Intel SPI Driver in Ubuntu 17.10 Release Might Corrupt Your Laptop Bios

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
13,500
Canonical has pulled their latest Ubuntu 17.10 release because there are many reports of Lenovo and other brand laptops experiencing corrupted BIOS when the OS is installed. At this time it looks like the Intel SPI driver is the culprit. Once a new kernel can be compliled without these drivers the OS will be made available for download once again. If you downloaded this before this morning then you shouldn't install on your laptop. In some cases only a motherboard replacement will fix the bad BIOS. Beware.

Should your BIOS be corrupted, you may need to replace your motherboard if there is not a removable flash chip. There are some reports that resetting the BIOS does work, but it's too early to know if that works for everyone. This issue has been confirmed for several different lines of Lenovo laptops including the Yoga and IdeaPad products. There is also the reports of it affecting a few Acer, Toshiba and Dell laptops.
 
After looking around it seems that SPI driver was written into the new Linux Kernal so it could be in more distro's than just the new Ubuntu they are just the most prominent. It seems it has something to do with how specific vendors implemented Legacy Bios.
 
Corrupted bios? What?

I thought that the motherboard pulled from flash, into a 'memory' (not ram) so this type of thing couldn't happen.
 
Way to go Intel.

I hope this wasn't what they had in mind when they said they were going to take more risks.
 
Since this is an Intel driver issue is really doesn't matter what OS it is on... And actually is Intel's problems not Ubuntu's...
Just like Puma 6 is an Intel problem, not an ARRIS or Netgear problem. Trust me, the end users who are affected don't see it this way.

And yes, this is a Linux problem, or at least an Ubuntu problem, because the driver is integrated into the kernel, so if you want that version of Linux, you will get the corrupted driver.
This issue appears to stem from the Intel SPI driver in the 17.10's Linux 4.13 kernel

Note that it is not a kernel module (as it should be, or as a separate install as it would be in Windows), but it is in the very core of what makes Linux Linux - the kernel.
 
:facepalm:The issue comes from the driver that Intel provided. They could have just as easily provided a bad driver for windows too.

Bad drivers through Windows update is a pretty common thing, especially when it comes to video, sound, and printer drivers.

Some of them have been so bad that you can't even boot into Windows once the update is installed.
 
Well this sucks. Cmon Cacnonical! Or... who's fault is this exactly? Why should an OS be able to muck up a BIOS like this? Something's not right here... Intel!
 
Note that it is not a kernel module (as it should be, or as a separate install as it would be in Windows), but it is in the very core of what makes Linux Linux - the kernel.

Well of course it's part of the kernel, the kernel is monolithic, that's the difference between Windows and Linux! Fact is, it's a driver component of the kernel, separate to the OS element of the kernel that's causing the issue.
 
Bad drivers through Windows update is a pretty common thing, especially when it comes to video, sound, and printer drivers.

Some of them have been so bad that you can't even boot into Windows once the update is installed.
That was my point. If Intel had supplied a bad driver for Windows we'd be seeing the issue there too. The whole issue is Intel's doing, not Linux or Windows.
 
Corrupted bios? What?

I thought that the motherboard pulled from flash, into a 'memory' (not ram) so this type of thing couldn't happen.

Modern flash chips are execute in place. There's also other things stored in the flash chip besides the main BIOS image. Stuff like NVRAM (particularly for laptops), password hashes, platform initialization variables, embedded controller code, etc that needs to be accessed before the processor can execute the main BIOS. To read and edit those things, there needs to be a driver included in the OS. And then there's the whole Intel ME stuff.
 
What add that only Laptop and only a few brands not Desktop
Many laptop providers and a select number of Desktop providers customize their Bios' in non standard ways to either provide or limit functionality, laptops tend more than not to be a race to the bottom in terms of cost so corners get cut all over the place which leads to implementations of hardware that work outside of spec, fewer available PCI lanes or increased number of daisy chains back to a common component that sort of stuff. Probably just the perfect storm of cost cutting and driver bugs, this one can just be chalked up as another one of the many faults in the 4.15 Kernel.
 
I guess you didn't even read the article in the link you posted. It is a generic catch-all web page with basically a whole lot of nothing on it. I dislike Windows, but come on now, at least post something of relevancy. I guess the people who liked your post didn't read the link you used either...

GREAT JOB. WHAT A BURN!

This. I bothered to read the article and found no mention of Windows or its drivers corrupting anyone's BIOS, just non specific advice to re-flash a corrupted BIOS.
 
Bad drivers through Windows update is a pretty common thing, especially when it comes to video, sound, and printer drivers.

Some of them have been so bad that you can't even boot into Windows once the update is installed.

