New Windows updates cause UNSUPPORTED_PROCESSOR blue screens
bleepingcomputer.com
bleepingcomputer.com
Finding issues like this is literally why the preview program exists. Should they have caught this with internal testing? Maybe, we don't have enough information to know.
Still, Microsoft gutted their internal testing, so they hold some blame for disastrous failure, even during preview.
The only people who "buy" Windows are those who build their own gaming machine. Even then, they usually buy a sketchy "volume license" key off ebay.
I bought windows retail, because I actually like the Windows kernel, and despise running pretty much any install of Linux that isn't just a command line to boot daemons and web servers from. I use my computer every single goddamn day, and don't ever want it to randomly break, which Linux has always done to me. The steam deck is the first Linux install that hasn't been utterly useless or shit the bed on me.
Every single problem I have had in Windows has been caused by shitty hardware, or AMDs abysmal drivers.
Enterprises pay a surprising amount of money to Microsoft usually in the form of monthly/annual pr user licenses(to the time of double digit percentages of the end user device budget) and yet no amount of badness seem to make anyone in those Enterprise departments think that they have any alternative and it's the same in the laptop space, where despite having to fork out whats often a double digit percentage of the retail price to Microsoft there is no serious appetite from hardware vendors or consumers to consider that maybe we could have a market where multiple Operating systems coexisted.
And your driver argument is exactly the weirdness here as people will absolutely hold anyone but Microsoft(especially smaller opensource competitors) responsible for bad drivers but with windows it's suddenly an valid excuse for bad experiences with windows.
I'm really happy with where MS is going, I think they have been killing it with vs code, chatgpt integration, and I was quite bullish on the surface pros.
But after having owned one for 1 month, they were still a real disaster in the OS area. I "subscribed" to the preview track, and had to reformat my pc twice. It was a fricking disaster. I'm back on osx now, having sold the surface pro.
Hardware and OS is still very much their weakpoint imo. It's like the whole OS is designed by product owners working on macbook pros (which apparantly isn't far from the truth)
I understand there are workarounds - such as buying a non-personal Windows license, and fiddling with the registry - but that's like selling me a car that'll shut its engine off on the highway and force me to listen to its own advertisements while I'm using a bluetooth device, and assuring me that I can adjust the behavior. If I know how. If I even know that's possible. If I'm willing to pay more.
"I subscribed to test betas and they were unstable. Thanks Bill Gates!"
Like, my brother in Christ, that's why there's the stable version of that OS.
What stability issues are you referring exactly?
There is a difference between having a few apps crash more often, and literally having the offical MS help tell you, the only solution is to reinstall everything.
And unsurprising.
They finally found out what everybody knew since Win 95 days: There is no problem in Windows(TM) that a complete reinstall cannot fix. /s
Are you tracking the pre-release versions of MacOS?
Is it therefore possible that this bug check is being implemented as part of the journey towards actually dropping "support" for these CPUs?
Indeed, it's a preview release, and the Insider program is amazing for getting insight into Windows development and it shouldn't be held up to the same level of journalistic scrutiny as a sanctioned general release, but I can absolutely see why this particular change may have raised some eyebrows.
That seems entirely possible. That being said, I'm not sure I'm going to be upset about them dropping support for a 20-year old CPU built on an architecture that I think is pretty widely accepted as a failure.
Windows 11 doesn’t support CPUs that are 6 years old (Intel 7th gen) that are still ok by today standards
A 7th or 6th gen Intel would run Windows 11 quite well, unlike a Pentium 4 which struggles a great deal with it. Despite that, without workarounds, you cannot install Win 11 on one of these CPUs. You must have an 8th gen or later.
You have to actually opt in and download the preview. Then you'll be part of the preview :-)
These always struck me as an odd middle ground between regular updates and Insider
fwiw my very 'unsupported' 11 Dell box seems okay with this update installed. All the Reddit folks getting the error are on Intel MSI boards
Yeah, but making this info clear in the title doesn't drive the pitchfork driven clickbait FUD that seems to be invading HN lately.
