BIOS got corrupted?

[Yousuf Khan]

Well, this is something I've never seen before. I got a Gigabyte
990FXA-UD5 motherboard, and yesterday, I left the machine running and
when I got back, I found it was sitting at an unusual screen. I read the
message on it, and it said that the BIOS got corrupted, and that it was
recovering the BIOS from the backup BIOS (this is one of those dual-BIOS
boards)! It brought the BIOS back into its original as-shipped version.
I had upgraded the BIOS at least a year ago, and it was running fine
until then. Not sure what could have corrupted the BIOS?

Anyways, it ran fine on the backup BIOS more or less, and I've reflashed
it back to the newer BIOS and it's running fine again on that version
too. Weird! No real question here, just thought it was something that we
don't see everyday.

Yousuf Khan

 

[Paul]

Well, this is something I've never seen before. I got a Gigabyte
990FXA-UD5 motherboard, and yesterday, I left the machine running and
when I got back, I found it was sitting at an unusual screen. I read the
message on it, and it said that the BIOS got corrupted, and that it was
recovering the BIOS from the backup BIOS (this is one of those dual-BIOS
boards)! It brought the BIOS back into its original as-shipped version.
I had upgraded the BIOS at least a year ago, and it was running fine
until then. Not sure what could have corrupted the BIOS?

Anyways, it ran fine on the backup BIOS more or less, and I've reflashed
it back to the newer BIOS and it's running fine again on that version
too. Weird! No real question here, just thought it was something that we
don't see everyday.

Yousuf Khan

It's possible for the flash to get "bit rot". It happens
to BIOS flash chips if they were flashed around ten years ago.
Sometimes a bit gets flipped in there. This would be more
of an issue during your next restart, when the corruption
could cause the BIOS to crash during POST.

My guess would be, the computer crashed, the crash left
some clock generator set to the wrong value. (Even though
the reset pulse should have caused everything to
initialize properly.) If the PCI clock generator
is set to the wrong frequency, it prevents the EEPROM
from reading out properly (timings not met). The BIOS
boot block computes the checksum of the main BIOS code
block, notes the signature is wrong, and things go downhill
from there.

If that happens to you under controlled conditions,
either keep resetting the machine until the BIOS crash
detection is triggered (three resets on an Asrock, in a row)
and the NVRAM is reset. Or just power off the machine completely,
o any transient settings can be cleared.

The puzzling part, is what crashed in the first place. I had
something like this happen recently, and the CMOS battery
was still in good condition. And somehow the BIOS settings
(NVRAM) were corrupted, preventing the thing from coming up
properly. The BIOS detected the corruption, returned
all settings to default, removing the Vnb boost I'm using
to keep the memory stable. So I had to go back into
the BIOS, and correct the custom settings, to return
everything back to "normal".

Paul

 

[Flasherly]

Weird! No real question here, just thought it was something that we
don't see everyday.

Never had a MB BIOS go out on me, but can't discount the wierdos.
Experiencing HDD lock ups recently, and it's starting to look like the
last PCI controller board caused them. Changed to another
controller/chipset and no more hiccups (dropped speeds/frozen copies -
everything else being the same). Then, again, steady thruput, I've
heard of modems similarly heating up and the chips failing over time -
though that's not at all about a BIOS. Believe I've provisions to
save my BIOS to some sort of HDD/sector backup/failsafe routine it has
(also a Gigabyte model). For whatever that's worth.

 

[Paul]

Maybe your computer was attacked by Mulism virus.
Christian were never had attack your system.

Al s-Barber 4, is that u ?

Wha chu due-en on AIOE?

Paul

 

[John Doe]

Well, this is something I've never seen before. I got a Gigabyte
990FXA-UD5 motherboard, and yesterday, I left the machine running and when
I got back, I found it was sitting at an unusual screen. I read the
message on it, and it said that the BIOS got corrupted, and that it was
recovering the BIOS from the backup BIOS (this is one of those dual-BIOS
boards)! It brought the BIOS back into its original as-shipped version. I
had upgraded the BIOS at least a year ago, and it was running fine until
then. Not sure what could have corrupted the BIOS?

