[FIXED] Windows 7 hangs/freezes on reboot after applying Meltdown/Spectre update (KB4056894, January 2018 Monthly Rollup)

This is the exact issue I'm having with restart/reboot.

Posted "RESTART no longer functions normally"

It appears it is related to the monthly update, still waiting for a response.

When can your topic be found?

Edit: found it: RESTART no longer functions normally (Windows 7)

I have exactly the same problem on multiple computers from multiple vendors: Dell Optiplex 790, Optiplex 390, and Lenovo M82s. All running Win7 x64, and exhibiting the problem whereby sometimes the rebooting process will hang just before the PC actually turns off, leaving it in this state:

* Monitor is inactive

* Keyboard has no USB power

* Power light and Power supply fans remain on

Reboots work fine in Safe Mode.

Uninstalling KB4056894 and KB4056897 resolve the problem. Installing the February preview brings the problem back.

On identically-imaged PCs, some seem to fail rarely if ever, and one fails almost every time.

I tried a great many experiments, including updating BIOS and drivers, uninstalling antivirus, Deep Freeze (interestingly, only computers that we have configured with Deep Freeze are failing, although Deep Freeze machines need many more reboots than regular PCs), leaving the domain and joining a different domain and a workgroup, selective startups, etc. etc.

Ultimately, I found one setting change in the BIOS that seemed to do the trick. I disabled C-State Control on an Optiplex 390, and now this one PC seems to reboot properly (at least it did 20 times ina  row). I will be testing additional PCs with this workaround soon, but wanted to float this to see if anyone else can confirm that this improves matters. Or if there's a better workaround.

@ FPL Guy,

I checked my C-State settings on my ASUS P8P67-M motherboard and they were set to the default settings in the ASUS manual. 

C1E [enabled]

C3 Report [disabled]

C6 Report [enabled]

Playing around by setting different combinations, I found the C1E enabled/disabled did not matter, so left it enabled.  However, by setting C6 Report to DISABLED, I have now been able to perform a normal restart/reboot from Windows 7.  Tried at least half dozen times and all good so far.

Thanks for investigating this and putting me on the right path to correct this issue.  Microsoft has caused me so many lost hours of troubleshooting problems they inject with updates you wonder if they have any quality control.

@Ben1907 ,

Thanks for verifying that the C-state workaround works for your system.

I can also confirm that C6 seems to be the problem. While my Optiplexes don't offer any granularity in their BIOS (C-States either all enabled, or all disabled), my Lenovo M82s have three settings: C1, C1+C3, and C1+C3+C6. My early tests last Friday suggested that the only state that didn't work was C6. So if I leave it as C1+C3, the system seems to reboot fine.

You used the phrase "correct this issue."  It's not so much a correction as a crutch. I measured the power consumption change in my Lenovos, and in my application, the system uses 3W more when C6 is disabled.  Haven't measured my Optiplexes yet, but I suspect the difference is more substantial, which is unfortunate. With my M82s, there was a full 10W difference at idle between C1 and C1+C3+C6.

It's quite mysterious why so few people seem to have this reboot problem. It has the feel of some sort of race condition. Half of my PCs seem to have no trouble whatsoever. Are you running anything unusual on your PCs? The affected ones that I have are domain-joined with lots of GPOs, roaming profiles, and are using the following applications:

Vipre antimalware, Faronics Deep Freeze, Cybrarian access/time control, MS Office 2013, Java, Irfanview, Flash, Reader

However, leaving the domain and uninstalling most of these programs has had no effect. I did not try a clean install of Windows to see if that helped, but that was next on my list to try (clean install, and gradually add things back to see if/when the problem recurred).

Do you see anything in common with your system setup?

Looks like you don't expect a reply from me but I'll reply anyways.

I can reproduce this bug on a brand new Windows 7 64 installation (of course with the "bad" update installed) and zero additional software aside from necessary drivers (NVIDIA GPU and RealTek HDA).

So, it doesn't seem like a race condition of any kind.

As for a low number of people reporting this issue: there are only a few Sandy Bridge / Ivy Bridge owners remaining and most of them have already upgraded to either Windows 8.1 or 10 which seem not to be affected by this regression (I guess ntoskrnl has changed quite a lot since Windows 7).

Thank you @ultrafine, that saves me some experimentation.

My Lenovo M82s experience this bug, and they are Haswell machines from late 2013.

I believe from what I've read that Windows 7 and Windows 10 have roughly similar market shares right now (one report says Win10 slightly higher, and another Win7 somewhat higher). But either way, there must be millions of similar (Haswell or earlier) PCs out there.

Does modifying the available C-States cause the problem to go away for your systems?

Disabling C6 made no difference but when I also disabled C3 the bug was gone. I'm too lazy to check whether it's just C3 + C6 or just C3 which affects my system - one thing is sure: along with KPTI they introduced a major regression in regard to reboot handling.

I'm astonished so few people are reporting the problem. At the moment it's just us, three, and the guy who reported a similar issue in a different thread.

And considering how different our devices are it's surely a low level kernel bug which has nothing to do with the rest of the installed software on your PCs.

I'm not ready to sacrifice power savings granted by these options, so I will continue to use my system with all the C-states enabled - beside, my primary OS is Linux and it works just fine. As for Windows - I use it only for gaming, so I'm OK with powering down my system or using the hardware reset button. ;-)

If any of you guys have a company-wide software contract with Microsoft you should file a bug report. I'm a home user, so I'm quite sure my report will be dismissed as erroneous or absolutely minor.

I have just my single home PC with Windows 7 - 64bit, no networks or anything odd.  If you read my original post on this issue, you can see all the things I did try to resolve it and nothing helped.

I suspect you won't see a lot of people reporting this issue since many may never even preform a restart unless a monthly update calls for it or some other update does.  Or they just simple power off/on or hit reset like we had to do.

I frequently perform a restart for software updates, or to switch booting to my XP drive ( I have a dual boot system), enjoy some older games that play better on XP then Windows 7 or some that use DOSBOX on my XP setup.

If around 3w additional is all that's used, guess I will keep the work-around, but I'm not holding my breath that the root cause will be fixed in a future update.

Cheers.

Yeah, I have about 35 PCs that are exhibiting this problem sporadically, and they reboot a lot more often than home PCs, since they do so automatically several times a week.

Looks like about 30 other people have this problem. It is amazing that there aren't more, you're right!