X32 X2 X4096 xl91 = 2. 00278016 ( Second) Figure 6 is a state flowchart of the path table aging module 35. As shown in the figure

第16頁 420961 五'發明說明（13) 不’ 救jgjm有七個狀熊:閒置狀態（idle)、 讀取老化指數狀態（RdIndx)、檢查老化指數狀態 CChkAl ive)、老化老化指數狀態（Wr Indx)、位址下數狀態Page 16 420961 Five 'Invention Note (13) No' Save jgjm has seven bears: idle state (read), read aging index state (RdIndx), check aging index state CChkAlive, aging aging index state (Wr Indx), address down status

UddrDn)、重讀老化旗標狀態（RdFUg)及寫入老化旗標狀 態（WrFlag)。 當老化功能停止時，路徑表老母模組35係維持在閒置 狀態。而當老化功能開始或觸發信號出現時，路徑表老化 ,纪35則離開閒置狀態奠進今讀取老化指數狀態。 _在讀取老化指數狀態中，路徑表老化模組3玉係讀取路 徑表1 5 0中各索引的j 3個字，用以得到其老化指數。 一待讀取老化指數後，在檢查老化指數狀態中，檢查路 杻表1 5 0中各索引的屬性。各索引的屬性分為三種：存 ’舌、老化、靜態。存活索引表示其老化指數大於〇 〇 H且小 於C0H。老化索引表示其老化指數等於_。#態索引則表 不其老化指數不在〇〇H~BFH之間。 在遞減老化指數狀態中，路徑表老化模組35會產生一 更新該索引老化指數的信號，而老化指數則在更新時下數 一個單位。 在位址下數狀態中，以匕位址邏輯電路會將欲存取素 老化旗標係表示索引的 老化旗標為1 Η。若索引屬性 化旗標為2Η或3Η表示該索引 索引的老化步驟都需要更新 屬性。若索引屬性為存活，則 為老化，則老化旗標為〇1|。老 非為老化或存活。並不是每個 老化旗標。路徑表老化模組35UddrDn), reread aging flag status (RdFUg), and write aging flag status (WrFlag). When the aging function is stopped, the path table old mother module 35 is maintained in an idle state. When the aging function starts or the trigger signal appears, the path table is aging, and Ji 35 leaves the idle state and reads the aging index state. _ In the state of reading the aging index, the path table aging module 3 jade reads the j 3 words of each index in the path table 150 to obtain its aging index. After reading the aging index, check the attributes of each index in Table 1500 while checking the aging index status. The attributes of each index are divided into three types: storage, aging, and static. The survival index indicates that its aging index is greater than 0 H and less than COH. The aging index indicates that its aging index is equal to _. The #state index indicates that its aging index is not between 〇〇H ~ BFH. In the state of decreasing aging index, the path table aging module 35 will generate a signal to update the index aging index, and the aging index will be counted by one unit when it is updated. In the address down state, the address logic circuit will set the aging flag of the desired accessor aging flag to 1 Η. If the index attribute flag is 2Η or 3Η, the aging steps of the index need to update the attributes. If the index attribute is alive, it is aging, and the aging flag is 〇1 |. Old is not aging or survival. Not every aging flag. Path table aging module 35