True, but in that case I boot to a USB or optical media and reinstall. In this case what do we do? Get a new motherboard? Replace the BIOS chip? This is next level shit.
 
True, but in that case I boot to a USB or optical media and reinstall. In this case what do we do? Get a new motherboard? Replace the BIOS chip? This is next level shit.

The configuration of the kernel is hugely irrelevant and there is every possibility such a scenario could play out under Windows just as effectively as the possibility exists for such a scenario to play out under Linux.

It's a driver issue, all operating systems use drivers for identical hardware.
 
The configuration of the kernel is hugely irrelevant and there is every possibility such a scenario could play out under Windows just as effectively as the possibility exists for such a scenario to play out under Linux.

It's a driver issue, all operating systems use drivers for identical hardware.

Indeed the bad driver is the cause, but Ubuntu released it in their OS apparently without enough testing. You'll never catch everything, but this is serious and Lenovo isn't exactly a small vendor.

https://en.wikipedia.org/wiki/Marke...puter_vendors#Top_vendors_market_share_(2016)

I agree that MS deserves all the blame they're due when they screw up, which is often, but apparently Linux, Ubuntu, and co get a pass from you guys.
 
Indeed the bad driver is the cause, but Ubuntu released it in their OS apparently without enough testing. You'll never catch everything, but this is serious and Lenovo isn't exactly a small vendor.

https://en.wikipedia.org/wiki/Marke...puter_vendors#Top_vendors_market_share_(2016)

I agree that MS deserves all the blame they're due when they screw up, which is often, but apparently Linux, Ubuntu, and co get a pass from you guys.

Geezus. If you want to point fingers and play the blame game when it comes to poorly tested software, I can assure you MS will loose every time in comparison to Ubuntu.

No one's giving anyone a 'free pass', but the fact has to be highlighted that a driver is to blame and the configuration of the kernel means bugger all.
 
Geezus. If you want to point fingers and play the blame game when it comes to poorly tested software, I can assure you MS will loose every time in comparison to Ubuntu.

No one's giving anyone a 'free pass', but the fact has to be highlighted that a driver is to blame and the configuration of the kernel means bugger all.

Geezus indeed. They want to compete with MS form the desktop to enterprise, yet their release bricks hardware from the #1 volume mfg. They didn't write the driver, but they included it in a release that they push as "safe".

If they want to win they need to build out a well known brand of trust and reliability. I'd love for them to do better because competition benefits us all, but they have to do better than this in the QA department. We've all seen how giants like Intel and MS do when the competition is weak.

I'm not against Linux, I want them to get the tough criticism they don't just deserve, but need to get anywhere. The kind that caused Mozilla to take their once revolutionary browser that got fat and lethargic and really dig in to turn out the fabulous release that was Quantum.

If you truly loved Linux/Ubuntu and wanted them to succeed you'd be banging on their door saying "Dude, WTF? We're counting on your guys!"
 
Geezus indeed. They want to compete with MS form the desktop to enterprise, yet their release bricks hardware from the #1 volume mfg. They didn't write the driver, but they included it in a release that they push as "safe".

If they want to win they need to build out a well known brand of trust and reliability. I'd love for them to do better because competition benefits us all, but they have to do better than this in the QA department. We've all seen how giants like Intel and MS do when the competition is weak.

I'm not against Linux, I want them to get the tough criticism they don't just deserve, but need to get anywhere. The kind that caused Mozilla to take their once revolutionary browser that got fat and lethargic and really dig in to turn out the fabulous release that was Quantum.

If you truly loved Linux/Ubuntu and wanted them to succeed you'd be banging on their door saying "Dude, WTF? We're counting on your guys!"

You obviously still don't get it.

When you install Windows 10 from scratch it automatically downloads and installs the best driver for the job as part of it's install process - This is one thing Windows 10 does well in direct comparison to Windows 7. Now if any one of those drivers had the potential to damage a machine's bios, the situation would be in no way any different from what's been reported in relation to Ubuntu - Especially now that Microsoft is trying to make Windows 10 a rolling release and failing dismally as the NT kernel is just not designed for it.

The fact that the kernel is monolithic and contains the necessary drivers alongside the OS kernel doesn't in any way indicate that Ubuntu tests every driver any more than Microsoft does - It's just that Linux in general experiences less issues as it's monolithic kernel is better suited to a rolling release testing/distribution model.

As stated, if you want to play the blame game in some attempt to piss on Linux, Microsoft (in particular Windows 10) holds the crown when it comes to problematic releases - No one on the planet can dispute otherwise.
 
Windows just slowly eats away at your sanity though. Like lead, it's used primarily when there are no other good options.

Or when the inability to learn better alternatives more suited to the job at hand due to outright laziness or manipulation from marketing departments prevails. ;)
 
You obviously still don't get it.

