US 5,388,267 AGrant
Method and Apparatus for Updating and Restoring System BIOS Functions While Maintaining BIOS Integrity
Issue Date:1995-02-07
•22 Claims
•2 Drawing Sheets
Abstract
A computer which carries its BIOS in a Flash EPROM. A UV-EPROM carries a redundant BIOS, which can be overlaid onto the BIOS address space by selection with a physical switch. The BIOS contains a small core software program, at the BIOS entry point, which checks BIOS integrity, and provides for reloading the Flash EPROM's BIOS if needed (from a floppy disk, or by copying the entire contents of the UV-EPROM).
Metadata
Assignee
- Dell USA, L.P.
Inventors
- Wai-Ming R. Chan
- Eric W. Schieve
- Charles P. Zeller
- Gary W. Abbott
Application Information
Application Number:US 7071213
Filing Date:1991-05-29
Priority Date:1991-05-29
Art Unit:236
Classifications
IPC:
G06F 906G06F 924
Field of Search:
37119
Patent Drawings (2 sheets)
Description
TABLE OF CONTENTS
BACKGROUND AND SUMMARY OF THE INVENTION
Bootstrap Programs
Basic System Software
"Basic Input/Output System" Software (BIOS)
Nonvolatile Memory Technologies
Functions Performed by the Software Core
Detecting Which Boot Memory is Active
Hardware Protection of the Software Core Sector
Floppy-Disk Handling
BRIEF DESCRIPTION OF THE DRAWING
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Preferred Hardware Context
Preferred I/O Controller Chip ("SLOB")
Use of Parallelled UVEPROM and Flash-EEPROM for BIOS
Use of Sector-Protected Flash-EEPROM
Hardware-Software Hybrid Switch
Preferred Memory Controller Chip ("TRANE")
Preferred Core Software Structure
Procedure 8KBOOT.sub.-- STRT
Procedure BOOTUVPROM
Procedure BOOTFLASH
Other Procedures
Further Modifications and Variations
CLAIMS
1. A method for operating a computer system
11. A method for operating a computer system
16. A method for operating a computer system
APPENDICES
PARTIAL WAIVER OF COPYRIGHT
All of the material in this patent application is subject to copyright protection under the copyright laws of the United States and of other countries. As of the first effective filing date of the present application, this material is protected as unpublished material.
Portions of the material in the specification and drawings of this patent application are also subject to protection under the maskwork registration laws of the United States and of other countries.
However, permission to copy this material is hereby granted to the extent that the owner of the copyright and maskwork rights has no objection to the facsimile reproduction by anyone of the patent document or patent disclosure, as it appears in the United States Patent and Trademark Office patent file or records, but otherwise reserves all copyright and maskwork rights whatsoever.
BACKGROUND AND SUMMARY OF THE INVENTION
The present invention relates to computer systems.
Bootstrap Programs
Any computer system must have some way to begin program execution after a cold start. The hardware architecture of a CPU (processor) will normally provide for a "reset" operation, which places all of the hardware circuits in a known electrical state; but it is still necessary to start the CPU on execution of a desired program. For example, in the very early days of computing, some computer systems would be manually configured to read in a "bootstrap loader" program at startup. This bootstrap program was a simple program which loaded in, and started execution of, another sequence of instructions, which were the beginning of the desired program. Bootstrap programs are often referred to simply as "boot" software.
To give a more recent example, the Intel 80.times.86 microprocessors, after a hardware reset, will always attempt to begin program execution from a specific memory address..sup.1 Thus, if a branch (or conditional branch) instruction is found at this address, the microprocessor will continue its program execution from whatever address is specified.
Thus, this initial target address is the entry point for every session of use. This address is normally used to enter execution of programs which must be run every time the computer is used.
Basic System Software
Whatever hardware is used will have its own procedures to return to a known state when a reset occurs. However, at some point these procedures end, and the CPU is ready to begin execution of instructions.
At this point the system performs various overhead tasks under software control. These may include surveying the system configuration, sanity checks, etc.
Basic Input/Output System Software (BIOS)
In modern personal computers, the initial target address is normally used as the entry point to a "basic input/output system" (BIOS) program. The BIOS program contains frequently-used routines for interfacing to key peripherals, for interrupt handling, and so forth..sup.2 In addition to these basic input/output routines, the "BIOS" software also includes some key pieces of overhead software, such as configuration update and the power-on-self-test (POST) routines..sup.3 The BIOS software will also launch the machine into the operating system software..sup.4 (Thus, the term "BIOS" has become somewhat broader nowadays, and normally refers to this whole collection of basic system routines.)
If the BIOS software were to become corrupted, the computer could become unusable. Thus, the BIOS software has conventionally been stored in read-only memory (ROM). When the microprocessor attempts to access the initial target address, it reads out software from the BIOS ROM.
In 1980 there was only one source for IBM-compatible BIOS software, and that was from IBM. However, during the 1980s, as IBM-compatible personal computers became more popular, modified versions of IBM-compatible BIOS ROMs were developed, and IBM-compatible BIOS ROMs were offered by multiple vendors. As of 1991, BIOS software is often modified to implement system-dependent features, especially in low-power systems.
Improvements in BIOS software means that sometimes it will be desirable to implement a BIOS upgrade. Dedicated users have successfully pried out and replaced ROM chips, but most users would not want this degree of hands-on contact.
Some attempts have been made in the past to provide capability for updating the basic system software. See, e.g., Bingham, D. B., "Achieving flexible firmware,+ 1978 MIDCON Technical Papers at 20/3/1-4 (1978), which is hereby incorporated by reference.
It is believed that some vendors may have offered upgradable-BIOS systems. However, insofar as is known to the inventors, no such system has offered the protection against corruption while updating BIOS, which is provided by the disclosed innovations.
Nonvolatile Memory Technologies
Without boot software, a computer could not run any program at all. Thus, corruption of the boot software can make a computer totally unusable. Thus, for security against such corruption, personal computers have commonly been manufactured with their BIOS software in nonvolatile memory chips. However, for rapid system development and upgrading, it is desirable to be able to rapidly provide new BIOS chips, or even to upgrade existing BIOS chips. Over the years, a variety of developments in semiconductor device technology have been used to reconcile these needs.
