WO2014127605A1

WO2014127605A1 - Mac address hardware learning method and system based on hash table and tcam table

Info

Publication number: WO2014127605A1
Application number: PCT/CN2013/080000
Authority: WO
Inventors: 周万涛; 徐剑辉
Original assignee: 烽火通信科技股份有限公司
Priority date: 2013-02-21
Filing date: 2013-07-24
Publication date: 2014-08-28
Also published as: CN103117931B; CN103117931A; MY170382A; CL2015000917A1

Abstract

Disclosed in the present invention are a MAC address hardware learning method and system based on a hash table and a TCAM table, relating to the field of MAC address learning. When there is no MAC address conflict, a learnt MAC address is stored in the hash table by means of SRAM or DRAM on hash table hardware; when there is a MAC address conflict, the conflicting MAC address is cached in the TCAM table by means of a TCAM memory on TCAM hardware. Because a parallel search is performed on the TCAM hardware, idle entries can be located by a single search; and the number of the entries of the TCAM is the actual number of conflicting MAC addresses capable of being cached. The present invention can be implemented on a general programmable exchange chip, does not need the support of a dedicated hardware circuit, has a high learning rate, occupies few memory resources, is flexible in application by adopting a general algorithm, and can achieve complete controllability of the conflict probability.

Description

MAC address hardware learning method and system based on hash table and TCAM table

The present invention relates to the field of MAC (Media Access Control) address learning, and particularly relates to a MAC address hardware learning based on a HASH (Hash) table and a TCAM (Ternary Content Addressable Memory) table. Method and system. Background technique

With the development of economic globalization, more and more enterprises are expanding in scope, and it is urgent for telecom operators to provide link connections, so as to include all branches of the enterprise and form a dedicated network within the enterprise. Under such market demand, VPLS (Virtual Private LAN Service) technology emerges as the times require. VPLS technology combines the advantages of Ethernet technology and MPLS (Multi-Protocol Label Switching) technology. The IP (Internet Protocol) / MPLS network connection provided by the operator is a geographically isolated LAN (Local Area Network), which simulates the traditional LAN functions and makes them look like a LAN. Work like that.

The VPLS guides the forwarding of user packets by looking up the MAC address forwarding table of the VSI (Virtual Switch Instance). The MAC address forwarding table is created through the MAC address learning function. Therefore, the MAC address learning function is a key factor in implementing VPLS.

The two trigger conditions for MAC address learning are:

(1) When a VSI receives an unknown MAC address, the source MAC address is unknown;

(2) A VSI receives a known source MAC address, but the corresponding forwarding entry needs to be updated. For example, the receiving VP (Virtual Port, virtual switching port in the VSI) has changed.

Address conflicts are unavoidable in MAC address learning because the MAC address has 48 bits, and the VSI is generally larger than lObit according to the device capacity (that is, at least lk VSIs are supported). In theory, if you want to avoid MAC address conflicts, you need to exhaust all MAC address combinations, at least 2 ^Λ 58 = 256 Ρ (computer storage unit, lPeta = 1024T) entries, obviously it is impossible to use hardware to cache such a large number of Entry. Currently, there are three methods for resolving MAC address conflicts:

(1) Relying on the central processing unit (Central Processing Unit) to help resolve address conflicts: A MAC address software learning method that solves the address conflict problem by implementing a "collision list" algorithm on the master CPU. Since the main control CPU is usually used to run complex routing protocols, the burden itself is already heavy. Therefore, valuable host CPU processing resources are wasted, and the learning rate of the MAC address software learning method is compared with the MAC address hardware learning method. Relatively low.

(2) Single hash table cache conflict method: A MAC address hardware learning method, which uses a hash table to cache MAC address conflicts, that is, by expanding the capacity of the hash table to reduce the probability of collisions. Simple, most common cache conflict method.

The hash algorithm maps binary values of arbitrary length to smaller binary values of fixed length to achieve data compression, and suppresses data conflicts generated by mapping by breaking and discretizing the original data. Commonly used hashing algorithms include MD2 (Message Digest Algorithm 2, Message Digest Algorithm Second Edition), MD5 (Message Digest Algorithm 5, Message Digest Algorithm Fifth Edition), CRC (Cyclic Redundancy Check) 32, etc. . In practical applications, CRC32 is often used as a hash algorithm. Taking the 64K address table as an example, the VSI and the source MAC are used to calculate the 32-bit hash hash value by using the CRC32 algorithm, and the high- and low- 16-bit XOR of the 32-bit result is followed by the lower 16-bit (2 ^Λ 16=64Κ) as the index value. , to index the entries in the AM. If the 16-bit discrete values of the two MAC addresses are the same, one of the MAC addresses cannot be cached. In this case, only the bitl7 of the CRC32 Hash value of the two MAC addresses can be expected to be different, and the table entry capacity is extended by 64k from 64k, thereby further Relieve conflicts.