Nearly half of Windows 11 related topics I read here lately were mistakenly titled or factually wrong, and yet full of rage comments of people who haven't even read the article, or don't even use Windows, but came by to tell everyone they use Arch or MacOS.
This isn't "lately", HN has been this way since at least 2017 when I started reading. If you didn't notice it, that's on you. Almost nobody has ever read the articles on HN.
I’ve been much happier with Debian 12 and KDE for my primary workstation.
I can’t think of a compelling to use windows outside of gaming.
My guess is that a lot of people are using the crappy Home edition.
What makes Home crappy? Other than the account thing, obviously. You can't join a domain with it either, but then most people don't have a domain at home. There's some other disabled features, like Hyper-V, but why should people have to pay for a higher tier Windows to use a local account if they don't need the other features?
you also don't have Bitlocker
no Group Policy
no Active Directory
- RDP
- Windows Sandbox (super neat and useful and under-appreciated!)
- BitLocker
- Active Directory
- Local accounts, apparently
- Various other little things and some local policies.
I use all of the above Pro-only stuff.
If my use of "crappy" offended you, please substitute the term "less-featureful".
Other than that, I didn't say what Microsoft should or shouldn't do for home users. That's on you.
It's my personal belief that multiple Windows 11 editions are insulting, and it should just be a single OS edition, with all features available, across the board.
I tried to domain join the third device to see what that flow is like. It asks:
1. Personal use / Work or School
2. You are asked to login to Azure AD. You have to click a small (10pt?) link for "Sign-in options." with no indication that will let you join to the domain.
3. Pick "Domain Join instead"
4. It asks you to setup a local account ("Even better, use an online account" :-)
5. _You finish setup without the device actually being joined to the domain!?_
6a. You now have to login (click)
6b. and go to the system settings (click)
6c. go to accounts (click, ad for M365),
6d. go to "Connect to work or school" (click),
6e. click another "Connect" button with no indication that it will join it to a domain,,
6f. Click another small link for "Join this device to a local Active Directory domain" to domain join the PC.
So, I'm sorry, but if you think the Pro SKU is better in this regard you are just incorrect. There is no excuse for these dark patterns, this is user-hostile behavior designed to convert you to being recurring SaaS revenue and not retail revenue.
As an added bonus I just tried to create a Windows 11 VM in hyper-v to confirm my memory, and _Microsoft Hyper-V_ doesn't even correctly configure itself to allow Windows installation to proceed by default. (Secure Boot enabled, but no vTPM.) It then proceeded to download updates to the installer and reboot without a prompt or any option to skip it. Microsoft _clearly does not give a shit_ about the OOBE if you are not using their cloud services.
And thanks to Steam deck/proton, that gap is also decreasing significantly.
Windows definitely has regressed in terms of user experience and updates/upgrades and online integration are two big pain points that really make me miss Linux (Ubuntu or Arch+lxde, if you want names) every time I'm there.
My nightmare, before the account shenanigans, was finding a mouse for my HP laptop, whose trackpad didn't work OOB. Then, once I'd managed to go through the installation, the account thing was a PITA because the Intel wifi card wasn't supported, and this laptop doesn't have a wired network port. I was actually going to use an MS account (this is a work computer, and they're all in on MS). But I couldn't set it up without jumping through the hoops of creating a local account first and then grabbing the wifi drivers by booting Linux.
This happened mid-2023 with a 2020 model laptop. Also, HP recommends Windows. Arch Linux worked fully out of the box.
Or maybe it was Intel who did a bazillion of cards with a minor diffs and/or didn't ask MS to add this flavour to the default driver set, shipped with Windows install media?
Oh and getting the installer to recognize the drivers was much harder than the usual "stick the files on a thumb drive, plug to laptop, tell the installer". I had to generate a custom Windows ISO with the drivers preinstalled, for which the process is described by a single indian dude on a website very sparsely indexed by google.
