Don started a thread yesterday with a title that may be too easily overlooked. The subject is an article about the suspected presence of viruses set to activate 01/01/0000. The article is from a reputable group (CSIS .. headed up by former Senator Sam Nunn). It's worth a look. This is the thread:

http://www.csis.org/html/y2kw9.html (don, don29681@aol.com, 1999-05-25)

-- Ron Rodgers (RonRodgers@Resilience2000.com), May 26, 1999

Answers

I'll repost my response here, then.

---

I was wondering if this CSIS article would ever turn up here. The line that intrigues me the most is this one:

"Three other malicious viruses will actually lock a processor in a divide by zero loop, which, if left running for a sufficient amount of time, will overheat the Central Processing Unit, causing it to melt down and effectively reducing the computer to scrap metal."

What on earth is a "divide by zero loop"? What CPU has ever been designed in such a way that the software (machine code) it "understands" is capable of overheating it, causing it to "melt down"?

Answers to these questions invited.

-- Richard Dymond (rjdymond@x-stream.co.uk), May 26, 1999.

"What on earth is a "divide by zero loop"? What CPU has ever been designed in such a way that the software (machine code) it "understands" is capable of overheating it, causing it to "melt down"?"

Doesn't sound right. Divide by zero can cause a software "interrupt", but I think that is about it.

This is not to say that the threat of viruses on New Year's Evil is not real. I just don't see microprocessors being "melted down". The only hardware damage I can expect in IBM-style PeeCees is a virus which over-writes the FLASH-BIOS on the motherboard. That can cause enough trouble that it is often cheaper to just replace the motherboard.

-- Anonymous99 (Anonymous99@Anonymous99.xxx), May 26, 1999.

A "divide by zero loop that'll melt the processor" is mostly B***S***. What truth there might be :

A processor's power consumption isn't constant, it varies depending on what instructions it's executing. The manufacturer specifies the maximum power drain, the PC manufacturer makes sure that the power supply and case and cooling can cope with this maximum.

It's just about possible that some cheapo PCs have cut corners, relying on the normal power drain being less than the maximum. Run a max-power instruction stream on one of these in a hot (30C+) room, and it might fail.

My experiences (and the physics) says that modern processors don't melt to destruction if the cooling fans fail. They just stop dead when they get too warm, work OK again when they cool down, and have a shortened life expectancy if they run overheated for any significant time (like days).

If a virus wants to kill a PC, it's best bet is to wipe the BIOS chip (which the Chernobyl virus does). How PCs came to be made with a permanently writeable BIOS is another example of extreme short- sightedness. A professionally designed computer should have a BIOS write-protect jumper or switch, so that software could alter the BIOS only if the user had explicitly allowed it to do so by using the jumper.

-- Nigel Arnot (nra@maxwell.ph.kcl.ac.uk), May 26, 1999.

I posted this reply on the other thread, but.... Re processors melting down - it is very easy nowadays because both the processor operating voltage (used to be 5 volts, now can be lower than 3 volts) and clock speed are under software control in a modern PC. The faster the clock, the more heat is generated. The higher the voltage the more heat is generated. So if you mess with these settings you CAN melt down the processor. What the effect of a CPU intensive looped instruction executed in high speed on-chip cache memory would be, I cannot say for definite, but such a loop would result in higher dissipation because of its operating without wait states, delays etc associated with referring to off-chip memory.

-- Mike (mike.booth@eu.anritsu.com), May 26, 1999.

There was a devide by zero bug in the Windows 95 OS.

If you where to run WIN 95 with some Cyrix(tm)processors(the media GX

comes to mind as the last one I have seen it with) and one or two

specific AMD(tm) processors you got ever so often a NDIS ( devide by

zero)error that caused the system to crash . This was fixed with a

patch and in WIN 98.

The only failure I could think that could happen would be a system

crash (failure to boot up but NO data lost)not a bios or MB failure

if the divide by zero was the only thing this virus is doing.

All bets are of if it is combined with a bios flash that writes wrong

data into the bios. This would be hard to do as it would be MB

specific. A pure erasing of the bios would not let the computer come

on at all and the MB would default to " initial save settings" that

are hard wired into The MB. A processor meltdown is not possible

without writing specific data to the bios and as I said before it would have to be MB specific.

A attempt to a processor meltdown would also

NOT work on any MB that is using jumpers to set voltages and other

parameters.

-- Rickjohn (rickjohn1@yahoo.com), May 26, 1999.

If "melting" the processor by software alone is possible, I marvel at the fact that no bug has yet done so accidentally. Perhaps the chances of doing accidental damage are negligible, though.

Now, can anyone provide an assembly language listing of a program that, when run, would "melt" a Pentium II, for example? I won't believe it until I see it.