The single hash bucket cache conflict method reduces the collision probability by extending the number of discrete values. For each lbit discrete value, the conflict table capacity needs to be doubled, so the conflict table resource is wasted. Moreover, since this method relies entirely on discrete values for the location of entries, no matter how the entries are expanded, it is not always possible to cache multiple MAC addresses with identical discrete values. In addition, when the MAC address sample changes, the probability of collision allowed in the system design cannot be guaranteed.

(3) Multi-bucket cache conflict method: A MAC address hardware learning method, which uses multiple buckets to mitigate conflicts. It is implemented on some ASIC (Application Specific Integrated Circuit) chips, that is, using hardware circuits to simulate multiple Conflict buckets, each conflict bucket has the ability to have certain cache conflicts. Taking the 64k address entry as an example, the CRC32 algorithm is used to discretize the MAC address, and the lower 16 bits of the 32-bit Hash value is taken as the discrete result. First, the lower 13 bits (2 ^Λ 13=8Κ) of the 16-bit discrete value are used to locate 8 collision buckets. There are 8 conflicting entries in each conflict bucket; then in the currently located conflicting bucket, Search for the free entries in the eight conflicting entries. If there are free entries, they are placed in the empty entries. If not, the conflicts cannot be cached.

The multi-bucket cache conflict method largely solves the problem that a single hash bucket cannot cache the same MAC address with the same discrete value, and the utilization efficiency of hardware resources is also relatively high. However, this method requires a large number of conflict buckets to work together, and requires dedicated The hardware circuit supports, and the search method of the idle entry in each conflict bucket requires a dedicated hardware circuit to implement the search algorithm. Since the algorithm implemented is not disclosed, the application is not flexible; in addition, since the method only utilizes the current conflict The free table entry resource in the bucket caches the conflict. Therefore, only up to 8 MAC addresses with the same discrete value can be cached. The ability to cache conflicts is limited for more severe cases with the same discrete value. Summary of the invention

The object of the present invention is to overcome the deficiencies of the above background art, and to provide a MAC address hardware learning method and system based on a hash table and a TCAM table, which are implemented on a universal programmable switching chip without the support of a dedicated hardware circuit. The learning rate is higher; the occupied memory resources are less, the general algorithm is adopted, the application is flexible, and the collision probability can be completely controlled.

The MAC address hardware learning method based on the hash table and the TCAM table provided by the present invention includes the following steps:

51. Calculate a hash value by using the virtual forwarding instance number and the source MAC address to which the packet belongs, and search the hash table by using the hash value as an index;

52. Searching for the key value of the TCAM entry: The virtual forwarding instance number is the virtual forwarding instance number to which the packet belongs; the MAC address is the source MAC address of the packet; the table entry full flag is set to 1, indicating that the TCAM table has been learned. Search in the item;

53. Determine whether the valid flag in the TCAM entry data is set, if set, go to S4; No Bay |J, go to S6;

54. Compare whether the virtual switch port in the TCAM entry data is equal to the packet receiving the virtual switch port. If yes, the entry does not need to be updated, and the process goes to S15; otherwise, the TCAM entry needs to be updated, and the process goes to S5;

55. Update the virtual switch port in the current TCAM entry data to actually receive the virtual switch port, and go to S15;

56. Determine whether the data valid flag in the hash table is set, if set, go to S8; otherwise, Go to S7;

57. The source MAC address of the packet, the virtual forwarding instance number of the packet, and the virtual switching port received by the packet are stored in the current hash entry, and the valid entry of the entry is set to 1, and the process proceeds to S15.

58. Compare the virtual forwarding instance number and the MAC address in the hash table with the virtual forwarding instance number and the source MAC address to which the packet belongs. If it matches, the current address has been learned in the hash table, and go to S9. Otherwise, go to S11;

59. Compare whether the virtual switch port in the current hash table is equal to the actual virtual switch port received by the packet. If they are equal, the hash entry does not need to be updated, and the process goes to S15; otherwise, the process goes to S10;

510. Update the virtual forwarding port number in the current hash entry to the virtual port number actually received by the packet, and go to S15.

