Background of the Invention
(1) Field of the Invention
The present invention relates to the field of data processing, and more specifically to methods and apparatus for programming an embedded memory module or on-chip cache of an integrated circuit.
(2) Prior Art
In the past, vendors and users of integrated circuits (IC's) having embedded memories (or on-chip caches) have programmed the memories of the IC's using custom built auto-programming circuits. A typical programming circuit comprises a socket having a plurality of pins configured for receiving the connection pins of a particular IC. This programming circuit is coupled to an external memory device having digital logic means such as a microprocessor or the like in addition to a programming routine stored in memory. The IC to be programmed is inserted into the socket of the programming circuit and the memory device instructed to call the programming routine for writing data stored in a specified memory location of the device to the embedded memory of the IC.
According to second (yet similar) method for programming an IC having embedded memory, a slave-load programmer is provided having a socket designed for receiving a particular family of IC's, dedicated circuitry and logic means for programming the IC, a hard drive or floppy drive (and diskette) for storing the data to be written and an input keyboard for inputting commands to the programmer. The IC to be programmed is inserted into the socket and the programmer is subsequently instructed via the keyboard to write specific data stored in either of the hard drive or the floppy diskette to a specified memory location of the IC.
In utilizing the conventional methods and apparatus described above for programming embedded memories of an IC, the problem arises that only a specific IC or a family of IC's can be programmed with a particular programming circuit or programmer. The reason for this is that the configuration and/or number of pins in addition to the timing specifications for accessing the embedded memories vary from IC family to IC family, and sometimes even between IC's within a particular family. Therefore, the wide variety of IC's available on the market requires a vendor or customer to invest in (if not to design himself) many different types of programming circuits and/or slave-load programmers in order to program the embedded memories of the different IC's which they use.
Accordingly, it would be desirable to provide a standardized method and apparatus for accessing and programming embedded memories (or on-chip caches) of all future IC's so as to eliminate the need for using a plurality of programming circuits and/or programmers for different types of IC's.
It is therefore an object of the present invention to provide a standardized method and apparatus for programming a variety of devices including micro-controllers, EPROMS, FLASHes and EPLDs.
It is another object of the present invention to provide a standardized method and apparatus for programming embedded memories of a plurality of different ICs while minimizing the number of interface pins required for the programming, thereby allowing the ICs to be less complex and less expensive.
It is a further object of the present invention to provide a method and apparatus which enables easy, in-system programming of a variety of ICs during the manufacturing process and reprogramming of the ICs when upgraded software becomes available.
Summary of the Invention
To accomplish these and other objectives, the present invention provides a standardized method and apparatus for programming embedded memories of a plurality of different ICs utilizing the IEEE Test Access Port (TAP) architecture and logic. The TAP architecture is used to serially shift address, data and command information into respective register chains of a RISM ACTION register located within a memory: interface unit of an IC. This register acts as a crucial interface between the TAP architecture and the memory interface unit of the IC for reading and writing data to and from specified memory locations.
The TAP architecture includes, among other things, a TAP port, a TAP controller and an instruction register. The TAP port comprises a plurality of input and output pins for serially transmitting and receiving data and control information to and from the TAP controller and instruction register. The TAP controller contains a synchronous finite state machine that controls the sequence of operations of the TAP circuitry. The instruction register is used to select the test to be performed or the test data register to be accessed or beth by serially shifting an instruction into the register and latching the instruction into the finite state machine at the completion of the shifting.
In utilizing the TAP architecture to perform accesses to embedded memory locations, the TAP controller is first instructed that a RISM ACTION operation is to be performed by serially shifting into the instruction register a RISM ACTION opcode via an input pin. The TAP controller reads the opcode and requests access to the memory bus via the MIU. The MIU disables the CPU as the bus master when the CPU is not performing a memory operation.
Next, the address, data and command bit strings for the memory access are serially shifted into the RISM ACTION register by placing the register between the input and output pins of the TAP port. After the shifting has been completed, the external system asserts an update signal via the TAP controller to the RISM ACTION register and the bit strings are transmitted in parallel to corresponding CORE registers. Normal operation of the IC is interrupted by an interrupt unit, and the interrupt unit transfers processor control from the user's code to RISM micro-code stored within an embedded memory module.
The RISM micro-code then reads the command bit string and executes the command accordingly. By means of sequentially inputting a series of commands in addition to appropriate data and addresses from the external system to the RISM ACTION register, the RISM micro-code can write the data transferred from the external system to the specified memory locations within the embedded memories. In this manner, data stored within a buffer in the external system can be serially transmitted to the IC and programmed into the IC's embedded memory modules.
