System and Method for Disabling or Re-Enabling PCI-compliant Devices in a Computer System by Masking the Idsel Signal With a Disable or Re-Enable Signal

This invention relates to a system and method for disabling Peripheral Component Interconnect (PCI)-compliant devices through hardware by controlling the Initialization Device Select (IDSEL) signal to a PCI-compliant device using system Input/Output (I/O) registers. The invention advantageously permits users to disable user-selected onboard PCI-compliant devices, such as video controllers or Small Computer System Interface (SCSI) controllers. Users may, for example, wish to substitute another PCI-compliant device in place of a particular PCI-compliant device provided by the computer manufacturer. In other cases, users may wish to disable a particular PCI-compliant device either temporarily or indefinitely.

At the present time, several computer manufacturers provide a user-initiated software method of disabling PCI-compliant devices. The software that accomplishes disablement of PCI-compliant devices is typically found in the Basic Input/Output System (BIOS) memory of a computer system. (BIOS is the firmware in a personal computer that interfaces directly with the hardware to perform I/O and other low-level functions. Typically this firmware is permanently stored in Read Only Memory or flash memory on the computer system's motherboard.) Subsequently, when the computer system is powered on, the system's Power On Self Test (POST) is executed. During POST, software contained in BIOS memory determines which PCI-compliant devices are disabled. Upon determining that a particular PCI-compliant device is disabled, the BIOS software does not initialize and execute the Read-Only Memory (ROM) associated with that device. However, the configuration space that is reserved by the system for each onboard PCI-compliant device is still accessible to system software, including operating system software. This configuration space remains accessible because the PCI specification does not recognize the concept of a PCI-compliant device being "disabled" by BIOS software.

As operating system software has become increasingly sophisticated, there has developed a clear trend toward automatically configuring all devices attached to the computer system, including PCI-compliant devices. Certain recent generations of operating systems--e.g., Microsoft Windows NT, and possibly the forthcoming Microsoft Windows 95 product--are likely to continue this trend. Specifically, the operating system software will endeavor to detect all accessible hardware associated with a computer system. The operating system software will detect and have access to the disabled PCI-compliant device's configuration space and possibly attempt to initialize it. Thus, despite a user's express request to disable a particular PCI-compliant device, the operating system software will detect the device's configuration space and re-activate it.

Therefore, what is needed is a hardware-based method for disabling PCI-compliant devices of a computer system to avoid the problem of operating system software detecting disabled devices and initializing them in contravention of the user's requests.

In a method in accordance with the invention, executable software contained in BIOS memory controls the state of the IDSEL signal to a PCI-compliant device using I/O registers of the computer system. Using utility software contained in BIOS memory, a user can request that one or more PCI-compliant devices be disabled. Upon request to disable a device, the software contained in the BIOS memory will control the state of the IDSEL signal to the affected PCI-compliant device, thereby disabling the device and preventing configuration space from being reserved for the device. IDSEL is a physical line in the PCI bus architecture--a PCI-compliant device generates an IDSEL signal in response to an I/O inquiry from, for example, the operating system. By providing a "gated" IDSEL, a PCI-compliant device can be disabled and re-enabled at the user's request via a BIOS routine. Once disabled via the method of the present invention, computer system software, including operating system software such as Microsoft Windows NT or possibly the forthcoming Microsoft Windows 95, will not detect the presence of the device and attempt to re-activate it.

FIG. 1 is a block-diagram view of a computer system including depictions of components used in a method in accordance with the present invention.

One implementation of the invention is described here for purposes of illustration, namely a hardware-based method of disabling one or more PCI-compliant devices in a computer system by using system I/O registers to control the state of the IDSEL signal to the target device(s). The machine-executed method is invoked at the user's request and is performed by executable computer software contained in BIOS memory. An actual implementation of such computer code might be executable on an Intel 80.times.86--based or Pentium.TM.--based computer system, or on other suitable processor-based computer systems.

(It will be appreciated by those of ordinary skill, of course, that references to the performance of method steps by computer software actually mean that a computer, or one of its components, is performing the method steps in response to the instructions encoded in the computer software.)

In the interest of clarity, not all features of an actual implementation are described in this specification. It will, of course, be appreciated that in the development of any such actual implementation (as in any hardware or software development project), numerous design and programming decisions must be made to achieve the developers' specific goals and subgoals (e.g., compliance with system-related and business-related constraints), which will vary from one implementation to another. Moreover, attention will necessarily be paid to, e.g., proper serialization to handle concurrent events. It will be appreciated that a development effort of this type might be complex and time-consuming, but would nevertheless be a routine undertaking of computer system design and development for those of ordinary skill having the benefit of this disclosure.