511. Generate a hash collision, and search for a key value of the TCAM entry to find an idle TCAM entry cache conflict: the virtual forwarding instance number in the key value is 0, the MAC address is 0, and the entry full flag is 0, indicating that the idle TCAM is in the idle TCAM. Search in the table entry;

512. Determine whether the valid flag in the TCAM data is set. If set, it indicates that the first idle entry is found, and the process goes to S14; otherwise, it goes to S13;

513, TCAM memory does not have a free entry, give up learning, go to S15;

514. Cache the MAC address into the current idle TCAM entry: The virtual forwarding instance number of the key value is filled in as the virtual forwarding instance number of the packet, and the mask is set to 1 _{; the} MAC address is filled in as the source MAC address of the packet, and the mask is set. 1; the entry full flag is set to 1, the mask is set to 1; the virtual switch port in the data is filled in as the packet actually receives the virtual switch port, go to S15;

S 15, MAC address hardware learning process ends.

On the basis of the foregoing technical solution, before step S1, the step of initializing the entry is further included: determining whether the index of the initialization entry is greater than the maximum value of the entry, and if yes, returning an exception; otherwise, determining whether the index value of the initialization entry is located in the hash. Within the table index range, if yes, initialize the hash table, directly locate the hash table with the initialization index value, and clear all the contents of the table entry; otherwise, determine whether the index of the initialization entry is the last item of the TCAM, if yes , initializes the last TCAM entry, all fields in the key value are initialized to 0, the mask is 0, all fields in the data are assigned a value of 0, indicating that the entry is the last entry of the TCAM; otherwise, the key value is The middle entry full flag is cleared to 0, and the mask is set to 1, indicating that the initial state of the current entry is empty, the remaining fields in the key value are cleared to 0, the mask is cleared to 0, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0. This entry is not the last entry, and initialization is complete. On the basis of the foregoing technical solution, the step of aging includes the step of aging: determining whether the index of the aging entry is greater than the maximum value of the entry, and if yes, returning an exception; otherwise, determining whether the index value of the aging entry is located in the hash table index If yes, the aging hash table is used to directly locate the hash table with the initial index value, and clear the contents of the entry to 0; otherwise, determine whether the aging entry index is the last item of the TCAM, and if so, Indicates that the last item of the TCAM is an entry matching any key value, and the aging is ended; otherwise, the full value of the table entry in the key value is cleared to 0, the mask is set to 1, the remaining fields are cleared to 0, the mask is cleared to 0, and the entry status is set. If the value is set to null, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating that the entry is not the last entry, and the aging ends.

Based on the above technical solution, the hash table is stored in SRAM or DRAM. Based on the foregoing technical solution, the TCAM entry is stored in a TCAM memory. The present invention also provides a MAC address hardware learning system based on a hash table and a TCAM table, including a search unit, a TCAM entry determination unit, a hash table determination unit, a hash table comparison unit, and a hash collision processing unit, where:

The searching unit is configured to: calculate a hash value by using a virtual forwarding instance number and a source MAC address to which the packet belongs, and search the hash table by using a hash value; organize the search for the key value of the TCAM entry: virtual forwarding instance The number is the virtual forwarding instance number to which the packet belongs. The MAC address is the source MAC address of the packet. If the entry full flag is set to 1, the search is performed in the learned TCAM entry.

The TCAM entry determining unit is configured to: determine whether a valid flag in the TCAM entry data is set, and if set, compare whether the virtual switch port in the TCAM entry data is equal to the packet receiving virtual switch port, and if yes, The table entry is not updated; otherwise, the virtual switch port in the current TCAM entry data is updated to actually receive the virtual switch port; if not set, the hash table judgment trigger signal is generated and sent to the hash table judgment unit;

The hash table determining unit is configured to: when receiving the hash table determining the trigger signal, determine whether the data valid flag in the hash table is set, if not set, the source MAC address of the packet and the virtual forwarding of the packet The virtual switch port received by the instance number and the packet is stored in the current hash entry, and the valid entry of the entry is set to 1. If set, the comparison trigger signal is generated and sent to the hash table comparison unit.

The hash table comparison unit is configured to: when receiving the comparison trigger signal, compare whether the virtual forwarding instance number and the MAC address in the hash table match the virtual forwarding instance number and the MAC address to which the packet belongs, and if yes, Indicates that the current address has been learned in the hash table. Compares whether the virtual switch port in the current hash table is equal to the actual virtual switch port received by the packet. If they are equal, the learning is ended. Otherwise, the update is performed. The virtual forwarding port number in the pre-hash entry is the actual virtual port number received by the packet, and the learning is ended; if not, the hash conflict processing trigger signal is generated and sent to the hash conflict processing unit;

The hash conflict processing unit is configured to: when receiving the hash conflict processing trigger signal, organize to search for a key value of the TCAM entry to find an idle TCAM entry cache conflict: the virtual forwarding instance number in the key value is 0, the MAC address 0, the entry full flag is 0, indicating that the search is in the idle TCAM entry; then it is determined whether the valid flag in the TCAM data is set, if set, the first free entry is found, and the MAC address is cached to the current In the idle TCAM entry, the virtual forwarding instance number of the key is filled in as the virtual forwarding instance number of the packet, the mask is set to 1, the MAC address is filled in as the source MAC address of the packet, the mask is set to 1, and the entry full flag is set to 1. The mask is set to 1. The virtual switch port in the data is filled in as the packet actually receives the virtual switch port, and the learning is ended. If it is not set, it indicates that the TCAM memory has no free entry, and the learning is abandoned.