420961 五、發明說明¢14) 只會在老化指數由01H下數至00H時才會更新老化旗標。 另外，資料交換電路30的交換功能則說明如下。資料 交換電路30的交換功能主要是根據輸入網路封包之來源位 址（SA)及目的位址（DA)以達成，其具有：路徑模組37及學 習模組3 9。學習模組3 9是用來記錄來源位址（S A )並設定來 源位址（S A)與輸入資料埠之間的關連。路徑模滅3 7則是根 據目的位址（DA)選定其對應資料埠，用以輸出該網路封 包。 )的資料流程 圖。 S.. 在這個實施例中’所有網路封包是由接收直接記憶體 存取80a、80b所接收。位址閂鎖電路82a、 82b(DASALatch)則是由接收直接記憶體存取8〇a、80b中取 出48位元的目的位址（DA)及來源位址（SA)。位址閂鎖電路 82a、8 2b取出目的位址（DA)及來源位址（SA)時會經由DA路 徑仲裁器84a、DA資料多工器86a產生一路徑選擇要求、及 經由SA路徑仲裁器84b、SA資料多工器86b產生一學習要求 分別啟動目的位址（DA)的路徑選擇模組37及來源位址（SA) 的學習模組39。路徑選擇模組37及學習模組39的路徑表 1 50存取則透過記憶體存取仲裁器9〇、記憶體存取指令控 制9 2、記憶體位址多工器9 4及記憶體資料多工器9 6以完 成。 學習模組39 來源位址（SA)的學習模组39係進行路徑表1 50的搜叠_420961 V. Description of the invention ¢ 14) The aging flag will only be updated when the aging index is counted down from 01H to 00H. The switching function of the data switching circuit 30 is described below. The exchange function of the data exchange circuit 30 is mainly achieved according to the source address (SA) and destination address (DA) of the input network packet, which has: a path module 37 and a learning module 39. Learning module 39 is used to record the source address (SA) and set the relationship between the source address (SA) and the input data port. Path annihilation 3 7 is to select its corresponding data port according to the destination address (DA) to output the network packet. ) Data flow chart. S .. In this embodiment, 'all network packets are received by the receiving direct memory access 80a, 80b. The address latch circuits 82a and 82b (DASALatch) receive the direct memory access 80a and 80b to extract the 48-bit destination address (DA) and source address (SA). When the address latch circuits 82a and 8 2b take out the destination address (DA) and source address (SA), they will generate a path selection request via the DA path arbiter 84a and DA data multiplexer 86a, and via the SA path arbiter 84b. The SA data multiplexer 86b generates a learning request to activate the path selection module 37 of the destination address (DA) and the learning module 39 of the source address (SA). The path table 1 of the path selection module 37 and the learning module 39 is accessed through the memory access arbiter 90, the memory access command control 9, the memory address multiplexer 94, and the memory data. Tools 9 6 to complete. Learning module 39 The learning module 39 of the source address (SA) performs the search of path table 1 50_

第18頁 420961 五、發明說明（15) 及資料更新。48位元的來源位址（SA)具有兩攔：麻商辨識 碼及序號。路徑表1 5 0的搜尋則是根據來源位址（s A )的序 號。24位元的序號可分為兩部分，其分別可用做路徑表的 索引位址。搜尋旗標定義路徑表150的搜尋方式。若參考 24位元序號的低權值（Least significant bits)之12位元 以搜尋路徑表150，搜尋旗標設定為〇 ;而若參考24位元序 號的高權值（Mo st significant bits)之12位元以搜尋路 徑表1 5 0 ’搜尋旗標則設定為1。為搜尋一個索引，必須讀 取3個字資料’包括：8位元的老化指數，24位元的廠商號 碼’ 1位元的網路埠號碼’ 1位元的位址旗標，2位元的老 化旗標及12位元的序號。老化旗標代表該索引的屬性。若 該索引屬性為存活，老化旗標為1H。若該索引屬性為老 化’則老化旗標為0H。2H及3H的老化旗標則表示該索引非 為存活或老化。屬性存活的索引的老化指數大於〇 〇 Η且小 於C0H。屬性老化的索引的老化指數等於〇〇Η。 網路埠號碼的字元數依網路埠的個數而定，本實施例 ，有網路埠40a、40b，則網路埠號碼之字元數僅須！位元。 右網路埠個數為64個時，則網路埠號碼須要6個位元來表 示〇 一個正確索引必須滿足下列第—條件， (1) 24位元的廠商辨識碼等於來源位址的⑵至位元 (2) 老化指數不等於〇qh且不大於gFjjPage 18 420961 V. Description of the Invention (15) and Update of Information. The 48-bit source address (SA) has two blocks: a hessian identification number and a serial number. The search of the route table 150 is based on the sequence number of the source address (s A). The 24-bit serial number can be divided into two parts, which can be used as the index address of the path table. The search flag defines a search method of the path table 150. If you refer to the 12-bit low-value (Least significant bits) of the 24-bit serial number to search the path table 150, the search flag is set to 0; and if you refer to the high-value (Most significant bits) of the 24-bit serial number The 12-bit search path table 1 5 0 'The search flag is set to 1. In order to search for an index, 3 words of data must be read. Includes: 8-bit aging index, 24-bit manufacturer number, 1-bit network port number, 1-bit address flag, and 2-bit Aging flag and 12-bit serial number. The aging flag represents the attributes of the index. If the index attribute is alive, the aging flag is 1H. If the index attribute is aging ', the aging flag is 0H. The 2H and 3H aging flags indicate that the index is not alive or aging. The aging index of the attribute surviving index is greater than 〇 Η and less than COH. The aging index of the attribute aging index is equal to 〇〇Η. The number of characters of the network port number depends on the number of network ports. In this embodiment, there are network ports 40a and 40b. The number of characters of the network port number is only required! Bit. When the number of right network ports is 64, the network port number needs 6 bits to indicate. A correct index must meet the following conditions. (1) The 24-bit manufacturer ID is equal to the source address. ⑵ to bit (2) aging index is not equal to 0qh and not greater than gFjj