One of the simplest nonvolatile memories is the mask ROM. By custom-patterning one level of a chip, a complex pattern of data can be permanently encoded..sup.5 With this technology, a new mask must be prepared whenever any bit of data is changed. In high volume, mask ROMs are reasonably cheap, but they are not suited for rapid upgrading.
A "programmable read-only-memory," or PROM, can be electrically written..sup.6 However, once the data has been written it is permanent.
An "electrically programmable read-only-memory," or EPROM, can be electrically written, and can be erased by ultraviolet light..sup.7 EPROMs are very commonly used, since they are cheap and their timing standards are familiar. EPROMs are also referred to as UVPROMs.
To write a cell, the channel is typically driven hard, with a high voltage on the control gate. Secondary carrier multiplication in the channel creates hot electrons, which are attracted into the floating gate. Thus, programmation of a conventional EPROM requires application of a high voltage (e.g. 10 or 15 V) to achieve the high currents needed. Moreover, programmation is typically quite slow, e.g. several milliseconds per bit.
To erase a cell, it is exposed to ultraviolet light. The energetic photons excite energetic carriers, which can pass through the dielectric layer to neutralized the charge on the floating gate. Thus, EPROMs normally need a package with a quartz window.
An "electrically erasable programmable read-only-memory," or EEPROM or E.sup.2 PROM, can be electrically written and electrically erased..sup.8 EEPROMs are less commonly used than EPROMs, since they tend to be more expensive and to require even higher voltages.
A more recent modification of the EPROM is the "Flash EPROM." This device, like the EEPROM, is electrically erasable, but only in blocks. Although this device does not have the bit-by-bit programmability of the EEPROM, it is still useful in many applications.
A wide variety of rewritable nonvolatile memory technologies have been proposed, and doubtless others will continue to be proposed. For example, the computers of the 1950s and early 1960s used "core" memory technology, which is, to some extent, nonvolatile and rewritable. The disclosed innovations are not limited to Flash EPROMs, but can be adapted to use with other memory technologies.
Safe In-Situ Upgrades for Boot ROM
The disclosed computer system contains several innovative features which permit the boot software to be easily upgraded, and also protect against any corruption of the boot software.
Redundant Boot Memories
The disclosed system uses a rewriteable nonvolatile memory as the primary boot memory. For further security against corruption of the boot software, the preferred system uses TWO boot memories. By selection with a hardware switch, either one can be connected as the boot source.
Overlaid Addressing
Since either of the two boot memories may need to be the target, both boot memories are mappable to the same address. Note that this causes some difficulty in writing to the volatile boot memory: if the volatile boot memory is not in the memory map, it cannot be written to.
In the presently preferred embodiment, this difficulty is avoided by mapping the core software out to RAM, and then commanding an address configuration change to access the other boot memory. If one boot memory is being copied onto the other, then the contents of the source memory must also be copied out.
As noted above, the default mapping of the boot memory address, at initial power-up, is controlled by a physical switch. However, thereafter, a peripheral memory controller chip controls the mapping of the boot memory address. Thus, by issuing a command to this peripheral chip, the core software can change this mapping.
Thus, to write to the boot memory, the core software copies itself out to a RAM location, and commands the microprocessor to branch into the RAM address. The bit to select the boot memory is then toggled, so that writes to the boot memory addresses are now directed to the other boot memory (i.e. to the boot memory which was not the source of the code being executed). The desired data can then be written into the target boot memory. Note that, if the source of the data is in the active boot memory, that code must be written out into RAM before the boot memory address toggle is switched.
Pre-POST Software Core
The starting address in the boot software is occupied by a small (and strongly protected) software core, which performs sanity checks, and also provides the needed supervisory functions for boot software upgrading and replacement.
Thus, the disclosed innovations solve the problem of performing BIOS restoration/upgrades in-situ (i.e. without removing the FLASH boot memory ROM from the system). The end-user (or a service technician) can program BIOS code into the FLASH without any specialized external hardware or software ROM-programming tools.
Functions Performed .by the Software Core
After any hard reset of the system, the core software detects whether the boot software is corrupted, and whether the user has entered a request for upgrade. If these conditions have occurred, the core software performs the FLASH programming from the appropriate data source.
In the presently preferred embodiment, the core software provides for the following three classes of cases. These cases together provide coverage for all conceivable field conditions:
1. Case: While executing reset code from the FLASH, the system BIOS and/or
video BIOS are found to have a bad checksum. In this case, power-on-self-test and booting is impossible, therefore, the boot code prompts the use.sup.9 to insert a diskette containing system and video BIOS code and reprograms the FLASH. When the system and video BIOS programming is successfully completed, the user is prompted (via SmartVu) to reset the system. Normal POST and boot should follow.
2. Case: While executing reset code from the FLASH, the boot code detects a CMOS.sup.10 flag setting indicating a user requested upgrade. The boot code then prompts the user (via SmartVu) to insert a diskette containing the upgrade system and video BIOS code. When the system and video BIOS programming is successfully completed, the user is prompted (via SmartVu) to reset the system. Normal POST and boot should follow.
3. Case: While executing reset code from the UVPROM, the boot code detects that a FLASH is present and that its protected 8 k boot sector is jumpered for programming. Since the UVPROM is intended as a secondary ROM device, this condition is understood as a request for FLASH programming and the entire contents of the UVPROM, including the boot code in the upper 8 k sector is copied into the FLASH. When the entire FLASH is successfully programmed, the user is prompted to reset the system. If the FLASH is selected as the primary ROM device, normal POST and boot should occur from the FLASH.
Detecting Which Boot Memory is Active
Note that the foregoing steps may require the CPU to ascertain, while it is running boot code, which memory is the source of the code. This is accomplished, in the presently preferred embodiment, by letting the CPU read a register in a peripheral chip.
Hardware Protection of the Software Core Sector
The preferred rewritable boot memory is a sector-protected Flash-EPROM. The sector which contains the software core is actually protected by a jumper, which must be physically moved before the software core can be overwritten. This provides additional robustness.
Floppy-Disk Handling
Because the core software may have to read from floppy disk, it includes the necessary overhead routines to perform this. These routines are generally similar to the BIOS functions under INT13h.