On the basis of the foregoing technical solution, the method further includes: an initialization unit, configured to: determine whether the index of the initialization entry is greater than the maximum value of the entry, and if yes, return an exception; otherwise, determine whether the index value of the initialization entry is in the index range of the hash table Inside, if yes, initialize the hash table, directly locate the hash table with the initialization index value, and clear all the contents of the table entry; otherwise, determine whether the index of the initialization entry is the last item of the TCAM, and if so, initialize The last TCAM entry, all the fields in the key value are initialized to 0, the mask is 0, all fields in the data are assigned 0, indicating that the entry is the last entry of the TCAM; otherwise, the key in the key entry The full flag is cleared to 0, and the mask is set to 1, indicating that the initial state of the current entry is empty, the remaining fields in the key value are cleared to 0, the mask is cleared to 0, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating the entry. Not the last entry, initialization is complete.

On the basis of the foregoing technical solution, the aging unit is further configured to: determine whether the index of the aging entry is greater than the maximum value of the entry, and if yes, return an exception; otherwise, determine whether the index value of the aging entry is in the index range of the hash table. If yes, the aging hash table directly locates the hash table with the initial index value, and clears the contents of the entry to 0; otherwise, determines whether the aging entry index is the last item of the TCAM, and if so, The last item of the TCAM is an entry that matches any key value, and the aging is ended; otherwise, the full value of the table entry in the key value is cleared to 0, the mask is set to 1, the remaining fields are cleared to 0, the mask is cleared to 0, and the state of the entry is set. If it is empty, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating that the entry is not the last entry, and the aging ends.

Based on the above technical solution, the hash table is stored in an SRAM or a DRAM. Based on the foregoing technical solution, the TCAM entry is stored in a TCAM memory. The advantages of the present invention over the prior art are as follows: (1) The MAC address hardware learning method of the present invention is completely implemented on a switch chip, and does not need to rely on the assistance of any host CPU, and the learning rate is relatively high.

(2) The present invention uses a hash table to store the learned MAC address when there is no MAC address conflict. The hash table hardware is implemented by SRAM or DRAM, and the capacity and bandwidth can be made larger, and the hash is preferred. Tables store address entries and take full advantage of AM's large-capacity hardware resources. When there is a MAC address conflict, the TCAM table is used to cache the conflicting MAC address. The TCAM table hardware is implemented by TCAM memory. Since the TCAM hardware is parallel search, the free entry can be located in one search, and the number of TCAM entries is It is the number of conflicting MAC addresses that can actually be cached. Therefore, not only can the advantages of TCAM fast parallel search be fully utilized, but also the collision probability can be fully controlled. In addition, since the hash table is not used to cache the conflicting MAC address, it occupies Less memory resources.

(3) The present invention can be implemented on a general-purpose programmable switching chip, does not require the support of a dedicated hardware circuit, and adopts a general-purpose algorithm, and the application is very flexible.

(4) The present invention uses the TCAM table to buffer the conflicting MAC address. The total number of conflicting MAC addresses that can be cached is less than the total amount of TCAMs opened, and the probability of collision is controllable. The number of conflicting MAC addresses that can be cached is determined. The constant, does not change with the change of the MAC address sample, the ability to cache the conflicting MAC address is guaranteed, even in the worst case of conflict, the TCAM table can still cache almost all MAC address conflicts. DRAWINGS

1 is a flow chart of initialization of an entry in an embodiment of the present invention.

2 is a flow chart of a MAC address hardware learning method based on a hash table and a TCAM table in an embodiment of the present invention.

3 is a flow chart of aging in an embodiment of the present invention. detailed description

The present invention will be further described in detail below with reference to the drawings and specific embodiments.

TCAM is a three-state content-addressable memory. Each bit of data in the TCAM has three states: In addition to the normal "0" and "1", there is also a "don't care" (ie, there is no need to match the bit). State, so called "three states." The "three-state" function is implemented by a mask, that is, the mask is 0, indicating "dont' t care "given key value; a mask of 1, indicating that an exact key value needs to be matched exactly. This feature of TC AM allows it to perform both exact and fuzzy matching. Another feature of TCAM is that "Parallel lookup", that is, given a key value, the TCAM will match all hardware entries at the same time, and return the index of the entry with the smallest address index value.

