Ubuntu alpha apparently breaking hardware, shattering dreams
By Donald Melanson 
posted Sep 26th 2008 12:56PM
posted Sep 26th 2008 12:56PM
Well, it looks like the good times that are the Ubuntu alpha testing process hit a bit of a snag recently, as one of the latest kernels apparently had the nasty side effect of irreparably damaging some users' hardware -- specifically, certain Intel network cards. So far, it seems that only laptops have been affected by the bug, which corrupts the NVRAM used to store data like MAC addresses, but folks don't seem to be ruling out the possibility that it could affect desktops as well. What's more, while a warning has been added to the ISO download, it hasn't been pulled altogether, with the only explanation given being that it would delay the 8.10 release schedule too much. Not entirely surprisingly, that has prompted at least a few folks to give up on testing alpha versions of the OS, at least until this whole thing gets sorted out.[Thanks, JagsLive]
You expect a few bad bugs in a OS alpha stage but not this bad.
Shocking.
Someone fucked up.... bad....
There's a difference between "a few bad bugs" and bugs that destroy hardware.
It could be worse....
http://lh6.ggpht.com/shaileshdoke/SDym7NzNupI/AAAAAAAAAq4/sLfF2cd_muQ/Hack%20and%20Blow%20like%20Bomb%5B5%5D.jpg
in my opinion, it is bad hardware design if it *can* be irreversibly damaged by software. there should be checks at the hardware level against that, as well as ways to factory-reset any piece of hardware that commercially ships.
These people do it for free, they owe you nothing.
Whoopsiedaisy. Lol
@dv
There is no reasonable way to do that with many types of hardware. You need to be able to write data to areas which contain the necessary code that allows data to be written to that area, because somtimes the code that communicates with external devices needs updating, so you cannot 'protect' that code in any way.
@ Curtis
@ Chris
I'm not sure if you guys really know much about Linux, but I've been running it over half my life, and one of the first things I learned is never run anything that's an alpha. They aren't meant for real hardware yet, unless your a developer (and this goes for the majority of them too) you should run it in a virtual machine. If you want a sneak preview, wait for a beta, if you want a current system, use a current build, and if you want a stable system, get an older "stable" build.
"You expect a few bad bugs in a OS alpha stage but not this bad. Shocking."
Well, I can tell you that I'm running the alpha build right now and I haven't... OW! What the hell was that?!?
@ dv
Why would intel need to protect their hardware against Operating systems it was not designed for? HMMmmm?
@rock99rock
Last I checked, hardware isn't usually designed AROUND operating systems. And moreover if Intel's industry standard networking hardware wasn't "designed" to run under Linux kernels, I doubt they would code their own drivers for them...
@rock:
considering both e1000 and e1000e drivers are actually written by Intel itself, I find your comment quite amusing
@rock
You comment would've held some credibility, if it wasn't for the fact that Intel Writes Their Own Modules
@Kamokazi:
Sure you can. Motherboard manufacturers have been doing it for YEARS. I've even un-bricked my WRT54G router that had a bad flash. It's all in the design.
Do people not know what Alpha version of a piece of software means? It means it's not even ready for testing. It's when developers are still adding to the schedule release. It is in Beta stage you have something worthwhile testing. Well so you might ask what's the point of Alpha versions? ( Mainly for integration testing, developer testing and not aimed for the general public). If you use Alpha or Beta that's too bad, you shouldn't be using those if you have the slightest clue what's going on. That's why there are official release date. This is good though that this bug was found in the Alpha stages of development rather than final gold master.
Intel network cards usually have problems with both Linux and Mac installs.
@dv "in my opinion, it is bad hardware design if it *can* be irreversibly damaged by software."
How do you suggest updating BIOS, firmware etc? The reason so many devices say "UPDATING FIRMWARE. DO NOT POWER OFF" is because it's difficult/expensive to design hardware which requires updates yet cannot be damaged irreversibly. You'd have to have a reliable redundant power source such as a battery should power go out during the write, or a secondary fallback firmware (as some motherboards/servers now have).
There are also various levels of irreversibility (not sure if that's a real word). The actual bug report says "recovery may be possible via a BIOS update". That's not so painful. Other techniques might include shorting pins (http://www.linksysinfo.org/forums/showthread.php?t=47259) or pulling off a chip and using an eeprom burner. At what point do you consider a device irreversibly damaged?