Would this have tripped on some obscure device specific to Dell laptops, sure, I would understand. (like their blimey fingerprint reader). But this is the standard NVMe controller on this generation of intel CPUs. The drivers are common for _most intel platforms_: the Intel Rapid Storage Technology.
As somebody else pointed out, Linux fares much better in this space. All devices are properly recognized out of the box, including this bog-standard NVMe controller.
This is a SATA AHCI/RAID controller. It has nothing to do with NVMe.
Also iRST was (and looks like still is) the most troublesome driver for Windows since it's introduction: there are bazillion of variants and vendor sub-variants with different VEN/DEV IDs, so it's CAN'T be identified by some generic driver!
And yes, it was (and from your description - is) the one stupid driver which couldn't be loaded at the running install, because for somegodfuckingreason it did shenanigans when loaded 'online' and Windows couldn't use it as the install destination.
This is not (and were not) a problem with any other SATA and SAS controller I ever encountered, from cheap-ass Silicon Image ones to $2000 LSI/HPE ones. Only Intel, only ICH/R.
There is no such thing as generic driver for a hundreds and thousands of variants of some vendor specific device if the vendor did not provided such driver.
Your loved Linux works there only because the kernel module which supports Intel ICH\R chipset (which provides iRST) has those bazillion of PCI VEN/DEV IDs assigned to it and this is why it works 'out of the box' on Linux kernel. Intel NEVER provided such driver for Windows and never would.
Eg: there is a thing called `HPE Smart Array SR Gen10`. It's a series of RAID controllers made by HPE.
Except `HPE Smart Array S100i SR Gen10 SW RAID`, aka S-class. Notice `SW` there? That's just a regular Intel ICH/R. You can manually install Intel ICH/iRST driver on that `HPE controller` and it would work, because it's just a regular Intel ICH/R. You know what wouldn't happen? Windows wouldn't install `the default` Intel ICH driver on it, because:
there is no such thing as `the default SATA` driver
there is no such thing as `the default Intel SATA` driver
all `Intel SATA` drivers built-in in Windows install media do have the PCI VEN/DEV IDs listed for INTEL (VEN8086), not HP/E, DELL, Lenovo or whoever.[0]
You can shit on Microsoft and Windows all you want, but the problem here is strictly with the vendor, Intel.
As soon as Intel would provide MS with the `default Intel SATA` driver what would have all VEN/DEV combos for all their licensed and white-labeled controllers - you would be able to install Windows client SKUs on all these controllers without slipstreaming the drivers in the install media.
Source: I know things, I lost too many hours of my life wrestlning with Intel ICH/R
[0] ffs, just open the .inf you slipstreamed and look for IDs there.
Yet it was somehow required to use the NVMe drive in this machine. This is braindead.
> This is not (and were not) a problem with any other SATA and SAS controller I ever encountered, from cheap-ass Silicon Image ones to $2000 LSI/HPE ones. Only Intel, only ICH/R.
Learnt this the hard way on a previous motherboard I had with two sata controller, the native Intel one, and an additional Marvell one. Sanity would encourage to use the intel one, except for all those friggin issues.
As for Linux, I know why it works :) But it always amazes me that this FOSS project can ship all those IDs, quirks and all, but not one of the largest company on the planet. What a time to live where Linux has less driver issues than Windows.
Thank you for all the details, I did not know iRST was such a hot mess. It's not that new, one would have guessed shit would have been figured out since. But no. Oh well...
Btw sorry for the tone, I thought by "vendor" you meant Dell and not Intel. They produce some of the worst driver I ever had to work with. Don't get me started on their drivers for the E810 line of datacenter network adapters.
The issue here is that MS plays the smart guy and installs his own buggy stuff, making it hard for the end user to use vendor supplied drivers, even reinstalling its own stuff with Windows update.
Or maybe the whole point of not supporting the touchpad was to weed off the faint of heart and send them off to Linux?
If they were 'pretty standard' then they would be in the default list of drivers shipped with install media.
