Method of Determining the Configuration of Devices Installed on a Computer Bus

1. Field of the Invention

The invention relates to computer busing systems, and more particularly to a method of and apparatus for determining the configuration and types of boards installed on a computer bus.

2. Description of the Related Art

The microcomputer industry has experienced tremendous growth over the last twenty years. From the days of its infancy when only a few interested "hackers" could fathom its quirks and nuances, the microcomputer has now evolved into a powerful business and personal tool found on virtually every office desk and in virtually every home.

The microcomputer's road to success has not been without its problems, however. While advances occur at an astounding pace, those advances must accommodate the standards found in the then existing base of microcomputer systems. This is known as upwards compatibility. To maintain such compatibility, the industry has seen one microcomputer standard laid on top of another, with a resulting hodgepodge of standards-within-standards that designers must maintain to allow existing users to upgrade their equipment. These multiple standards gradually shed their oldest layers, replacing them with new layers reflecting the state-of-the-art. In this way, only the very oldest microcomputer systems become obsolete.

One early idea to enhance microcomputer systems was the addition of hardware enhancing boards. These boards were generally plugged into a system bus to provide added functionality, such as telecommunications, disk storage, and improved video. These boards obviously had to conform to some standard. With the introduction of the IBM PC by International Business Machines Corp., and the later introduction of the PC/AT by IBM, the AT system bus soon became a de facto standard known as the Industry Standard Architecture bus, or the ISA bus. The AT bus accommodated both the 8-bit boards of the PC and newer 16-bit boards developed for the AT. Third-party manufacturers could economically design standard boards compatible with the wide variety of IBM PC and AT compatible microcomputer systems.

Further advances in microprocessor technology, however, pushed the ISA bus to its limits. For this reason, another "layer" was added to the ISA bus standard. This added layer became known as the Extended Industry Standard Architecture bus, or the EISA bus. Boards designed for the EISA bus had more pins, providing a wider data path for information to flow through the microcomputer system bus, analogous to adding lanes to a highway. The EISA bus also added more address lines to the standard, permitting more memory locations to be individually specified, much as would adding more digits to a phone number or a zip code.

One limitation of the ISA bus involved its method of handling I/O addressing. An address enable signal (AEN) was driven low by an ISA bus master to indicate to all of the cards that the currently asserted address was an I/O address or a memory address rather than a direct memory access (DMA) operation. But because AEN was asserted low to all cards, each card had to be physically configured to respond to a different range of I/O or memory addresses to avoid conflicts. This address differentiation was usually accomplished when installing the boards by setting microswitches on dual in-line packages (DIP) or by connecting jumpers on each board. Improperly setting these switches could result in conflicts on a read or write to a particular I/O or memory address and could even result in physical hardware damage.

While the ISA standard provided 16 bits of I/O addressing, in developing boards for PC-compatible computers, vendors often only used or decoded the lower 10 bits. Thus, to be fully compatible with the available boards, the I/O address space of the ISA bus effectively was only from 0 to 03 FFh. Thus, a large portion of the I/O space was unusable.

Another corresponding limitation of the ISA bus involved the DMA channels and interrupt lines. The PC offered several DMA channels and interrupt lines for third party expansion boards to utilize, however, because each DMA channel and interrupt line would have to be physically configured, the potential existed for conflicts in this area also. System resources would also reserve several of the DMA channels and interrupt lines for themselves, thus further limiting the number of DMA channels and interrupts line available to third party boards, and further compounding the problem.

The EISA bus standard has resolved this problem to some extent. The EISA bus definition provides for a conflict-free I/O address space for each slot. This is fully described in U.S. Pat. No. 4,999,805 and the EISA Specification, Version 3.1, which is Appendix 1 of U.S. Pat. No. 4,101,492, both of which are hereby incorporated by reference. The expansion board manufacturers include a configuration file with each EISA expansion board, and optionally, with switch programmable ISA products. A configuration utility program provided by the system manufacturer uses the information contained in the configuration files to determine a conflict-free configuration of the system resources. The configuration utility stores the configuration and initialization information into non-volatile memory and saves a backup copy on diskette. Details of this configuration process are provided in Ser. No. 07/293,315, entitled "Method and Apparatus for Configuration of Computer System and Circuit Boards," allowed on May 10, 1993, which is hereby incorporated by reference. The system ROM power up routines use the initialization information to initialize the system during power up, and device drivers use the configuration information to configure the expansion boards during operation.

However, this slot specific addressing does not help with ISA boards. Slot specific ISA board disabling can prevent such physical conflicts between two boards during their initialization. Briefly, a mask register is provided to mask off the AEN signal to selected slots. Details are provided in Ser. No. 08/145,400, entitled "Method of and Apparatus for Disabling Individual Slots on a Computer Bus," filed Oct. 29, 1993, which is hereby incorporated by reference.

Further, the slot specific addressing is of no assistance with memory operations, as the EISA bus standard does not provide for slot specific memory spaces for ISA cards.

Determining what addresses that board responds to is not trivial. Unlike EISA boards, ISA boards do not provide an identification register. Thus, the occupied address space of an ISA board must be determined in some other way.

During the booting or power on self test (POST) operation, the computer determines if any option ROMs are present on the installed circuit boards. If so, an initialization routine in the ROM is executed. It is at this time that the circuit board enables or reserves its selected DMA and interrupt channels by properly programming the DMA and interrupt controllers. Because ISA boards do not include standardized ways to indicate the selected DMA and interrupt channels, they cannot be determined by reading a location on the circuit board. Thus, even the availability of slot specific operations and selective addressing does not reveal what DMA and interrupt resources are used.