Nooooooo!
Apparently, yes.
This is a bit discouraging. I'm currently running Ubuntu on my laptop. But, this news really scare me.
Hopefully it gets corrected soon.
as long as its not this alpha version your fine, if your not sure then i'd say you have a final version and nothing to worry about
Can this bug affect the hardware even if you turn it off (disable) the LAN card in the BIOS?
It's an alpha.
Alphas always have horrible bugs that make them unusable to the end user. They should not be played with unless your developing.
Just wait for the current build and it'll run fine.
@Mark:
I doubt it would affect disabled cards, since the issue most-likely stems from faulty drivers. If those drivers aren't loaded, then the card isn't used, and the probability of issues occurring decreases dramatically.
While I didn't RTFA, that's just my opinion.
"Hopefully it gets corrected soon."
Don't count on it. Once they find a bug, they usually leave it alone until the final version. That's what alpha and beta releases are for: to let bug proliferate.
Wow...and I was looking at installing the alpha today on my laptop since it's nearing the end of alpha testing. Glad I read this first...
Wow...never heard of an OS permanently damaging hardware. Thought only bad flashing & overclocking could do that....
So that's why it's called an alpha... RIP faithful warriors... your p0rn will be safe with us.
Raul,
I'm a software developer, and although the concept of Alpha and Beta software is relative, in general I believe that Alpha means "If you run this software the way it is, there's a very high chance of a bug being found at this stage, and a high chance that the software will crash", when in Beta the crashes should be pretty much erradicated and this stage is what i call the "bug hunting" stage.
Software that corrupts hardware is definitely pre-public release material.
So Alex... you have every piece of hardware ever made to test your software on? Eventually it has to be pushed public to see how it performs. Sure, it's a driver and only affecting Intel cards, but using the generic driver model to support multiple cards can sometimes have quirks.
In all honesty, the hardware (if it requires a flash ROM of some sort) should have protection against catastrophic failure from overwriting the ROM. Motherboards have been implementing stuff like this for some time and even before that, they had methods of reprogramming them.
Nice way to make it a problem on one distro, when in fact its a bug in intel ethernet driver in 2.6.27 release candidate kernel. (don't remember exact driver name), and affects all distros using that kernel.
Its the e1000e Intel network driver and only affects that chip. Its a kernel driver bug (and as dv said, shouldn't even have been possible in the hardware in the first place) that affects all distributions that are using a 2.6.27-rc kernel.
Ultimately, though, I believe "alpha" releases do (or should, at least) imply a certain level of readiness. I personally don't try out an Ubuntu pre-release until its final, and I have years of personal and professional linux experience.
YMMV
I despise stories like this. I feel its a blatant abuse of power when an otherwise respectable news outlet plasters one vendor's logo and name all over a bad news story. The offense is particularly egregious when several other sources have already reported on the problem and subsequently tracked it to its root. At best this story was badly researched and in my opinion, that reflects badly upon the author.
This man speaks the truth.
Wasn't there a Slashdot article about another distro doing this a few days ago? It was also wrong to label that distro at fault when it is really a Kernel Issue
This bug applies to all distros using the as of yet unreleased 2.6.27 version of the linux kernel.
This is not strictly an ubuntu bug/mistake
So this is sort of a blow to Linux in general... Damn.
While you could argue that this is alpha software and carries risks, the problem is that many people who would be testing this and future alpha versions and reporting bugs will now hesitate to run early release software, ultimately slowing down the release process.
One could argue "you should be running alpha in VMs" but VMs have a very limited virtual hardware set. You can't reliably test hardware compatibility in VMs.
It is not an Ubuntu bug only.
It is a Linux kernel bug, for all 2.6.27 releases before rc7.
It damages e1000 Intel ethernet cards, writing in their EEPROM.
how does a fuk up like this even happen? werent there any safe gaurds to checking what rom it could write into?
Linux kernel: "im in ur eeprom, ruining ur bits"
Dang it captsaltyjack, I was thinking the same exact thing!
Good thing I don't run any alpha software. I'm a firm believer in not distributing alpha software altogether, unless you absolutely hate somebody! >:-)
since i'm already low rated all around for my valid questions and points.