(3) 老化旗標等於1H (4) 網路埠號碼等於存取該來源位址之位址閃鎖電路(3) The aging flag is equal to 1H. (4) The network port number is equal to the address flash lock circuit accessing the source address.

第19頁 42〇961 五、發明說明（16) ' ' 對應之網路埠號碼 (5) 位址旗標等於〇 (6) 12位元的序號等於來源位址中24位元序號的高權 值之1 2位元 (7) 搜尋旗標等於0 或下列第二條件 - (1) 24位元的廠商辨識碼等於來源位址的25至48位元Page 19 42〇961 V. Description of the invention (16) '' Corresponding network port number (5) The address flag is equal to 0 (6) The 12-bit serial number is equal to the power of the 24-bit serial number in the source address 1 of 2 (7) The search flag is equal to 0 or the following second condition-(1) The 24-bit vendor ID is equal to 25 to 48 bits of the source address

(2) 老化指數不等於ooh且不大於bfH (3) 老化旗標等於1 η (4) 網路埠號碼等於存取該來源位址之位址閂鎖電路 對應之網路埠號碼 (5 )位址旗標等於1 (6) 12位元的序號等於來源位址中24位元序號的低權 值之1 2位元 (7) 搜尋旗標等於1 首先’學習模組39將來源位址（SA)中24位元序號的低 權值之12位元作為路徑表搜定址參考搜尋對應的第一索 引。而後’讀取該索引的3個字，並依據上述第一種條件 辨別是否為正择的索引。若是，學習模組39會將老化指數 更新為BFH ’並使其他資料維持不變；否則，學習模組39 便根據來源位址（SA)中24位元序號的高權值之12位元搜尋 對應的第二索引。 與第一索引的搜尋相同，第二索引的3個字隨後會被 讀取，並依據上述第二種條件辨別是否為正確的索引。若(2) Aging index is not equal to ooh and not greater than bfH (3) Aging flag is equal to 1 η (4) Network port number is equal to the network port number corresponding to the address latch circuit accessing the source address (5) The address flag is equal to 1 (6) The 12-bit serial number is equal to the low weight of the 24-bit serial number in the source address. 1 2 bits (7) The search flag is equal to 1. First, the learning module 39 sets the source address. The 12 bits of the low weight of the 24-bit serial number in (SA) are used as the first index corresponding to the path table search and reference search. Then, the three words of the index are read, and whether the index is a positive index is discriminated according to the first condition described above. If it is, the learning module 39 updates the aging index to BFH 'and keeps other data unchanged; otherwise, the learning module 39 searches for the 12-bit high-weight value of the 24-bit serial number in the source address (SA). Corresponding second index. Similar to the search of the first index, the 3 words of the second index are then read, and whether the correct index is determined according to the second condition described above. If

