Method and apparatus for automatically configuring a configurable integrated circuit

[0002] The invention is in the field of configuring integrated circuit devices to perform particular functions.

[0003] In electronic systems, various system functions are typically performed by specific circuits. The specific circuits may be integrated circuit chips manufactured to perform the specific function. For example, digital signal processor (DSP) chips can be purchased to process digital signals. Various kinds of DSP chips can be purchased to fill particular needs. Some functions, however, are too complex or varied for a manufacturer to supply off-the-shelf circuits for each possible variation of the function. For example, a circuit for controlling all communication between a network and a personal computer (PC) would have to be configured for the particular PC, the particular network, and also for optional capabilities. One response to the challenge of designing integrated circuits for complex, varied and changeable applications is to produce integrated circuits that are designed for a particular type of function, yet have some programmability to make them more flexible. One example is application specific integrated circuits (ASICs). An ASIC is an integrated circuit designed to perform a particular function by defining the interconnection of a set of basic circuit building blocks drawn from a library provided by the circuit manufacturer.

[0004] ASICs include registers that may be loaded with data that affects the behavior of the circuit. ASICs are usually configured by a skilled person “manually” loading specific registers so that the circuit will perform its intended function precisely as required, for example, by particular types of hardware with which it will interact. An ASIC may have some capability to be automatically configured by circuitry that loads data into registers of the ASIC from some source location. One method of configuring an ASIC embedded in a larger system is to store the configuration data in a non-volatile storage device, such as an electrically erasable programmable read only memory (EEPROM), and retrieve it when the system is initialized. This configuration data is traditionally in some fixed format and each location represents specific information about the device. Each time the system is initialized, the configuration registers of the ASIC are loaded with the data from the EEPROM. The source location in the EEPROM is “hardwired” to the ASIC registers. Typically, read and write pulses are sent to the serial EEPROM to read out each location in turn and write the contents to a corresponding register in the ASIC. This requires the ASIC to “know” what each location in the EEPROM represents. A disadvantage of this scheme is that it is very difficult to adapt the ASIC to different applications while embedded in the same or a similar system. For example, if it is desired to load a particular set of ASIC registers that are not supported by the EEPROM it is necessary to make hardware changes to provide the required support. In practice this usually involves replacing the ASIC altogether. Also, the addition of new register loads from the EEPROM dramatically increases the number of gates, and hence the semiconductor area, in the ASIC. This limits the ability of the ASIC to be configured automatically for any unforeseen application.

[0005] A method and apparatus for automatically configuring a configurable integrated circuit is described. One embodiment comprises a method for automatically loading data including configuration data to a configurable integrated circuit upon initialization of a system in which the configurable integrated circuit is embedded. The method of one embodiment comprises storing a plurality of commands and a plurality of data elements in a non-volatile memory of the system. The method further comprises reading contents of an initial address in the non-volatile memory. If the initial address contains a command, depending upon a type of the command, the method comprises writing contents of a next address in the non-volatile memory to a register space of the configurable integrated circuit, to a configuration space of the configurable integrated circuit, or to a command space of the configurable integrated circuit.

[0006] In one embodiment, a network interface card in a network device, such as a personal computer (PC), is configured by automatically writing data in input/output (I/O) registers, configuration registers, and command registers each time the network device is initialized. In one embodiment, a programmable memory device, such as an electrically erasable programmable read only memory (EEPROM), is loaded with commands and data. Commands and register addresses are read from the EEPROM by a state machine when the network device is initialized. Data is also read from the EEPROM and written to registers of the network device as specified by the commands. In this way, the network interface card can be reconfigured merely by writing additional or different commands, register addresses and data into the EEPROM.

[0007] FIG. 1 is a block diagram of a portion of a network.

[0008] FIG. 2 is a block diagram of a network interface controller.

[0009] FIG. 3 is a block diagram of a serial electrically erasable programmable read only memory (EEPROM).

[0010] FIG. 4 is a block diagram of network interface controller control circuitry.

[0011] FIG. 5 is an illustration of a state machine of the network interface controller control circuitry.

[0012] Various embodiments of the invention are described below. First, a system view of one embodiment is described in which the invention is part of a device in a network. Then, an embodiment of the invention as a method and apparatus for automatically configuring a network interface card (NIC) upon initialization of the network device that uses the NIC is described.