@msalivar
So what? I would like to know more about computers and what is possible with them. I doubt someone who wants to cause physical harm through hacking and virus's is gonna use this site to gather info.
so all in all I'm disappointed at you guys, I've had questions and statements replied to in a kind and helpful manner in the past or just ignored. But this is ridiculous.
I've been interested in if it's possible to kill hardware with software. I'd like to know if there are methods of killing other specific types (like a brand or model) or general hardware with software.
So now people who ask real questions get low ranked?
and not one reply even relates to what I said, why?
Because you sound like someone who's waaaaaaaaay too interested in hacking someone.
So anyone interested in anything immediately wants to put that knowledge to use then is what your saying. I'm interested in many things dangerous and basically illegal, but I'm never gonna do anything with all that I know.
Besides I couldn't hack a computer unless you hand it to me.
Even if it's not your intention, it could be someone's whos reading this thread.
The bug in the e1000e driver has been around for some days in kernel 2.6.27-rc1 to rc4. This is not ubuntu specific, but a problem of all upcoming linux versions like ubuntu intrepid, suseLE 11.1, fedora "rawhide", see here f. ex.: http://lists.opensuse.org/opensuse-announce/2008-09/msg00017.html
In versions 2.6.27-rc5 (sept. 23rd) and newer module e1000e is IMHO blacklisted by default. But you better check that yourself before you install any alpha on your system.
Everyone seems to forget that this isn't an UBUNTU specific problem. It effects ALL distros using the latest kernel (2.6.27-3 I believe) Don't be so quick to assign blame to Ubuntu!
The first distro I saw get blamed for this was actually Fedora..
First, this is not a specific bug to Ubuntu. The bug exists in the e driver for the Intel ethernet drivers. The bug is part of the kernel and they are working on it as it is marked critical in the bug tracker.
I agree that it is a serious issue but there is no need to panic. They uploaded a fix that blocks the e1000e kernel driver from loading so that should prevent further damage. If you want to learn the fact and avoid speculation, check out the bug report here < https://bugs.launchpad.net/ubuntu/+source/linux/+bug/263555>.
Intel make their own open source drivers that go into the linux
kernel. Theyre quite vocal about this fact. This means that Intel are most
probably at fault here.
I motion that Engadget rename the article to Unreleased Linux kernel 2.6.27 breaking hardware, shattering dreams. Throwing dirt on Ubuntu's name doesn't help Linux, it harms it.
I second this motion, and urge everyone else who does to reply to this post and make it long enough for Engadget to notice and hopefully take action.
You have my vote. Engadget team -- please correct this misleading title!
Thanks!
Throwing dirt on Ubuntu's name harms Ubuntu. Destroying hardware harms Linux (and the hardware, I guess). When I first read the posting from Engadget, I thought "Oh, it's just Ubuntu. If I wanted to try Linux, I could just avoid Ubuntu." Now everyone's saying it's ALL distros using a specific kernel, I'm thinking "Linux? Yeah...not so much."
And before anyone says "It's just a specific kernel version!" Unless you can prove that frying Intel NICs was intentional and made known to users beforehand, then I'm going to go ahead an assume that it was a bug, and it cropped up unexpectedly, frying the NICs of some poor, unsuspecting folks. That being said, I wouldn't feel comfortable putting something on my machine that could potentially break it. Say what you will about Windows, but it's never broken any hardware.
I can imagine the shitstorm that would come from the Linux fanbois if MS ever did this, even with an alpha. The fact that those same folks think this is somehow acceptable and expected due to alpha status makes me think twice about Linux. Think I'll stick with MS for now...
@EricC
You mean when some Windows NT and 2000 service packs (meant to solve bugs, not create them) crashed your machine irreversibly? Or when Windows Home Server, final release, used to corrupt files with an "undefined list of applications"?
You know, I'd rather spend $20 on a new network card than thousands of dollars in time for restoring a customer's machine and loss of revenue for downtime, or losing all my beloved photos and files forever because of a file corrupting bug in the OS.
BTW, I'm not a Linux fanboy; and if your idea was to give Linux a try with an Alpha, you're simply dumb.
@ giuliop:
None of those examples are examples of Windows harming hardware, which is my point. Data can be replaced if it's backed up. If it's not backed up, you're a fool and probably deserve to lose your data.