It would be desirable to provide the functionality of EISA configuration software for ISA boards. That is, it would be desirable for the system to be able to determine what ISA boards were installed in which slots, and to appropriately respond to any conflicts or mismapping of those devices. Finally, it would be desirable to determine where an ISA board is mapped in the address and memory space and to configure the system accordingly. To do this, it is necessary to obtain all of the information possible from a given ISA board. The related applications detail techniques do determine active address locations, but do not provide DMA or interrupt information, which can further characterize a board and allow conflict free setup. So it would be desirable to be able to determine the DMA and interrupt resources utilized by as given circuit board to be able it to be characterized and matched.

In a computer system constructed according to the invention, code executed at power-on determines if an expansion board requires a particular DMA channel and/or interrupt line. Usage requirements are determined by monitoring register changes in the DMA controller and interrupt controller made by the initialization steps of an option ROM. The system first locates an option ROM. Then before loading the option ROM, the register values of the DMA controller and the Interrupt controller are stored in memory. The code contained in the option ROM is then executed. If the expansion board containing the option ROM required DMA or Interrupt controller services, the registers will have changed. When control returns from the option ROM, the registers are read and the routine determines whether any DMA and/or Interrupt controller utilization was required. If usage requirements are found, the information is stored for later use by the configuration software. The routine continues cycling through all the slots and all pertinent addresses searching for option ROMs. When finished, the system completes it's booting process.

After the system has booted, user initialization software determines whether an expansion board is present in a selected slot and what input/output (I/O) addresses that selected expansion board drives. The system first disables all the expansion slots except for the one under test. The system then cycles through all the relevant I/O addresses, creating a map of the I/O addresses to which an expansion board in the enabled expansion slot responds. A response other than a normal undriven bus value indicates an installed board in the enabled slot is driving the data bus in response to a read from that I/O address. The returned value is stored in the map. For each I/O address from which the system reads a normally undriven bus response (0FFh in typical systems), this is stored in the map as an "undriven" response for further testing.

The system further tests each address returning such an "undriven" response to determine whether an installed board is actually driving the bus to that normally undriven value. First, if the expansion slot under test asserts certain bus control lines in response to an I/O read, that indicates that an installed device is responding to that I/O read. If so, that response is stored in the map.

If not, the system then analyzes the data line response time to an I/O read at the address being tested. Specifically, the system measures how long the data lines take to rise to an undriven state of 0FFh from an artificially driven state of 00h. Hardware first drives the data lines to zero at the beginning of an I/O read cycle. The hardware then times the rise time of the data lines. Because the response time for a driven bus is less than that for an undriven bus, a timer value less than the normal response time indicates that a device is actually driving the normally undriven value onto the data bus. Thus, the particular expansion slot under test both contains an expansion board and that board is driving the bus in response to a read from the I/O address under test.

In this way, the system creates a map of the I/O address locations used by a board in a selected expansion slot and the actual response values of that board. This information can be combined with the previously stored DMA and interrupt information. Configuration software then determines the actual board type by comparing the I/O map, and DMA channel and/or interrupt line requirements with "signatures" compiled for various known expansion boards. Once the board type and its address map are known, the configuration software passes that information to an extended version of EISA configuration software, which both sets up device drivers and otherwise configures the system. Further, the EISA configuration software determines whether there are any conflicts between the expansion boards installed into the various slots of the system bus. If so, the EISA configuration software can warn the user, can suggest to the user the proper switch settings, or can even disable the board entirely.

Further according to the invention, a corresponding memory map of an ISA board is created in a manner similar to that for creating an I/O address map. The technique is similar, except memory "response time" need not be checked to determine if a memory address is being driven, as instead, two different predetermined values are written to that memory location and then read back. If those values are properly stored, the corresponding value will be returned on a read. If the values are not properly stored, the read will return a value other than what was written on at least one of the reads. This map is also then used by the configuration software.

A better understanding of the present invention can be obtained when the following detailed description of the preferred embodiment is considered in conjunction with the following drawings, in which:

FIG. 1 is a block diagram of a computer system incorporating the method and apparatus according to the invention;

FIG. 2 is a flowchart of configuration software that determines the system configuration according to the invention;

FIGS. 3 and 4 are flowcharts of software used to determine address utilization on the system bus according to the invention;

FIGS. 5A and 5B are exemplary address maps created by the software of FIGS. 3 and 4; and

FIGS. 6A-F and 7A-D are schematics of the control line status latching circuitry used according to the invention.

FIGS. 8A, 8B and 9 are flowcharts of software used to determine direct memory access or interrupt controller utilization on the system bus according to the invention.

Turning now to the drawings, FIG. 1 is a block diagram of a microcomputer system 100 in which the method and apparatus according to the invention is implemented. The microcomputer system 100 includes a host bus 102 and a system bus 104. A microprocessor 106, memory 108, and a video subsystem 110 internally communicate at high speed with each other over the host bus 102. The host bus 102 is designed for speed rather than expansion, so the host bus 102 has little, if any, provision for expansion boards. Optionally, a local bus could be present in addition to an expansion bus, in which case the method of the present invention could be applied to it also.

A system bus controller 112 provides an interface between the host bus 102 and system bus 104. The system bus controller 112 is located on the system board of the microcomputer system 100. The system bus controller 112 is preferably implemented using an application specific integrated circuit (ASIC) but could be implemented using discrete components. Internal to the system bus controller 112 are a Direct Memory Access (DMA) controller 130 and a Interrupt Controller (Interrupt controller) 132. The DMA controller 130 is preferably an Intel 8237 compatible DMA controller and the Interrupt controller is preferably an Intel 8259 compatible Interrupt controller. More advanced DMA controllers and Interrupt controller's could also be used.

The system bus 104 is typically an EISA bus, but could be another bus using similar addressing protocols. The system bus controller 112 implements the functions of an EISA bus controller. It is also within the system bus controller 112 that the data bus response time circuitry according to the invention is preferably implemented.

