Slashdot Mirror
Windows 8.1 Update Crippling PCs With BSOD, Microsoft Suggests You Roll Back
MojoKid writes Right on schedule, Microsoft rolled-out an onslaught of patches for its "Patch Tuesday" last week, and despite the fact that it wasn't the true "Update 2" for Windows 8.1 many of us were hoping for, updates are generally worth snatching up. Since the patch rollout, it's been discovered that four individual updates are causing random BSoD issues for its users, with KB2982791, a kernel-mode related driver, being the biggest culprit. Because of the bug's severity, Microsoft is recommending that anyone who updated go and uninstall a couple of the specific updates, or rollback using Windows Restore. You can uninstall these updates in much the same way you uninstall any app; the difference is that once you're in the "Programs and Features" section, you'll need to click on "View installed updates" on the left. While it's mostly recommended that you uninstall 2982791, you may wish to uninstall the others as well, just in case.
For some it was just a plain black screen with no errors displayed (win 8.1 x64) , same fixes though:
http://answers.microsoft.com/e...
I'm using Windows 7 and I was affected by this. I can't fathom the depths of ineptitude required to release such an update, to be perfectly honest.
Microsoft announced they will change from calling their software 'software' to 'cute puppy' in bid to distance themselves from their reputation as they have now run out of feet to shoot and stick in their mouth.
I'm going to have to reinstall Windows 7, and I can't tell you how excited I am about this.
Chicago pretty much did (ignoring driver flaws) ... until they added backwards compatibility.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
Here is the better workaround for the problem from Microsoft: https://support.microsoft.com/kb/2982791
I've learned the hard way over the years. Never let Windows Update install a driver of any kind. Ever.
I've had them blow out network cards, video cards, sound cards, and low level on-board devices. I've had them completely bork systems to the point where they were unbootable.
Go to the vendor and get the official updates.
I don't know how they do it, but Windows Update perpetually mis-identifies hardware and installs the wrong drivers, delivers broken drivers, and otherwise screws up when it comes to drivers. Yet the official vendor's drivers (such as Intel) work just fine.
Go figure. One would think Microsoft is just redeploying those same drivers, but years of being burned have taught me that's not the case.
I do not fail; I succeed at finding out what does not work.
I thought for sure that I had missed the [Satire] tag
-- Sig under construction...
One of the bits of logic used for recent layoff and reorgs has been something like 'component/security/etc testing had become so mature at Microsoft (!) and ingrained into normal dev processes, that such a large population of SDETs (testers) across OS and key office products is unecessary.' Just chew on that for a second, and ponder how intensely stupid that seems.
But nevermind my opinion; I guess we're getting some at-scale empirical testing of whether getting rid of testers en masse was a good idea.
I think not...(*poof*)
I hadn't realised it was an update which caused the error, so when I finally resorted to system restore it just auto-updated immediately and broke again. At which point a second System Restore decided it would fail to modify a file and thus refused to work. Four hours later, I had to format to get Windows back.
One thing I learned: Disable fast boot, if it's enabled, on your Windows machine (powercfg -h off will disable hibernation entirely). Apparently a Ubuntu boot dvd cannot mount an NTFS partition with write enabled if a hiberfile.sys is present (apparently windows leaves its mounts active and stored in said file, so modifying the file system would cause problems). You can mount it as read-only and get your data, but if you run into a problem that could be fixed by modifying or deleting files then you're out of luck if fast boot is enabled and the action required cannot be performed from the windows boot environment (you can't disable fast boot from it, the required services aren't loaded).
Your startup time will be a little slower, but you might just save that time if something ever goes wrong with your Windows install and system restore fails.
One thing I have learned over the course of MS OS's lives, is to NEVER update the computer at within a week of the updates being released. MS had a nice reputation for putting out crappy patches every now and then.
You are the product tester and you get to pay for it. So be smart, let the stupid people get the BSOD's so you don't have to.
Be seeing you...
Since some of the updates were for security fixes, this gives hackers time to analyze and reverse engineer the original fault, then use it against systems before there is a fix available.
The real "Libtards" are the Libertarians!
It's been M$ since the 1970s, back when Microsoft was a publisher of BASIC interpreters and names of string variables in BASIC ended with $.
http://support.microsoft.com/kb/2982791
Reference Known issue number 3.
The linked report is rather dull, lacking all of the OMG DEAD COMPUTERS EVERYWHERE, aspect.
It does confirm the problem, and states downloads were removed, followed by what appears to be (only glanced at and did not actually read) detailed instructions for 7/8 to remove the updates if you were negatively affected.
This is why I always wait one week and do my updates manually on the advice of a wizened old computer guru I knew years ago.
Donald Trump, on a crusade to make Nixon look respectable
The Windows Serviceability team (dealing with updates) was decimated in the middle of last month, losing about 30% of their testers. This outcome is not surprising at all. Expect things to get much worse soon.
Since most people on this planet are not tech-savvy and don't read sites like /., I wonder what percentage of Windows 8 users will actually do the things Microsoft recommends.
-- Cheers!
One of the Window 7/8 glitches basically went like this. I changed the kernel mode drivers and now I'm going to reboot. Hey, it looks like a glorified temp file called fntcache.dat that's a cache for font-related stuff is corrupt. Should I keep loading the OS? OH HELL NO! Fuck everything! I quit! *blue screens*
Mega case of the Mondays. Seriously, who coded it to crash the entire OS if a font cache seems to be a bit off instead of regenerating or renaming it?
This is a perfect example of why Microsoft should go back to doing Service Packs and not these seemingly random "feature updates" that have become the norm with Windows 8.x and Office 2013 (non-MSI / "click to install"). There's no standard codebase anymore and feature updates are just being installed willy-nilly, with no real support window for delayed installations. (At least with a SP, you had a year to test & work around a problem before MS pulled the support plug). This is another reason why companies don't want Win8.x--kernel-level updates with only a few days warning. (Articles were still talking about "Windows 8.1 Update 2" as recently as 2 weeks prior to August's Patch Tuesday). I'd hate to be an NT administrator fretting over all my 2012R2 installations right now.
Instead of getting a SP for Windows 8, we now have 8.1. Instead of getting SPs for Windows 8.1, we now have 8.1 Update 1 and 8.1 August Update. We have updates that come through the "Store" app. This is one of the reasons (granted, not the primary one) why the uptake of Windows 8.x is now slower than Vista's uptake some ~2 years post-RTM, and why Windows 7 is gaining market share, at the expense of XP and Vista. Companies don't want this model and the headaches that go along with it.
So, for Win9, just go back to a Service Pack model and make everybody happy. Yes, SPs cost a lot of money to put out, and yes MS ends up looking old-school, but the rigor with testing is (presumed to be) significantly higher than some rushed, "little" update. Windows 8.x is broken, and Microsoft keeps pitching a newer, faster cycle of feature updates, but this just proves they are incapable of properly handling such a model... Microsoft: you are not Apple, and you don't have to try to emulate them.
As for myself, so far my two Win8.1 installations (one x86, one x64) and one of 2012R2 in a VM are not showing problems from these updates... But I have only myself to blame for not waiting a few extra days. Of course, now MS will have to come up with an out-of-band fix (with even less testing) within the next ~3 weeks or will have to have 2 sets of patches for September's Patch Tuesday--one for those who haven't uninstalled these updates and one for those who have. Pure stupidity...
Windows 3.1x calc: 3.11 - 3.10 = 0.00