3 ms·
Be happy that it only crashed. It could be worse. We once got a couple of high end motherboards for AMD processors, back in the Win98 era, and I tried to insta
by tzs 6y ago
Be happy that it only crashed. It could be worse.
We once got a couple of high end motherboards for AMD processors, back in the Win98 era, and I tried to install Win95 on one of them. The box said that Win98 was required, but it couldn't hurt to try, right? Maybe there would not be drivers for some of the peripherals, but all I needed was the CD-ROM and hard disk to work.
Install is going fine until it is time for it to reboot. That failed. It didn't even get to the BIOS initialization screen. It appeared to just be dead.
Figuring we just got unlucky and got a defective motherboard or processor, I tried installing on the other one so I could get on with my work.
That one died too.
Eventually I found something about this on the motherboard maker's support site. The problem was with the device probing during install.
What I'm about to say is not made up. As unbelievable as this might sound to people who grew up with more modern PCs, at one time they really did work as I'm about to describe.
The early PC buses had no built-in way for the host to identify what cards were plugged into the expansion slots. Typically, an expansion card would have a set of jumpers or DIP switches that could chose between several sets of possible addresses for the card's registers to appear in I/O space.
The user was expected to keep track of the settings of all cards they installed, and adjust the jumpers appropriately to avoid conflicts, and record the settings in a config file that the drivers would read to find out where their card was.
Later buses, such as EISA and later PCI, provided ways for the host to find out what is there and how it is configured. But operating systems still needed to support the old bus, and they wanted to make this as user friendly as possible.
So systems like Win95 would have a device probe during install that would try to identify what is on your old bus. They would do this by very carefully probing the I/O address space.
For example, suppose you know that a particular network card if present has to be at one of 8 addresses, and you know that after power on or reset that certain bits will be set in its status register and certain bits will be clear. You can read those 8 possible addresses, looking for the right bit pattern. If you don't find it, that particular network card is not present. If you do find it, you can do more tests to confirm it.
Some of those other tests might involve writing to the device registers, and seeing if it responds the way that network card should.
This is obviously risky. What if it is not that network card, but rather a disk controller card that just happens to have a register that after reset has the same bit pattern you expect in the network card status register? The thing you write then to verify it is the network card might be the "FORMAT DISK" command to that disk controller.
And so you had to be very careful with these probes. They had to be done in a safe order. You'd need to probe for that disk controller before you probed for that network card.
Those new motherboards contained peripherals that Win95 did not know about, and so the Win95 probe procedure did not know how to avoid doing bad things to them.
One of those peripherals was the built in interface for flashing the BIOS EEPROM. The Win95 device probe ended up overwriting the BIOS.
- 555-5555 6y agoIt's unfortunate (though understandable) that you bricked both before discovering the root cause. I remember hot-swapping EEPROM chips to repair a bricked motherboard from a similar era. While attempting to upgrade the BIOS, something went wrong. Most likely there was a bad sector in the boot floppy I used. The result was an unbootable machine. Solution? Swap in a working EEPROM chip from a compatible motherboard. Boot to a floppy disk that has a BIOS imaging utility and image file. Hot-swap the bad EEPROM chip back in. Re-flash the BIOS. Or, if you had money, you could sometimes purchase a pre-flashed replacement EEPROM chip. I don't miss those days, but am fortunate to have experienced them. It forced us to learn more about how a computer really works.