BRIEF DESCRIPTION OF THE DRAWING
The present invention will be described with reference to the accompanying drawings, which show important sample embodiments of the invention and which are incorporated in the specification hereof by reference, wherein:
FIG. 1 is a flow chart which schematically shows key portions of the methods used in the computer system of the presently preferred embodiment.
FIG. 2 shows the hardware configuration which permits switching between alternative boot memories, in the computer system of the presently preferred embodiment.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The numerous innovative teachings of the present application will be described with particular reference to the presently preferred embodiment. However, it should be understood that this class of embodiments provides only a few examples of the many advantageous uses of the innovative teachings herein. In general, statements made in the specification of the present application do not necessarily delimit any of the various claimed inventions. Moreover, some statements may apply to some inventive features but not to others.
Preferred Hardware Context
The preferred embodiment is an 80486-based EISA-bus PC system. General features of the EISA bus are described in Glass, "Inside EISA," Byte magazine November 1989, pp. 417ff, which is hereby incorporated by reference, and in the EISA specification, which is hereby incorporated by reference. General features of the Intel 80486 are described in Sartore, "The 80486: A Hardware Perspective," Byte magazine IBM Special Edition, Fall 1989, pp. 67ff, which is hereby incorporated by reference. Further detailed background on the 80486 may be found in the "80486 Programmer's Reference" and the "80486 Hardware Reference Manual," both available from Intel and both hereby incorporated by reference.
Many of the architectural features of this system are conventional in modern PC systems. However, several unusual features are used. One of these is the provision for dual boot memories, with hardware and software switching between them. Another notable feature is the use of the "SLOB" controller chip described below.
Preferred I/O Controller Chip ("SLOB")
In the presently preferred embodiment, the system motherboard includes a custom chip, referred to herein as the "SLOB" chip, which performs a variety of useful functions. These include reset control, X-bus transceive.sup.11 control, and control of the boot memories and CMOS nonvolatile memory.
Use of Parallelled UVEPROM and Flash-EEPROM for BIOS
The software core, in the presently preferred embodiment, resides in the protected 8 k boot sector of a FLASH ROM or in the upper 8 k of a 128 k UVPROM. Selection of the primary ROM device is accomplished via a physical switch.
FIG. 2 shows more detail of how this is implemented. The SLOB chip supplies chip enable (CE ) signals separately to the FLASH EPROM 220 and to the UVEPROM 230 (and also to the NVRAM 240). An output enable line OE is connected to both boot ROMs, and address lines are also provided to both. A hardware switch provides a line ROMIN to the SLOB chip, to define which boot memory is the initial default target.
Use of Sector-Protected Flash-EEPROM
In the presently preferred embodiment, the FLASH EPROM has sector-by-sector protection. (This feature is not available in most FLASH EPROMs, but is available in the newly introduced 28F001B from Intel, which is used in the presently preferred embodiment.)
In the presently preferred embodiment, one sector 220A, of only 8K bytes, is dedicated to the core software. This sector is protected by a hardware jumper, so corruption of the core software is unlikely.
Hardware-Software Hybrid Switch
To achieve automatic programming of the Flash memory from a plug-in ROM, the hardware must provide a switching mechanism that allows software to boot from one source and then toggle between sources. This switchability is also required to access EISA configuration from Flash if running BIOS out of a plug-in ROM.
A user settable hardware switch determines which boot memory the boot code will be fetched at power-up or cold boot. This switch sets the polarity of a bit called ROMIN# which is decoded by the hardware control logic to enable the selected boot source. This bit is readable from an I/O port, so that software can determine the source of the boot code, i.e. ROM (bit=0) or Flash (bit=1). Software then has the ability to toggle between accessing ROM or Flash by setting a bit that assumes the opposite state of ROMIN# after a cold boot. This bit is called ROMEN and enables Flash when 0 and ROM when 1. This bit is readable as well as writable via an I/O port. This mechanism allows hardware to boot from ROM then gives software the ability to determine the presence of Flash memory and program Flash automatically by transferring the contents of the BIOS in ROM to the Flash.
Preferred Memory. Controller Chip ("TRANE")
The chip referred to herein as the TRANE chip is a custom memory controller. Besides performing normal DRAM management functions, this chip also provides selection, by memory domains, of which memory areas will be cached or not.
Preferred Core Software Structure
The actual implementation of the core software will now be described in great detail. However, it must be understood that this specific implementation is merely illustrative, and is subject to change, and does not by any means delimit the scope of the inventions claimed.
Functionality
FIG. 1 is a flow chart which schematically shows key portions of the methods used in the computer system of the presently preferred embodiment.
When the 80486 first comes out of reset, it comes up in real mode.sup.12 (but the high address bits are held high, so that the processor can go to the top of the 4-Gigabyte (32-bit) memory space. The processor accesses address FFFFFFF0h.
Preferred Assembly Language Implementation
The actual implementation of key portions of the core software, in the presently preferred embodiment, will now be given. The following listings in documented assembly language also contain a large number of informal comments. These comments do not necessarily define the scope of the invention, but will help to explain the motivation, structure, and workings of the presently preferred embodiment.
Some of the procedures actually used, in the presently preferred embodiment, will now be described in detail. Of course, it should be understood by those skilled in the art that the very specific implementation details given are not by any means necessary to the invention. The following wealth of detail is provided merely to assure compliance with the best mode requirements of U.S. patent laws.
Procedure 8KBOOT STRT
Procedure 8kboot.sub.-- strtis the power-on-reset entry point into the 8 k boot sector code for a 128 k FLASH EPROM. This code along with a reset vector is located in the upper 8 k boot sector of a 128 k FLASH EPROM. The reset vector will be located at 1fff0h (physical ROM address), and the start of this routine will be located at 1e000h (start of the 8 k boot sector). The reset vector will contain a NEAR jmp to this code, so execution will start in processor real mode, ROM native mode, at logical address 0ffffe000h. This procedure implements most of the steps in the flow chart of FIG. 1. Specifically, this procedure:
Saves the state of EAX and DX for later use by POST.
Initializes a base address for the DRAM controller (the TRANE chip, in the presently preferred embodiment).
Disable interrupts (NMI and INTR).
If warm boot, jmp into POST SHUTDOWN routine.