第20頁 A20961 五、發明說明（17) 是’學習模組39會將老化指數更新為BFH，並使其他資料 維持不變。 待兩次搜尋後，若學習模組39仍未找到正確索引，則 學習模組3 9會依據第一索引及第二索引的老化指數，將來 源位址的資料寫入索引指數較小的索引中。 第8圖係學習模組3 9的狀態流程圖。如圖中所示，學 習模組39具有：閒置狀態（IDLE)、第一搜尋狀態（PART 1)、第二搜尋狀態（PART 2)、讀取索引狀態（RD1W， RD2W， RD3W)、學習狀態（LEARN)、寫入索引狀態（WR1W， WR2W，WR3W)及完成狀態（DONE)。 當未有學習要求時，學習模組係維持在間置狀態。 在第一搜尋狀態中，學習模組3 9係設定搜尋旗標為 0 ’而來源位址的2 4位元序號之低權值之1 2位元則用以搜 尋路徑表的索引。 在第二搜尋狀態中，學習模組3 9係設定搜尋旗標為 1，而來源位址的24位元序號之高權值之12位元則用以搜 尋路徑表的索引》 在讀取索引狀態（RD1W， RD2W， RD3W)中，學習模組 39係分別讀取搜尋索引的第1，第2，第3個字。 、' 在學習狀態中’學^模H39係辨別搜尋索弓丨是否為正 確索引。 " 在寫入索引狀態中，是習分別寫入選定索 的第1’第2，第3個字。 、 而完成狀態則是學習步驟完成後的狀態。Page 20 A20961 V. Description of the invention (17) Yes ’Learning module 39 will update the aging index to BFH and keep other information unchanged. After two searches, if the learning module 39 still cannot find the correct index, the learning module 39 will write the data of the source address into the index with the smaller index index according to the aging index of the first index and the second index. in. Fig. 8 is a state flowchart of the learning module 39. As shown in the figure, the learning module 39 has: an idle state (IDLE), a first search state (PART 1), a second search state (PART 2), a read index state (RD1W, RD2W, RD3W), and a learning state (LEARN), write index status (WR1W, WR2W, WR3W) and completion status (DONE). When there are no learning requirements, the learning module is maintained in an interposed state. In the first search state, the learning module 39 sets the search flag to 0 ', and the 12-bit low-weight value of the 24-bit serial number of the source address is used to search the path table index. In the second search state, the learning module 3 9 sets the search flag to 1, and the high-weight 12-bit of the 24-bit serial number of the source address is used to search the index of the path table. In the states (RD1W, RD2W, RD3W), the learning module 39 reads the first, second, and third words of the search index, respectively. , 'In the learning state', the learning mode H39 identifies whether the search cable is the correct index. " In the writing index state, it is Xi writing the 1 ′, 2nd, and 3rd words of the selected line respectively. The completion status is the status after the learning steps are completed.

第21頁 ^20961 五、發明說明（18) 路徑選擇模組3 7 目的位址的路徑選擇模組3 7則只進行路徑表丨5 〇的搜 尋工作。具有兩欄：廠商識別碼及 序號。路徑考擇模組37則根據目的位址（DA)的序號進行搜 尋。24位元的序號分為兩個部分，分別可在路徑表搜 尋時用以定址。搜尋旗標係用以第一次路徑表搜尋及第二 次路徑表搜尋。若路徑選擇模組37係參考24位元序號的低 權值之12位元以搜尋路徑表丨5〇，搜尋旗標設定為〇 ;而若 路徑選擇模組37係參考24位元序號的高權值之12位元以搜 尋路徑表1 5 0，搜尋旗標則設定為1 ^讀取每個字後，目的 位址的路徑選擇模組3 7會檢查該字是否正確。第}個字具 有1位元的輸入資料埠號碼，1位元的位址旗標，2位元的 老化旗標及12位元的序號。老化旗標定義索引的屬性。若 屬性為存活’老化旗標為1 H。若屬性為老化，老化旗標為 0H。而2H及3H的老化旗標則表示該索引非為存活或老化: 對一個正確索引’第1個字必須滿足下列第—條件， (1 )老化旗標為1 Η (2 )位址旗標為〇 (3) 12位元的序號等於24位元目的位址序號的高權值 之1 2位元 (4) 搜尋旗標等於〇 或下列第二條件 (1 )老化旗標為1 Η (2 )位址旗標為1Page 21 ^ 20961 V. Description of the invention (18) Path selection module 37 7 The path selection module 37 of the destination address only performs the search of the path table 丨 50. It has two columns: manufacturer identification number and serial number. The path selection module 37 searches based on the serial number of the destination address (DA). The 24-bit serial number is divided into two parts, which can be used for addressing during path table search. The search flag is used for the first path table search and the second path table search. If the path selection module 37 refers to the low-weight 12-bit number of the 24-bit serial number to search the path table 丨 50, the search flag is set to 0; and if the path selection module 37 refers to the high 24-bit serial number, The 12-bit weight is used to search the path table 1 50, and the search flag is set to 1 ^ After reading each word, the path selection module 3 7 at the destination address will check whether the word is correct. The first character has a 1-bit input data port number, a 1-bit address flag, a 2-bit aging flag, and a 12-bit serial number. The aging flag defines the attributes of the index. If the attribute is alive, the aging flag is 1H. If the attribute is aging, the aging flag is 0H. The 2H and 3H aging flags indicate that the index is not alive or aging: For a correct index, the first word must meet the following conditions: (1) the aging flag is 1 Η (2) the address flag 0 (3) 12-bit serial number is equal to the high weight of the 24-bit destination address serial number 12 bits (4) The search flag is equal to 0 or the following second condition (1) the aging flag is 1 Η ( 2) The address flag is 1