Depicted now in FIG. 1 is a high-level block diagram of the computer system operating environment in which the method of the present invention may be implemented. The elements of a computer system not necessary to understand the operation of the present invention have been omitted for simplicity. The computer system includes CPU 100 which is coupled to host bus 105. Host bus 105 includes address, data, and control portions. The host bus 105 is coupled to PCI/Host Bridge 110, which, in turn, is coupled to PCI bus 115. PCI/Host Bridge 110 is a microchip that controls PCI bus 115.

PCI bus 115 is a 32-bit local bus supporting relatively high data-transfer speeds (up to 132 megabytes per second at 33 megahertz.). Advantages of the PCI bus standard include greater speed than that obtainable from Industry Standard Architecture (ISA) and Extended Industry Standard Architecture (EISA) buses, as well as automatic configuration of PCI-compliant devices that are added to, changed, or removed from a computer system utilizing the PCI bus architecture. Automatic configuration of PCI-compliant devices is accomplished, in part, by information provided in registers contained in the PCI-compliant devices themselves. The content and addressable locations of the registers contained in the PCI-compliant devices are dictated by PCI bus architecture standards.

PCI Bus 115 is coupled to one or more PCI-compliant devices 120. PCI/Host Bridge 110 produces a signal across PCI bus 115 on the IDSEL pin of PCI-compliant device 120 when said device 120 is the target of a configuration command (read or write). In this sense, the IDSEL pin of a PCI-compliant device functions very much like a classical "chip select" signal.

The present invention allows the user to disable a PCI-compliant device via hardware-based control of the IDSEL signal. In one implementation, the IDSEL signal from PCI bus 115 directed to the target PCI-compliant device 120 is gated through AND gate 130. Thus, the IDSEL signal from PCI bus 115 serves as one input to AND gate 130. The second input to AND gate 130 is CONFIG ENABLE signal from a bit set in system I/O registers 125. In the present illustration, system I/O registers 125 can be system-specific registers with addressable locations known only to the manufacturer of a given computer system, but not generally known to third party hardware or software developers. Because the addressable locations of system I/O registers 125 are in undocumented locations, system software developers, including developers of operating system software such as Microsoft Windows NT or possibly the forthcoming Microsoft Windows 95, do not attempt to access them. Alternatively, the addressable locations of system I/O registers 125 could be fully documented and made known to third party developers, but be implemented as "write once" registers, as is well known in the art. If implemented as "write once" registers, the bit corresponding to CONFIG ENABLE would be set once during POST, and then remain unchangeable until the computer system is re-booted. Under either of these implementations, the CONFIG ENABLE signal is generated from a bit set in system I/O registers 125, said bit being set by software contained in the computer system BIOS 135.

Responsive to a request by the user to deactivate or re-activate a particular PCI-compliant device 120, software contained in the computer system BIOS 135 can either set a bit in system I/O registers 125 to either a `1` or `0.` This bit is identified as CONFIG ENABLE, and provides the second input to AND gate 130. Whether the system I/O registers 125 are implemented in undocumented addressable locations or as "write once" registers, the software contained in BIOS 135 will manipulate the CONFIG ENABLE bit in response to requests for deactivation and re-activation by the user. If system I/O registers 125 are implemented in undocumented addressable locations, third party software developers, including operating system software developers, will typically not be able to address these system I/O registers 125. If system I/O registers 125 are implemented as "write once" registers, third party software developers will not be able to manipulate the contents of these registers because a write operation to these registers will occur during POST.

Because the present invention makes use of system I/O registers for the purpose of gating the IDSEL pin of a PCI-compliant device, existing and future operating system software will not override or otherwise ignore a user's affirmative request to disable or re-enable a particular PCI-compliant device. Once a PCI-compliant device is disabled using this hardware-based solution, operating system software will not attempt to assign system resources to a device which a user has specifically disabled.

Those of ordinary skill in the art will recognize that them are many alternative implementations to gate the IDSEL signal to a PCI-compliant device. The use of an AND gate combining as inputs the IDSEL signal coming from the PCI bus and the CONFIG ENABLE signal coming from a bit set in system I/O registers is but one specific implementation of gating IDSEL. Therefore, it will be appreciated by those of ordinary skill, having the benefit of this disclosure, that numerous variations from the foregoing illustration will be possible without departing from the inventive concept described therein. Accordingly, it is the claims set forth below, and not merely the foregoing illustrations, which are intended to define the exclusive rights claimed in this application.