The system bus 104 consists of address lines 114, data lines 116, and control lines 118. Connected to the system bus 104 is a system board slot 120. The system board slot 120 is not a separate physical connection, but instead logically connects "devices" integrated into the system board of the microcomputer system 100 itself to the system bus 104. Further connected to the system bus 104 are slots 122. The slots 122 are physical connectors for inserting expansion boards compatible with the standard of the system bus 104 to provide the added functionality to the microcomputer system 100. Shown inserted in the first, fifth, and seventh of the slots 122 are respectively a hard disk controller 124, a floppy disk controller 126, and a network interface board 128. Additionally, a non-volatile memory (NVRAM) for storing configuration information about the computer system 100 is connected to the system bus 104.

The lower byte of the data lines 116, denoted as SD[7..0], are pulled up by pull-up resistors 117. These pull-up resistors 117 ensure that the undriven data lines 116 return a value of 0FFh. The EISA standard specifies that these pull-up resistors 117 should be 8.2 k ohm.

As is further discussed below, the pull-up resistors 117 can instead be pull-down resistors. In such a case, the value returned by an I/O read of an undriven data bus is then 00h instead of 0FFh.

Each device connected to the system bus 104, whether it be a device plugged into one of the slots 122 or a system board device corresponding to the system board slot 120, includes an individual slot specific address enable line SSAEN[Z], where Z equals 0 to 7. These signals correspond to the AENx signals of the EISA specification or AEN signal for ISA systems, but further implementing slot specific disabling, as is described in Ser. No. 08/145,400, previously incorporated by reference. This permits each individual slot to be queried for the presence of an expansion board without interference from other potentially conflicting expansion boards.

The DMA controller 130 provides a mechanism for peripheral devices residing on the system bus 104 to directly transfer data to or from memory without microprocessor intervention. Peripheral devices communicate with the DMA controller 130 over what is called a DMA channel with the DMA controller 130 having several channels. The DMA controller 130 has a number of DMA signals referred to as DMA Request (DRQ) and DMA Acknowledge (DACK) signals. A channel represents a particular pair of the DRQx/DACKx signals, with x denoting the channel number. Certain DMA channels are reserved for system use, for example, the floppy disk is allocated channel 2. Most of the channels supplied by the DMA controller 130 are provided to each slot 122 for the expansion boards to utilize, thus the potential exists for two different expansion boards to attempt to use the same DMA channel or even one of the reserved channels.

The interrupt controller 132 provides a mechanism for receiving interrupts from peripheral devices, prioritizing them and presenting the interrupt to the microprocessor. When a peripheral device needs servicing by the microprocessor, that device can issue an interrupt signal on the interrupt line. Those skilled in the art will appreciate that certain system resources are reserved certain interrupt lines and the remaining interrupt lines are left for the expansion boards. Interrupts can be shared, but this requires another layer of software that is not easily managed between boards made by different manufacturers. The concern over backwards compatibility with older boards also limits the use of interrupt sharing since some of these boards are not designed for this. Thus, with limited interrupt lines available, the potential for conflicting use exists.

The DMA channels and interrupt lines used by the expansion boards are sometimes initialized by software supplied with the expansion board in the form of an option ROM (Read Only Memory). Other types of expansion boards may have accompanying software in the form of a device driver loaded when the operating system is booted. Both of these ways will cause the software to communicates with the registers of the DMA controller 130 and the Interrupt controller 132 to initialize the DMA controller 130 and Interrupt controller 132 for particular expansion board requirements. However, for purposes of detecting DMA controller 130 and Interrupt controller 132 utilization, the preferred method of detecting utilization is by monitoring the loading of option ROMs. It should also be noted that since some expansion boards come without an option ROM, such as those that provide standard parallel port and serial port peripherals, those expansion boards will also not be responsive to the DMA/Interrupt controller method disclosed herein.

Referring now to FIGS. 8A and 8B, a flowchart is shown illustrating the method used to determine the usage of DMA channels and interrupt lines by expansion boards. The software is preferably part of the system BIOS (Basic Input/Output Services) software and contained in ROM. After powering on the computer system 100, or a cold reset, the microprocessor 106 will begin executing from the system BIOS at step 700. Control proceeds to step 702 where the microprocessor 106 performs the Power On Self Tests (POST). These tests indicate whether the hardware is operating properly before continuing. Control then proceeds to step 704 where the microprocessor 106 begins to initialize the computer system 100. At this point, the reserved DMA channels and interrupt lines are initialized for use, thus the DMA controller 130 and Interrupt controller 132 are in a predictable state.

One of the final steps in the initialization process is to run the initialization sequence of any installed option ROMs. Conventionally this is simply done by scanning the defined option ROM space for a defined signature and executing the initialization code if an option ROM is found. However, operations are slightly different in the preferred embodiment. Control proceeds to step 706 where the routine enables the first slot for testing as described below.

Control then proceeds from step 706 to step 708 where the first option ROM address is determined. Option ROMs are placed at 1 k-byte boundaries in memory space between C0000h and EFFFFh. In the present method, since options ROMs are preferably loaded from lowest to highest address, the preferred method of cycling through slots and addresses is to cycle through the slots while keeping the address fixed, then increment to the next higher address. An alternative method would be to cycle through the addresses while keeping only one slot enabled, then move to the next slot. Once the first address is set, control then proceeds to step 710 where a single slot is read at the preset address for the presence of an option ROM.

Control then proceeds from step 710 to step 712 where the routine determines if an option ROM was found. This is typically accomplished by looking at each 1K boundary of memory space for a 55AAh value. If an option ROM is not found then control proceeds to step 728, which is discussed below. It should be noted, that if a board is present but an option ROM is not installed on that expansion board, then the routine is unable to predict whether any expansion boards would utilize any DMA channels or interrupt lines. If at step 712 an option ROM is found at the present slot number/memory address, then control proceeds to step 716 where the routine reads the DMA controller 130 and Interrupt controller 132 registers. Control then proceeds to step 716 where the routine temporarily stores the register values in memory 108.