[0013] FIG. 1 is a block diagram of a portion of a network 100. Network 100 in various embodiments, may be a wide area network (WAN), a local area network (LAN), or any other collection of electronic devices that communicate via a physical or non-physical signal carrier. The network 100 has a physical network signal carrier 120. Other embodiments may use satellite or radio frequency signal transmission. In the network 100, any network protocol may be used. For example, transmission protocol over internet protocol (TCP/IP), Ethernet protocol, asynchronous transfer mode (ATM), integrated services digital network (ISDN), and digital subscriber line (DSL) may all be used.

[0014] Network signal carrier 120 connects multiple network devices. For simplicity, only one network device, personal computer (PC) 122, is shown in FIG. 1. PC 122 includes a central processing unit (CPU) 102 and a memory 104. PC 122 also includes several peripheral devices, such as keyboard 110, video card 112, audio card 114, display device 116 and mouse 108. The CPU 102, memory 104 and all peripheral devices communicate via bus 118. The bus 118 may be any type of PC bus according to a specific architecture. In one embodiment, the bus 118 is a peripheral component interconnect (PCI) bus. The PC 122 further includes a network interface card (NIC) 106 that is connected to the PCI bus 118 and to the network signal carrier 120. The NIC 106 includes circuitry for controlling communication between the PC 122 and any other entity on the network 100. The circuitry includes application specific integrated circuit (ASIC) 202, serial electrically erasable programmable read only memory (EEPROM) 204, receive first-in-first-out buffer (FIFO) 206, and transmit FIFO 208. In other embodiments, NIC 106 could be external to the PC 122 or it may reside on a different bus than PCI bus 118. For example, in other embodiments with different bus architectures, PC 122 may include a video bus for the video card 112 and a network bus for the NIC 106.

[0015] FIG. 2 is a block diagram of the NIC 106. Receive FIFO 206 temporarily stores incoming data received from other devices on the network 100. Transmit FIFO 208 temporarily stores data that is to be transmitted onto the network from the PC 122.

[0016] ASIC 202 includes circuitry that controls the operation of the NIC 106. ASIC 202 is designed to control communications between the network signal carrier 120 and the PC 122 for the particular architecture of PC 122 and the particular network protocol used in the network 100. ASIC 202 has a degree of flexibility in its operation. Some data may be loaded into registers of the ASIC 202 to change its operation. A certain number of registers are available for input/output (I/O) data, configuration data and command data. The data in particular registers is used by the ASIC 202 to perform its communication control functions. Changing the value of the data will change the behavior of the ASIC 202. According to the present invention, the ASIC 202 is made more flexible by enabling the values of I/O data, configuration data and command data to be easily changed on initialization of the PC 122. In other embodiments, the circuitry of the ASIC 202, the memory comprised in the EEPROM 204 and the FIFOs 206 and 208 may be part of one integrated circuit or may be divided up variously among different integrated circuits.

[0017] In one embodiment, I/O data, configuration data, and command data is stored in the EEPROM 204. Whenever the PC 122 is initialized, ASIC 202 automatically loads the configuration data from the EEPROM 204 to the ASIC 202. In one embodiment, a state machine in the ASIC 202 controls the loading of configuration data. In other embodiments, the state machine may reside in other elements of the NIC 106 (not shown), or even apart from the NIC 106. In contrast to traditional schemes, the EEPROM 204 stores both commands and data. The commands indicate what type of data is to follow in the EEPROM and also indicate an address in the ASIC in which the data should be stored. The EEPROM addresses are not “hardwired” to corresponding ASIC addresses and therefore the ASIC can be flexibly configured. To change the configuration of the ASIC, it is merely necessary to reprogram the EEPROM. In other embodiments, the programmable memory function performed by the EEPROM 204 may be performed by any other type of nonvolatile memory whose contents may be changed, such as battery backed random access memory (RAM).

[0018] FIG. 3 is an example of data stored in EEPROM 204. The first address (0) in the EEPROM 204 corresponds to an I/O register write (“Reg Write”) command which includes the I/O space register address in the ASIC 202. The next address (1) in the EEPROM 204 is the data to be written to the I/O register address specified in address 0. Similarly, addresses (3) and (7) of the EEPROM 204 correspond to data to be written to the I/O register address specified in address (2) and address (6) respectively. Address (5) of the EEPROM 204 corresponds to data to be written to the peripheral component interface configuration (“PCI Config”) address specified in address (4). Address (8) of the EEPROM 204 contains an automatic initialization done (“AutolnitDone”) command which indicates the end of the EEPROM 204 initialization. All addresses in the EEPROM after (8) are shown as “don't cares”. In various embodiments, the addresses subsequent to address 8 may contain additional commands and data to be loaded to the ASIC 202, or may be used for any other purpose.

