Slashdot Mirror
LG CD-ROMs Destroyed by Mandrake 9.2
An anonymous reader writes "The latest offering of Mandrake's distribution, 9.2, has been found to not only be incompatible with some LG CD-ROM drives, but to destroy them during the installation process. Mandrake have posted information on their errata page and further information can be found on this thread [google]. Along with over 350Mb of updates within a week of release, it's not been a good start for this latest release."
Someone find a 1st level Cleric so they can cast Protection from Evil on these Lawful Good CD-ROMs. That should keep the evil Mandrakes from destroying them.
At least they should get a freakin' saving throw. What a harsh DM.
You see? You see? Your stupid minds! Stupid! Stupid!
...and software is capable of destroying your products, you're fucking fired.
Do you think there might be a problem with your hardware if it can be destroyed solely with software?
"Smoking helps you lose weight - one lung at a time" -- A. E. Neumann
As for Mandrake, I'm sure that the updates are a good thing, unless they're stupid bugs that should have been fixed before release.
I know quite a few people who have had big problems with LG drives. I think that they are very unreliable anyway. Lots of people I know also have LG burners that mess up cds when burning. Stay away. Anybody else know people with faulty drives?
People started noting this a week ago:
9.2 FRIED my CDROM drives
Whenever the offence inspires less horror than the punishment, the rigour of penal law is obliged to give way...
appears to be a kernel patch
I found this post.
I've had this exact problem with the Cool Linux CD: http://emergencycd2.sourceforge.net/ This article just confirms the problem. I was using a Dell OptiPlex GX1 and the system would just halt on boot. Then on reboot the drive was no longer detected. The drive would not even respond to an eject - I had to do it manually to get my CD back. Unfortunately, I assumed that a CD-ROM could not be damaged by software and that this drive just happened to fail as I was booting...so I tried it on another system! Now I've got two dead CD-ROM drives waiting to be returned to Dell. Now time to play stupid about why the drives failed to get an RMA!
Well, after RTFT, I came across this:
So, I guess if you tend to use bleeding edge kernels, beware. Mandrake sometime tosses in non-"Linus blessed" things, I believe, so this might have been something you'd only get if you went looking for it.
I've installed 9.2, and it's been a mess. The missing kernel source package in the download version ws a major pain in the ass. Since I'm a silver Mandrake Club member, I was able to get the PowerPack edition as a download as well, but that kept messing up when trying to install. The checksums all checked out, so I have no clue why I got the various problems I did. I've finally gotten it stable, and able to do a few things I haven't been able to get working in the past, like DVD viewing (no, not through the stock Mandrake stuff... only through additional non-Mandrake packages).
Sad to say, Mandrake messed up this release big time. It just wasn't soup yet. It's really too bad, since I've had pretty good luck with them in the past.
If you're a hardware manufacturer...and software is capable of destroying your products, you're fucking fired.
So how do you propose putting firmware updates into CD-ROM drives, DVD drives, modems, etc.? Just about any peripheral which has flashable firmware can be rendered unusable by software.
Dude, software should never be able to damage hardware. Not in 2003. Part of this is just common sense -- how could anyone design hardware that bad? But beyond that, it is only a matter of time before someone writes a virus that includes this cute little effect. It is no longer possible to blow up a CRT by giving it an out-of-range signal, or to call halt-and-catch-fire, or to blow up your car's engine by overreving it (assume you haven't screwed with the rev limiter). It is not okay for normal usage to damage hardware, and in the computer world 'normal usage' means any data at all, even malicious or (in the case of Mandrake, it seems) really bad data.
I've had this sig for three days.
> If Windows would do this to your drive there would be a public outcry. Here on /. it is more like "ah well, shit happens, it's mentioned in the errata so suck
>
> it up and get over it".
But windows could do this. All it would have to do is send one of the two normal APATI commands to this cdrom drive, and it will fry just the same.
LG stated the bug is in their cdrom drive, and one of two commands sent to it will execute the buggy routine in firmware, causing it to dump its firmware totally.
They cant be fixed because to flash firmware, you have to use a program that is in the firmware in the first place.
I wonder how many folks here would be bitching out LG if it was XP that was trashing the hardware?
1984 was supposed to be a warning, not an instruction manual.
I just read a post from Alan Cox, it appears that if you send a flush cache command to the specific LG drives or their compaq rebadged ones, the drive gets fried. So this really has nothing to do with Mandrake and everything to do with a poorly designed drive.
If Tyranny and Oppression come to this land,
it will be in the guise of fighting a foreign enemy. -James Madison
Already been done. "Back in the day", there were plenty of virus written that would throw your monitor out of sync, simply baking the tube.
Some others that whould smash the read/write heads of your HDD into the spindle destroying the drive, that's why it's controlled at the hardware level now. That was back when the heads actually required a seperate program to park them. That was alot of fun.