Else, if power-on check to see if we're running out of a UVPROM or the FLASH.
If FLASH, jump to the "bootFLASH" procedure which programs the flash if tile BIOS is corrupted or if the user has requested an upgrade.
If UVPROM, jump to the "bootUVPROM" code which copies the UVPROM to the FLASH if a FLASH exists.
If either "bootFLASH" or "bootUVPROM" does not program the flash, control returns to "do.sub.-- reboot" which displays a SmartVu message and halts. NOTE: Because of the redundancy between this code and early parts of POST, (RESET, SHUTDOWN) the BIOS is NOT dependent on this code. This means that a working BIOS can be built for a 64 k non-FLASH ROM architecture by simply removing this module from the build process. The original 64 k BIOS reset/initialization logic has been left untouched.
Procedure BOOTUVPROM
Procedure bootUVPROM contains the 8 k power-on logic executed when a UVPROM is installed. When cold-booting from a UVPROM, this code is executed to look for a FLASH ROM and if present, attempts to copy the entire UVPROM contents into the FLASH. If FLASH programming is unsuccessful for any number of reasons, e.g., FLASH is not present, 8 k boot sector is not jumpered for erase/programming, etc., then this routine just bails back into POST.
Procedure BOOTFLASH
Procedure bootFlash contains the 8K power-on logic executed when the system boots from the Hash.
The steps are as follow:
Checksum the system anal video BIOS (96K).
If checksum is good, go to I.
If checksum is bad, go to II.
I. Does the user request a BIOS upgrade?
If upgrade is selected, go to II.
If upgrade is not selected, we are done with this routine and shall go to do regular POST.
II. Reset the upgrade option to prevent an endless loop.
Initialize a lot of stuff.
Initialize 106 so that we have SmartVu.
Slob initialization.
Trane initialization and get at least 2 meg of DRAM.
Initialize the refresh counter.
Set refresh page to 0 through the DMA controller.
Copy code to RAM and execute from RAM.
Set up stack.
Test and initialize timers.
Calibrate count before first call to DLY100.
Set up the DMA controllers.
Set up the interrupt descriptor table.
Set up the interrupt controllers.
Initialize the floppy subsystem.
Copy the BIOS from floppy.
If unsuccessful, display error message and halt.
Make sure we indeed have a valid BIOS in RAM.
Set the Flash up for programming.
Copy the BIOS from RAM to Flash.
If unsuccessful, display error message and halt.
Inputs:
Native-mode, protected-mode, Trane has grabbed its I/O address.
Outputs:
Either halts or go to do regular POST.
If the BIOS checksum is good and no upgrade is requested, this routine will jump to do regular POST.
If the BIOS checksum is bad or an upgrad is requested, then this routine will halt at the end.
Other Procedures
The actual assembly language realization used contains numerous other procedures, which will now be detailed. Most of these procedures are completely conventional. However, any of these procedures which have any particular relevance to the claimed inventions are actually listed below.
INITSTUFF
This procedure initializes the "SLOB" and "TRANE" chips.
RAM.sub.-- init
Procedure RAM.sub.-- initperforms RAM initialization for the 8 k boot code, initializes TRANE, identifies memory, initializes it, turns on refresh and points RCR0 to a block from 0-2 Mb.
Inputs:
HSM.sub.-- xx . . . Host state machine initialization tables.
FS . . . Must point to a 4 Gb data selector based @0. bp contains the return address. NOTE: It doesn't matter what type of SIMM we have identified. The presence test assumes that it's a 256 kbit chip and initializes the RCR accordingly. If it's actually a 1 Mbit or a 4 Mbit SIMM it should still behave ok if it's programmed as a 256 kbit. Since all we need is 128 kb of RAM for copying the UVPROM, who cares what size the SIMM really is?
Outputs:
SUCCESS . . . If successful, returns the following
All SIMMs identified as present or not and tabulated in CMSMTPL below. CMSMTPL . . . CMOS byte containing the SIMM map.
RCROM points to 2 Mb of DRAM. SIMMs for this RAS programmed as 256 kbit devices.
Refresh enabled.
FAILURE . . . If no RAM can be found, issues `fRAM` to SmartVu and waits forever.
$INIT.sub.-- VL106
Procedure $init.sub.-- vl106 initializes the vl106. The VL106 is a chip which, among other functions, replaces the 8742 keyboard handler.)
PMODE
This procedure is a Protected mode entry routine.
Inputs:
DS . . . points to the segment containing "gdt.sub.-- ptr" and NULL.sub.-- IDT.
RMODE
This routine is created for the floppy code to run in real mode.
Inputs: none. Outputs: real mode.
PMODE2
This routine is created for the floppy code. This will switch the execution back to protected after the floppy code is executed. [Uses different IDT and GDT tables ub low RAM.]
Inputs: none. Outputs: protected mode.
CMOSSUM
This routine checksums the CMOS range 10 h thru 2 Dh.
Inputs:
DI . . . contains NEAR return offset.
Outputs: none.
FLASH.sub.-- MSG
Procedure flash.sub.-- msg displays the SMARTVU message in EAX, waits CX seconds and returns to the NEAR address in DI.
Inputs:
SS . . . Writable segment pointing to code.
DI . . . Contains the NEAR return address.
CX . . . The number of seconds to delay after displaying the SMARTVU message.
EAX . . . Contains the SmartVu message.
DELAY
Procedure delay waits for CX seconds. This procedure uses the RTC to implement a delay specified by the count in seconds in CX. If the RTC battery is bad, we just execute a fixed delay loop of DUMMY.sub.-- DELAY iterations.
Inputs:
cx . . . Countains the number of seconds to delay.
GET.sub.-- SEC
This routine reads the seconds count from the RTC (real-time clock), and waits if a time update is in progress.
Inputs:
SI . . . Contains the NEAR return address.
Outputs:
AL . . . Contains the RTC seconds count.
BEEP
This routine beeps!
Inputs:
cx=frequency.
di=near return address.
Outputs:
none.
INIT.sub.-- TIMER
This routine tests and initializes the timers.
SETUPDMA
Procedure setupDMA will set up the DMA controllers just like the regular POST. The code are identical so as not to introduce any problem with the INT13H routine.
