WO2016101551A1

WO2016101551A1 - Method and device for writing data into ternary content addressable memory

Info

Publication number: WO2016101551A1
Application number: PCT/CN2015/081093
Authority: WO
Inventors: 姜海明; 杨远田
Original assignee: 中兴通讯股份有限公司
Priority date: 2014-12-26
Filing date: 2015-06-09
Publication date: 2016-06-30
Also published as: CN105791125A; CN105791125B

Abstract

A method and a device for writing data into a ternary content addressable memory (TCAM). The method comprises: acquiring a link state of a physical bus between a network processor (NP) and a TCAM; if the link state of the physical bus is good, controlling a working mode of a write operation to be a first working mode, the first working mode being that a query response of the TCAM is not waited for, or after a query response of the TCAM is obtained, the query response of the TCAM is not parsed and processed ; and when an operation of writing data into the TCAM is detected, if the working mode is the first working mode, directly writing data into the TCAM.

Description

Method and device for writing data in three-state content addressing memory

Technical field

This paper relates to the field of computers, and in particular to a method and apparatus for writing data in a three-state content addressed memory.

Background technique

Today's network development is amazing, network traffic growth and the emergence of new services require network equipment to have wire-speed and flexible processing capabilities. At present, the network chip includes two major categories: ASIC (Application Specific Integrated Circuits) and NP (Network Processor). With its high-speed processing and flexible programmability, network processors have become an effective solution for data processing in today's networks.

There are many different kinds of memory in the network processor, such as SRAM, TCAM, SDRAM, etc. These memories store various service items such as port table, MAC table, routing table and so on.

TCAM (ternary content addressable memory) is a three-state content-addressable memory. It is mainly used to quickly find ACLs (Access Control Lists) and routes. The main feature of TCAM is fast parallel search. The application on the network processor is usually that the microcode composes a KEY (key value) from the message extraction information, and sends a search command to the TCAM through the NP TCAM controller, and processes the result according to the result. Message.

FIG. 1 is a schematic diagram of a network processor connecting a TCAM in the related art. The CPU accesses the external TCAM through the TCAM adaptation interface of the NP, and the NP and the TCAM are interconnected via the Interlaken bus.

The Interlaken physical bus consists of multiple lanes. During the operation of the device, one of the LANs is aging or the link quality is poor. The TCAM write table may have errors such as CRC (Cyclical Redundancy Check) or Framing (framing). . Therefore, the control plane accesses the TCAM, and waits for each read and write, regardless of whether it is read or written, waits for the TCAM to return and parses it, and if there is an error, performs a failure rewrite.

If an occasional synchronization error or CRC occurs during the Interlaken link operation, the TCAM can be successfully written by repeated overwriting.

Currently, there are many network services that use TCAM, such as IPV4/IPV6 route lookup, ACL, flow mirroring, and flow statistics. These services often write tables in batches, and may include up to tens of thousands of rules. If each rule is written to the TCAM, it must wait for the TCAM to return, and if the result is parsed, it will be a waste of time, so the current large-capacity ACL The configuration takes a long time and the service takes effect slowly. Therefore, how to improve the speed of TCAM write operations is a technical problem to be solved.

Summary of the invention

This document provides a method and apparatus for writing data in a tri-state content addressed memory to increase the speed of TCAM write operations.

A method for writing data in a three-state content addressing memory TCAM, comprising: acquiring a link state of a physical bus between a network processor NP and a TCAM; and controlling a write operation if a link state of the physical bus is good The working mode is a first working mode, wherein the first working mode is a query response that does not wait for the TCAM, or, after obtaining the query response of the TCAM, does not perform parsing and processing of the query response of the TCAM; when detecting to the TCAM When the operation of writing data is performed, if the working mode is the first working mode, the data is directly written into the TCAM.

Optionally, after the working mode of controlling the write operation is the first working mode, the method further includes: if an error of the link of the physical bus occurs, switching the working mode of the write operation from the first working mode to the second working mode The second working mode is a query response waiting for the TCAM, and after obtaining the query response of the TCAM, performing the parsing of the query response of the TCAM; when detecting the operation of writing data to the TCAM, if the working mode is In the second working mode, the query response of the TCAM is obtained, and according to the query response of the TCAM, the cause of the error is obtained, and the write operation of the data is managed according to the error reason.

Optionally, the managing the write operation of the data according to the error cause includes: if the error cause is a cyclic redundancy check CRC, rewriting the data until the correct Until the data is written.

Optionally, the obtaining the link state of the physical bus between the network processor NP and the TCAM comprises: reading the error status register inside the TCAM and the information recorded in the link error count register, and acquiring the relationship between the NP and the TCAM. The link state of the physical bus.