Brief Description of the Drawings
The objects, features and advantages of the present invention will become apparent from the following detailed description in which:
FIG. 1 is a block diagram of an Intel architecture micro-controller integrated circuit of the 80C296 product family showing a generalized implementation of the TAP architecture and its connections to the RISM ACTION registor.
FIG. 2 is a more specific block diagram of the RISM ACTION register and the corresponding CORE registers within the memory interface unit of the C296 micro-controller.
FIG. 3 is a block diagram of the RISM ACTION register chains and the interface with the CORE registers,
FIG. 4 is a state diagram for the TAP controller pursuant to the IEEE Test Access Port Standard
FIG. 5(a) is a schematic diagram of a bit cell for a update only type register utilized in the RISM ACTION ADDR register chain.
FIG. 5(b) is a schematic diagram of a bit cell for a capture and update type register utilized in the RISM ACTION CMD and RISM ACTION DATA register chains.
Detailed Description of the Invention
A method and apparatus for programming embedded memories of a variety of integrated circuits using the IEEE Test Access Port is described. In the following description, numerous details such as specific programming steps, specific electronic components and circuitry, etc., are given in order to provide a thorough understanding of the present invention. It will be obvious, however, to one skilled in the art that not all the details given are required to practice the present invention. In other instances, well-known system components, features, methods and the like are not set forth in detail in order to avoid obscuring the description of the present invention.
According to the present invention, an integrated circuit (IC) such as a microprocessor or a micro-controller (CPU) having an on-chip cache or an embedded memory (i.e., an EPROM, a FLASH or an EPLD) is provided with an input/output (I/O) architecture and associated logic in addition to at least one interface register. The I/O architecture and logic enable the serial input and output of information to and from the IC, while the interface register provides a serial to parallel interface between the I/O architecture and the embedded memories of the IC.
According to a preferred embodiment of the present invention, the CPU-based IC to be programmed comprises an 80C296 micro-controller (C296 controller) manufactured by Intel Corporation. As shown in FIG. 1, the C296 controller comprises embedded memory modules including a FLASH memory, a code RAM and a register RAM. The I/O architecture is shown as a functional block entitled "TAP Architecture" coupled to a memory interface unit (MIU) of the C296 controller and having external connections via a "TAP port" to an external system 32 (shown in FIG. 2) having digital logic means. The MIU controls accesses to memory outside of the C296 controller register file (Reg.sub.-- file) by means of a memory bus (MemData and MemAdd) which allows the core of the C296 controller to interface with the embedded memory modules and the bus controller.
De-bugging micro-code comprising a reduced instruction set monitor (RISM) is stored within the test ROM of the FLASH memory module. The RISM micro-code in addition to a RISM register block 2 preferably (but not necessarily) residing within MIU enable reads and writes to memory locations and jumps to other codes stored anywhere in memory for execution. As shown in FIG. 3, the registers of the RISM register block 2 comprises a RISM ACTION register 5 having three register chains 4,6,8 which interface with five RISM special function registers of the IC. The RISM ACTION register 5 forms the interface register of the present invention and includes a RISM CMD register chain 4 (8 bits) coupled in parallel to a corresponding CORE CMD register 21, a RISM ADDR register chain 6 (24 bits) coupled in parallel to a corresponding CORE ADDRESS register 22 and a RISM DATA register chain 8 (32 bits) coupled in parallel to a corresponding CORE DATA register 23.
The RISM ADDR register chain 6 holds the address for memory location reads and writes or the address for branch operations, whereas the RISM DATA register chain 8 holds the data for a write operation or the result data of a read operation. The RISM CMD register chain 4 is used to hold 6 bit macro-instructions ("RISM commands") to be executed by the RISM micro-cede. As shown in the table below, the RISM commands actually form the addresses (shifted fight by two bits) in a break interrupt service routine (ISR) representing the memory locations where the RISM commands are stored.
In implementation of the RISM ACTION register 5 according to the present invention. as shown in FIG. 2, the RISM ADDR chain 6 of the register 5 comprises "update only" type bit cells (shown in FIG. 5(a)) to form an "UPDATE ONLY" shift register chain capable of serial shifting and parallel writing of bits. The RISM CMD and DATA chains 4, 8, on the other hand, comprise "capture and update" type bit cells (shown in FIG. 5(b) to form "CAPNUP" shift register chains capable of parallel loading, serial shifting and parallel writing of bits. As can be seen in FIGS. 5(a) and 5(b), each the UPDATE and CAPNUP type bit cells preferably comprise an edge triggered master/slave D-flip-flop 27 and a level sensitive latch 29 coupled together as shown for performing the above-described functions. For the CAPNUP type bit cells, a multiplexor 30 is further added for selecting between the parallel data in (DIN) from the corresponding CORE registers 21, 23 and the serial data in (TDI) from one of the instruction register IR and a previous bit cell. To control the operation of the bit cells of each of the RISM ACTION register chains 4, 6, 8, control signals in the form of a shift signal (SHDR), an update signal (UPDR) and a clock signal (CLKDR) are transmitted from the external system 32 to each of the bit cells of the respective register chains 4,6,8.