In the embodiment of the present invention, the AM storage hash table is used as the main memory for MAC address learning, and the characteristics of the "three-state storage" and the "parallel search" of the TCAM memory are combined to cache the conflicting MAC address. In terms of function partitioning, the hash table acts as the main memory for the MAC address, and the TCAM table is responsible for the core functions of the cache address conflict. The hash table and the TCAM table cooperate with each other to complete the function of MAC address learning, and the implementation requires initialization, address learning, and address aging to achieve the three.

The hash table is stored in SRAM or DRAM, and the TCAM entry is stored in the TCAM memory. As shown in Table 1, the data structure of the hash table includes the virtual forwarding instance (Vsi), the Mac address, the table valid flag (Valid), and the virtual switch port (Vp).

Table 1. Data structure definition table of hash table

As shown in Table 2, the data structure of the TCAM entry includes the key value (Key) and the data (Data). The key value includes the virtual forwarding instance (Vsi), the Mac address, and the table entry full flag (FlagFull), and the data ( Data) includes the entry valid flag (Valid) and the virtual switch port (Vp).

Table 2. Data structure definition table of TCAM entries

Referring to FIG. 1, the process of initializing an entry in the embodiment of the present invention is as follows:

Step 101: Determine whether the index of the initialization entry is greater than the maximum value of the entry, and if yes, go to step 102; otherwise, go to step 103;

Step 102, the index of the initialization entry exceeds the maximum value of the entry, returns an exception, and proceeds to step 108; Step 103, determining whether the initialization table entry index value is within the hash table index range, and if so, then proceeds to step 104; otherwise, proceeds to step 105;

Step 104: Initialize the hash table, directly locate the hash table entry by using the initialization index value, and clear all the contents of the entry to 0, and go to step 108;

Step 105: Determine whether the initialization entry index is the last item of the TCAM, and if yes, go to step 106; otherwise, go to step 107;

Step 106: Initialize the last TCAM entry, all fields in the key value are initialized to 0, and the mask is 0, that is, the entry is a "catch all" (match any key value) entry; all fields in the data If the entry is 0, it indicates that the entry is the last entry of the TCAM and cannot be used to cache address conflicts. The last entry of the TCAM is: (1) When the service forwards the entry, the entire TCAM table can be judged. The MAC address is not cached in the entry, so the hash table needs to be searched to guide the service forwarding. (2) When the MAC address learns to search for the idle entry, the TCAM conflict cache is full and the cache cannot be cached again. The MAC address is; therefore, the total amount of conflicts that can actually be cached is the total number of TCAM entries minus 1, go to step 108;

In step 107, the full value of the entry in the key value is cleared to 0, and the mask is set to 1 (indicating that the initial state of the current entry is empty), the remaining fields in the key value are cleared to 0, the mask is cleared to 0, and the valid entry of the entry in the data is set to 1. The remaining fields are cleared to 0, indicating that the entry is not the last entry, and can be used to cache the conflicting MAC address, and proceeds to step 108; Step 108, the process of initializing the entry ends.

Referring to FIG. 2, an embodiment of the present invention provides a MAC address hardware learning method based on a hash table and a TCAM table, including the following steps:

Step 201: Calculate a hash value by using a virtual forwarding instance number and a source MAC address to which the packet belongs, and search the hash table by using a hash value index.

Step 202: Search for the key value of the TCAM entry: The virtual forwarding instance number is the virtual forwarding instance number to which the packet belongs; the MAC address is the source MAC address of the packet; and the table entry full flag is set to 1, indicating that the TCAM has been learned. Search in the table entry;

Step 203: Determine whether the valid flag in the TCAM entry data is set, if set, go to step 204; otherwise, go to step 206;

Step 204, comparing whether the virtual switch port in the TCAM entry data is equal to the packet receiving virtual switch port, if yes, the entry does not need to be updated, go to step 215; otherwise, the TCAM entry needs to be updated, go to step 205; Step 205, update the current TCAM entry data in the virtual switch port as the packet actually receives the virtual switch port, go to step 215;

Step 206, determining whether the data valid flag in the hash table is set, if set, go to step 208; otherwise, go to step 207;

Step 207: The source MAC address of the packet, the virtual forwarding instance number of the packet, and the virtual switching port received by the packet are stored in the current hash entry, and the valid entry of the entry is set to 1, and the process proceeds to step 215;

Step 208: Compare the virtual forwarding instance number and the MAC address in the hash table with the virtual forwarding instance number and the source MAC address to which the packet belongs. If the matching, the current address has been learned in the hash table, and then go to Step 209; otherwise, go to step 211;

Step 209: Compare whether the virtual switch port in the current hash table is equal to the packet actually receiving the virtual switch port. If they are equal, the hash table item does not need to be updated, go to step 215; otherwise, go to step 210; Step 210, update The virtual forwarding port number in the current hash entry is the actual virtual port number received by the packet, and the process proceeds to step 215.