(I promise not to use the information to go and write a virus!)

-- Richard Dymond (rjdymond@x-stream.co.uk), May 26, 1999.

How do you do italics on here?

-- Richard Dymond (rjdymond@x-stream.co.uk), May 26, 1999.

Divide by zero loop:

xor di, di
mov es, di
mov word ptr es:[di], offset new_0_handler
mov word ptr es:[di+2], cs
new_0_handler:
mov ax, di
mov cx, di
mov dx, di
div cx

... but this will NOT "melt" the processor. You'll probably have to switch the machine off and back on to reboot, but other than that, no real harm will be done.

There are SOME (not all, by any means) motherboards which permit setting the processor voltages and clock speeds by software control, and there are some that don't have write-protect on the flash BIOS. Personally, I wouldn't buy such a motherboard, myself (and would always check before I DID buy).

Caveat emptor applies here. And if you're worried about viruses, download F-Prot; it's free for non-commercial use, and it's very, very good. (You can believe that there have been plenty of other people downloading and/or buying it -- and other AV software -- after the Win95.CIH disaster[g].)

-- Stephen M. Poole, CET (smpoole7@bellsouth.net), May 26, 1999.

Italics on: [left arrow]i[right arrow]

Italics off" [left arrow]/i[right arrow]

Example: italics set

[View][Source] of any html document to see the coding.

-- Tom Carey (tomcarey@mindspring.com), May 26, 1999.

Divide by zero loop and overheat CPU is an old and venerable virus hoax, I've seen warnings about it a lot. This is the first time I have seen it connected to a Y2K virus warning.

-- walt (walt@lcs.k12.ne.us), May 26, 1999.

Stephen: Thanks for the listing. The only assembly language I'm fully familiar with is Z80 assembly language, but your Intel x86 job should become comprehensible to me after a quick look in the reference books...

Tom: Thanks for the hint on producing italics. I think my first stab at it was pretty successful, though, don't you? :)

-- Richard Dymond (rjdymond@x-stream.co.uk), May 26, 1999.

Anyway, back to hardware-damaging viruses. What surprises me is that I've seen the CSIS cited as a credible source of info. I don't think this toss about melting CPUs does much for their reputation.

-- Richard Dymond (rjdymond@x-stream.co.uk), May 26, 1999.

I have to side with the skeptics here. In the course of doing my job, I regularly deal with a fairly large number of computer viruses, trojan horses, and other such malicious program.

I have yet to see a virus that can 'melt' a processor. The worst ones attack the BIOS or corrupt your hard drive. They are a serious threat mostly to those people who refuse to implement a good backup procedure and who also do not want to be troubled with maintaining good virus protection software.

Where I work, it is interesting to note that the Internet is not the primary source of such 'infections'. Most viruses came from data exchanges between ourselves, our vendors and our parent/sibling companies. Most common at our specific location are MS Word and Excel 'macro' viruses. A notable exception to this was the recent HAPPY99 'trojan' which attaches itself to your email and mails copies of itself around. Our gateway detected and stopped a couple dozen copies of this malicious program every day for a few weeks. It seems to have slowed down some now.

I would be very interested in any documentation demonstrating physical damage to a PC caused by malicious programming. While I have significant experience with them, viruses are not my specialty.

The BIOS wiping viruses can cause very serious problems (your machine won't boot), but they do not cause physical damage to the hardware. (One could of course argue that if the BIOS chip has had its programming wiped, the chip has been rendered inoperable and thus it is effectively 'physical damage'.)

-- Arnie Rimmer (Arnie_Rimmer@usa.net), May 26, 1999.

While "exception processing" is very rare, when compared to normal instruction processing, I'm having a real hard time swallowing this melted processor theory. Divide by 0 results in an interrupt 0, which is treated just like a JMP instruction by the processor. I'm going to drop Intel support an e on this when I get home. I'll post any reply here. <:)=

-- Sysman (y2kboard@yahoo.com), May 26, 1999.

Oh I'm sure there will be thousands (millions ?) of "accidental" (number of some kind)/(another number of some kind = 0) accidents.

They will be mixed in as part of the generic process and hardware "locking" accidents abounding as testing continues, and as the turnover itself starts "eating" bad data - that is, "bad" data meaning data significantly different than what was expected.

Viruses at the same time? Hasn't been publicized much in open press. Can't discount it - especially since Clinton wants to declare cyber warfare in the Balkans. (We're part of the "cyber" back over here - by the way.)

Metdown - can't see it happening physically - as an exaggerated description of failure by locking up computer? Sure. (The original quote probably came from somebody who saw "War Games" a couple of dozen times too often....)

Don't seriously

-- Robert A. Cook, PE (Kennesaw, GA) (cook.r@csaatl.com), May 26, 1999.