In order to establish a communications interface between the external system 32 and the RISM ACTION register 5, the RISM ACTION register is located such that it is accessible through a single vertical window setting. In the preferred embodiment of the present invention, this is accomplished using the serial interface described in the IEEE document entitled IEEE Standard Test Access Port and Boundary-Scan Architecture (IEEE Standard 1149.1-1190), this document being incorporated herein by reference. This IEEE standard developed by the Test Technology Technical Committee of the IEEE Computer Society sets forth a mechanism (referred to as the "TAP architecture") for accessing the internal workings of an IC in order to 1) test the interconnections between integrated circuits once they have been assembled onto a printed circuit beard, 2) test the IC itself and 3) observe or modify the activity of the IC during its normal operation.
The objective of the Standard is to provide a means for accessing and a means for controlling design-for-test features which are built directly into the IC's themselves. This is accomplished by providing an I/O architecture that enables test instructions and associated test data to be serially fed into the IC, while the results of the execution of such instructions are serially read out. The test logic is designed such that the serial movement of instructions and data, controlled by an external system, is not apparent to those circuit blocks whose operation is controlled by the instruction.
With reference to FIGS. 1 and 2, the TAP port comprises a test clock input pin (TCK) for clocking the TAP controller asynchronously with respect to the CPU's driving clock, a test mode select pin (TMS) for controlling the operation of the TAP controller, a test data input pin (TDI) for serially transmitting data and instruction signals to the internal TAP logic, a test data output pin (TDO) for serially receiving data and instruction signals output from the internal TAP logic and a test reset input pin (TRST) for providing asynchronous initialization of the TAP controller.
The TAP controller contains a synchronous finite state machine that responds to changes of the TMS pin on rising edges of the TCK pin and controls the sequence of operations of the TAP circuitry. The TAP controller is coupled to the plurality of RISM registers and the instruction register IR for controlling the timing and operation of the registers and the serial input and output to and from the TAP architecture. The instruction register IR is used to select the test to be performed or the test data register to be accessed or beth by serially shifting an instruction into the instruction register IR and latching the instruction into the finite state machine at the completion of the shifting. In this implementation, the instruction register IR forms a four bit shift register serially connected to the TDI input, each of the plurality of RISM registers and the TDO output in addition to being connected in parallel to the TAP controller.
Referring again to FIG. 1, the C296 micro-controller further comprises an interrupt unit (INTERRUPT) having break logic for interrupting the controller's current stream of instructions and entering an interrogation mode. To do this, the break logic issues a non-maskable interrupt (NMIE) to the interrupt unit upon detecting an external break interrupt from the external system (i.e., toggling the NMIE pin of the IC) or upon detecting execution of a RISM ACTION command.
In order to enter the interrogation mode by means of executing a RISM ACTION command, the command is serially shifted into the RISM ACTION register 5 via the TAP architecture when the IC is still in normal operation. As shown in FIG. 2, the RISM ACTION command to be shifted in comprises a series of bits forming a RISM data portion (ignored for read operations), a RISM address portion, and a RISM command portion including a FGO bit and a break bit. In order to cause the transition upon execution of the RISM ACTION command, the break bit in the RISM CMD register chain 4 must be set so as to invoke the NMIE.
Upon assertion of an update signal from the TAP controller to the RISM ACTION register 5, the bits in the RISM ACTION register are written in parallel to the corresponding CORE registers 21,22,23. The break logic detects the set break bit upon updating the RISM ACTION register 5 and issues the NMIE to the interrupt unit. The interrupt unit transfers processor control from the user's code to the RISM micro-cede stored at the NMIE vector address within the test ROM. The RISM micro-code then reads the RISM command in the RISM CMD register chain 4 and executes the command accordingly. By means of sequentially inputting a series of RISM commands (such as write.sub.-- word) in addition to appropriate data and addresses from the external system 32 to the RISM ACTION register 5, the RISM micro-code can write the data transferred from the external system 32 to the specified memory locations within the embedded memories. In this manner, data stored within a buffer in the external system 32 can be serially transmitted to the IC and easily programmed into the IC's embedded memory modules.