Inputs:
Native mode, protected mode.
Outputs:
If error, let regular POST reports the error.
If no error, DMA controllers are now ready for the INT13H routine.
SETUPPIC
Procedure setupPIC will set up the PICs just like the regular POST. The code is identical to that used in the POST, so as not to introduce any problem with the INT13H routine.
Inputs:
Native mode, protected mode.
Outputs:
The PICs are now ready for the INT13H routine.
COPY
Procedure copy copies the new BIOS from the floppy to RAM. The steps are as follow:
Find out if file exists. The file has to be on a 1.2M floppy in the root directory with a filename of "DELLBIOS.BIN".
Inputs:
none.
Outputs:
Carry flag set if error.
FILEEXIST
Procedure fileExist reads in the root directory, and searches to see whether file "DELLBIOS.BIN" is in the directory.
Inputs:
None.
Outputs:
Carry clear if file exists.
ax has the first disk cluster if file exists.
bx:dx has file size.
SETTYPE
Procedure setTypedetermines the type of floppy drive and sets the drive parameters accordingly.
Inputs:
none.
Outputs:
Parameters are initialized.
COPYFILE
Procedure copyFile copies the desired file from drive 0 to RAM.
The steps are as follow:
Make sure the file size is what we expected.
Read the FAT table.
Copy the file to DRAM.
Inputs:
ax has the first disk cluster of the file.
bx:dx has file size.
es=GDTD.sub.-- BUFFER.
Outputs:
Carry flag set if error.
SANITYCHECK
Procedure sanityCheck will try to verify that the BIOS copied is a legitimate Dell BIOS.
Inputs:
BIOS (112K) spans from 5000:0000 thru 6000:C000.
Outputs:
Carry set if we do not have a legi BIOS.
NEXT
Procedure next gets the next link from a 12-bit FAT.
Input
ax=current entry number.
Output:
ax=next element in the chain.
REL2ABS
Procedure rel2abs converts the relative sector number to the absolute sector location.
Input
ax=relative sector number.
Output:
ch=track number.
cl =sector number.
dh=head number.
dl=0=drive number.
GETFAT
Procedure getFat reads in the FAT table.
Inputs:
es=GDTD.sub.-- BUFFER.
Outputs:
Carry set if error.
If successful, FAT resides from 4000:0000 thru 4000:0E00.
PROGRAM
Procedure program writes data into the flash ROM.
Inputs:
none.
Outputs:
Carry set if error.
ERASE.sub.-- SECTOR
Procedure erase.sub.-- sectorperforms the following steps:
Clear the status register.
Set the FLASH to read status mode.
Set the FLASH for erase.
Do the erase.
Verify the erase is successful by reading the status register.
Set the carry flag if erase unsuccessful.
Clear the status register.
Set the FLASH to read array mode.
Inputs:
es:edi=address of sector to be erased.
FLASH program/erase better be enabled before calling this routine.
Outputs:
If successful, carry flag cleared.
PROGRAM.sub.-- SECTOR
Procedure program.sub.-- sectorwrites a sector of the flash ROM.
Inputs:
ecx=number of bytes to program.
ds:esi points to beginning of data.
es:edi points to beginning of sector.
The steps are as follow:
Clear the status register.
Set the FLASH to read status mode.
Go and program the sector.
Abort if error with the carry flag set.
Clear the status register.
Set the FLASH to read array mode.
Outputs:
If successful, carry flag cleared.
ENABLE.sub.-- fPROGRAM
Procedure enable.sub.-- fProgram enables program/erase operations on the FLASH ROM.
Inputs: none. Outputs:
FLASH is ready to be programmed or erased.
DISABLE.sub.-- fPROGRAM
Procedure disable.sub.-- fProgram disables FLASH program/erase.
Inputs: none. Outputs: FLASH cannot be programmed or erased.
USECWAIT
Procedure usecWait creates cx microseconds of delay.
Inputs:
cx=number of usec.
Outputs:
cx destroyed.
CLR.sub.-- FLASH.sub.-- STATUS
Procedure clr.sub.-- Flash.sub.-- status clears the FLASH status register. This routine was created because the new spec requires the address to be at 00000h of the Flash.
Inputs: none. Outputs: Flash status register cleared.
READ.sub.-- FLASH.sub.-- STATUS
Procedure read.sub.-- Flash.sub.-- statusreads the FLASH status register at 00000 h of the Flash.
Inputs: none. Outputs: Flash status register read in al.
NMI.sub.-- ISR
Procedure NMI.sub.-- ISR is a dummy interrupt handler, which swallows any NMMI interrupt.
DMATST
This routine tests the DMA address and word count registers.
Input
CX=#ff of ports to test
DX=first port to test
SI=increment between ports
Output: flags ?
AL=?
AH=?
CX=0
CALB.sub.-- SPD
Procedure CALB.sub.-- SPD ("calibrate speed") sets DLLY.sub.-- CNT with count for 100 us
FDISR
Procedure fdISR is the floppy drive controller interrupt handler. The steps are as follow:
Send EOI to master PIC.
Set interrupt bit in DRVSTAT to indicate an interrupt has occurred.
INT13H
Procedure int13h determines and executes the desired int13h procedure functions.
RESETFLOP
Procedure resetFlopresets the floppy system. This is part of the INT13H procedure.
Inputs:
ah=0
dl=drive number (0-3), bit7=0 for floppy.
Outputs:
Carry flag set if error.
ah=status/error code.
ds:ERRSTAT(FDATA)=ah.
READSECTOR
Procedure readSectorreads al sectors from the floppy.
Inputs:
ah=02h.
al=number of sectors.
ch=track number.
cl=sector number.
dh=head number.
dl=drive number.
es:bx=address of buffer.
Outputs:
Carry flag set if error.
al=number of sectors transferred, ah=status=ERRSTAT.
SETCARRY
Procedure setCarrysets the carry flag on the stack. This is part of the INT13H procedure.
CLRCARRY
Procedure clrCarry clears the carry flag on the stack. This is part of the INT13H procedure.
Inputs: none.
Outputs:
Carry flag in the flag register on the stack is cleared.
Interrupt flag also set.
DSKRESET
Procedure dskResetresets the floppy system. This is part of the INT13H procedure.
