Windows operating system security update block for some AMD based devices | Hacker News Reader

Windows operating system security update block for some AMD based devices

support.microsoft.com

41 pointsrdtsc33 comments

The important part: "After investigating, Microsoft has determined that some AMD chipsets do not conform to the documentation previously provided to Microsoft to develop the Windows operating system mitigations to protect against the chipset vulnerabilities known as Spectre and Meltdown."

Which shows, that no matter how certain you are that your code matches the specifications, it still pays to test it.

Looking for which hardware is affected, I only found a statement form AMD [0] tldr, Ryzen CPUs are fine

> "AMD is aware of an issue with some older generation processors following installation of a Microsoft security update that was published over the weekend. AMD and Microsoft have been working on an update to resolve the issue and expect it to begin rolling out again for these impacted shortly."

[0] http://www.techradar.com/news/meltdown-and-spectre-windows-p...

I'm surprised by this because I know that Microsoft has a test lab with various hardware configurations and software to help verify that updates don't break existing software.

Doesn't Microsoft have the resources to run automated tests on enough hardware configurations to have caught this?

How far back do the machines in Microsoft testing lab go?

I heard the chips having the problems were Athlons from 2005-2006, which are 12-13 years old now. Apple doesn't have to worry about this kind of stuff when they patch OS X because they are able to drop support for anything they want after a certain time.

They do, but there’s a huge number of possible configurations of both hardware and software to choose from and they’re always going to miss some.

It would be nice to know which older CPUs are affected. Do I have to worry about my Fam 10h processors getting bricked, or just my K-7 and K-8 systems?

Related, I just shared that Microsoft is going to make the mitigations for Windows Server optional https://news.ycombinator.com/item?id=16108356

This should make it very clear that other vendors downplaying the performance impact were misleading users.

yep they even have a KB article where they said where the fix is NECESSARY and where it might be ok to not install it:

> https://support.microsoft.com/en-us/help/4072698/windows-ser...

Part: "Enabling protections on the server"

it even has a registry key to enable/disable it. Basically they say that:

> Your server is at increased risk if it is in one of the following categories:

> Hyper-V hosts

> Remote Desktop Services Hosts (RDSH)

> For physical hosts or virtual machines that are running untrusted code such as containers or

> untrusted extensions for database, untrusted web content or workloads that run code that is

> provided from external sources.

Breaking windows is not bricking a machine, can we please stop abusing this term. Brick means irrevocably render an electronic device inoperative (making it as useful as a brick), short of something like JTAG or soldering.

This is especially important surrounding hardware bug concerns like this, because its at least conceivable that interactions with hardware bugs really could brick machines.

EDIT: for posterity post title is "Windows AMD Spectre patches brick some machines" atm.

I do agree with you, but at the same time, a non-booting OS does essentially mean a heavy-but-useless device for the vast majority of computer users. They'll have to spend their own money taking it in to get the OS reinstalled, and a lot more if they want to try to recover the data still on the file system.

Or use some other computer and a USB key fob to create a bootable USB disk that can fix it, and reading the directions on how to boot from it on their phone while actually at the computer in question. Of have their kids do as much. We're probably a few days at most from something like that being put together (possibly by MS or AMD themselves) if it hasn't already.

It's not 2005, there's a much higher population of people that have at least enough experience to follow directions such as that.

I used to repair smartphones and this was a common phrase when someone would try and install a custom ROM. They would do it wrong and essentially brink the phone.

It's weird for me to hear it in software circles, since I've always thought of software as being more malleable. Correct a few lines of code here or there and you're up and running again.

On smartphones, if you do the custom ROM installation badly, there's a chance you will wipe SPL, meaning no way to boot anymore, until you reprogram your flash in external programmer.

This is not a case on the PCs; updating or reinstalling operating system will not affect your BIOS or UEFI ability to boot anything in the future[1]

[1] with exceptions like Ubuntu 17.10 Intel SPI bug. But hey, exceptions confirm the rule ;).

Fair point, though for a typical end user this might be indistinguishable

That's fair, I updated the title. I though bricking was a shorter title but your point is true.

You appear to be talking about soft brick vs hard brick? neither hard brick or soft brick is physically broken though...

Brick means "to render a device permanently inoperable." There is no hard or soft involved. Bricked is bricked. If you can recover the device without replacing hardware or components, you didn't brick it, you just fucked it up.

