Re: maybe a bit OT: EEPROMs and PCs


Jure Sah wrote:

[bare metal]
| Ahh, okay. I supposed I will need to make sure I know where every opcode
| is in the binary (right offsets).

Exact.

[EPROM/FlashRam]
| Well the problem here is that I have these computers, that use EEPROM
| BIOS chips. Considering what you just told me about pins, I gather I
| can't just replace them with another chip type.

I don't know by heart yet, but I'm confident for there were a few offers
on pin-compatible FlashRam replacements for some of the 27hc.. ROM-family.

| > Good EPROM burners(>500U$) got the ability to read/detect the EPROM-type,
| > but cheap EPROM clones may/can not report a correct Vendor-ID.

| Well, what is the problem in this? I know what the chips are, there is a
| label on them and according to what you've told me there's just these I
| can use.

Be aware of not mixing up programming algorithm.
On request (by raising the one pin to 12V) an EPROM reports a Vendor-ID
(which implies the needed programming algo and capabilities,
ie: pgm-voltage, pulse duration, fast-verify).


| > No, after reset: at 0100h will not be what you expect.
| Ahh... hmm so if 0100h is rearanged, which of the jumps will work fine?

As Frank already said, all relative call/jump wont be affected, but your
compiler may assign code relative variables to a wrong address then.

For all ROM code I'd use
ORG: F0000
and have all data code-relative (DS=CS)

[NewBasicASM]

| NewBasic is an Assembler as is really, no high level components short of
| the EQU 'opcode' it's even far more primitive than FASM, which is why I
| liked it (produces tiny tiny files and is quite straightforward).

Seems I need to look at it ;)

| I'll use FASM now. I get most things, just what are the things in square
| brackets? Offsets?

FASM/NASM/MASM/TASM and also ZILOG,KESYS and more use square brackets
to indicate a memory operand.
The things inbetween these brackets are the memory address.

| > Yes it's easy to destroy an EPROM (fastest way is upside down inserting).
| > As above, check EPROM-type before you apply programming power to the
| > pins.

| Ahh... gosh. (I hate the tech world for these kinds of tricks, sometimes
| "you'll screw up for missplacing a byte" and "you'll screw up for
| sticking muffins in your CD drive" are considered to be quite simmilarly
| common in all comunities.)

:) yes.

[ IRQ, EXC and INT ]
| Ahh, yes sorry. I meant INTs. So if I call INT 10h as one of the first
| things of a BIOS I will be able to display things on the screen it will
| already point to VBIOS or are there catches?

Not by itself :)
The very first job of the BIOS is to detect and check existing RAM.
Then it need to setup and create one whole IDT (real-mode at 0:0),
it must have exception- and IRQ handlers (even just in a native form)
and then the BIOS can search/find the presence of a display-ROM and
put the here found INT10 entry-point into the IDT
(the VBIOS may do this on its own if initialised as bootable device).

[]
| Yes, I expect to be able to stick out of the excessively complicated
| stuff, just to be on the safe side.

Good, or even better ;)

| > Some of 'these bits' may be vital hardware initial sequences,
| > and even you may not use all hardware on your board, at least you should
| > disable the unused (I'd keep the BIOS-setup routine for all this).

| Well in this particular case, the BIOS-setup program is a text-based B&W
| user interface to putting 1s and 0s in the 24 CMOS bits (litterately)
| and an elaborate animated diagnostics program that can test keyboard,
| memory, cpu, fpu, LTP printer and graphics when instructed to. I've no
| idea if there even is any POST. It's an XT computer some five years
| older than myself.

I see, you work for a museum :)
Many years passed ...
IIRC there once were an IBM-XT hardware standard defined.

[Disassembling BIOS]
| Since I don't have enough schooly education of programming yet, I fail
| to understand why most people fear things such as pointers and stacks so
| greately. So a memory area is being used instead of pushes and pops,
| where is the problem?

Not a problem and no fear, just:
the BIOS cannot setup any stack until an existing and checked memory
range can be assigned for the stack.
So ROM-code which doesn't have any memory available cannot use
push/pop nor call/ret and need to be 'tricky' here, and this tricks
make most disassemblers big headache.

[RAM]
| The computer has RAM built-in onto the motherboard... not using it would
| be somewhat impractical. But, speaking of RAM one thing I need to ask:
| Without the presence of a technology such as ROM shadowing or caching,
| does the code actually execute from address space that actually means
| the EEPROM chip itself? How does all of this (shadowing, caching and
| neither) appear from the ASM programmer's perspective?

Regardless of enabled 'shadows', the BIOS ROMs got their physical
locations in memory and remain there.

main BIOS: f000:xxxx
BIOS extensions: e000:xxxx
VBIOS: c800:xxxx

ROM-shadowing just perform:
map a physical RAM-area to the same address as the ROM.
copy the ROM to the RAM (actually to itself)
then disable the ROM-cs (the Chip-Select pin)
finally make this memory range write protected (or not).
the reason for doing it is access speed only, ROM are awful slow.

Shadows and caching are just speed-up thingies and wont affect
the address of the parts.

Only paging can make the story somehow weird and confusing here.



[boot ROM]

| Yes that certiantly is a good idea, even tho I wanted to do some playing
| around with the system BIOS just to be 1337er than thou and cut the boot
| up time to the extreemes and take everything standard off of a PC to
| make it look like a purpose built device such as say a telephone or a
| radio.

May be good for learning purpose...
I can buy a top featured phone-set including a radio for less than U$10 :)

| But anyhow, with NIC boot-ROM... what is the starting address there?

I still owe a few old ISA-net cards with jumpers on it to select
the ROM-base address. Newer ones got software setup fot this.

| Every single NIC I have has a slot for a ROM chip but I never saw the
| documentation provided provide anything about how to have one of those
| burned.

This ROM-BOOT-options have to follow the same rules as all BOOT-devices
with the '55aa' sign and a header and some initialising code..
The VBIOS is a good example of a ROM-BOOT-device.

__
wolfgang



.

Relevant Pages

  • Re: [Clax86list] Why reset address in x86 processors is at end of total address space
    ... The 8080 did reset to zero, but, in those days, the BIOS was in real ... reasonable to set the starting address to a ROM address. ... There is nothing inherent about RAM being low and ROM being high, or vice versa, and an number of CPUs have picked that choice either arbitrarily or to suit specific memory management regimes. ... One could "jam" in an arbitrary instruction on the bus at this point to direct it elsewhere, or one could let it fetch the first instruction, usually a RST instruction, from memory location zero. ...
    (comp.lang.asm.x86)
  • Re: Console buried in Tribble ROM?
    ... Only thing with a claimed memory range was my ATI GUP. ... Heck, the system BIOS takes 128K, why not the SCSI adapter... ... ADPItem 1 ROM Address Range ... Only one SCSI Adapter will have the ROM ...
    (comp.sys.ibm.ps2.hardware)
  • Re: old misc.c
    ... The code indicates something special about the memory area 0xFA0000 ... 640K and 1M are forwarded to all kinds of devices and to the BIOS ROM. ...
    (comp.os.minix)
  • Re: Random Restarts - Please Help
    ... I upgraded my BIOS a long time ago ... Beginners Guides: Diagnosing Bad Memory ... How to diagnose bad RAM ...
    (microsoft.public.windowsxp.help_and_support)
  • Re: Please diagnose my problem!
    ... The problem is I cannot access anything (BIOS, ... > Try swapping memory SIMMs to get the machine to boot. ... > set of diagnostics for the machine. ... > tool to assist diagnostics in finding intermittent defects. ...
    (alt.comp.hardware.pc-homebuilt)