I'd say that Mandrake is responsable for the replacement of those drives.
Om, nomnomnom...
While it is nice that Slashdot posts this as a service to the community, it could have been an idea to at least try and get more facts before posting this.
Firstly, it seems to be only (or mostly) CD-ROM drives, and not CD-RW drives or CD/DVD drives, however Mandrakesoft is compiling a list of the affected model numbers.
Secondly, not all drives of the same model number are affected, since some drives of the same model, but with differing firmware revisions, have different results.
Thirdly, this is a hardware/firmware defect, which seems to be triggered by the packet writing patch (I believe SuSE has shipped with this patch for some time, so LG drives could be affected under SuSE). If your drive is still under warranty, LG should replace it.
It may also be possible to reflash the drives with a working firmware, but no-one has reported success with that yet.
Instead of posting a link to alt.os.linux.mandrake, maybe next time Slashdot can link to the thread on the cooker mailing list which has been posted to by the Mandrakesoft people investigating the issue? But I guess that's too much to ask of Slashdot.
Software != Firmware.
The mandrake problem doesn't have anything to do with firmware as far as I can tell, you just send a flush command to the drive, and it fails.
A simple software command should never, EVER be able to fry hardware. Screwing with the firmware is another problem entirely.
Then put a separate copy of the original firmware into read-only memory at manufacturing time and provide a physical button that writes the known good firmware over the current firmware...
Well, what I think will happen is that there will be more effort put into eliminating our choice as to what software we run on our PCs and what we use them for. After all, that's really what Palladium and "trusted computing" is all about. More FUD is on its way about how these "rogue" Linux systems can't be trusted not to burn up your equipment, etc. etc.
This will only fuel their "See, you lost a CD-ROM drive and because it's open-source, there's no one to cry to" argument.
Of course, practically speaking there is never anyone to cry to when hardware fails other than the hardware manufacturer, or your local retailer. This problem could easily have shown up in a Microsoft product first, since it is using a documented feature of the drive! There are reasonable limits you can expect software vendors to go to in testing hardware, given the vast number of products on the market. In any event, even if Windows did toast my drive (and I've had a couple mysteriously croak under Windows although I never suspected it was a firmware issue) I can't see Microsoft sending me a new drive, or for that matter ever admitting it was their fault! All the pro-Microsoft apologist trolls here on Slashdot can grumble all they want, but at least here the accountability trail is very complete (a definite plus for open source) and we'll be able to verify when and how the problem is fixed. Try doing that with Windows.
The higher the technology, the sharper that two-edged sword.
That is stupidiest BS I've seen. OS can e.g. override firmware of the disk drive. If it writes bogus firmware, the disk will be permanently damaged. Just like OS can screw your BIOS and computer would not boot anymore. Current hardware is highly configurable by software, and if software damages hardware, it's software fault.
I think that LG should be getting busy soon with making sure this doesn't happen in the future.
I think Mandrake should be busy about it.
MSDOS: 20+ years without remote hole in the default install
My MB does. Many Gibgabyte boards have a dual bios feature. And yes, it does have a restore factory defaults option.
Look on it as a lesson in life.
Treat other people with respect, be part of a "community", and they'll forgive you the odd unfortunate mistake.
Spend your life screwing over other people, think about nothing else except "number one" or "the bottom line" and, rightly or wrongly, any unfortunate mistake you made gets jumped on.
Lete not get all worked up, we all know what is going to happen.
1. LG continues to deny any responsibility.
2. The usual suspects will float a few pieces on the ZD rags and perhaps C|Net spreading FUD that Linux is dangerous.
3. One of the Linux IDE Gods will become sufficently annoyed that a proper investigation will happen, the flaw in LG's firmware will be documented in overkill detail.
4. The PR war will turn against LG, they will repent and issue a firmware update, stick a penguin somewhere deep on their support site and declare their eternal love of all things Linux. But it will strictly be for PR.
5. Once understood, a workaround will keep Linux from destroying unpatched drives. Probably something as simple as not checking for packet writing capacity unless basic RW support has already been detected.
6. No longterm changes anywhere. Nothing to see here, move along.
Democrat delenda est
hahaha reminds me of a time i put a severely scratched cd inside a 16x drive only to have it shake the case violently. so i hit the eject button and the cd rose straight up out of the drive tray, flew at my coworker and then veered off to the floor. I didn't leave it in long enough for any marks, but it was warm to the touch. ahh fun times with broken hardware
-dk
I guess LG had a different idea what "flush drive" meant.
"Trademarks are the heraldry of the new feudalism."
...but they should have advised the users of certain models of their drives to check and possibly update the drive firmware.
The thing which kills the drives is - wait for it - setting them up for packet writing. The hackers who made the patch to do this (included starting with Mandrake 9.2rc1) may be able to figure out a way to do it without triggering LG's bug, or may not, in which case any Linux kernel which features this packet writing code will kill a broken LG drive.