When you install Windows 10 from scratch it automatically downloads and installs the best driver for the job as part of it's install process - This is one thing Windows 10 does well in direct comparison to Windows 7. Now if that driver had the potential to damage a machine's bios, the situation would be in no way any different from what's been reported in relation to Ubuntu - Especially now that Microsoft is trying to make Windows 10 a rolling release and failing dismally as the NT kernel is just not designed for it.

The fact that the kernel is monolithic and contains the necessary drivers alongside the OS kernel doesn't in any way indicate that Ubuntu tests every driver any more than Microsoft does.

As stated, if you want to play the blame game in some attempt to piss on Linux, Microsoft (in particular Windows 10) holds the crown when it comes to problematic releases - No one on the planet can dispute otherwise.

Wow, you're really hurt by this. If you think I'm trying to "piss on Linux" you're the one who doesn't get it.

I've never said that this is impossible on Windows. I criticize lots of MS stupidity (and there's plenty of it, but you'll completely ignore that I said this), but you think anyone that gives anything less than unabashed praise to Linux hates it and loves MS/Windows. Wrong, wrong, and wrong.

I don't even think you get my basic premise. You're too busy being butt hurt and combative, made obvious by all the "but MS does this!" arguments, even though its a weak sauce "they have software problems" when Ubuntu is bricking motherboards. You're grasping at straws trying to compare temporary problems with permanent ones.

Bottom line is more testing should have been done on the part of Ubuntu, especially on major PC vendor hardware like Lenovo.

I've said it before and I'll say it again, the Linux community is its own worst enemy.
 
Wow, you're really hurt by this. If you think I'm trying to "piss on Linux" you're the one who doesn't get it.

I've never said that this is impossible on Windows. I criticize lots of MS stupidity (and there's plenty of it, but you'll completely ignore that I said this), but you think anyone that gives anything less than unabashed praise to Linux hates it and loves MS/Windows. Wrong, wrong, and wrong.

I don't even think you get my basic premise. You're too busy being butt hurt and combative, made obvious by all the "but MS does this!" arguments, even though its a weak sauce "they have software problems" when Ubuntu is bricking motherboards. You're grasping at straws trying to compare temporary problems with permanent ones.

Bottom line is more testing should have been done on the part of Ubuntu, especially on major PC vendor hardware like Lenovo.

I've said it before and I'll say it again, the Linux community is its own worst enemy.

Mate, I don't give a shit! I'm running Ubuntu MATE 16.04 LTS with kernel 4.12.14 just fine, my system is far more stable than any Windows based system before it with far less issues!

Bottom line is: 17.10 is not LTS and issues are to be expected. Just like Windows and their new release model, any release that is not CB is for bleeding edge testing purposes only. Likewise, any release that is not LTS under Linux is 'use at your own risk'.

This is a testing model that you can't seem to grasp.
 
Last edited:
Mate, I don't give a shit! I'm running Ubuntu MATE 16.04 LTS with kernel 4.12.14 just fine, my system is far more stable than any Windows based system before it with far less issues!

Bottom line is: 17.10 is not LTS and issues are to be expected. Just like Windows and their new release model, any release that is not CB is for bleeding edge testing purposes only. Likewise, any release that is not LTS under Linux is 'use at your own risk'.

This is a testing model that you can't seem to grasp.

This!!!
 
The configuration of the kernel is hugely irrelevant and there is every possibility such a scenario could play out under Windows just as effectively as the possibility exists for such a scenario to play out under Linux.

It's a driver issue, all operating systems use drivers for identical hardware.

But is it purely an Intel driver issue? Same basic software as would be released for Windows, and presumably OS X, but only the driver included with this Linux kernel is causing problems. You don't have to be so sensitive to think we view this as an inherent problem with Linux, but if you can't admit when you've made a mistake, you lose credibility. Canonical has admitted they made a mistake, albeit a very bad one, and is trying to rectify the situation. The more important consideration is how to prevent this from happening in the first place.

If you can only see this as "just an Intel" problem (and they sure have had their fair share of big problems the last couple of years) you are setting yourself up to have a similar event happen again.

Now here is a technical question to consider - does this driver only work in the 4.13 kernel, or does it work in other kernel versions, without having the same problem?
 
Is the blame entirely in the right place? It just seems odd that in 2017 you could let the operating system straight up break the bios. Aside from ensuring the BIOS/UEFI specifications are implemented correctly and prevent bricking scenarios, maybe we need to bring back the "backup bios" that was a thing after the CIH/Chernobyl virus.

Intel should put on its Superman cape and find a way to use the Management Engine to fix this.. that'd get them some good will points. :D
 
Back
Top