Control then proceeds to step 718 where the routine transfers control to the option ROM code by making a call to an address 3 bytes above the boundary address where the option ROM was found. This causes control to be transferred to step 740 (FIG. 9) where the option ROM initialization code begins. Control then proceeds from step 740 to step 742 where the option ROM initialization code is executed. At this point, if the expansion board requires the use of a DMA channel and/or interrupt line, the option ROM initialization code will attempt to initialize the registers affecting that particular DMA channel and/or interrupt line. When the option ROM initialization code has finished executing, control then proceeds to step 744 where control is returned to the BIOS routine at step 718 and control then proceeds to step 720.

At step 720, the routine again reads the register values of the DMA controller 130 and the Interrupt controller 132. The routine then proceeds to step 722 where the values previously stored in memory 108 are compared to the values read at step 720.

The routine then proceeds to step 724 where the routine determines if a DMA channel and/or interrupt line was initialized by looking for changes in the register values. If the register values are unchanged, then the routine assumes that no DMA channels and/or interrupt lines are required by that expansion board and control then proceeds back to step 728, which is discussed below. If the register values are changed, then the routine stores the slot number and usage information into NVRAM 136 for later use by the CONFIG 400 routine, discussed below, and control proceeds to step 728.

At step 728 the routine determines if the last slot has been tested. If so, then the routine proceeds to step 732. If the last slot has not been tested, then the routine proceeds to step 730 where the slot number is incremented so the next slot can be tested. Control then proceeds from step 730 back to step 710, discussed above. At step 732 the routine determines if the last 1 k-byte address has been tested. If so, then control proceeds to step 736 where the remaining BIOS routines are executed and the operating system is booted. If the last 1 k-byte address has not been tested, then control proceeds to step 734 where the memory address is incremented so the remaining memory addresses can be tested on the same slot. Control then proceeds from step 734 to step 710, discussed above to continue testing the remaining slots and memory addresses. Thus when completed, the NVRAM 136 will contain an indication of which slot utilized which DMA and interrupt resources.

Turning now to FIG. 2 is a flowchart of a routine CONFIG 400 that uses address maps provided by a routine ADDRESS.sub.-- UTILIZATION 200, discussed below in conjunction with FIGS. 3 and 4, to determine board presence and to check for conflicts among the slots 122. Beginning at step 402, the routine CONFIG 400 calls the routine ADDRESS.sub.-- UTILIZATION 200, retrieving an address map for each of the slots 122, as is discussed below in conjunction with FIGS. 5A and 5B. The routine CONFIG 400 then proceeds to step 404, where it determines the presence or absence of an expansion board in each of the slots 122.

This is done by checking the address map discussed below in conjunction with FIGS. 5A and 5B to see if any address locations were responsive to I/O reads when that particular slot 122 was enabled. That address map preferably indicates the values read from each I/O address, the values on the control lines 118 on a read from each I/O address, and the rise times of the data lines 116 if the address otherwise appears unresponsive. Based on these values, the address map also indicates whether the particular I/O address is responsive. If all of the I/O addresses were indicated to be unresponsive, then no ISA expansion board or EISA expansion board with ISA functions occupies that particular slot 122.

The routine ADDRESS.sub.-- UTILIZATION 200 also determines the presence of EISA boards in slots by reading from the EISA identification registers. If an EISA board does in fact occupy a slot 122, the address maps discussed above and below in conjunction with FIGS. 5A and 5B need not be checked, as the EISA configuration registers both indicate board presence and uniquely identify those expansion boards.

From step 404, the routine CONFIG 400 proceeds to step 405, where it reads the slot number and usage values generated by the DMA/Interrupt controller routine from NVRAM 136. This information will indicate if an expansion board in a particular slot has initialized a particular DMA channel and/or interrupt line. The routine then merges this information into the address map to provide a more complete set of information about the expansion boards.

From step 405, the routine CONFIG 400 proceeds to step 406, where it attempts to determine the board types and address spaces using signature analysis routines. The system software includes a library of ISA boards and their address space and DMA/interrupt signatures. A "signature" is a definition of the possible address spaces and DMA/interrupt channels occupied by a board used to (hopefully) uniquely identify both the functions on and the types of boards installed. For example, a parallel port may return a unique configuration of bits on an I/O read. That configuration is the "signature" of the parallel port function on an expansion board. An expansion board implementing four such ports might have four such parallel port "signatures" spaced four I/O addresses apart each. Such parallel port "signatures" plus the port spacings would then form that expansion board's signature. Each board may have multiple signatures, in that each board may include more than one logical device or function. In that case, the signature of the board would represent more than a single device, and both of those determined "devices" could be passed to further configuration software for configuration of each of those devices. Using this library, the routine determines the various ISA boards installed in the system.

From step 406, the routine CONFIG 400 then passes the identifications, base addresses, DMA channel utilization and interrupt line utilization of the ISA boards, along with the identifications of the installed EISA boards, to expanded EISA configuration software EISA.sub.-- CONFIG at step 408. This EISA configuration software can further resolve conflicts or inform the user of such conflicts, and can complete the configuration of the system, including setting up the device drivers to properly drive each installed board. This EISA configuration software is more fully described in Ser. No. 07/293,314, entitled "Method and Apparatus for Configuration of Computer System and Circuit Boards," which is hereby incorporated by reference. That EISA configuration software relies on the board IDs returned by the identification registers of EISA boards and on saved configurations for ISA boards.

FIG. 3 is a flowchart of the routine ADDRESS.sub.-- UTILIZATION 200 that determines both the presence of an expansion board in a slot 122 and, if present, the particular addresses that device uses.

Beginning at step 202, the routine ADDRESS.sub.-- UTILIZATION 200 sets a variable SLOT# equal to one. This variable corresponds to the specific slot 122 under test on the system bus 104. Proceeding to step 204, the routine enables the slot 122 that corresponds to SLOT# by writing to an arbitrary I/O address P.sub.-- SSAEN used to disable the remaining slots 122. Writing zeros to all the bits of P.sub.-- SSAEN except the bit corresponding to SLOT# prevents all of the slots 122 except SLOT# from responding to any I/O operations. Even if enabled, the slot 122 SLOT# only responds if a board is both installed and mapped to a particular I/O address under test.