Anyways, it ran fine on the backup BIOS more or less, and I've reflashed
it back to the newer BIOS and it's running fine again on that version too.
Weird! No real question here, just thought it was something that we don't
see everyday.

Yousuf Khan

Maybe your computer was attacked by Mulism virus.
Christian were never had attack your system.

 

[Yousuf Khan]

It's possible for the flash to get "bit rot". It happens
to BIOS flash chips if they were flashed around ten years ago.
Sometimes a bit gets flipped in there. This would be more
of an issue during your next restart, when the corruption
could cause the BIOS to crash during POST.

Yeah, but the motherboard is no more than a year old, and I would hope
they didn't use 10 year old flash to hold its bios.

My guess would be, the computer crashed, the crash left
some clock generator set to the wrong value. (Even though
the reset pulse should have caused everything to
initialize properly.) If the PCI clock generator
is set to the wrong frequency, it prevents the EEPROM
from reading out properly (timings not met). The BIOS
boot block computes the checksum of the main BIOS code
block, notes the signature is wrong, and things go downhill
from there.

This would make some sense.

If that happens to you under controlled conditions,
either keep resetting the machine until the BIOS crash
detection is triggered (three resets on an Asrock, in a row)
and the NVRAM is reset. Or just power off the machine completely,
o any transient settings can be cleared.

The puzzling part, is what crashed in the first place. I had
something like this happen recently, and the CMOS battery
was still in good condition. And somehow the BIOS settings
(NVRAM) were corrupted, preventing the thing from coming up
properly. The BIOS detected the corruption, returned
all settings to default, removing the Vnb boost I'm using
to keep the memory stable. So I had to go back into
the BIOS, and correct the custom settings, to return
everything back to "normal".

Paul

Yeah, I don't know what caused the crash to happen in the first place,
the only entry in the event log was a "The previous shutdown happened
unexpectedly". That means that it had no recording of what caused it to
go down in the first place.

And in fact, even restarting the computer took extra special effort this
time. Usually just touching the power button is enough. Sometimes
holding it down for a few seconds is needed to do the trick. This time
none of that was working. I had to actually reach behind the computer at
the main power switch on the PSU, and flick that this time.

Yousuf Khan

 

[Paul]

Yeah, but the motherboard is no more than a year old, and I would hope
they didn't use 10 year old flash to hold its bios.


This would make some sense.


Yeah, I don't know what caused the crash to happen in the first place,
the only entry in the event log was a "The previous shutdown happened
unexpectedly". That means that it had no recording of what caused it to
go down in the first place.

And in fact, even restarting the computer took extra special effort this
time. Usually just touching the power button is enough. Sometimes
holding it down for a few seconds is needed to do the trick. This time
none of that was working. I had to actually reach behind the computer at
the main power switch on the PSU, and flick that this time.

Yousuf Khan

Some power conversion circuits have "latchoff under fault".
It's to prevent the power converter from trying
over and over again, to restart, until it is ruined.

The other power converter design mode, is "putt putt" mode,
where the converter tries every once in a while, to restart.
The restart interval is very quick, if there is a dead short,
and the idea is, the circuit doesn't get too hot if things
are not going well. But "putt putt" mode was more
popular back in the PIII or earlier era, when stuff was
running at 35W or less.

To release latchoff circuits, you need to switch off at the
back. And in some cases, even wait a while for "things
to drain".

In one case, there was a timing issue with Antec power supplies,
and Asus VCore. The Antec was a little slow to get 12V up
to full voltage, the Asus VCore would stop ignoring overcurrent
after too short an interval, and the VCore would latch off
(leaving the CPU with no VCore power). Users who replaced their
brand new Antec power supply with something else, were
greeted by fully working computers. So the latch off idea
has a few holes in it, if the timing isn't right. And in
that case, all it takes is a slightly delayed power supply
output to cause a problem.

Paul