420961 _ 五、發明說明（19) 一 " (3) 12位元的序號等於24位元目的位址序號的低權值 之1 2位元 (4) 搜尋旗標等於1 而第2及第3個字則必須如學習模組3 9之比對條件， 即: (1) 24位元的廠商辨識碼等於來源位址的25至48位元420961 _ V. Description of the invention (19)-(3) The 12-bit serial number is equal to the low-weight 1 of the 24-bit destination address serial number. (4) The search flag is equal to 1 and the second and second The 3 words must be compared with the comparison conditions of the learning module 39, that is: (1) The 24-bit vendor identification code is equal to 25-48 bits of the source address

(2) 老化指數不等於00H且不大於BFH 首先，路徑模組3 7根據2 4位元目的位址序號的低權值 之1 2位元搜尋第一索引’並讀取其中第1個字。然後，根 據上述第一條件判斷該索引是否正確。若是，則路徑模組 37會繼續讀取該索引的第2及第3個宇以判斷是否正確。若 是，取出第1個字中的網路埠號碼並進入完成狀態。反 之’路徑模組37則根據24位元目的位址序號的高權值之12 位元搜尋第二索引’並根據上述第二條件判斷該索引是否 正確。若是’則路徑模組37會繼續讀取該索引的第2及第3 個字以判斷是否正確。無論第二索引是否正確，路徑模組 37都會進入完成狀態，並回到閒置狀態。 第9圖係路徑選擇模組3 7的狀態流程圖。如圖令所 示，路徑選擇模組37具有：閒置狀態（IDLE)、第一搜尋狀 態（PART 1)、第二搜尋狀態（part 2)、讀取索引狀態 (RD1W ’ RD2W， RD3W)、索引學習狀態（3WLRN)及完成狀 態（DONE)。 當未有路徑要求時，路徑模組3 7係維持在閒置狀態^ 在第一搜尋狀態中，路徑模組37係設定搜尋旗標為(2) The aging index is not equal to 00H and not greater than BFH. First, the path module 37 searches for the first index according to the low weight of the 24-bit destination address sequence number, and reads the first word. . Then, it is judged whether the index is correct according to the first condition. If so, the path module 37 will continue to read the second and third fields of the index to determine whether it is correct. If yes, take out the network port number in the first word and enter the completion state. On the other hand, the “path module 37 searches for the second index according to the 12-bit high weight value of the 24-bit destination address sequence number” and judges whether the index is correct according to the above-mentioned second condition. If it is', the path module 37 will continue to read the second and third words of the index to determine whether it is correct. Regardless of whether the second index is correct, the path module 37 will enter the completion state and return to the idle state. FIG. 9 is a state flowchart of the path selection module 37. As shown in the figure, the path selection module 37 has: an idle state (IDLE), a first search state (PART 1), a second search state (part 2), a read index state (RD1W 'RD2W, RD3W), an index Learning status (3WLRN) and completion status (DONE). When there is no route request, route modules 37 and 7 remain idle. In the first search state, route module 37 sets the search flag to