For example, if SLOT# equals 3, the routine writes 00001000b (08h) to I/O address P.sub.-- SSAEN. A write to that address stores that value in a mask register which then prevents the AENx signal from going low to the disabled slots. This selective slot disabling is further described in Ser. No. 08/145,400, referenced above.

In the first time through the routine ADDRESS.sub.-- UTILIZATION 200, SLOT# equals one, enabling the first of the slots 120. Because the designers of the system 100 know the I/O address map of the system board slot 120, the system startup software can skip mapping the 'system board slot 120 (SLOT# equal to zero) and proceed straight to SLOT# equal to one.

Proceeding to step 205, the routine ADDRESS.sub.-- UTILIZATION 200 determines whether an EISA board is installed in the slot 120 SLOT#. This is done by reading I/O addresses containing EISA board identification information, as defined by the EISA Specification, Version 3.1, referenced above. An EISA board need not be mapped because the identification registers uniquely define that board. So, if an EISA board is detected, the routine proceeds to step 206 where it saves that identification information for later use by the EISA configuration software, and then to step 222 to process the next slot 120.

If no EISA board is detected at step 205, the routine ADDRESS.sub.-- UTILIZATION 200 then proceeds to step 207, where it reads all significant I/O addresses and stores the read values in an address map. In this embodiment, the routine only needs to read I/O addresses an ISA expansion board would use, or the address range 0100h to 03FFh. ISA systems only effectively employ ten significant address bits SA[9..0] on the address lines 114 of the system bus 104. Further, bits SA[9..8] equalling zero corresponds to an ISA system board address in the system board slot 126 or to an EISA address. ISA expansion boards should not respond to system board addresses, so the routine need only examine those 10-bit addresses in which SA[9..8] do not equal zero. The significant address range is therefore 0100h to 03FFh. Addresses with bits SA[15..10] other than zero are also disregarded, because in an ISA system these addresses are generally considered to be aliases of the I/O addresses located in the ten address bit expansion board range. Thus, by checking an address range of 0100h to 03FFh, the routine maps all I/O addresses used by an ISA board.

As noted above, addresses in which bits SA[9..8] equal zero can correspond not only to ISA system board addresses but also to EISA addresses. Under the EISA standard, EISA board addresses are handled through circuitry that enables a specific slot enable SSAEN[X] when SA[9..8] equal zero, with X corresponding to the four high order bits of the full 16-bit address, or SA[15..12]. Thus, EISA devices installed in the slots 122 will not conflict with each other because each slot 122 has its own separate EISA address range.

In reading the significant addresses at step 206, the microprocessor 106 will read an 0FFh if no device is installed in the slot 122 under test or if the device is not mapped to the I/O address read. This results from the pull-up resistors 117 pulling the data lines 116 high. Conversely, reading a value other than 0FFh indicates the slot 122 has an enabled device mapped to the I/O address read. In such a case, the routine stores a "true" flag in the address map indicating this I/O address is occupied by a device in the slot 122 SLOT# under test. In either case, the data value read is stored in the address map.

For simplicity, memory locations are not tested. Occupation of certain I/O addresses alone should be enough to establish a unique signature for a particular board under test. At that time, a memory check can be performed if desired.

However, simply because an I/O read returns 0FFh does not mean that a device in the slot 122 under test is not driving the data lines 116. A device installed in the slot 122 may be driving the data lines 116 to the normally undriven value of 0FFh. The routine ADDRESS.sub.-- UTILIZATION 200 later, at step 212, determines this by checking whether the device under test asserts certain control lines in response to an I/O read from the address under test and, if not, whether the response time of the data lines 116 to an I/O read is quicker than the response time of an undriven bus.

At step 208, the routine ADDRESS.sub.-- UTILIZATION 200 determines the response time of the undriven data lines 116. First, the routine enables a rise time measurement mode by writing an enabling bit to a status/control register at an arbitrary I/O address P.sub.-- CTRLSTAT. The routine then disables all of the slots 122 by writing 00h to P.sub.-- SSAEN. The routine then reads from an ISA expansion board I/O address in rise time measurement mode. This read operation is special as the data lines 116 are first driven to zero and then a timer is started. The timer is stopped upon the data lines 116 reaching predefined voltage levels or upon reaching a predetermined timer value limit. In this case, because all slots 122 are disabled, no installed device can respond. The routine then reads the value of the timer from an arbitrary I/O address P.sub.-- TIMER. This returns the rise time in HCLKs, or host bus 102 clock cycles, for a read of an undriven I/O address after first driving the data bus to zero. The hardware enabled by the rise time mode bit written to P.sub.-- CTRLSTAT performs the precharging of the data lines 116 to logic zero and then the timing of the bus rise time to a value other than logic zero. This hardware is preferably implemented in the system bus controller 112 in FIG. 1, and is further described in Ser. No. 08/145,339, entitled "Detecting the Presence of a Device on a Computer System Bus by Measuring the Response Time of Data Signals on the Bus," which is hereby incorporated by reference. This undriven data line 116 response time is saved for later comparisons.

At step 210, the routine ADDRESS.sub.-- UTILIZATION 200 stores in an address pointer variable ADDRPTR the next I/O address determined to be "undriven". The first time through this inner loop, the first I/O address is the first I/O address in the address map created at step 206 that returned a normally undriven I/O read value of 0FFh. The routine 200 then proceeds to step 212, where it checks for the assertion of control lines 118 in response to a read and checks the rise time of the data lines 116 during a read from the I/O address pointed to by ADDRPTR. This is done in a routine CHECK.sub.-- PRESENCE 212, discussed below in conjunction with FIG. 3.