Step 211: Generate a hash conflict, and organize the search for the key value of the TCAM entry to find the idle TCAM entry cache conflict: the virtual forwarding instance number in the key value is 0, the MAC address is 0, and the entry full flag is 0, indicating that it is idle. Search in the TCAM entry;

Step 212, determining whether the valid flag in the TCAM data is set, if set, indicating that the first free entry is found, go to step 214; otherwise, go to step 213;

Step 213: The TCAM memory has no free entries to cache conflicts, abandon learning, and go to the step

215;

Step 214: Cache the MAC address into the current idle TCAM entry: The virtual forwarding instance number of the key value is filled in as the virtual forwarding instance number of the packet, and the mask is set to 1; the MAC address is filled in as the source MAC address of the packet, and the mask is Set to 1; the entry full flag is set to 1, the mask is set to 1; the virtual switch port in the data is filled in as the packet actually receives the virtual switch port, go to step 215;

Step 215: The MAC address hardware learning process ends.

Referring to FIG. 3, the process of aging in the embodiment of the present invention is as follows:

Step 301: Determine whether the index of the aging entry is greater than the maximum value of the entry, if yes, go to step 302; otherwise, go to step 303;

Step 302: The index of the initialization entry exceeds the maximum value of the entry, and returns an exception. Go to step 308. Step 303: Determine whether the index value of the aging entry is in the index range of the hash table, and if yes, then Go to step 304; otherwise, go to step 305;

Step 304, aging the hash table, directly using the initialization index value to locate the hash table entry, and clear all the contents of the table entry, go to step 308;

Step 305: Determine whether the aging entry index is the last item of the TCAM, and if yes, go to step 306; otherwise, go to step 307;

Step 306, the last item of the TCAM is a "catch all" (match any key value) of the entry, once initialized does not need to operate on it, go to step 308;

Step 307: Clear the key of the key in the key value to 0, the mask is set to 1, the other fields are cleared to 0, the mask is cleared to 0, the state of the entry is set to null, the valid flag of the entry in the data is set to 1, and the remaining fields are cleared. , indicating that the entry is not the last entry, can be used to cache the conflicting MAC address, go to step 308;

Step 308: The MAC address aging process ends.

The embodiment of the invention further provides a MAC address hardware learning system based on a hash table and a TCAM table, including an initialization unit, a search unit, a TCAM entry determination unit, a hash table determination unit, a hash table comparison unit, and a hash collision. Processing unit and aging unit, wherein:

An initialization unit, configured to: determine whether the index of the initialization entry is greater than the maximum value of the entry, and if yes, return an exception; otherwise, determine whether the index value of the initialization entry is within the index range of the hash table, and if so, initialize the hash table Directly use the initialization index value to locate the hash table entry, and clear all the contents of the table entry; otherwise, determine whether the initialization entry index is the last item of the TCAM, and if so, initialize the last TCAM entry, all in the key value The fields are initialized to 0, the mask is 0, and all fields in the data are assigned a value of 0, indicating that the entry is the last entry of the TCAM; otherwise, the full value of the entry in the key value is cleared to 0, and the mask is set to 1, Indicates that the initial state of the current entry is empty, the other fields in the key value are cleared to 0, the mask is cleared to 0, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating that the entry is not the last entry, and the initialization ends;

The search unit is configured to: calculate a hash value by using the virtual forwarding instance number and the source MAC address to which the packet belongs, and search the hash table by using a hash value index, and the hash table is stored in the SRAM or the DRAM; the TCAM entry is stored. In the TCAM memory, the organization searches for the key value of the TCAM entry: the virtual forwarding instance number is the virtual forwarding instance number to which the packet belongs; the MAC address is the source MAC address of the packet; the table entry full flag is set to 1, indicating that it has learned Search in the TCAM entry;

The TCAM entry determining unit is configured to: determine whether the valid flag in the TCAM entry data is set, and if set, compare whether the virtual switch port in the TCAM entry data is equal to the packet receiving virtual switch. The port, if yes, does not update the entry; otherwise, the virtual switch port in the current TCAM entry data is updated to receive the virtual switch port for the packet; if not set, the hash table judgment trigger signal is generated and sent to the hash. Table judgment unit;