第23頁 420961 五、發明說明（20) 0，而目的位址序號的低權值之1 2位元則用以搜尋路徑表 的索引。 在第二搜尋狀態中’路徑模組3 7係設定搜尋旗標為 1 ’而目的位址序號的高權值之12位元則用以搜尋路徑表 的索引。 在讀取索引狀態（RD1 W， RD2W， RD3W)中，'路徑模組 37係分別讀取搜尋索引的第1，第2，第3個字。 在索引學習狀態中，路徑模組3 7係辨別搜尋索引是否 為正綠索引。 而完成狀態則是路徑步驟完成後的狀態。 [網路埠40a、40b] 為連結不同規格的區域網路，如100 BASE及10BASE的 區域網路’資料埠4〇a、40b可由介質獨立界面組成。 [緩衝管理單元50及緩衝表52] 緩衝管理單元5 0及緩衝表5 2係用以管理記憶體裝置 。緩衝表52經由資料埠40a、40b及記憶體介面控制裝置 記錄記憶體裝置1 〇的使用狀態，而緩衝管理單元5 0則耦接 緩衝表52，並根據緩衝表52所記錄的狀態存取記憶體裝置 10的網路封包緩衝器12a。 [發光二極體顯示單元6〇] 發光二極體顯示單元60係用以顯示此乙太網路交換器 之狀態。 [外部控制界面7 〇 ] 外部控制界面係用以接受外部控制信號。Page 23 420961 V. Description of the invention (20) 0, and the lower 12 bits of the destination address number are used to search the index of the path table. In the second search state, the 'path module 3 7' sets the search flag to 1 'and the 12 bits of the high weight of the destination address number are used to search the index of the path table. In the read index state (RD1 W, RD2W, RD3W), the 'path module 37 reads the first, second, and third words of the search index, respectively. In the index learning state, the path module 37 determines whether the search index is a positive green index. The completion status is the status after the path step is completed. [Network ports 40a, 40b] are connected to different specifications of local networks, such as 100 BASE and 10BASE local networks. Data ports 40a and 40b can be composed of media-independent interfaces. [Buffer management unit 50 and buffer table 52] The buffer management unit 50 and the buffer table 52 are used to manage the memory device. The buffer table 52 records the usage status of the memory device 10 through the data ports 40a and 40b and the memory interface control device, and the buffer management unit 50 is coupled to the buffer table 52 and accesses the memory according to the status recorded in the buffer table 52 The network packet buffer 12a of the body device 10. [Light Emitting Diode Display Unit 60] The light emitting diode display unit 60 is used to display the status of this Ethernet switch. [External control interface 7 〇] The external control interface is used to receive external control signals.

第24頁 420961 五、發明說明（21) 综上所述，本發明可提供—種乙太網路交換器及交換 方法，其經路·射句._ϋ·读於务區段之間，__並一 且，可 JL 不-届格的 1^-isS^a-^~如：將-44m_A S F.,略系 統連結至 網 5 路系統_。 雖然本發明已以較佳實施例揭露如上，然其並非用以 限定本發明，任何熟習此項技藝者’在不脫離本發明之精 神和範圍内，當可作些許之更動與湖飾’因此本發明之保 護範圍當視後附之申請專利範圍所界定者為準。420961 on page 24 V. Description of the invention (21) In summary, the present invention can provide an Ethernet switch and a method of exchange, its path and sentence. _Ϋ · read between the service sections, _ _Also, it is possible to connect 1 ^ -isS ^ a- ^ ~ of the JL non-standard case, such as: -44m_A S F., slightly connected to the network 5-way system_. Although the present invention has been disclosed in the preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes and lake decorations without departing from the spirit and scope of the present invention. The protection scope of the present invention shall be determined by the scope of the attached patent application.

第25頁Page 25

Claims (1)

