Solved Crash/reboots on custom HTPC: ntoskrnl.exe ntkrnlmp.exe symptoms

[kc25]

Hi I built this custom htpc a few months ago and have experienced regular crashes/reboots. The system is based on an MSI Z97-GD65 mobo with i5-4690S processor in a fanless Streacom case designed for the purpose with extensive heatsinking. To ensure temperatures are not excessive I have kept the processor within the Streacom recommendations, nothing is overclocked and it uses only Intel's integrated graphics with no other video card. All components are new. The only "unusual" item has been an Asus Xonar STX sound PCI-e card. The system is used for TV (windows media center), as a Plex server and a Foobar server - not much else.

By way of troubleshooting, after the first few weeks I did a clean reinstall to remove any issues I might have caused as I iterated my way to the optimum setup, but this made no difference. I also replaced the PSU with no effect. I have run memtest on the memory - this produced 2 errors in the new "hammer test", but G-Skill apparently discount this test and when I run memtest86+ which does not include such a test it shows no errors. Temperatures are ok - typically 30-40 degrees C with occasional 50 (and up to 60 only when doing extreme tasks like Memtest) and the crashes do not seem to happen when anything intensive is going on.

I have been blaming the issue on the Asus Xonar STX sound card as google suggests some people find this causes BSOD issues and the Asus drivers are considered buggy and I had been using the alternative UNI drivers. I have therefore recently now removed the card and installed the mobo's built in realtek 1150 sound. However, I have already had one reboot.

These reboots can be while watching TV using Windows Media Center, or when updating a library in Plex, or simply when putting Flash video to full screen in Chrome browser.

Rather than attempt another clean rebuild with no more likelihood that I will eliminate the problem this time than last time, I would appreciate any help you folks can give me in tracking down the problem so I can fix it. I have now switched minidump reporting one but the last reboot had already resulted in a MEMORY.DMP file and one minidump file so I am attaching the requested zip file.

Thanks!

 

[Arc]

Some modules are unloaded there, which resisting the crash dump recording mechanism to record the complete data there.

fffff880`0330e098  fffff880`058a5b8e <Unloaded_spsys.sys>+0x3cb8e
fffff880`0330e0d8  fffff880`058a41f4 <Unloaded_spsys.sys>+0x3b1f4
fffff880`0330e0f0  fffff880`0586c720 <Unloaded_spsys.sys>+0x3720
fffff880`0330e0f8  fffff880`0587b510 <Unloaded_spsys.sys>+0x12510
fffff880`0330e100  fffff880`0587b510 <Unloaded_spsys.sys>+0x12510
fffff880`0330e108  fffff880`058a4270 <Unloaded_spsys.sys>+0x3b270

So crash dump analysis is also not being helpful.

Free up the startup. Windows does not need any other program to auto start with it, but the auto start programs often conflicts and causes various problems including BSODs.

1. Click on the Start button
2. Type “msconfig (without quotes), click the resulting link. It will open the System Configuration window.
3. Select the “Startup” tab.
4. Deselect all items other than the antivirus.
5. Apply > OK
6. Accept then restart.

Then let us see the zip for the next BSOD, that may contain some information. To replicate a BSOD in your situation, put the computer in hibernate/sleep and wake it up.

 

[kc25]

Thanks! I have done as directed and will monitor for BSODs and report back. Interestingly I have not been able to replicate using sleep/wake (BTW the machine is normally set to never sleep) - I will set to allow hibernate and try that too.

Also FYI prior to following your directions I have experimented with a few changes: I had noticed a new BIOS update for the MSI mobo so I updated that this week and made a few driver updates. On the chance that this is ultimately video related I also used Driver Uninstaller to uninstall the Intel graphics and reinstalled. The system did seem more stable for 24hrs but then I had a batch of reboots which were easy to produce doing things like stopping play in or just menu navigating in Plex Home Theater. (Note I had previously suspected that Plex Home Theater/Media Server might be responsible for my problems so had done complete uninstalls on both for a few weeks, but the reboots did not stop so I reinstalled). As you suggest, none of these reboots resulted in new dump files. Finally, after reading about problems with the Haswell 4600 HD graphics and the need to increase voltage to the graphics, I installed Intel Extreme Tuning Utility and set the Graphics voltage to 0.875v: I have had no reboots since that change, but it's early days.

Anyway, hopefully after following your directions we will get some dump files if there are further reboots and can take it from there. In the meantime I will make no further changes.

Thanks again for your help!

 

[kc25]

OK three crashes in a row. The crashes seem most easily (but not exclusively) triggered by opening or navigating Plex Home Theater, but after three crashes in a row it seems fine again.

Microsoft Security Essentials is the only program auto-starting in msconfig: I still don't see any new dump files but I am uploading a new zip file in case there is anything new in it that helps. Please let me know if you see anything. Thanks!

 

[Arc]

Can you let us see a camera snap of the next BSOD screen, if it happens again in the near future? Perhaps the onscreen codes will give us some idea about the the BSOD and the non-recording of the crash dump.

 

[kc25]

Thanks, ok I've just disabled autostart on system failure so let's hope that results in something to see/photograph next time.

 

[kc25]

Although I have disabled autostart on system failure in Windows and have nothing autostarting in msconfig other than Microsoft Security Essentials, the machine continues to restart immediately on crashing and I get no BSOD to photograph or updated dmp files. Unless you have a better idea I think I therefore have no choice but to do another clean reinstall when time permits. This time I will first set up only Windows Media Center for TV and see whether the system is stable before moving on to Foobar2000 and finally Plex. In the unlikely event all that works I will then add in a pci-e sound card. If it doesn't prove stable and/or provide useful troubleshooting information again I will assume a hardware problem: I have a cheaper used motherboard and will also get some cheap used memory and some more sata cables so I can start swapping things out systematically to eliminate cables/memory/mobo/processor. I will report back though in case it is useful to someone. Thanks for your help Arc and if you have any suggestions please let me know.

 

[Arc]

As there is no recorded crash dump, and you cannot provide any minimum information about the crash, it is merely impossible to form an idea as to what is causing the issue.

There is apparently no way but hitting around the bush, unfortunately.

Remove the added hardware components one by one, see if it becomes stable at any stage. Do it till you have only a single RAM stick and the keyboard-mouse connected.

If you fail to determine the issue by this process too, do a vanilla installation of the OS. Then add the drivers and software one by one. Observe at what stage the issue comes back.

 

[kc25]

OK over the past weeks I did several clean reinstalls and then replaced the motherboard, hard drives, optical drive, SATA cables, USB front panel header, front panel power switch and memory. All to no effect random reboots remained. As I had originally suspected and replaced the PSU that only left the processor from my original build. However the processor passed Intel's diagnostics utility so I decided to go back to the PSU, which had been my original suspect. I therefore replaced it with a different make and model. One week on, and after considerable test use and repeated attempts to crash it, I have not had any kind of problem. I have therefore concluded that either both examples of the original PSU were defective or the model was simply inadequate for the processor/motherboard combination used. The moral is that sometimes, when you think the problem is hardware, it really is and that after you've built a few PCs your first instinct is usually the right one. It's a shame I didn't have a spare different PSU to try right back at the beginning...

Anyway, thanks Arc for your help and I think I'll risk marking this one solved.