Note that this happens when the drive is init'ed, not when you write a CD with one, so you'll kill a drive just as effectively even if you install over the network or whatever.
As to responsibility, well... the drive software is broken, end of story. If your LG drive dies, take it back and make a warranty claim.
For those who assert that Mandrake should have tested 9.2 on every known drive before releasing it, the answer is that Mandrake did indeed test 9.2 on these models of LG drives, but none of their testers happened to have the broken firmware revision(s). <shrug>
For those speculating about what would happen if it had been MS-Windows-XP's problem instead, the only differences would have been that more than 80% of all broken LG drives would have been killed by now due to semi-forced upgrades, Penguinistas would have been gleefully rejoicing that their software didn't kill drives, and Microsoft would still be ignoring the problem and we'd expect them to for at least another two weeks.
I don't know whether it's possible to flash a killed drive's firmware and resurrect it, or whether the broken firmware actually destroys hardware.
Got time? Spend some of it coding or testing
Mandrake actually tested on several broken models of LG drive, including one I own. It didn't kill any of them. Why not? Well, it turns out that none of the drives tested had the broken firmware revision(s).
Using your reasoning, Mandrake should have tested every single firmware release of every single model of every single piece of hardware that their OS interacts with - in all possible combinations - with every single subrelease of their own kernel. Got a spare aeon or two?
Got time? Spend some of it coding or testing
In that case, however, it didn't damage the hardware...it crashed the OS. And then when it was rebooting, it reads from the CD drive and crashes again...and you can't eject the CD manually, because on Macs no removable media can be ejected manually. There isn't even a pinhole. Stupid design, but nothing is ruined, if you know what you are doing. (You can eject from open firmware)
ASCII stupid question, get a stupid ANSI
LG doesn't support Linux, so this problem doesn't exist in their eyes.
When I bought my LG CD burner, they claimed compatibility with Linux - Slackware 2.0, but Linux - on the box.
With his problem, if the Mandrake installer is conforming to standards when accessing the drive, and the drive fails because it doen't meet those standards, then it's the drive at fault. If however, the Mandrake installer is pushing something too far and stepping outside the boundries the standard specifies, then Mandrake would be at fault.
It appears at this point that they (Mandrake) are still looking into which of the two above it is.
I AM, therefore I THINK!
Somehow this reminds me of Hofstadter's illustration of Godel's incompleteness theorem in Godel, Escher, Bach... wherein Achilles has a phonograph which he claims can reproduce any sound, so the tortoise gives him a record with a sound which destroys phonographs...
Well, it was funny to me.
Where are you going?
It is not Linux, it is Mandrake who put an *experimental* kernel patch into a *production* release. It was very stupid. The patch was meant for ide cd-rw drives in which case you want to flush. That is why the LG CD-RW drives are not affected, only the normal CD-ROM drives. This is one of the reasons I *never* use Mandrake. I have had too many problems with their distribution. I stick to Red Hat. It is *far* more stable and Red Hat has 6 of the top 10 kernel developers working for them.
If Tyranny and Oppression come to this land,
it will be in the guise of fighting a foreign enemy. -James Madison
A flush command which happens while deciding whether they're a writer or not has been redefined (<thwack!>) by LG to mean "upload firmware" (with predictable results). To quote Juan Quintela from the Cooker list, "Yep, whoeved decided at LG that reusing for UPLOAD_FIRMWARE command FLUSH_CACHE comand should be shoot. Twice."
Got time? Spend some of it coding or testing
I submitted an article about this after I fried THREE CD-ROMs in about 2 hours installing 9.2 on a rack of 5 machines. In the article I submitted were the exact model numbers of the dead puppies. All that remains of the article now is:
- 2003-10-23 20:40:24 Mandrake 9.2 Eats CD-ROM Drives On Install (articles,mandrake) (rejected)
When I get back to work Monday I'll post that info (and the firmware versions, if I can get them) to the Mandrake Club Install forum. Of course, that's where I should have posted it in the first place. I'll know better next time, but I tried to warn 'ya!
Any technology distinguishable from magic is insuficiently advanced.
Any technology distinguishable from magic is insufficiently advanced. - Geek's corollary to Clarke's law
I disagree in part. Sending a command to a device without knowing it is supported is not good ATA practice at all. The patch they applied should have checked but didn't.
Shipping easy to fry drives isnt bright either and I suspect LG know this without any help, especially when they get lorry fulls of faulty drives back. Not only can a wrong command occur due to an error on the cable (very unlikely) so should be handled tolerantly, but every virus writer on the planet now knows how to toast all the LG ROMs (and rebadged LG ROMs).
I just hope Mandrake have the decency to recall any boxed sets sitting in warehouses and heading to shops and replace the CD's in them.