If they weren't in the install media then:
your media is older than the drivers
your device isn't 'standard'
device vendor doesn't give a shit if you can reinstall Windows and get the working WiFi out of the box
> , I don't see why windows couldn't ship with drivers for that
Because Intel does not give a fuck. It's that simple and it is borderline ridiculous how linux fanboys can't accept what Windows has a completely different driver model compared to built-in-the-kernel way of Linux.
And again, where did you got your install media for Windows? Directly from https://www.microsoft.com/software-download/windows10 ? Somewhere else?
> Or maybe the whole point of not supporting the touchpad
Can you ask Synaptic (or whoever made that touchpad) why there is no drivers for their touchpad in the default Windows install media or why it's not just a simple HID pointing device? Or again that's Nadella's fault or whoever at the lead now?
If you don't know why Linux-way doesn't work in Windows and you want to know - you can look at my sibling comment.
> Your combative behavior here is entirely unwarranted
What is really unwarranted is 'but it works in Linux1111' without thinking.
> If the drivers work fine in a standard Linux install
Ah, yes, work so good you need to disable RST completely to install Linux: https://askubuntu.com/questions/1233623/workaround-to-instal...
Somewhere else: https://www.microsoft.com/fr-fr/software-download/windows11 This was Windows 11, downloaded in June 2023 or something like that. I always create a new USB drive when I reinstall.
> your media is older than the drivers
No it's not, see below.
> Because Intel does not give a fuck.
Fair enough. Then how come windows magically knows to install all those drivers, Wi-Fi and touchpad, if I plug in a USB network adapter, and it only connects to Windows update? This has worked ever since I bought the PC three years ago. So no, the devices "aren't brand new", intel didn't "just send them to MS yesterday" or "not at all". And if I manually install the drivers out of band by downloading them from intel's website, windows will still install its own over them, even when they are older.
> Somewhere else
Heh.
> Then how come windows magically knows to install all those drivers, Wi-Fi and touchpad, if I plug in a USB network adapter, and it only connects to Windows update
Because it asks WU about all PCI\VEN&DEV&SUBSYS&REV found in the system and WU responds with the specific drivers it has for these devices.
Remember, there is no 'generic' drivers in Windows, like in Linux.
> So no, the devices "aren't brand new", intel didn't "just send them to MS yesterday" or "not at all".
Well, yes.
Now the question why these aren't included in the OS. Don't have Win11 at hand (and probably wouldn't at least for some time, lol).
Back in the day MS gathered info about the most common devices at the market, to include them in the OS distro. So with a wireless chipset made in '19 the question is who to blame, vendor or OS mfg. Considering how hostile Intel to the customers I wouldn't be surprised it's them. BTW you can open Device Manager pretend to replace the driver, select 'I would select manually' and see all the drivers OS has. On your running system you would see those you installed too, but on a clean system you would see only bundled ones.
On my notebook I have a Realtek NIC, so the Intel drivers are the ones what was bundled with Win10 install media (and maaaybe some updates?) and I have there:
Intel(R) Wi-Fi 6 AX101
Intel(R) Wi-Fi 6 AX200 160MHz
Intel(R) Wi-Fi 6 AX201 160MHz
> I always create a new USB drive when I reinstall
Do you just download an .ISO or do you use the 'Installation Assistant'? There are some shenanigans with that tool.
> And if I manually install the drivers out of band by downloading them from intel's website, windows will still install its own over them, even when they are older
Or something had rotten in Seattle or somehow Windows bundled drivers are more specific ones than the ones from the vendor site.
I heard about that but I don't remember ever witnessing this myself. If anything, I needed to that manually, because Intel drivers for 'Intel(R) Wireless-N 7260' are so shitty what the ones bundled with Windows (with Intel copyright from 2011) are way more stable (but not stable enough, sigh) than ANY later revisions from the Intel site.
ADD:
Okay, for the sake of it: I've got my hands on Win11 and it DOES have some drivers for AX201: https://imgur.com/a/ozob6kM
So the question about the PCI ID of your card.