420961 六、申請專利範圍 1. 一種乙太網路交換器，包括： 複數路I，接收或傳送網路封包； 第一記憶體裝置，儲存該等網路封包來源位址及該等 網路埠之關連訊息； 第一 §己憶體裝置，储存該等網路封包所接收之網路封 包資料； - 第一記憶體控制裝置，連接至該第一記憶體裝置，控 制該第一記憶體裝置之寫入及讀取； 第二記憶體控制裝置，連接至該第二記憶體裝置， 制該第二記憶體裝置之寫入及讀取； 一交換裝1，連接至該複數個網路埠及該第—記憶 控制裝置，建立該等網路封包來源位址與該等網路埠 連訊息及依據該第一記憶體裝置之内容建立該等網路 目的位址與該等網路埠之關連訊息，並管理該第—掊= 裝置之内容；以及 體 第二記憶體管理裝置，連接至該複數個網路 二記憶體控制裝置，管理該第二記憶體裝置之内容。該第 2. 如申請專利範圍声丄遵^述之乙太網路交換 中，該等複數個網路埠包括第一網路埠及第二網路埠其 3. 如申請專利範圍义丄項所述之乙太網路交換芊。 中’該等複數個網路埠具有介質獨立介面，用 ’其 介質規格之區域網路。 疋得不同 其 用以 4.如申請專利範圍第2項所述之乙太網路交換 中’該第一料琿及第二網料具有介質獨立介沒420961 VI. Scope of patent application 1. An Ethernet switch, including: a plurality of channels I, receiving or transmitting network packets; a first memory device, storing the source addresses of the network packets and the network ports Related information; the first § self-memory device stores the network packet data received by the network packets;-a first memory control device connected to the first memory device and controlling the first memory device Write and read; a second memory control device connected to the second memory device to write and read the second memory device; an exchange device 1 connected to the plurality of network ports And the first-memory control device, establishing the network packet source address and the network port information, and establishing the network destination address and the network port based on the content of the first memory device Relevant information and manage the content of the first-掊 = device; and a second memory management device connected to the plurality of network two memory control devices to manage the content of the second memory device. The second. If the scope of the patent application is in accordance with the Ethernet exchange described in the following, the multiple network ports include the first network port and the second network port. The Ethernet exchange mentioned above. China ’s multiple network ports have media-independent interfaces and use the local network of their media specifications. It wo n’t be the same. It is used for 4. In the Ethernet network exchange described in item 2 of the scope of the patent application, the first material and the second material have independent media. 420961 六、申請專利範圍 連接不同介質 5. 如申請 中’該交換裝 體裝置之内容 6. 如申請 中，該交換裝 裝置内容之有 7. 如申請 中，該交換裝 換路徑選擇模 封包來源位址 更新一索引以 路埠之 目的位 該網路 關連； 址搜尋 埠以轉 如申請 交換裝 個網路 中，該 該複數 順序。 9. 提供一路 該等封包來源 提供一路 徑表之 一種乙 内容 規格之區 專利範圍 置包括一 〇 專利範圍 置包括一 效期間。 專利範圍 置包括一 組；該封 搜尋該第 記錄該等 該封包交 該第一記 送或過濾 專利範圍 置包括一 埠接收該 域網起^之乙太網路交換器，其 °槔組，用以重置該第一記憶 項所碟之乙太網路 化拉組，用以衰減該第一記體 封勹j '述之乙太網路交換器，其 ’匕父換路徑學習模細 八 包交接站模組及一封包交 「記愧體袭置之内= 網路封4·η 错以建立或 換路該等複數個網 該等、，藉以獲得對應之 ΐ1 包項交:述：乙太網路交換器，# 等網跟4、處理仲裁模組，用以仲裁 子包時之網路封包交換處理420961 VI. Apply for a patent to connect different media 5. If you apply for 'The contents of the exchange loading device 6. If you apply for, the contents of the swap loading device 7. If you apply for, the exchange loading path selects the source of the mold package The address updates an index to connect the network with the destination of the port; the address search port is transferred to the network in order to apply for the exchange in the plural order. 9. Provide all the way. These packet sources provide a path table of a B content specification area. Patent scope includes 10 patent scope includes one validity period. The patent scope includes a group; the search for the first record and the packets are sent to the first record or filtering. The patent scope includes a port to receive an Ethernet switch from the domain network, its group, It is used to reset the Ethernet pull set of the first memory item, and to attenuate the Ethernet switch described in the first memory seal j ', and its learning mode is changed. The eight-pack transfer station module and one package "within the body of the ashamed body = network cover 4 · η is wrong to establish or reroute these multiple networks, etc., to obtain the corresponding ΐ1 package delivery: description : Ethernet network switch, # 等 网 与 4, processing arbitration module, used to arbitrate sub-packet network packet exchange processing 徑表，該 位址與該 徑學習裝 建立該路Trail table, the address and the trail learning equipment to build the route 路換方法’包括： 笪to 係由複數個索引組成以記錄 寻網跟接 略埠之關連訊息； 置》 現此 徑表 應該等封包來源位址及該路 中之一新索引或更新該路徑表 inra I 420961 六、申請專利範圍 ------- 中之·一索引；以及 提供一路徑選擇裝置，因應該等封包目的位址 徑表之内容，決定該等封包目的位址與該等網 =路 連。 埤之關 ίο.如申請專利範圍第9項所交換 交換方法，進一步包括：提供一路徑表重置裝查，二、'包 置該路徑表之索引内容。 以重 11.如申請專利範圍第9項所_述之乙太 交換方法，進一步包括：提供一路徑表索引有效周，2 裝置，用以衰減該路徑表之索引有效周期。 