Predator99 wrote on 2020-01-20, 11:34:

What is the size of your eprom? Also 32Kb?
Maybe the NIC performs a check if the ROM is valid?
Can you test NIC with ROM in another (standard) Mainboard?

Yes the EEPROM was also 32 Kbyte. Is it OK to use EEPROM's larger than the firmware size?

If I test the NIC in another computer, what should I be looking for on the boot screen - the AIC 7880 BIOS looking for SCSI drives? Should I try it with the 64 Kbyte Adaptec-supplied BIOS as well?

feipoa wrote on 2020-01-21, 01:19:

Yes the EEPROM was also 32 Kbyte. Is it OK to use EEPROM's larger than the firmware size?

If I test the NIC in another computer, what should I be looking for on the boot screen - the AIC 7880 BIOS looking for SCSI drives? Should I try it with the 64 Kbyte Adaptec-supplied BIOS as well?

If the EEPROM is bigger then you need to fill it full, some allow all 00 or all FF to fill blank space but standard practice is to duplicate the code thru the PROM if using double, quad size, etc. So if using a 64k ROM with a 32k file you have two serialized images, one starting at 00000 and the other at 08000.
You need to see the NIC as an item and assigned an IRQ under the PCI/PNP section at end of POST in order to tell if your NIC can use a odd boot rom (as you noticed in post above)
You should see the Adaptec BIOS right at end of POST (most add-on ROM come after main BIOS, just like the regular 2940UW or a NIC with BOOT rom).
You could try the 64k file to see if it works.
I have 4 or 5 good PCI NIC's that allow me to use an odd bootrom BIOS and are still seen w/o using any software, they show up as an item and assigned an IRQ under the PCI/PNP section at end of POST, have some that do not with an odd BIOS and are useless for this type of test. The ones that work for me are: 3com 3c905-tx rev A (28pin), D-Link DFE-500TX rev C1 (28pin socket, max 27x256), TrendNet TE-PCI/W (with RTL8029, 28 pin socket <not RTL8019a>), SMC TW 8432BT (DEC 21041BT chip, 32 pin socket)

Added: not all NICs are created equal, many do not have the additional support circuitry needed and rely on the driver to tell it what to do....

Maybe I'm mistaken, but doesn't EEPROM software fill the blanks with FF by default when you write the file to it?

Sounds like you have plenty of NIC's to run this test. When do you think you'll have time to run it?

I have an original 3com 3c905, a 3c905B, and 3c905C. The original 3c905 performed the best so I placed it in a K5-PR200 system, but the B and C are not being used. Have you tried these? For DIP-32, I'd have to use a 1 mbit EEPROM though, which means 128KB - 32KB of FF's.

Some software does auto burn all FF's to blank space but depending on things that is not enough and you need to dupe the rom in the prom, Predator can tell you more and why.
I cannot run any tests as no flashable roms left and am heading out of town so will be away from any hardware till Thurs nite.
All the NICs that I have tried and work are full of chips, the one chip "wonder" cards never work for me.
If any of those 3c905 look like those in top of pic then probably ok, if they look like those in bottom of pic then most likely not (not enough support hardware) but could be wrong as there are many variables (exact card make and model, motherboard, main BIOS, etc)...

added: I suggest you get Uniflash and test to see if the op-rom bios in a NIC can be seen (if it does not show at boot) then run Uniflash from DOS, no drivers added but Himem.sys ok. If you use an ISA vid card and have no other PCI or AGP cards with bios then only the nic will show when you run: uniflash -pcirom -save back.bin. If nothing shows to select other than 0 in the list then op-rom will not load as an op-rom with out other config or driver setup if at all. If it shows a pci device you still may have run a config util to set it to boot. If it sees it save the pci rom to disk and then use a Hexeditor to check it to make sure it is the same file you flashed to the op-rom.

This evening I tried a 3C905-TX (same card as in your photo), but I did not get any type of Adaptec boot ROM screen; it booted as if there was no network card. I also tried using a 64 Kbyte EEPROM with the Adaptec v1.34 BIOS, but same thing - nothing. Also tried using "AIC-7880 v1.24 ROM from orig. NEC BIOS (48000-50000H).bin", but same thing - nothing. Also tried using a 3c905B-TX without result. The 3Com configuration programs see the cards just fine.

Lastly, I soldered on an IDE header to the motherboard, but I ran out of time to test it.

This is a reply to a previous PM. W27E257 is an EPROM-replacement class chip, so it's pin-compatible with 27256 type. No problems there.
So this is a dual-CPU mobo? With built-in VGA? The way it handles BIOS extensions might be non-standard then. Are you able to boot and run NSSI? Working floppy drive is enough. You can see if the option ROM is even detected in Details->BIOS menu. NSSI uses it's own code to search for compatible ROM code so if it does see the network card payload but the BIOS doesn't boot it then it very well might be some sort of BIOS issue/limitation.

HELP! I'm currently at a stand-still. Since I wasn't having any luck with the NIC ROM's, I went ahead and flashed SSTV2's BIOS.

