1

I recently installed Windows 10 Enterprise 64-bit on a new computer. Since then, I have experienced UI hangs 2-3 per day where applications bundled with Windows will stop responding to mouse clicks and stop updating their data. Specifically, here is what I've seen once the system is in a bad state:

  • The Start Menu and Search bar stop responding to mouse clicks.
  • Task Manager (if it was already open) stops updating its data. It allows you to switch tabs, but the data in the tabs is outdated.
  • Because the UI is hung, I am unable to launch new applications including the Task Manager.
  • Attempting to log out the user or rebooting the system hangs and I am forced to power cycle.

Strangely enough, any applications that were not bundled with Windows (e.g. Chrome or VMWare) continue to function just fine. As such, I don't believe this is a hardware problem.

I took a quick look at the Event Log but didn't see anything obvious.

Are other people experiencing this problem? Any ideas on how to fix it?

My system configuration is:

UPDATE: More clues... I spent days testing my hardware using MemTest86 and BurnInTest but found no problem. I tried disabling power savings on the hard-drive. This didn't help.

I noticed that when the system hangs, I can access all hard-drives just fine (dir /s does not hang) but anything that requires kernel access does. Attempting to open any process with admin privileges hangs immediately after I select "Yes" for granting Admin access. Listings or killing processes hangs, but Task Manager's overview tab does not. Resource Management launches fine, but never gets populated with any data. In short: it sounds as if the Windows kernel is deadlocked. Alternatively, some kernel level driver (e.g. anti-virus) is deadlocked and it is taking down the entire system with it.

Gili
  • 1,901

1 Answers1

2

It took me almost a year to figure this out, but I tracked down the problem to HWMonitor. I was running version 1.1.7.0 which apparently was not compatible with Skylake nor Windows 10. I had no idea that userspace applications could take down an operating-system this way, but apparently they can. Upgrading to the latest version of HWMonitor fixed the problem. Please note that CPU-Z is affected by the same bug (the latest version fixes this as well).

References:

I learned another lesson while debugging this problem: NTFS journaling only protects the file-system metadata. It will not prevent your data from getting corrupted! https://devblogs.microsoft.com/oldnewthing/20130101-00/?p=5673

When stress-testing your system, always, ALWAYS, disable hard-drive write caching. If you do not, system crashes will result in file contents getting zeroed in spite of Windows reporting that the disk is clean.

UPDATE: After fixing the aforementioned HWMonitor problem, I stopped experience UI hangs (where one process would hang while others would run). However, I did continue experiencing system-wide hangs where the mouse/keyboard would become unresponsive. These hangs would occur randomly, sometimes when the system was active, other times when it was idle.

It took me another couple of months to track this down. It turns out that I had bent CPU socket pins on the motherboard. I haven't experienced a single hang or crash since replacing the motherboard.

Gili
  • 1,901