Utilizing the TAP controller states shown in state diagram of FIG. 4 (which are described in detail in the IEEE 1149.1-1990 Standard), read and write operations to embedded memories of a CPU-based IC via the RISM ACTION register 5 can be performed by sequentially applying binary 1 and 0 signal states to the TMS pin during each dock of the TCK pin to navigate the finite state machine of the TAP controller through the state diagram.
To select the RISM ACTION register 5 as the current, operational register, a RISM ACTION opcode is shifted into the instruction register IR by traversing the TAP controller states of Test-Logic Reset, Run-Test/Idle, Select-DR-Scan, Select-IR-Scan, Capture-IR and Shift-IR. Upon entering the Shift-IR state, a logical 0 signal applied to the TMS pin for four clock cycles so that the opcode bits will be serially shifted in through the TDI input. Next, the TAP controller is instructed to request the memory bus (both MemAdd and MemData) for the RISM ACTION register 5 (via a bus request to the bus controller) by traversing the states of Shift-IR, Exit1-IR, Pause-IR, Exit2-IR, Update-IR and Run-Test/Idle. The bus request actually occurs when the TAP controller reaches the state of Update-IR where the opcode bits within the instruction register IR are transmitted in parallel to the TAP controller which executes the opcode instruction and issues a bus request to the bus controller.
In order to perform a write operation, the TAP controller must be instructed to traverse the states of Select-DR-Scan, Capture-DR and Shift-DR. When the Shift-DR state is reached, a logical 0 signal is applied to the TMS pin for 64 clock cycles so that a 24 bit address string followed by a 32 bit data string (i.e., the data to be programmed) followed by an 8 bit command string (i.e., write.sub.-- word) is received by the RISM ACTION register 5 from the external system 32 via the TDI input. The write operation will not be requested until the TAP controller enters the UPDATE-DR state where the bits within the flip-flops 27 of the RISM ACTION register chains 4,6,8 are loaded in parallel into the associated level-sensitive latches 29 and subsequently transmitted to the corresponding CORE registers 21,22,23. At this time, the RISM micro-code writes the data bit string from the CORE DATA register 23 to the memory location specified by the address bit string. This process is repeated as many times as is needed to write the entire contents of the external system's storage buffer to the desired memory module.
In order to verify that the data was properly written to the specified memory location, it is possible to read the data back out of the memory module in which it was written. In this case, if the RISM ACTION opcode is not then in effect, the TAP controller will have to be navigated to the SHIFT-IR state and held there for four cycles to shift in the opcode for the RISM ACTION register 5. Next, the TAP controller is instructed to request the memory bus by navigating it to the Run-Test/Idle state in the same manner as was done for the write operation. The TAP controller is then instructed to traverse the states of Select-DR-Scan, Capture-DR and Shift-DR. When the Shift-DR state is reached, a logical 0 signal is applied to the TMS pin for 64 clock cycles so that a 24 bit address followed by a 32 bit data string (ignored for read operations) followed by a RISM command (i.e., read.sub.-- word) is input to the RISM ACTION register 5 from the external system 32. Finally, the read operation is requested upon entering the update state of the TAP controller. Several clock cycles after the request, the MIU will return the data bite stored in the specified memory location and load them into the CORE DATA register 23.
The returned data bits within the CORE DATA register 23 in addition to the command bits within the CORE CMD register 21 are then loaded into the RISM DATA chain 8 and the RISM CMD chain 4, respectively, by entering the CAPTURE-DR state (where the parallel DIN input to the multiplexor 30 is selected). The TAP controller is next moved to the SHIFT-DR state and held there for 64 clock cycles by the application of a logical 0 signal state to the TMS pin during each cycle. In this manner, the entire contents of the RISM ACTION register 2 is serially output through the TDO output to another storage buffer within the external system 32 where the data read from memory can be compared with that which was previously written to memory.
Accordingly, the present invention provides the ability to program a wide variety of IC's using a single apparatus and a standardized programming method in addition to a serial interface which requires only four active pins. Consequently, future ICs can be made less expensive and can be programmed during the manufacturing process or reprogrammed at a later time when upgraded software becomes available.
While the invention has been described in conjunction with the preferred embodiment, it is evident that numerous alternatives, depictions, variations and uses will be apparent to those skilled in the art in light of the foregoing description. Thus, it is understood that the invention is not to be limited by the foregoing illustrative details, but rather is to be defined by the appended claims.