I think it's fair to say that different people have different thresholds for "bricked" depending on their awareness, willingness, and ability with regard to various repair techniques that may be to hand. To my mother, a virus that causes her laptop to bluescreen upon boot is bricked. There's nothing she can do about that. To me, you gotta do something like wreck the firmware, and even that's not quite permanent, as I am not afraid of soldering tools.

> I think it's fair to say that different people have different thresholds for "bricked" depending on their awareness, willingness, and ability with regard to various repair techniques that may be to hand.

I think it's fair to say that people misuse terms through ignorance often, but that doesn't mean we should always just say "okay, let's extend the definition to include that too". It's useful to resist the erosion of terms until it becomes unuseful to do so. It often gets you marked as a pedant in a derogatory way, but it serves a very real and useful purpose, which is the preservation of definition which helps clear communication.

It's a losing battle, but it's a worthwhile fight.

Your grandma is not going to say the machine is “bricked”. She will say it’s not working.

Bricked is a technical term used by techies, and it has one specific meaning.

Bricked is bricked.

to me, soft bricked means that you have to do something out of the ordinary to fix it, something that most people would not know how to do, even if it's just a software problem (maybe it's a new problem that had no known solutions yet)... http://unbrick.itcse.com/soft-brick-vs-hard-bricked-vs-broke...

Common use of the term predates Android phones by at least a decade, so I'm going to resist this new "soft-brick" designation as both misleading and unuseful.

Having your phone go into a boot loop or corrupting your MBR has historically not been considered "bricked", but just having a corrupted install. It's generally not that hard to fix or find someone to fix, which is why it wouldn't be considered bricked.

Getting a device to a state where it's unrecoverable, or at least requires custom equipment and lots of knowledge about it to fix (and I'm not talking USB key fob and how to reset an MBR, or even a serial port connection, I'm talking JTAG connectors).

People familiar with putting experimental linux hobbyist distros on devices with no support, such as the Sharp Zaurus or wireless routers circa 2000 know what this means.

JTAG connector is not that much different from USB at at stage... you just have to remove the case to access it or solder a connector to existing pins? I would call that hard brick... and if JTAG doesn't work, then it might be considered broken by some and need electronic part(s) replaced

Are you comparing plugging in a USB fob to soldering a connector to pins, and that's what you're referring to as "not that much different"?

I would consider any time you have to solder a custom connector onto the device to fix it as being bricked. The equipment, knowledge and capability to do that are very rare.

but it's not physically broken... it's just a different way of connecting to it... so yes I would call that hard bricked, but not broken, since you don't need to replace components

> since you don't need to replace components

I think the fact that you have to manually add a component for connection by soldering, even if you can remove it later, makes that point somewhat moot.

Let's look at it from another point of view. The combination of needed skills is rare, and if you had to pay someone to do this, it would likely cost to fix it is more than the cost of a new device (since they would have to research how to fix and apply that knowledge. If there was an existing market you could easily tap it may be cheaper). If the cost to fix is more than the cost to replace, it's a brick.

I get your point, but if soldering is the only problem, just get a spring loaded solderless connector for JTAG pads? then it is about the same as USB for many more people?

I just think you'll have to go quite a bit farther to convince me that using a USB key fob in an existing USB port, which I can probably find and buy at a 7-11 down the street, is the same as acquiring a spring loaded JTAG pad or soldering a JTAG connection in place to a device I have opened to expose the circuity, and then configuring the software required to interface with it.

In one case, I plug a USB stick into a working computer, which virtually everyone has already done at some point before, run a utility included in Windows[1], and maybe download and drag some files around to copy them. In the other I do a whole bunch of stuff that isn't common to any neophyte or even most advanced computer users.

1: https://www.google.com/search?q=how+to+make+a+bootable+usb

"Bricked" refers to that your device can now be used as a brick and nothing else... at least that's how I always understood it.

I posted these two images: https://imgur.com/a/QANol https://imgur.com/a/h3MhG Notice the variables they set up that RtlInterlockedPopEntrySList don't use.

I had an AMD machine in on Monday that I installed in 2009, was chugging along fine until this update. Now has a stop error on boot.

New machine time - which actually gives Microsoft money in reward for their mistake but is the best course of action for the client.

It’s just expense for the the client and inconvenience for all involved.

Unbootable, they say. So I think they actually mean it!

Reply on news.ycombinator.com