Proceeding to step 214, if the routine CHECK.sub.-- PRESENCE 212 determined a device actually drove 0FFh onto the data lines 116, then at step 216 the routine ADDRESS.sub.-- UTILIZATION 200 stores a true value into a flag in the address map that indicates the address was responsive to a read. Otherwise, the routine ADDRESS.sub.-- UTILIZATION 200 instead proceeds to step 218, where it stores a false value in that flag, indicating the address was non-responsive to a read. That is, any device, if present, in the slot 122 under test is not mapped to this address.

From both steps 216 and 218, the routine ADDRESS--UTILIZATION 200 proceeds to step 220, where it determines whether any I/O addresses that initially returned 0FFh remain in the address map. If so, the routine proceeds again to step 210, where it stores the next "undriven" I/O address in the map into ADDRPTR. If no such addresses remain, the routine proceeds from step 220 to step 222.

At step 222, the routine ADDRESS.sub.-- UTILIZATION 200 determines whether SLOT# equals MAXSLOT. In an EISA system, this could be up to 14, the greatest number of non-system board slots EISA systems support. Because the system designers actually know how many slots 122 are present in the system 100 the designers can set MAXSLOT equal to the appropriate number in the configuration software. This eliminates the mapping of slots 122 that are not present.

If SLOT# is not equal to MAXSLOT, the routine ADDRESS.sub.-- UTILIZATION 200 proceeds to step 224, where it increments SLOT#, and then proceeds to step 204 to enable that next slot 122. The entire loop is repeated, creating another address map of the I/O addresses to which a device in the next slot 122 responds.

If at step 222 SLOT# equals MAXSLOT, then the routine ADDRESS.sub.-- UTILIZATION 200 is done checking for I/O address utilization and has created its address maps, so it returns to the configuration software CONFIG 400.

FIG. 4 is a flowchart of the routine CHECK.sub.-- PRESENCE 212. This routine uses the rise time circuitry as shown and described in Ser. No. 08/145,339, previously referenced, to determine the response times of the data lines 116. This routine also uses latching circuitry described in conjunction with FIGS. 6 and 7 to determine expansion board presence.

The routine CHECK.sub.-- PRESENCE 212 begins at step 300 by enabling a latching mode by writing a bit to the arbitrary port P.sub.-- CTRLSTAT previously discussed. On an EISA or ISA system, some expansion boards respond to I/O reads by asserting certain control lines among the control lines 118. They may do so even if they drive the data bus to 0FFh. If these control lines are asserted on an I/O read, the selected expansion board of slot 122 is mapped to that I/O address, so no further testing need be done. The hardware for implementing this latching mode is described below in conjunction with FIGS. 6 and 7; it essentially latches the values of certain control lines 118 at appropriate times in a I/O read cycle.

At step 300, with the latching mode now enabled, the routine CHECK.sub.-- PRESENCE 212 then performs an I/O read from the address under test pointed to by ADDRPTR. The routine then reads the arbitrary register P.sub.-- CTRLSTAT, which returns the latched values of the particular control lines mentioned above. The routine then stores that returned value in the address map.

Proceeding to step 304, the routine CHECK.sub.-- PRESENCE 212 examines the latched value read at step 302 to see if an expansion board has driven the control lines 118. If so, the routine proceeds to step 306, where it sets a return parameter to true, indicating that a read from this address results in a response from an expansion board in the slot 122 that is enabled. The routine then returns that parameter at step 308.

If at step 304 it was determined that no expansion board asserted any of the relevant control lines 118 in response to a read, the routine CHECK.sub.-- PRESENCE 212 proceeds to step 310. Even though an expansion board, if any, in the slot 122 under test failed to assert these lines, this does not necessarily mean that an expansion board in the slot 122 under test is not driving the data lines 116 in response to a read from the I/O address ADDRPTR. The routine at step 310 further determines expansion board presence by enabling the rise time mode by writing a particular bit to the arbitrary port P.sub.-- CTRLSTAT discussed above.

The routine then proceeds to step 312, where it checks the rise time of the data lines 112 in response to a read from the address under test ADDRPTR. This is accomplished by the hardware in Ser. No. 08/145,339, referenced above. To summarize, this hardware first drives the data lines 116 to 00h and then counts the number of HCLKs until one of those data lines 116 changes to one. The routine then reads that timer at an arbitrary I/O address P.sub.-- TIMER. As discussed in Ser. No. 08/145,339, the choice of using pull-up resistors 117 and driving an initial value of 00h onto the data lines 116 is not the only way to implement this hardware. Values other than 00h could be used in conjunction with the data bus compare register, and the rise time could be based on the time it takes all, rather than just one, of the data lines 116 to change from their initial values. In any case, a response time at least a certain number of clock cycles (such as 2 HCLKs) less than the rise time of the undriven data lines 116 as determined and stored at step 208 above indicates an expansion board is driving the normally "undriven" value onto the data lines 116. Thus, the slot 122 under test contains an expansion board that is responding to an I/O read from the location ADDRPTR. This rise time is stored in the address map and the rise time mode is disabled.

Proceeding to step 314, this rise time is compared to that undriven bus rise time stored at step 208. If this rise time is less than the undriven rise time by a certain number of HCLKs added for random fluctuations in rise time, the routine proceeds to step 306, setting its return parameter to true. Otherwise, the routine CHECK.sub.-- PRESENCE 212 proceeds from step 314 to 316, where it resets its return parameter to false, indicating a device is not responding to a read from ADDRPTR. The routine then returns the true or false parameter at step 308. Before returning, the routine disables the rise time mode.

The specific order in which the address map is created is arbitrary. For example, the routine CHECK.sub.-- PRESENCE 212 could be executed as each address is initially mapped at step 207, rather than performing a separate loop to detect expansion board presence at "undriven" addresses. Similarly, the latch control line checking of steps 302 and 304 could be combined with the initial mapping of step 207, because such latching does not interfere with an I/O read.