The management data input/output MDIO interface reads the error status register inside the TCAM and the information recorded in the link error count register, and acquires the link state of the physical bus between the NP and the TCAM.

An apparatus for writing data in a TCAM of a three-state content addressing memory, comprising: an obtaining module, configured to: acquire a link state of a physical bus between a network processor NP and a TCAM; and a control module configured to: if the physical If the link state of the bus is good, the working mode of the control write operation is the first working mode, wherein the first working mode is a query response that does not wait for the TCAM, or the TCAM query is not performed after the TCAM query response is obtained. The parsing and processing of the response; the first write operation module is configured to: when the operation of writing data to the TCAM is detected, if the working mode is the first working mode, the data is directly written into the TCAM.

Optionally, the control module is further configured to: if an error of the link of the physical bus occurs, switch an operation mode of the write operation from a first working mode to a second working mode, where the second working mode In order to wait for the query response of the TCAM, and after obtaining the query response of the TCAM, perform the parsing of the query response of the TCAM; the second write operation module is set to: when the operation of writing data to the TCAM is detected, if the working mode is In the second working mode, the query response of the TCAM is obtained, and according to the query response of the TCAM, the cause of the error is obtained, and the write operation of the data is managed according to the error reason.

Optionally, the second write operation module includes: a rewriting unit, configured to: if the error cause is a cyclic redundancy check CRC, rewrite the data until the correct write is performed Until the data is described.

Optionally, the acquiring module is configured to: read the error status register inside the TCAM and the information recorded in the link error count register, and acquire a link state of the physical bus between the NP and the TCAM.

Optionally, the acquiring module reads the information recorded by the error status register and the link error count register inside the TCAM by the management data input/output MDIO interface, and acquires the link state of the physical bus between the NP and the TCAM.

A computer readable storage medium storing computer executable instructions for performing the method of any of the above.

The embodiment of the present invention obtains the link state between the TCAM and the NP. When the link state is good, the control write operation does not need to wait for the TCAM to return the result, and does not need to parse the return result, so the large-capacity write table can save the situation. A lot of time, thus shortening the time required to complete the write operation and increasing the speed of the TCAM write operation.

BRIEF abstract

1 is a schematic diagram of a network processor connecting a TCAM in the related art;

2 is a schematic flow chart of a method for writing data in a three-state content addressing memory according to an embodiment of the present invention;

3 is a schematic diagram of a network processor connecting a TCAM according to an embodiment of the present invention;

4 is a schematic structural diagram of an apparatus for writing data in a tri-state content addressable memory according to an embodiment of the present invention;

Fig. 5 is a schematic view showing another structure of the apparatus shown in Fig. 4.

Embodiments of the invention

Embodiments of the present invention will be described below with reference to the accompanying drawings. It should be noted that, in the case of no conflict, the features in the embodiments and the embodiments herein may be arbitrarily combined with each other.

FIG. 2 is a schematic flowchart of a method for writing data in a three-state content addressable memory according to an embodiment of the present invention. The method shown in Figure 2 includes:

Step 201: Obtain a link state of a physical bus between the network processor NP and the TCAM.

Step 202: If the link state of the physical bus is good, control the working mode of the write operation to be a first working mode, where the first working mode is a query response that does not wait for the TCAM, or obtains a query response of the TCAM. After that, the parsing and processing of the query response of the TCAM is not performed;

Step 203: When an operation of writing data to the TCAM is detected, if the working mode is When it is in the first working mode, the data is directly written to the TCAM.

The method embodiments provided by the present invention are further described below:

After the working mode of the control write operation is the first working mode, the method further includes:

If an error occurs in the link of the physical bus, the working mode of controlling the write operation is switched from the first working mode to the second working mode, wherein the second working mode is a query response waiting for the TCAM, and the query of the TCAM is obtained. After the response, the TCM query response is parsed;

When the operation of writing data to the TCAM is detected, if the working mode is the second working mode, the query response of the TCAM is acquired, and according to the query response of the TCAM, the cause of the error is obtained, according to the error reason, The writing of the data is managed.

The managing the write operation of the data according to the error cause includes:

If the cause of the error is a cyclic redundancy check CRC, the data is rewritten until the data is correctly written.

Taking the Netlogic series TCAM widely used in the industry as an example, the TCAM returns a Packet Error Response format containing a 64-bit Error Code. To write to the TCAM, wait for the TCAM to return a Response result. Parsing the Error Code. If a link CRC is found, the corresponding action is taken, such as failure rewriting.

It should be noted that when the working mode is the second working mode, the TCAM write operation can be performed according to the existing process to ensure correct writing of data.

By controlling the TCAM write operation in different working modes, the data writing speed is improved under the premise of ensuring correct data writing.

The obtaining the link state of the physical bus between the network processor NP and the TCAM includes: reading the error status register inside the TCAM and the information recorded in the link error count register, and acquiring the physical bus between the NP and the TCAM. Link status.