The steps are as follow:
Disable interrupts.
Clear DRVSTAT and ERRSTAT.
Get floppy motor status.
Issue reset command.
Turn off reset command.
Enable interrupts and wait for reset result interrupt.
Check result.
Inputs:
dl=drive number (0-3), bit7=0 for floppy.
Outputs:
Carry flag set if error.
ds:ERRSTAT(FDATA)=ah=error code if error, else 0.
WAITINT
Procedure waitInt waits for a FDC interrupt. This is part of the INT13H procedure. This routine uses a regular wait instead of int15h.
Inputs: none.
Outputs:
Carry flag set if timeout error, else not.
ERRSTAT bit7=1 if timeout.
FDCRDY
Procedure fdcRdy waits for the floppy controller to be ready. This is part of the INT13H procedure.
Inputs: none.
Outputs:
Carry flag set if timeout.
SISSTAT
Procedure sisStat senses the interrupt status. This is part of the INT13H procedure.
Inputs: none.
Outputs:
Carry flag set if error, ERRSTAT set, ah=FDC status, else Carry flag clear and al=present cylinder.
WRTCMD/WRTCMDA
Procedures wrtCmd and wrtCmdA write commands to the FDC, and are part of the INT13H procedure.
Inputs:
al=command.
ds points to FDATA.
Outputs:
Carry flag and ERRSTAT set if error.
ah, dx destroyed.
RSLTRD7/RSLTRD
Procedure rsltRd7reads 7 result bytes from the FDC. Procedure rsltRd reads cx result bytes from the FDC. These are part of the INT13H procedure.
Inputs:
cx=number of result bytes to read.
ds points to FDATA.
Outputs:
al=FDC status.
ah=FDC ST0, only valid if no carry.
ERRSTAT bit 7=1, carry set on timeout.
ERRSTAT bit 5=1, carry set if FDC results cannot be flushed.
Carry set if less results available than expected.
DSKST=FDC ST0, only valid if no carry.
dx non-zero.
I82077EXIST
Procedure i82077exist determines if a i82077 FDC is installed.
This is part of the INT13H procedure.
Inputs: none
Outputs:
Carry flag set if i82077 installed and the FIFO enabled.
stk.sub.-- tmp bit0=1 if i82077 installed and bit1=1 if FIFO enabled.
MLTCMD2/MLTCMD
Procedure mltCmd2 writes 2 commands to the FDC. Procedure mltCmd writes cx commands to the FDC. These are part of the INT13H procedure.
Inputs:
cx=number of bytes to write (mltCmd).
bx=starting table offset.
Outputs:
es:si=disk parameter block.
ERRSTAT bit7=1 on timeout, bit5=1 on controller error.
Carry flag set on timeout or controller error, else not set.
DPBADR
Procedure dpbAdr puts the address of the disk parameter block into es:di. This is part of the INT13H procedure.
Inputs: none.
Outputs: es:si points to disk parameter block.
CONFIGURE
Procedure configure issues the configure command to enable the FIFO in the i82077 FDC. This is part of the INT13H procedure.
Inputs:
ah=byte 2 of configure command.
Outputs:
Carry flag set if FDC error, else cleared.
DSKMOT
Procedure dskMot turns the floppy drive motor on. This is part of the INT13H procedure.
Inputs:
dl=drive number.
Outputs: none.
CHKCHG Procedure chkChg checks the change line on the floppy drive. This is part of the INT13H procedure.
Inputs: none.
Outputs: Carry flag set if error.
GETDRVINFO
Procedure getDrvInfogets drive information from HDCFLG. This is part of the INT13H procedure.
Inputs: none.
Outputs:
Low 3 bits of al=drive information.
CLRCHG
Procedure clrChg tests and clear the change line. This is part of the INT13H procedure.
Inputs:
dl=drive number.
Outputs:
Carry flag set if timeout.
ERRSTAT=timeout or change line error.
DOSEEK/DOSEEK2
Procedures doSeek and doSeek2 will seek to track specified by ch. This is part of the INT13H procedure.
Inputs:
ch=track to seek.
Outputs:
Carry flag and ERRSTAT set if error.
DRIHEDSEL
Procedure driHedSel loads the head (HDS) and the drive (DS1, DS0) parameters to the FDC. This is part of the INT13H procedure.
Inputs: none.
Outputs: none.
INITDMA
Procedure initDMA will initialize the DMA controller for floppy operations. This is part of the INT13H procedure.
Inputs: none.
Outputs:
Carry flag and ERRSTAT set if error.
HEADSETTLE
Procedure headSettle conditionally waits for the head to settle. This is part of the INT13H procedure.
Inputs:
dx non-zero if head settle wait needed.
Outputs:
es:si points to DPB.
SPINUP
Procedure spinup conditionally waits for the floppy to spinup. This is part of the INT13H procedure.
CHKTYP
Procedure chkTyp checks the media type. This routine is called prior to read/write operation to verify and update, if necessary, FDMED. Returns with the carry flag set if no remaining valid FDMEDs to try. This is part of the INT13H procedure.
Inputs:
FDMED, ERRSTAT (if non-zero, it is a retry).
bx=drive.
Outputs:
FDMED updated:
ERRSTAT=carry flag=0 if any more valid rate/step combos.
FDMED restored:
FDOPER=0.
ERRSTAT and carry flag set if no more valid rate/step combos.;
FDTYPIF
Procedure fdTyplf gets floppy disk type from CMOS if CMOS is valid. This is part of the INT13H procedure.
Inputs: none
Outputs:
Carry flag set if CMOS invalid, else
al=drive type, zero flag set if no drive or type unknown,
top of ah non-zero if another drive.
FDTYPE
Procedure fdType gets floppy disk type from CMOS. This is part of the FDTYPIF routine, and part of the INT13H procedure.
Inputs: none.
Outputs:
al=drive type, zero flag set if no drive or type unknown,
top of ah non-zero if another drive.
SETRATE
Procedure setRate sets the transfer rate. This is part of the INT13H procedure.
Inputs:
bx=drive number.
Outputs:
Rate control port updated.
CHKRES
Procedure chkRes reads and checks the results from the FDC. This is part of the INT13H procedure.
Inputs: none.