I never said I was interested in trying an alpha version of anything. I'm saying that fried NICs came from a bug, and there's no guarantee that the bug can't or won't crop up in a final release. Today it's NICs, tomorrow it's some other piece of hardware. This very fact is going to keep me, personally, away from Linux on my home machines. My hardware has enough to worry about without me installing some rogue OS that's going to arbitrarily decide it doesn't need to work anymore.
Hey, does anyone here know if this bug is specific to Ubuntu or affects all distros?
Um, read above comments ----^
Wait Wait, is this bug in Ubuntu only? just checking, want to make sure, its not really clear from the comments
//ubuntu 8.10 Intel Incinerator???
I completely agree with Ubuntu not pulling it's alphas. The problem is 100% a problem in the pre-release kernel that Ubuntu and other distros are building on to put into their next version. Ubuntu should not set their development schedule back for one kernel module that is broken (although rather badly) simply because it will be fixed before the 2.6.27 kernel is finished. Caveat Instalor.
This isn't a Ubuntu specific problem.
I just wanted to be the millionth person to mention that. :)
congrats you win a free copy of Ubuntu 8.10 !!
Anyone who is seriously alpha testing an OS should be prepared for the consequences. Anyone who is doing it for kicks should run it inside a virtual machine. If you are the latter and you ran it on good hardware then you had this coming...
Even beyond the end users, the hardware manufacturer should take responsibility if their hardware doesn't have fail safes to prevent irreversible damage from software. In all honesty my standards for Intel have been set pretty low at this stage. Their internal dev process with network hardware must be pretty piss poor to begin with considering that every time they release new PRO/Wireless drivers they stop working with our Aironets and occasionally bork the cards in the computers... I don't think it's fair to pin the blame on the OS developers when the hardware manufacturers make making fatal changes to the hardware so easy...
This issue involving the Intel nic (e1000e) is related to the 2.6.27 kernel and not to anyone Linux OS. If you do want to try out alpha and have a Intel e1000e nic, before installing do ethtool -e ethx > ~/mye1000eeeprom.txt so you will have a copy of your EEPROM which also includes the MAC address for your card.
I just wished engadget would have done more research regarding this. All the other sites like Lxer, Phoronix, OSNews to name a few did their research.
So you guys are saying that this is related to 2.6.27 kernel and not Ubuntu. Can someone clarify.
...and everyone else wishes you would have read the comments of the other 20 people who posted the exact same info above you.
Research doesn't bring in the rabid people and their page hits (thus revenue.) Sensational stuff like this is what has kept The National Enquirer and several other publications in business. People want drama.
I read comments about the bug in the kernel....
My point of view is a bit different : that piece of hardware is a piece of shit. I strongly with Kamokazi's comment :
"There is no reasonable way to do that with many types of hardware. You need to be able to write data to areas which contain the necessary code that allows data to be written to that area, because somtimes the code that communicates with external devices needs updating, so you cannot 'protect' that code in any way."
There is always a way, it's just that sometimes people are not smart enough.
Can someone edit the article name to not put the blame on Ubuntu, who is clearly not responsible for this unfortunate situation?
No, because throwing names about = more page views = more money.
This is why I don't run alpha releases; there's a reason beta comes after alpha.
The bug is in the new kernel release. At least Linux providers have come forth, admitted to a major bug, warned people not to use this version of Linux and are trying to correct the problem now.
If this was Windows, Microsoft would have told everyone they had bad hardware and kept the bug in their system, breaking even more computers! Remember, the end user is always the blame on a Windows system.
Why is this even news?
Alpha means "WILL EAT YOUR BABIES" (and maybe your network card).
No normal user will experience this problem because normal users will not use ALPHA SOFTWARE!
A: This isn't news.
B: The software is developed by a completely open community. You have sources for real information; use them!
Thanks to the people who've been victims the problem in this BETA. These problems will undoubtedly create a better product in the long run. At least your issues are appreciated by the Ubuntu fans out here.
"First, this is not a specific bug to Ubuntu."
Yes, but then Engadget has never really understood that any other form of Linux exists. Witness my forlorn efforts to get them to mention Mandriva Flash, or any netbook preloaded with Mandriva...ever.