For example, the PCS layer of the Netlogic TCAM chip stores error status and error counts. The NL11K chip Common Status Register has the ability to detect errors such as CRC, and there is a CRC error count in the RX PCS Counter register.

By using the characteristics of the register storage link information in the TCAM chip, the information stored in the register is read and used as the basis for switching the write operation without increasing the hardware cost, and the implementation is simple.

Of course, in practical applications, it is also possible to set a register to store the bus state between the TCAM and the NP as a basis for switching the operating mode.

In order to further save the hardware cost, this paper proposes to read the information recorded in the TCAM internal error status register and link error count register through the Management Data Input/Output (MDIO) interface to obtain the physical relationship between NP and TCAM. The link status of the bus.

FIG. 3 is a schematic diagram of a network processor connected to a TCAM according to an embodiment of the present invention. In the diagram shown in Figure 3, the internal CPU interface in the NP is connected to the TCAM using an external MDIO interface to achieve the purpose of reading the information in the register in the TCAM. If the read result finds that the error status register and the link error count register have a link error such as CRC, it is timely aware that there is a problem with the link between the NP and the TCAM.

The internal registers of the TCAM are periodically read through the MDIO interface to sense the Interlaken link state between the NP and the TCAM. The advantage of borrowing the MDIO interface is that the MDIO signal line is an out-of-band resource and therefore does not occupy the Interlaken bandwidth resources between the NP and the TCAM.

The following is an example of an application:

Preset two TCAM write table modes:

Mode 1: TCAM writes the table, does not wait for the TCAM response, and does not need to analyze and judge the TCAM response result according to the response format specified by the TCAM manual. This mode works when the link between NP and TCAM is in good condition.

Mode 2: The TCAM write table needs to wait for the TCAM response. The TCAM response result is parsed and judged according to the response format specified by the TCAM manual. If there is a CRC error, it needs to be rewritten until it is successfully written to the TCAM. This mode works when the link status between NP and TCAM is not good.

The switching between mode one and mode two is controlled by a software global variable.

After the chip initialization is completed, a software thread is opened, and the error status register and the error count register related to the link error (such as CRC) inside the TCAM are continuously scanned through the MDIO interface.

Step A01: determining whether an error is detected;

The result of the software thread scan, ie the value of the global variable, determines which operating mode is used. If the software thread scans the register and finds that a link error such as CRC exists, it proceeds to step A02; otherwise, it proceeds to step A03.

Step A02, mode one write table;

In this mode, there is no need to wait for the TCAM to return the result, and there is no need to parse the returned result and perform a failure rewrite. This mode is the usual fast write table mode.

Step A03, mode 2 write table;

This mode needs to wait for the TCAM to return the result, parse the returned result, and if there is a link problem such as CRC, it will fail to rewrite until it is correctly written.

The above application example discloses a fast write table mode for the problem that the current batch write TCAM is effective. The TCAM's MDIO out-of-band interface reads the link quality-related error status register and error counter inside the TCAM to determine which operating mode to use. Normally, when the quality of the TCAM link is normal, there is no need to wait for the TCAM to return the result, and there is no need to parse the returned result. Therefore, the large-capacity write table can save a lot of time, thus shortening the service effective time.

FIG. 4 is a schematic structural diagram of an apparatus for writing data in a three-state content addressable memory according to an embodiment of the present invention. The schematic shown in Figure 4 includes:

The obtaining module 401 is configured to: acquire a link state of a physical bus between the network processor NP and the TCAM;

The control module 402 is configured to: if the link state of the physical bus is good, control an operation mode of the write operation to a first working mode, where the first working mode is a query response that does not wait for the TCAM, or After the TCAM query response, the TCAM query response is not parsed and processed;

The first write operation module 403 is configured to: when the operation of writing data to the TCAM is detected, if the operation mode is the first operation mode, the data is directly written into the TCAM.

Fig. 5 is a schematic view showing another structure of the apparatus shown in Fig. 4. The device in the schematic shown in Figure 5 includes:

In this embodiment, the control module 402 is further configured to: if an error occurs in the link of the physical bus, switch the working mode of the write operation from the first working mode to the second working mode, where the second The working mode is to wait for the query response of the TCAM, and after obtaining the query response of the TCAM, perform the parsing of the query response of the TCAM;

The second write operation module 404 is configured to: when detecting the operation of writing data to the TCAM, if the working mode is the second working mode, acquiring a query response of the TCAM, and according to the query response of the TCAM, The cause of the error is obtained, and the write operation of the data is managed according to the cause of the error.

The second write operation module 404 includes:

The rewriting unit is configured to: if the error cause is a cyclic redundancy check CRC, rewrite the data until the data is correctly written.