I'm pretty sure Linux doesn't have "generic" wifi cards. See all the issues people have getting to work brand-new models. I had already seen the AX201 in a separate machine for some time before buying this laptop, so it wasn't brand-new when I first tried installing Windows. Didn't check the PCI IDs to confirm it was precisely the same, though.
I don't have that specific PC on hand to check the IDs right now.
For installing, I create the USB drive with the "media creation tool" I download right before from that site. I only use the direct ISO download to install on VMs, since my usual approach of writing it with dd from linux doesn't work for some reason.
I've noticed the driver reinstall issue because at one point, there was some kind of issue with my external monitor which worked with the Intel drivers but not with the Windows Update ones, and every so often the display output would break again until I manually reinstalled the ones from the Intel site. The Windows Update drivers have been fine for a few months now. Wi-Fi never had any issue once it got going after the first reboot and Windows update run.
Fun fact: if I look for drivers, in the "recommended drivers" list there are two drivers by Intel and one by Microsoft. https://i.imgur.com/jpghrgd.png
- Office
- Adobe
- Ableton
- Visual Studio
That being said, I'm terribly frustrated with the direction Microsoft is taking Windows and I hope to stay off of 11 for as long as I can.
Too bad that audio is total joke on Linux. There's very good software, like BitWig and Reaper, but the audio subsystem is total mess doing random stuff all the time. PipeWire seems promising but there are still problems. Maybe one day.
https://learn.microsoft.com/en-us/windows-hardware/drivers/a...
>- Adobe
>- Ableton
>- Visual Studio
Find alternative software. If those companies aren't compatible on Linux, use one that is.
Make sure you send them an email explaining why you are canceling your subscription.
> - Office
In a corporate env maybe. Libreoffice is enough for home use.
> - Adobe
Only reader. For special pdfs.
> - Ableton
Never heard about it.
> - Visual Studio
What's wrong with vim ?
- Visual Studio
Why would someone use both in a corporate env ?
They all gave up on that trainwreck of an OS, and either their final Mac died and they replaced it with an even better laptop by a different vendor (Framework, et al), or they still own a Mac but it now runs Windows or Linux (and when that one dies, it will be their last, and they will become post-Mac, too). I know exceptionally few tech people with M1, and they bought it knowing they had to suffer the rough edges of Asahi; I do not know a single tech person who bought an M1 and left OSX on it.
The number of Macs in existence only goes down, not up; doubly so in the tech world.
I don't get why theres a few people on HN that persist in the FUD+FOMO combo that OSX is gonna make a comeback and win because "its a Unix!" and "Apple takes security seriously!" and "Safari is world's best browser!" and "Swift is a great language to code in, even if it doesn't work on other OSes!" and "I use XCode unironically!"
Just stop already, we've all heard it. Repeating it doesn't make it true.
I've gone so far as to deploy M1 Mini's in factory environments to run production processes for very large scale printing operations (think Amazon, Walgreens, tons of orders daily) and guess what I haven't been called in over a year because it is all rock solid.
Better news, installing Windows on Intel Macs improves battery life under common use cases.
Best news, Linux on that same Mac is even longer.
Hilariously, as much as Chromebooks suck, there are many Chromebooks that have half the Wh of M1/M2 99Wh Macbooks, cost half as much, and meet or exceed battery life under common tasks (implying they're both twice as efficient per watt, and twice as efficient per dollar). These Chromebooks can run Linux, and aren't stuck with ChromeOS.... so if you're really into ultra-thin laptops with excessive battery life, well, they're an option.
I also semi-recently saw a review for the LG Gram 17, an Intel laptop that has 3/4th the Wh of Macbooks, yet has the same battery life in common tasks, and a 17 inch (!) screen; costs about the same as a Macbook. So, even with the Intel CPU holding it back, its keeping up with those Macs.
So yeah, Apple hasn't kept up with the battery life game at all, another myth cranked out by the FUD+FOMO Apple PR machine.
You should realize “I don’t know people who do X” is a very weak argument.
OSS is alive and well in the enterprise!