Outputs:
Carry flag set if error.
al=number of sectors transferred.
AFTERTRY
Procedure afterTrysets the zero flag if there is still more retry after an unsuccessful operation. This is part of the INT13H procedure.
Inputs: none.
Outputs:
Zero flag set if there is still more retry.
RESTYP
Procedure resTyp forces FDMED[bx] media known and restores compatible low bits. This is part of the INT13H procedure.
Inputs:
ah=top three bits of FDMED[bx].
Outputs:
FDMED[bx] bit4=1.
STODRVINFO
Procedure stoDrvInfo puts the drive information in al into HDCFLG. This is part of the INT13H procedure.
Inputs:
al=drive information.
bx=drive number.
Outputs:
Drive information in [bx] half of HDCFLG.
INITFD
Procedure initFD will initialize the floppy subsystem and all the variables used by the INT13H routine. This is part of the INT13H routine.
Inputs: none.
Outputs:
DRVSTAT, FMOTS, FDTIMO, ERRSTAT, DSKST, HDCFLG, FDMED, FDOPER, WTACTF, FDRATE, are all initialized.
DSKTEST
Procedure dskTest will test and determine drive type.
Inputs:
ss:bp points to stack frame for disk driver.
stk.sub.-- dl[bp]=drive number.
Outputs:
FDMED[drive#], HDCFLG initialized.
Carry flag set if no drive.
SDRVST
Procedure sDrvStsenses the drive status.
Inputs: none.
Outputs:
Carry flag set, ERRSTAT set if error.
Carry flag clear, ah=FDC's ST0 if no error.
Zero flag set if track 0 reached.
CKFDCFG
Procedure ckFdCfg checks the floppy configuration data against the actual installed hardware.
Inputs:
byte ptr stk.sub.-- dl[bp]=drive number.
si points to the media type for that drive number.
Outputs:
Carry set if different configuration.
CMREAD
Procedure CMREADis a Subroutine to Read a CMOS register.
Input
AL=CMOS register address to read
Output:
AL=CMOS register value
Interrupt Flag cleared
Carry Flag cleared
Zero Flag cleared.
CMWRT
This is a Subroutine to Write a CMOS register.
Input
AH=CMOS register address to write, Bit 7=0 enables NMI
AL=Value to write to CMOS register
Output:
Interrupt Flag cleared
Carry Flag cleared
Zero Flag cleared.
DLY100
This procedure will delay for 100 us. The loop count for the delay function is taken from DLY.sub.-- CNT. DLY.sub.-- CNT is set during ATPOST, and whenever turbo speed is changed by the routine CALB.sub.-- SPD. It takes no inputs and gives no outputs.
Further Modifications and Variations
It will be recognized by those skilled in the art that the innovative concepts disclosed in the present application can be applied in a wide variety of contexts. Moreover, the preferred implementation can be modified in a tremendous variety of ways. Accordingly, it should be understood that the modifications and variations suggested below and above are merely illustrative. These examples may help to show some of the scope of the inventive concepts, but these examples do not nearly exhaust the full scope of variations in the disclosed novel concepts.
In particular, the disclosed innovations are not by any means limited to DOS systems, nor to systems using 80.times.86 microprocessors, nor to systems using a single microprocessor as the CPU. The disclosed innovations provide generally applicable architectural techniques, which can be applied to a wide variety of computer systems, including high-performance systems and multiprocessing systems.
It should also be noted that the source for the user-supplied BIOS code does not have to be a floppy drive. For example, a tape cartridge or a dial-in telephone interface could be used instead.
In addition, although the Flash EPROM is the preferred device technology for the second nonvolatile boot memory, it should be recognized that other nonvolatile storage technologies can be used if they become practical. Thus, many of the innovative features of the presently preferred embodiment can be directly adapted to a system using EEPROMs, battery-backed SRAM chips, ferroelectric RAMs, or future technologies.
It should also be noted that BOTH of the boot memories can be rewriteable nonvolatile memories if desired. Even though this arrangement provides slightly less robustness than the preferred embodiment, it still provides a large added margin of safety over the use of a single rewriteable nonvolatile boot memory.
As will be recognized by those skilled in the art, the innovative concepts described in the present application can be modified and varied over a tremendous range of applications, and accordingly the scope of patented subject matter is not limited by any of the specific exemplary teachings given.
APPENDICES
Additional details of the presently preferred embodiment are set forth below, to ensure the fullest possible compliance with the best mode requirements of US patent law.
Appendix A is a Product Specification for the computer system of the presently preferred embodiment. This computer system is internally referred to as "Hammerhead." short summary of the most recent changes to
Appendix B is a specification for the "SLOB" chip referred to in the foregoing specification. This chip performs a variety of control functions.
Appendix C is a specification for the "TRANE" memory controller chip referred to in the foregoing specification.
Although these detailed specifications are the most current available as of the filing date of the present application, it must be noted that these are actual engineering documents prepared for internal use, and may contain errors or omissions. Thus, users are warned that the these examples are not definitive and not necessarily complete, and should be used only with caution. ##SPC1##
Claims
What is claimed is:
1. A method booting of a computer system including a CPU, a first rewritable nonvolatile boot memory storing basic system software, a second nonvolatile boot memory, and a third memory, having a predetermined data location storing a value which indicates whether a user has requested an upgrade of said basic system software, the method comprising the computer implemented steps of: the CPU performing a hard reset; the CPU executing software from one of either the first rewritable nonvolatile boot memory or the second nonvolatile boot memory, depending on a predetermined indicator, wherein (1.A) If said CPU is executing software from the first rewritable nonvolatile boot memory, then: (1.A.i) performing a checksum operation on the basic system software in said first rewritable nonvolatile boot memory, and, if a checksum error is found, then: (1.A.i.a) prompting the user to provide a data source for the basic system software, and thereafter (1.A.i.b) reprogramming said first rewritable nonvolatile boot memory from the data source provided by the user; (1.A.ii) Reading said value from said predetermined data location in said third memory to determine if a user has requested an upgrade of said basic system software, and, if said value indicates that the user has requested an upgrade of said basic system software, then: (1.A.ii.a) prompting the user to provide a data source for the basic system software, and thereafter (1.A.ii.b) reprogramming said first rewritable nonvolatile boot memory from the data source provided by We user; (1.B) If said CPU is executing software from the second nonvolatile boot memory which is not the same as said first rewritable nonvolatile boot memory and if said first rewritable nonvolatile memory is not currently write-protected, then writing the contents of said second nonvolatile boot memory into said first rewritable nonvolatile memory.