The obtaining module 401 is configured to:

Read the error status register inside the TCAM and the information recorded in the link error count register to obtain the link status of the physical bus between the NP and the TCAM.

The obtaining module 401 reads the information recorded by the error status register and the link error count register in the TCAM through the management data input/output MDIO interface, and acquires the NP sum. The link state of the physical bus between the TCAMs.

One of ordinary skill in the art will appreciate that all or a portion of the steps of the above-described embodiments can be implemented using a computer program flow, which can be stored in a computer readable storage medium, such as on a corresponding hardware platform (eg, The system, device, device, device, etc. are executed, and when executed, include one or a combination of the steps of the method embodiments.

Alternatively, all or part of the steps of the above embodiments may also be implemented by using an integrated circuit. These steps may be separately fabricated into individual integrated circuit modules, or multiple modules or steps may be fabricated into a single integrated circuit module. achieve.

The devices/function modules/functional units in the above embodiments may be implemented by a general-purpose computing device, which may be centralized on a single computing device or distributed over a network of multiple computing devices.

When each device/function module/functional unit in the above embodiment is implemented in the form of a software function module and sold or used as a stand-alone product, it can be stored in a computer readable storage medium. The above mentioned computer readable storage medium may be a read only memory, a magnetic disk or an optical disk or the like.

Industrial applicability

Claims

A method for writing data in a TCAM of a three-state content addressing memory, comprising:

Obtaining a link state of a physical bus between the network processor NP and the TCAM;

If the link state of the physical bus is good, the working mode of controlling the write operation is the first working mode, wherein the first working mode is not waiting for the query response of the TCAM, or after obtaining the query response of the TCAM, Performing analysis and processing of the query response of the TCAM;

When an operation of writing data to the TCAM is detected, if the operation mode is the first operation mode, data is directly written into the TCAM.
The method of claim 1, wherein after controlling the working mode of the write operation to be the first working mode, the method further comprises:

If an error occurs in the link of the physical bus, the working mode of controlling the write operation is switched from the first working mode to the second working mode, wherein the second working mode is a query response waiting for the TCAM, and the query of the TCAM is obtained. After the response, the TCM query response is parsed;

When the operation of writing data to the TCAM is detected, if the working mode is the second working mode, the query response of the TCAM is acquired, and according to the query response of the TCAM, the cause of the error is obtained, according to the error reason, The writing of the data is managed.
The method of claim 2, wherein the managing the write operation of the data according to the cause of the error comprises:

If the cause of the error is a cyclic redundancy check CRC, the data is rewritten until the data is correctly written.
The method of claim 1, wherein the obtaining a link state of a physical bus between the network processor NP and the TCAM comprises:

Read the error status register inside the TCAM and the information recorded in the link error count register to obtain the link status of the physical bus between the NP and the TCAM.
The method according to claim 4, wherein the MDIO is input and output through management data The interface reads the error status register inside the TCAM and the information recorded in the link error count register to obtain the link status of the physical bus between the NP and the TCAM.
A device for writing data in a TCAM of a three-state content addressing memory, comprising:

Obtaining a module, configured to: obtain a link state of a physical bus between the network processor NP and the TCAM;

a control module, configured to: if the link state of the physical bus is good, control an operation mode of the write operation to a first working mode, where the first working mode is a query response that does not wait for the TCAM, or After the query response, the TCAM query response is not parsed and processed;

The first write operation module is configured to: when the operation of writing data to the TCAM is detected, if the operation mode is the first operation mode, the data is directly written into the TCAM.
The apparatus of claim 6 further comprising a second write operation module, wherein:

The control module is further configured to: if an error occurs in the link of the physical bus, switch the working mode of the write operation from the first working mode to the second working mode, wherein the second working mode is waiting for the TCAM Query the response, and after obtaining the query response of the TCAM, perform the parsing of the query response of the TCAM;

The second write operation module is configured to: when detecting an operation of writing data to the TCAM, if the working mode is the second working mode, acquiring a query response of the TCAM, and responding according to the query of the TCAM Obtaining the cause of the error, and managing the write operation of the data according to the cause of the error.
The apparatus of claim 7, wherein the second write operation module comprises:

The rewriting unit is configured to: if the error cause is a cyclic redundancy check CRC, rewrite the data until the data is correctly written.
The apparatus of claim 6 wherein said obtaining module is configured to:

Read the error status register inside the TCAM and the information recorded in the link error count register to obtain the link status of the physical bus between the NP and the TCAM.
The apparatus according to claim 9, wherein the acquisition module reads the information recorded in the error status register and the link error count register inside the TCAM by the management data input/output MDIO interface, and acquires the physical bus between the NP and the TCAM. Link status.
A computer readable storage medium storing computer executable instructions for performing the method of any of claims 1-5.