Only Samba implements part of the above. OSS has a way to go, yet.
To be clear, AD never has, never was, and is incapable of being “just LDAP”.
For instance shadow rights / MIM / PAM does not to my knowledge have an OSS equivalent.
Windows Hello for Business is the only moderately secure take on biometric auth into kerberos that I have heard of.
LAPS is the only secure, native LDAP take I have heard for managing root passwords, which fills a big compliance need for a lot of orgs.
And there is quite a lot about GPO that works wonderfully when paired with sssd, e.g. access control.
Unless you absolutely must play ea games then even that can be done on linux rather nicely.
Fortunately, the advancements in Proton and better cross-platform support for a lot of software has let me use Linux on my gaming computer and notebook. There are a few games which I can't play, but honestly there's so many options available that I don't miss them much. I don't think I'll ever completely eliminate Windows from my life, I think around 2021 was The Year Of Linux On The Desktop for me.
>The company also added that the problematic cumulative updates "might automatically uninstall to allow Windows to start up as expected."
"Might"? What a weaselly way to put it. Either Windows will, or it won't. I realise there's tens of millions of LoC but if we're talking about an update that crashes machines shouldn't you know what the failure mode is before talking to the press?
I myself have used "might" when speaking to customers when the criteria by which we would differentiate is something that isn't necessarily accessible to the audience.
The term might in marketing department vetted communications often cover over "the techies droned out for 10 minutes using words we did not understand" and few large brands feel that comfortable letting anyone not under some kind of confidentiality agreement talk to anyone in the technical teams.
https://learn.microsoft.com/en-us/troubleshoot/windows-clien...
Here's last month's article from the same author giving an example of how KIR can be used to mitigate a video showstopper at the time:
https://www.bleepingcomputer.com/news/microsoft/microsoft-fi...
My follow-up consisted of following the journalist back, and that might give an idea about what he had in mind.
Thank you!
> To do that, you will have to go through the following steps:
> Launch Feedback Hub by opening the Start menu and typing "Feedback hub", or pressing the Windows key + F
Something seems off here.
But _you_ don't install updates. The thing autoupdates itself.
These execs and managers should be criminally charged
Testing is expensive. That's what users are for.
Just look at the SW made by MS and Google. It looks that they work only based on change requests. If the bugs are not rewritten, they are fixed only when there is a change request.
The default new tab on bing is a bunch of trashy news. It’s just icky.
That feature is sending an insanely large amount of personal data to Microsoft, and it's horribly irresponsible to include this feature as it is.
Then I turned to the dark side and bought a used cheese grater Mac Pro (dual CPU Xeon) for a few years before and I always found OSX weird - dropping to command line on occasion to get things done and definitely not very friendly but tolerated it for the sake of the hardware.
When I started the SAP part of my career, I got a Windows 7 laptop for SAPGUI and realized it was nicer - the Mac Pro went to a new home after I had to hack the boot firmware to install Sierra and realized the bells was tolling on the hardware.
Could not afford a Mac Pro replacement so I bought a Dell - A decent specced Mac Studio M2 (64GB) would cost me more than my car is worth.
I don't understand what the point of preview updates are if MS says they're not beta updates.
Is it just a place for say corporate IT to test to decide if they want to do the next monthly update?
The list of affected operating systems supposedly includes Windows 10, version 22H2.
The workaround for that isn't some weird hack, it's an explicit registry value that they made.
And I'm not talking about old crap like Pentium 4s or something. I'm talking about ones that are barely 5yrs old.
Presumably, for Windows 12, remote attestation capability (Pluton) will be required.
This already a thing with Windows 10 in S Mode. So definitely not that.
There are quite a few CPUs that implement the necessary TPM crap and aren't supported. So not that. There are several new CPU features they want, but half or 3/4s of the CPUs on the supported list don't have them. So not that either.
Only valid reason left is that's arbitrary and not based on anything logical.
We besides the user, even the CPU has to be on drugs to run Windows 11, something is teribly wrong with this world. /s