Not that I'm bitter.
Wait, yes I am.
(In case it's not obvious, I work for MDV).
It would be useful if people could actually report this properly as it would be useful for everyone affected to know about the issue. This isn't an issue with just Ubuntu. Its an issue with the 2.6.27 kernel which means it affects all distrobutions. Fedora, Ubuntu, Debian, Suse, Mandrake etc etc. Any distro that is shipping (fortunately at this time it is shipping just in development and not shipping releases). And it was/is Intel that fucked up it seems. They know of the problem and are working to fix it.
Hy everyone!
I just wanted to point out two things out when reading the comments here:
1st - If some new Windows "alpha" update would damage hardware, hell would broke loose and the people would go crazy, bashing Windows like there would be no tomorrow
2nd - The hardware *should* be able to catch Software errors to protect the hardware itself, but there are just too many ways to damage something and no hardware can be tested through for anything, it is just too expensive. That is why driver programming is normally left to the manufacturers only.. It is just not a good way of the Linux programmers to pass on drivers for different hardware versions which was not tested. They should have chosen the typical "this driver/hardware is not supported yet" message. Hardware programming is just a area which does not allow bad programming and passing on untested driver code is just careless
This post is missing one fact that the problem is not only with Ubuntu 8.10 Alpha release but with all new distro releases including Open Suse's. In fact the warning from the Kernel team first arrived in the Open Suse mailing list and then appeared in Ubuntu bugs list. Another fact that alpha releases always comes with "something bad might happen, but we're trying our best not to let something like that". Unfortunately this time it did and the fault is not just with the distro but with the driver with came with the upstream kernel. Indeed the Kernel devels who wrote the driver has started working with the downstream distribution's kernel teams. As a precautionary measure, the driver has been removed in the latest kernel update and hence should be safe enough to try out now.
Wow, I see a lot of people in denial.
Firstly, yes Alphas break things. Partitions, OS installs and most of all data. Hardware though? I think if testers expected an alpha to brick a bit of their system they'd be unlikely to be using it. Seriously, to use the excuse "well alpha's break things" is a bit ridiculous. No one would be testing alpha's for free if they really expected it to break their computer.
As for where the fault lies? A lot of people seem to be saying Ubuntu are whiter than white and this all lies with Intel. Up until the first report of this bug came in I'd agree with you, but Ubuntu took over 3 weeks to do anything about it. Read the bug report on launchpad and you'll see how long it took for any action to be taken in order to protect the hardware of the people who are devoting their free time to help Ubuntu improve their product.
The latest Alpha iso is still dangerous and it was only a few days ago a warning was added to the ISO download sites and that a kernel with the problem module disabled was pushed to the repositories. In my opinion they should have pulled the ISOs the moment the bug was confirmed and released a kernel with the module disabled. Then they could have released an updated ISO (Alpha 6.1). Instead they ummed and arred about what to do.
In my opinion the headline is 100% correct and anyone trying to make excuses for Ubuntu's inaction has no idea what they're talking about.
Ubuntu need to learn a lesson from this and never let an issue like this go unresolved for this long again.
Oh and I'm an ubuntu user using the alpha on intel hardware (with the module disabled).
Man, I'm usually first in line for beta releases (it makes me feel special) but holy crap, I guess that is the downside for being first sometimes.
For cryin' out loud people, this is exactly what 'alpha' means. It means 'hands off, except for testing systems'. Just because the last few alpha versions have been exceptionally good and stable doesn't change the fact that you're taking a risk if you run alpha versions.
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=history;f=drivers/net/e1000e/ethtool.c;hb=4a7703582836f55a1cbad0e2c1c6ebbee3f9b3a7
The main person to work on these drivers is: Bruce Allan an Intel worker
Now since this is an Intel card they are in a VERY good situation to write drivers for their hardware
Since this kernel release is STILL ALPHA!!! there is expected to be bugs
So again
1) This isn't an Ubuntu issue
2) this aint some ZOMG linux is rubbish issue
3) This is healthy development
It's just lucky I don't use Intel network cards...
The problem lies with Intel because it is a driver fault and that has been proven as the hardware doesn't get damages when the module is disabled. Intel really need to get up to speed here and fix that kernel module before hell breaks loose!