1 2 ·如申请專利範圍第沭之乙太_^|^交換器封包 交換方法，進一步包括：提供一路徑學習與選擇仲裁裝i 置’用以仲裁該複數個網路埠接收該等網路封包時之網 封包交換處理順序》 ‘ 1 3.如申请專利範圍j 9項所述之乙太網路交換器封包 交換方法’進一步於該索引中提供一老化指數，一來源^ 路埠編碼’一位址旗標及一辨識碼。 14.如申請專利範圍多9項所述之乙太網路交換器封包 交換方法’其中，該路徑學習裝置建立該路徑表中之_索 引或更新該路徑表中之一索引時，執行下列步驟： a)設定位址旗標為第/位址旗標，設定搜尋索引位址 為第一索引位址，設定辨識碼為第一辨識碼，設定來源網 埠編碼為第一網路埠及設定有效之老化指數範圍為第一老 化指數範圍；The method of route change includes: 笪 to is composed of multiple indexes to record the connection information between the network search and the access port; the current route table should wait for the source address of the packet and a new index on the route or update the route Table inra I 420961 6. Scope of patent application ------- Index of one of them; and provide a path selection device, according to the contents of the address table of the packet destination address, determine the destination address of these packets and the Waiting network = Lulian.埤 之 关 ίο. If the exchange method is applied for the item 9 in the scope of patent application, the method further includes: providing a path table resetting and checking, and second, 'including the index content of the path table. Focus on 11. The Ethernet exchange method as described in item 9 of the scope of patent application, further comprising: providing a path table index valid period, and 2 means for attenuating the path table index valid period. 1 2 · If the scope of the patent application is the first Ethernet_ ^ | ^ switch packet exchange method, further comprising: providing a path learning and selecting arbitration device 'to arbitrate the plurality of network ports to receive the networks Packet Processing Sequence for Network Packet Exchange at the Time of Packets "'1 3. The Ethernet Switch Packet Exchange Method described in item 9 of the patent application scope' further provides an aging index, a source ^ port code 'in the index An address flag and an identification code. 14. The Ethernet switch packet exchange method according to more than 9 items in the scope of patent application, wherein when the path learning device establishes an index in the path table or updates an index in the path table, execute the following steps : A) Set the address flag as the first / address flag, set the search index address as the first index address, set the identification code as the first identification code, set the source network port code as the first network port, and set The effective aging index range is the first aging index range; 第28頁 42096ί 六 申請專利範圍 b) 依第一索引位址於該路徑表讀取該索引； c) 比較该索Μ之老化指數是否在第一老化指數範圍之 内； d)比較該索引之來源網路埠編碼，位址旗標及辨識碼 是否分別相同於第一網路埠’第一位址旗標及第一辨識 碼；以及 e )因應步驟C )及步驟d)之結果以更新該索引之老化指 數或建立一新索引。 曰 15.如申請專利範圍篇_9項所述之乙太網路交換器封包 交換方法’其中’該路徑選擇裝置決定該些封包目的位址 與該些網路埠之關連時’執行下列步驟： (a) 設定位址旗標為第二位址旗標，設定搜尋索引位 址為第二索引位址’設定辨識碼為第二辨識碼，設定有致 之老化指數範圍為第二老化指數範圍； (b) 依第二索引位址於該路徑表中讀取該索引； (c) 比較該索引之老化指數是否在第二老化指數範園 之内； (d) 比較該索引之位址旗標及辨識碼是否分別相同於 第二位址旗標及第二辨識碼；以及 (e) 因應步驟（c)及步驟（d)之姓里 ^ 、u j I、'.σ果，決定該索引中 來源網路埠編碼與該網路封包目的位址之關連。Page 28 42096ί The scope of the six patent applications b) Read the index in the path table according to the first index address; c) Compare whether the aging index of the cable M is within the first aging index range; d) Compare the index Whether the source network port code, address flag and identification code are the same as the first network port's first address flag and first identification code, respectively; and e) update according to the results of step C) and step d) The aging index of the index may create a new index. 15. According to the _9 item of the scope of the patent application, the packet switching method of the Ethernet switch 'wherein' when the path selection device determines the destination address of the packets and the connection of the network ports' performs the following steps : (A) Set the address flag as the second address flag, set the search index address as the second index address', set the identification code as the second identification code, and set the consistent aging index range as the second aging index range ; (B) read the index in the path table according to the second index address; (c) compare whether the aging index of the index is within the second aging index range; (d) compare the address flag of the index Whether the target and identification code are the same as the second address flag and the second identification code, respectively; and (e) determine the index according to the results of steps (c) and (d) ^, uj I, '.σ The source network port number is related to the destination address of the network packet.