The hash table determining unit is configured to: when receiving the hash table determining the trigger signal, determine whether the data valid flag in the hash table is set, if not set, the source MAC address of the packet and the virtual forwarding instance number of the packet And the virtual switch port received by the packet is stored in the current hash table entry, and the entry valid flag is set to 1; if set, the comparison trigger signal is generated and sent to the hash table comparison unit;

The hash table comparison unit is configured to: when the comparison trigger signal is received, compare whether the virtual forwarding instance number and the MAC address in the hash table match the virtual forwarding instance number and the MAC address to which the packet belongs, and if they match, the current The address has been learned in the hash table. It compares whether the virtual switch port in the current hash table is equal to the actual virtual switch port received by the packet. If they are equal, the learning is ended. Otherwise, the virtual forwarding port number in the current hash table entry is updated. Ending the learning by actually receiving the virtual port number for the packet; if not, generating a hash conflict processing trigger signal, and sending the hash conflict processing unit to the hash conflict processing unit;

The hash conflict processing unit is configured to: when receiving the hash conflict processing trigger signal, organize the search for the key value of the TCAM entry to find the idle TCAM entry cache conflict: the virtual forwarding instance number in the key value is 0, and the MAC address is 0. If the entry full flag is 0, it means searching in the idle TCAM entry; then determining whether the valid flag in the TCAM data is set, if set, indicating that the first free entry is found, and the MAC address is cached to the current idle TCAM. In the entry: The virtual forwarding instance number of the key is filled in as the virtual forwarding instance number of the packet, the mask is set to 1, the MAC address is filled in as the source MAC address of the packet, the mask is set to 1, and the entry full flag is set to 1. The code is set to 1, the virtual switch port in the data is filled in as the packet actually receives the virtual switch port, and the learning is ended; if not set, it indicates that the TCAM memory has no free entry, and the learning is abandoned;

The aging unit is configured to: determine whether the index of the aging entry is greater than the maximum value of the entry, and if yes, return an exception; otherwise, determine whether the index value of the aging entry is in the index range of the hash table, and if yes, the aging hash table Directly use the initial index value to locate the hash table entry, and clear all the contents of the table entry; otherwise, determine whether the aging entry index is the last item of the TCAM, and if so, the last item of the TCAM is a match for any key value. If the entry is cleared, the mask is set to 0, the remaining fields are cleared to 0, the mask is cleared to 0, the entry status is set to null, and the entry in the data is valid. 1. The remaining fields are cleared to 0, indicating that the entry is not the last entry, and the aging ends.

A person skilled in the art can make various modifications and variations to the embodiments of the present invention, and if such modifications and variations are within the scope of the appended claims and their equivalents, Within the scope of protection of the invention.

The contents not described in detail in the specification are prior art known to those skilled in the art.

Claims

A MAC address hardware learning method based on a hash table and a TCAM table, characterized in that the method comprises the following steps:

53. Determine whether the valid flag in the TCAM entry data is set, if set, go to S4; No Bay 1J, go to S6;

511. Generate a hash collision, and search for a key value of the TCAM entry to find an idle TCAM entry cache conflict: the virtual forwarding instance number in the key value is 0, the MAC address is 0, and the entry full flag is 0, indicating that the idle TCAM is in the idle TCAM. Search in the table entry; 512, determining whether the valid flag in the TCAM data is set, if set, indicating that the first free entry is found, go to S14; otherwise, go to S13;

513, TCAM memory does not have a free entry, give up learning, go to S15;

S 15, MAC address hardware learning process ends.

The method for learning the MAC address hardware based on the hash table and the TCAM table according to claim 1, wherein: step S1 further comprises the step of initializing the entry: determining whether the index of the initialization entry is greater than the maximum value of the entry. If yes, an exception is returned; otherwise, it is determined whether the index value of the initialization entry is within the index range of the hash table, and if so, the hash table is initialized, the hash table entry is directly located with the initialization index value, and the contents of the entry are all Clear 0; otherwise, determine whether the index of the initialization entry is the last item of TCAM. If yes, initialize the last TCAM entry. All fields in the key value are initialized to 0, the mask is 0, and all fields in the data are assigned. If it is 0, it indicates that the entry is the last entry of the TCAM; otherwise, the full value of the entry in the key value is cleared to 0, and the mask is set to 1, indicating that the initial state of the current entry is empty, and the remaining fields in the key are cleared to 0. The code is cleared to 0, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating that the entry is not the last entry, and the initialization ends.