[0019] The command format of one embodiment is illustrated below.

[0020] [15:13] Encoded Command

[0021] 111-AutolnitDone: EEPROM has completed initialization

[0022] 110-Reserved for future expansion

[0023] 101-PCI Configuration Space Write

[0024] 100-Reserved for future expansion

[0025] 011-I/O Register Write

[0026] 010-Reserved for future expansion

[0027] 001-Reserved for future expansion

[0028] 000-Reserved for future expansion

[0029] [12] Byte/Word Access

[0030] 0-Byte

[0031] 1-Word

[0032] [11] Reserved for future expansion

[0033] [10:0] Register address within the ASIC

[0034] The most significant three bits [15:13] indicate the encoded command as shown in the command format above. The encoded command can be a I/O register write, a PCI configuration space write, an ASIC command space write, or it could be an AutoInitDone command which indicates the end of EEPROM 204 initialization. The next bit [12] indicates whether a byte or word is being written. Bits [10:0] indicate the register address in the ASIC 202. The word following the command is the data to be written to the specified location.

[0035] FIG. 4 is a block diagram of ASIC 202 showing control circuitry 402 and state machine 406. State machine 406 is coupled to EEPROM 204 for generating read and write pulses to read and write information from the EEPROM 204. The EEPROM 204 is coupled to the I/O registers 408, the configuration registers 410, and the command registers 412 for writing information to the respective registers from the EEPROM 204.

[0036] FIG. 5 is an illustration of one embodiment of the state machine 406. State 502 is an idle state that is automatically entered on power up of the PC 122. AutolnitDone is tested in state 502. If AutolnitDone is true, then the EEPROM 204 initialization is complete and the state machine 406 remains in idle state 502. If AutolnitDone is not true then the state machine 406 enters the read address state 504. A command word is read from EEPROM 204 and tested to see if it is an I/O register command, a configuration command, an ASIC command or an AutolnitDone command. If the command is an AutolnitDone command, then idle state 502 is entered. If the command is an I/O register command, then read I/O register data state 506 is entered. If the command is a configuration command, then read configuration data state 508 is entered. If the command is an ASIC command, then read command data state 510 is entered.

[0037] In the read I/O register data state 506, the I/O data word is read from the EEPROM 204 and the state machine 406 transitions to the write I/O register address state 512. In the read configuration data state 508, the configuration data is read from the EEPROM 204 and the state machine transitions to the write configuration address state 514. In the read command data state 508, the command data is read from the EEPROM 204 and the state machine 406 transitions to the write command address state 516.

[0038] In the write register address state 512, the data word read in the read I/O register data state 506 is written to the address acquired in the read address state 504. The state machine 406 then enters the read address state 504.

[0039] In the write configuration address state 514, the data word read in the read configuration data state 508 is written to the address acquired in the read address state 504. The state machine 406 then enters the read address state 504.

[0040] In the write command address state 516, the data word read in the read command data state 510 is written to the address acquired in the read address state 504. A command write pulse is generated and the state machine 406 then enters the wait state 518 until the ASIC command has finished executing. The state machine 406 then enters the read address state 504.

[0041] According to the described embodiments, it is possible to easily reconfigure a NIC by merely entering additional or different NIC register addresses and data in the EEPROM 204. This is an advance over the prior art, in which reconfiguration of the NIC requires hardware solutions such as replacing the ASIC 202 and dramatically increasing the number of gates on the ASIC 202 to support additional register loads from the EEPROM 204.

[0042] A method and apparatus for automatically configuring a configurable integrated circuit has been described with reference to particular embodiments shown in the figures. Other embodiments may be envisioned by one of ordinary skill in the art. For example, the invention is applicable to any device that may be configured by loading data to registers in the device whether or not the device is part of a network. One of ordinary skill in the art may make modifications to the embodiments without departing from the scope of the invention as set forth in the following claims.