FIGS. 5A and 5B illustrate a typical address map created by the routine ADDRESS.sub.-- UTILIZATION 200. As shown in FIG. 5A, this address map is an array containing four values for each address between 100h and 03FFh. For each address location, the routine ADDRESS.sub.-- UTILIZATION 200 stores a main value in a flag DRIVEN, stores the initial read value from the I/O address in a location READ, and may or may not store values in locations LATCH corresponding to the value latched on a latched read and RISE corresponding to the timer value on a rise time check.

The initial address map as shown in FIG. 5A is created by the routine ADDRESS.sub.-- UTILIZATION 200 at step 207. As can be seen, the routine has read all significant I/O addresses (100h-3FFh) and stored those read values in the second column (READ) of the address map. In memory, this is stored in the second byte of the four bytes of memory reserved for each I/O address tested. Also at step 207, the DRIVEN flag in the address map has been set to true (0FFh) for each I/O address that returns a value other than the undriven value of 0FFh. This indicates that it is known that this address location is driven. As can be seen, on the initial read, locations 104h-106h have returned values other than zero on the read, so the corresponding DRIVEN flag byte has been set to 0FFh.

FIG. 5B shows the address map after the routine CHECK.sub.-- PRESENCE 212 has been executed for all of the "undriven" memory locations. This can be seen in relation to address 100h which, as seen in FIG. 5A, returned an "undriven" read value of 0FFh. In FIG. 5B, location 100h has returned a latched read value LATCH of 02h, as determined at step 300. This indicates that the slot 122 under test has an expansion board which has latched certain control lines in response to an I/O read, so DRIVEN for location 100h is also set to true. Note that no rise time measurement is made on location 100h, because the latch read has already indicated a utilized address.

For address 101h, both the read of the address at step 207 has returned an undriven value of 0FFh and the latch read has returned an undriven value of 00h. So, the rise time is checked at steps 310 and 312. In this case, a rise time of 05 h is returned. Assuming the standard undriven rise time of the system is 01 Ch, this rise time of 05h indicates that an expansion board is driving the data lines 116 to 0FFh in response to an I/O read from address 101h. Thus, DRIVEN for address 101h is set to true. Addresses 102h and 103h have also returned an undriven value of 0FFh on an I/O read from those addresses and a latch value of 00h. Those values have returned rise time values of 01Bh and 01Ch, however, and since the undriven bus rise time is 01 Ch, both of these values correspond to an undriven rise time. (A rise time of 01 Bh is within two clock values of the undriven rise time value, and is considered as being within an error margin for undriven values.) Thus, the locations DRIVEN corresponding to addresses 102h and 103h stay false. The same holds true for location 107h and 3FFh (and presumably for all intervening values).

The latched read value and rise time value need not be checked on I/O reads of location 104h-106h, as a read has returned driven values in response to an I/O read from those locations. The latch read is preferably performed for those locations anyway, as it might give extra information to assist the signature checking algorithms to determine the type of ISA board installed in the slot 122 under test.

Not shown are the corresponding address maps for the other slots 122 under test. A corresponding address map is created for each of these, and then those maps are passed further along to the higher level configuration routine CONFIG 400, as discussed above in conjunction with FIG. 2.

At this point, the routine ADDRESS.sub.-- UTILIZATION 200 has created an address map indicating responding I/O addresses that each slot 122 drives when enabled.

Turning to FIG. 6, that figure is a schematic of the hardware for latching the control lines 118 on an I/O read at step 302 above. FIG. 6 shows latch control signals P.sub.-- LATCH[3..0], which are driven onto the data lines 116 on an I/O read from P.sub.-- CTRLSTAT. The hardware provides four signals SS.sub.-- LNOWS*, SS.sub.-- LIO16*, SS.sub.-- LM16*, and SS.sub.-- LCHRDY as these latch control signals, each of which is further described below. These signals correspond to the sampled values of the ISA bus no wait state signal NOWS*, input/output command signal IO16*, memory command signal M16*, and channel ready signal CHRDY when those signals are valid on an I/O read. An ISA board sometimes asserts these signals in response to an I/O read. If asserted, these signals thus indicate that an expansion board in the slot 122 under test is responding to a read from the particular I/O address under test.

These signals provided as P.sub.-- LATCH [3..0] are generated by the circuitry of the schematics shown in FIGS. 6A-F and 7A-D. SS.sub.-- LCHRDY is provided by the Q* output of a D flip-flop 500. The D flip-flop 500 is reset by a reset signal SS.sub.-- RST*, discussed below, and is clocked by an inverted BCLK signal BCLK*. BCLK is a standard ISA clock signal, and generally runs at 8 MHz. An OR gate 502 drives the D input of the D flip-flop 500. The OR gate 502 receives as inputs the Q output of the D flip-flop 500 and the output of an AND gate 504.

The inputs to the AND gate 504 include the EISA bus command signal CMD* after being inverted by an inverter 506, a signal ISARDY (discussed below) after being inverted by an inverter 508, a latched host bus 102 HLDA signal LHLDA*, which is latched on the host bus clock signal HCLK by circuitry not shown, and the output of an AND gate 510. HLDA is true during memory refresh, DMA, and bus master cycles, when all of the circuitry of FIGS. 6A-F and 7A-D should be disabled. The AND gate 510 has as inputs a signal DIS.sub.-- LEN (discussed below) and a status latching enable signal SS.sub.-- LEN. SS.sub.-- LEN reflects the value of a corresponding bit in the I/O register P.sub.-- CTRLSTAT.