2. The method of claim 1, wherein said first rewritable nonvolatile boot memory consists essentially of a Flash EPROM.
3. The method of claim 1, wherein said first rewritable nonvolatile boot memory consists essentially of an electrically erasable programmable-read-only-memory.
4. The method of claim 1, wherein said second nonvolatile boot memory consists essentially of a programmable-read-only-memory which is not electrically erasable.
5. The method of claim 1, wherein first rewritable nonvolatile boot memory is write-protected in sectors, and said core software is write-protected independently of other portions of the contents of said rewritable nonvolatile boot memory.
6. The method of claim 1, wherein said data source is a floppy disk drive.
7. The method of claim 1, wherein said step (1.B) copies the entire contents of said second boot memory to overwrite the contents of said first rewritable nonvolatile boot memory.
8. The method of claim 1, wherein said CPU is a microprocessor.
9. The method of claim 1, wherein said step of reading said value from said predetermined data location in said third memory comprises reading said value from said predetermined, data location in a batter-backed static memory.
10. The method of claim 1, wherein said first rewritable nonvolatile boot memory includes a portion storing core software and wherein said core software performs said steps 1.A, 1.A.i, 1.A.ii, and 1.B, and wherein said first rewritable nonvolatile boot memory is write-protected in sectors, and said core software is write-protected, independently of other portions of the contents of said rewritable nonvolatile boot memory, by a hardware electrical connection which must be manually changed in order to overwrite said core software.
11. A method of booting a computer system including a CPU, a first rewritable nonvolatile boot memory storing basic system software, a second nonvolatile boot memory, and a third memory having a predetermined data location storing a value which indicates whether a user has requested an upgrade of said basic system software, the method comprising the computer implemented steps of: the computer system undergoing a power-up transition to change from a non-powered state to a powered state the computer system executing a software from one of either the first rewritable nonvolatile boot memory or the second nonvolatile boot memory depending on a predetermined indicator ( 11.A) ascertaining whether said CPU is currently executing boot software from the first rewritable nonvolatile boot memory or from the second nonvolatile boot memory which is not the same as said first rewritable boot memory but which is mappable onto the same address as said first rewritable boot memory; (11.B) If said CPU is executing software from said first rewritable boot memory, then: (11.B.i) performing a checksum operation on the basic system software in said first rewritable nonvolatile boot memory, and, if a checksum error is found, then: (11.B.i.a) prompting the user to provide a data source for the basic system software, and thereafter (11.B.i.b) reprogramming said first rewritable nonvolatile boot memory from the data source provided by the user; (11.B.ii) Reading said value from said predetermined data location in said third memory to determine if a user has requested an upgrade of said basic system software, and, if said value indicates that the user has requested an upgrade of said basic system software, then: (11.B.ii.a) prompting the user to provide a data source for the basic system software, and thereafter (11.B.ii.b) reprogramming said first rewritable boot memory from the data source provided by the user, and thereafter prompting the user to reboot the system; (11.B.iii) and otherwise executing a power-on-self-test routine from said first rewritable boot memory; (11.C) If said microprocessor is executing software from said second boot memory and if said first rewritable nonvolatile memory is not currently write-protected, then copying the entire contents of said second boot memory to overwrite the contents of said first rewritable nonvolatile boot memory, and thereafter prompting the user to reboot the system; and otherwise executing a power-on-self-test routine from said second boot memory.
12. The method of claim 11, wherein said first rewritable nonvolatile boot memory consists essentially of a Flash EPROM.
13. The method of claim 11, wherein said second nonvolatile boot memory consists essentially of a programmable-read-only-memory which is not electrically erasable.
14. The method of claim 11, wherein said first rewritable nonvolatile boot memory is write-protected in sectors, and said core software is write-protected independently of other portions of the contents of said rewritable nonvolatile boot memory.
15. The method of claim 11, further comprising the additional step, after said step (11.B.iii), of launching execution of a predetermined operating system software.
16. A method of booting up a computer system, the computer system comprising a first rewritable boot memory including first boot instructions and a second boot memory including second boot instructions, comprising the computer implemented steps of: determining whether boot operations are occurring from the first boot memory or the second boot memory; determining if the first boot instructions from the first boot memory are operational if the boot operations are occurring from the first boot memory; updating the first boot instructions in the first boot memory if the first boot instructions are not operational; and performing a power on self test if the first boot instructions are operational.
17. The method of claim 16, further comprising: determining if an upgrade is requested if the first boot instructions from the first boot memory are operational, said step of determining if an upgrade is requested occurring prior to said step of performing a power on self test, wherein said step of updating is performed if an upgrade is requested.
18. The method of claim 17, wherein said step of updating comprises: prompting a user for an external data source comprising an upgrade of the first boot instructions; transferring the upgrade of the first boot memory instructions from the external data source to the first boot memory.
19. The method of claim 18, further comprising: determining if said step of updating was successful after said step of updating; displaying an error code if said step of updating was not successful; and rebooting the system if said step of updating was successful.
20. The method of claim 19, further comprising: resetting an upgrade bit in a nonvolatile memory if said step of updating was successful.
21. The method of claim 16, further comprising: determining if the first boot memory exists if boot operations are occurring from the second boot memory; determining if programming of the first boot memory is requested; transferring said second boot instructions from the second boot memory to the first boot memory if programming of the first boot memory is requested; and performing a power on self test if programming of the first boot memory is not requested.
22. The method of claim 21, further comprising: determining if said step of updating was successful after said step of updating; displaying an error code if said step of updating was not successful; and rebooting the system if said step of updating was successful.
Patent Citations (3)
Non-Patent Literature (2)
- Glass, Brent; "The IBM PC BIOS"; Byte, 1989, Apr.; pp. 303-310.
- Bingham, Douglas; "Achieving Flexable Firmware"; 1978 MIDCON Technical Papers; 20/3/pp.1-4, 1978.