The method for learning the MAC address hardware based on the hash table and the TCAM table according to claim 1, wherein: step S15 further comprises the step of aging: determining whether the index of the aging entry is greater than the maximum value of the entry, if The exception is returned. Otherwise, it is determined whether the index value of the aging entry is in the index range of the hash table. If yes, the aging hash table is used to directly locate the hash table with the initial index value, and all the contents of the table are cleared. Otherwise, it is determined whether the index of the aging entry is the last item of the TCAM. If yes, it means that the last item of the TCAM is an entry matching any key value, and the aging is ended; otherwise, the full flag of the key in the key value is cleared to 0. The mask is set to 1, the other fields are cleared to 0, the mask is cleared to 0, the status of the entry is set to null, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating that the entry is not the last entry, and the aging ends.

The MAC address hardware learning method based on the hash table and the TCAM table according to any one of claims 1 to 3, characterized in that the hash table is stored in an SRAM or a DRAM.

The method for learning a MAC address hardware based on a hash table and a TCAM table according to any one of claims 1 to 3, wherein: the TCAM entry is stored in a TCAM memory. 6. A MAC address hardware learning system based on a hash table and a TCAM table, comprising: a search unit, a TCAM entry determination unit, a hash table determination unit, a hash table comparison unit, and a hash collision processing unit, among them:

The hash table comparison unit is configured to: when receiving the comparison trigger signal, compare whether the virtual forwarding instance number and the MAC address in the hash table match the virtual forwarding instance number and the MAC address to which the packet belongs, and if yes, Indicates that the current address has been learned in the hash table. The virtual switch port in the current hash table is equal to whether the packet actually receives the virtual switch port. If they are equal, the learning is ended. Otherwise, the virtual forwarding in the current hash table is updated. The port number is the packet actually receiving the virtual port number, and the learning is ended; if not, the hash conflict processing trigger signal is generated and sent to the hash conflict processing unit;

The hash conflict processing unit is configured to: when receiving the hash conflict processing trigger signal, organize to search for a key value of the TCAM entry to find an idle TCAM entry cache conflict: the virtual forwarding instance number in the key value is 0, the MAC address 0, the entry full flag is 0, indicating that the search is in the idle TCAM entry; then it is determined whether the valid flag in the TCAM data is set, if set, the first free entry is found, and the MAC address is cached to the current In the idle TCAM entry, the virtual forwarding instance number of the key is filled in as the virtual forwarding instance number of the packet, the mask is set to 1, the MAC address is filled in as the source MAC address of the packet, the mask is set to 1, and the entry full flag is set to 1. The mask is set to 1. The virtual switch port in the data is filled in as the packet actually receives the virtual switch port, and the learning is ended. If it is not set, it indicates that the TCAM memory has no free entry, and the learning is abandoned. The device for learning a MAC address based on a hash table and a TCAM table according to claim 6, further comprising: an initializing unit, configured to: determine whether an index of the initialization entry is greater than a maximum value of the entry, and if so, Return an exception; otherwise, determine whether the index value of the initialization table entry is within the index range of the hash table, and if so, initialize the hash table, directly locate the hash table entry with the initialization index value, and clear all the contents of the table entry; Otherwise, it is determined whether the index of the initialization entry is the last item of the TCAM, and if so, the last TCAM entry is initialized, all fields in the key value are initialized to 0, the mask is 0, and all fields in the data are assigned a value of 0. Indicates that the entry is the last entry of the TCAM; otherwise, the full value of the entry in the key value is cleared to 0, and the mask is set to 1, indicating that the initial state of the current entry is empty, and the remaining fields in the key are cleared to 0, and the mask is cleared. The valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating that the entry is not the last entry, and the initialization ends.

The device for learning a MAC address based on a hash table and a TCAM table according to claim 6, further comprising: an aging unit, configured to: determine whether an index of the aging entry is greater than a maximum value of the entry, and if so, The exception is returned. Otherwise, it is determined whether the index value of the aging entry is in the index range of the hash table. If yes, the hash table is aged, and the hash table entry is directly located by using the initialization index value, and all the contents of the entry are cleared to 0. Otherwise, it is determined whether the index of the aging entry is the last item of the TCAM. If yes, it indicates that the last item of the TCAM is an entry that matches any key value, and the aging is ended; otherwise, the full value of the entry in the key value is cleared to 0. The code is set to 1, the other fields are cleared to 0, the mask is cleared to 0, the status of the entry is set to null, the valid entry of the entry in the data is set to 1, and the remaining fields are cleared to 0, indicating that the entry is not the last entry, and the aging ends.

The MAC address hardware learning system based on the hash table and the TCAM table according to any one of claims 6 to 8, wherein the hash table is stored in an SRAM or a DRAM.

The MAC address hardware learning system based on the hash table and the TCAM table according to any one of claims 6 to 8, wherein the TCAM entry is stored in the TCAM memory.