DIS.sub.-- LEN, which disables status latching on reads from P.sub.-- SSAEN and P.sub.-- CTRLSTAT, is provided by the Q* output of a D flip-flop 512. The D flip-flop 512 is reset by a low signal out of an AND gate 514, which has as inputs SS.sub.-- RST* and a system reset signal RST*. CMD* clocks the flip-flop 512. The D input to the D flip-flop 512 is provided by an AND gate 516, which as inputs receives SS.sub.-- LEN, and the output of an OR gate 518. The inputs to the OR gate 518 are the Q output of the D flip-flop 512 and the output of an AND gate 520. The AND gate 520 receives as inputs LHLDA*, a decode P.sub.-- SSAEN.sub.-- DEC* for a read or write to P.sub.-- SSAEN, and a decode P.sub.-- CTRLSTAT.sub.-- DEC* for a read or write to P.sub.-- CTRLSTAT.

SS.sub.-- LIO16* is provided by the Q output of a latch 22, which is gated by an AND gate 524 with IO16.sub.-- LEN, SS.sub.-- LEN, and DIS.sub.-- LEN as inputs. The D input to the latch 522 is provided by an AND gate 526, which has as inputs LHLDA* and IO16*.

SS.sub.-- LM16, is provided by the Q output of a latch 528, which is gated by an AND gate 530 with inputs BALE, SS.sub.-- LEN, and DIS.sub.-- LEN. BALE is a standard ISA signal that indicates the validity of address values. The D input to the latch 528 is provided by an AND gate 532, which has as inputs LHLDA* and the ISA signal M16*. SS.sub.-- LM16* can be ignored if only I/O reads are being mapped.

SS.sub.-- LNOWS* is provided by a D flip-flop 534, which is reset by SS.sub.-- RST* and is clocked by BCLK. The D input to the D flip-flop 534 is provided by an OR gate 536, which has as inputs the Q output of the D flip-flop 534 and the output of an AND gate 538. The AND gate 538 has as inputs CMD* after inversion by an inverter 540, LNOWS (the inverted latched value of an ISA no wait state signal NOWS*), LHLDA*, and the output of an AND gate 542, which has as inputs DIS.sub.-- LEN and SS.sub.-- LEN.

Turning to FIG. 7A-D, ISARDY is provided by the Q output of a latch 600. BCLK gates a latch 600, which as its D input receives the Q output of a D flip-flop 602. CHRDY drives the reset input and the D input of the D flip-flop 602, and BCLK clocks the D flip-flop 602. A latch 604 provides LNOWS, and LNOWS at its Q and Q* outputs, is gated by BCLK, and as its D input receives NOWS*.

SS.sub.-- RST* is provided by a NAND gate 606, which has as inputs P.sub.-- CTRLSTAT.sub.-- DEC* after inversion by an inverter 608, an ISA I/O read command signal IORC*, and the Q output of a D flip-flop 610. The D flip-flop 610 is clocked by BCLK* and has its D input pulled low. The D flip-flop 610 is reset by the output of an 0R gate 612, which has as inputs P.sub.-- CTRLSTAT.sub.-- DEC* and IORC*. SS.sub.-- RST* thus resets the latching circuitry when the latched values are read from P.sub.-- CTRLSTAT.

IO16.sub.-- LEN is provided by an AND gate 620, which has as inputs the Q* outputs of a D flip-flop 622 and a D flip-flop 624. The D flip-flop 622 is clocked by BCLK*, and its D input is driven by the Q* output of the D flip-flop 624. The D flip-flop 624 is clocked by BCLK, and its D input receives the output of a NAND gate 626, which has as inputs the signals ESD1, ESQ2*, and ESQ0, which are outputs of a state machine that is not shown. This state machine tracks the command state of the ISA bus. These signals cause IO16.sub.-- LEN to go high on the first rising edge of BCLK after IOWC* or IORC* is asserted, and to then go low on the next falling edge of BCLK. This circuitry is well known to one of ordinary skill in the art of expansion bus design. IO16* is then latched on the falling edge of IO16.sub.-- LEN.

An eight-line, four-input multiplexer 614 provides as outputs the signals DO[7..0]. These signals are driven onto the data lines 116 by circuitry not shown in response to I/O reads of P.sub.-- CTRLSTAT, P.sub.-- TIMER, or P.sub.-- SSAEN. The lower 4 bits of the C input to the multiplexer 614 receive P.sub.-- LATCH[3..0], with ground tied to the other bit inputs. The D input receives TIMER[6..0], with ground tied to the eighth bit input. The select lines for the multiplexer 614 are driven by a NAND gate 616 to the select line 1 and a NAND gate 118 to the select line 0. The NAND gate 616 receives as inputs P.sub.-- CTRLSTAT.sub.-- DEC* and P.sub.-- TIMER.sub.-- DEC*, while the NAND gate 618 receives as inputs P.sub.-- SSAEN.sub.-- DEC* and P.sub.-- TIMER DEC*.

The multiplexer 614 provides DO[7..0] corresponding to the signals P.sub.-- LATCH[3..0] when P.sub.-- CTRLSTAT.sub.-- DEC* is low, and the signal TIMER[6..0] is provided as DO[7..0] when a read is detected by a decode of the signal P.sub.-- TIMER.sub.-- DEC*, as both the NAND gate 616 and the NAND gate 618 are driven high. This hardware thus allows the microprocessor 106 to read the values of the latch register and the timer register.

Memory addresses are mapped in a way similar to that illustrated in FIGS. 2-5B, with the exception that instead of determining presence as shown in FIG. 4, two distinct values, such as 00h and 0FFh, are consecutively written to and read from the memory address under test. If the memory address on each subsequent read returns the written value, that indicates the memory address is present. If the memory address does not return one of those values on a read, that indicates the memory address is not driven by the board under test. The corresponding address map is set up to FIGS. 5A and 5B, with the exception that the "latch" and "rise" values need not be stored, and instead just a presence or absence flag is stored. This address map is then used by the CONFIG routine 400 of FIG. 2.

The foregoing disclosure and description of the invention are illustrative and explanatory thereof, and various changes in the size, shape, materials, components, circuit elements, wiring connections and contacts, as well as in the details of the illustrated circuitry and construction and method of operation may be made without departing from the spirit of the invention.