The system reports that the Asus AIC-7880's firmware is v1.32. Unfortunately, the system hangs here:

I've waited several minutes, but it doesn't find a hard drive. I've tried disconnecting the hard drive, but it still hangs here. The problem now is that I cannot boot from a floppy disk to get the original v14 BIOS back on because the system must pass up the AIC-7880 firmware before it will boot from a floppy diskette. And the only way to disable the AIC-7880 is by using the EISA Configuration Utility, which must be access via floppy disk.

I didn't see any mention of a BIOS Recovery Mode jumper sheet. Is there not one? Attached is a list of jumper options. Also, nothing to disable the SCSI controller.

Only idea I have currently to fix the issue would be if I had another motherboard. I could set it up such that the AIC-7880 was disabled, then move the RTC module into the injured motherboard. Does it work this way? Unfortunately I only have one of these boards. Another option is to desolder the EEPROM and program it in an external programmer. This would be a pain. Would this even work though? Didn't SSTV2 mention that part of the firmware was compressed? Perhaps I can set some pin on the AIC-7880 chip to ground to force it disabled? Or perhaps there is a way to trick the RTC's NVRAM into thinking the AIC-7880 is disabled? I'm at a loss here.

In better news, before I bricked the BIOS I ran some tests on the IDE port. It works with both hard drives and CD-ROM drives, which certainly increases options CD-ROM drives. Unfortunately, hard drive size is limited to 8455 MB. W2K had no problem using the full size of the HDD though.

Argh. Can you still enter the BIOS Setup? What options do you have there?

You need to get another device booted before the AIC-7880 is initialized (PCI-NIC, XT-IDE, PCI-SCSI,...), but unlikely this happens.

BIOS is still accessible, but I suspect nothing in here will help me. All screens are provided below.

I think there is the explanation why the ROM on the NIC didnt work? "Execute user ROM" is disabled...

Yes, perhaps. I didn't notice that option until now. Didn't think such an option would be under "Security". Anyway, doesn't help me now.

I tried putting in another AIC-7880P-based PCI card, hoping it would confuse the system, but it still hangs on the Asus SCSI BIOS.

It still may help you. If you manage to get an option ROM on an expansion card to be executed before the PC hangs you will be able to boot the PC...

However, I dont know how the order the ROMs are executed is determined...

In my experience with NIC ROMs they usually boot after SCSI. The system is dead set on loading these SCSI ROM's in the beginning. If by chance the NIC ROM booted first, what ROM would I use to tell it to boot from the floppy drive?

You could e.g. set up a network boot server and boot via NIC...

Well, the chance that the NIC would boot first is nearly zero. Anyone know if swapping the RTC from another Proserva system with disabled SCSI will help? My best bet is probably to get another motherboard. Unless someone has any other ideas? I tried looking for an AIC-7880P white paper, but couldn't find one. Was hoping there was a means to disable the 7880 via grounding some pin.

It's still worth trying the network card payload route (once the BIOS option is enabled to allow it) - if it's executed first, and works, then there's a chance it will be mostly skipped in the mobo ROM. Playing around with cache and shadow options might also help, plus don't give up on the "hang" too early. Some SCSI timeouts are very long due to retries. Try with the cable terminated and not terminated.

If nothing helps but the network card paylod is run then it would be possible to put a custom code there to either re-write or at least damage the header of the 7880 BIOS module so it's not run during boot. Then boot and properly re-flash the entire BIOS with the backup.
If that's not possible, how about adding another VGA card? Does that work? If you can use external VGA with it's own BIOS then this can be the payload delivery method.

Do you have a link to some BIOS ROM's for either the VGA or NIC which may help in this case?

How long should I wait for the SCSI timeout? I figured 6 minutes was long enough. I can try it with just the cable terminated. I already tried it with no HDD.

Yes, external PCI graphic cards work. I'm using one now. I can switch to using the internal VGA if you know of some ROM I can install to a graphics card which will let the system boot to floppy.

Any chance the BIOS code has the recovery boot block option in it and some high or low set pin can trigger this mode?

Sorry, I'm not aware of any ready-to-use images that would help. I mean there's that XT extender that allows IDE booting but will that even work on a new mobo such as this?
I'd leave the mobo be for an hour or two during boot. It's not like you have to turn it off after a few minutes, I do not think it would damage anything. Maybe it goes past. If not, at least you know it's properly stuck.

I should be able to hack that PCI card ROM and add a piece of code to mess with the main BIOS image - that's assuming it has a DIP socket for the part and you have external programmer to prepare a substitute. It's risky though as I can't exactly test-run it. And even for that I need to know what chip, exactly, the mobo uses (I assume it's a Flash ROM) and how it's mapped to the PC memory space. That's because to write to it there is usually a specific unlock procedure you have to follow (in software) and to that I need all that info. All I know for now it's a 512KiB part.

Good idea with the VGA...

What one could do: Add XT-IDE ROM to the VGA-ROM. Its only 8 kb and there should be a card with enough free space.

Modify the VGA - INT 10 routine: Intercept a specific call that is done before this SCSI-hangup. At this time inititalize the XT-IDE ROM and let it boot...

Think its possible in theory (at least in a AT system, dont know how it works in a Pentium...) but would take much time to program...you have desoldered the chip faster than that 😉