Windows 10: BSOD (Critical_Structure_Corruption) once every few days or so

Sorry I had to actually do a bit of research on this as I never encountered this problem before. First I have to applaud you for being so curious. It seems like you want to just know more and more about your system than anything else.

Anyway,

Give this a shot and it should fix your issue so you can change your page file to a separate hard drive. How to move page file in Windows 10 to another disk

 

Thank you for the help as well! I am interested in learning more about my system because I want to work in this field at some point.

Anyway, I looked at the link you provided but it still does not give advice regarding how to set it (only how the person writing the guide set it for his PC, not a general recommendation). He does say you should use the page file on an SSD if you have one, which I'm already doing.

The issue is: use system managed page file or custom size (and in this case how big)? I ask again because when I switch to Automatic memory dump, it gives me the error with page file needs to be at least 800 MB, when page file is set to system managed and not custom.

That is my question as I found multiple guides, some suggesting custom size, others system managed, and the least suggesting page file off (page file off is probably the worst out of all options though).

 

Hi RaduAvr,

The recommended option is to let the system handle the pagefile, this way the system will ensure that enough virtual memory is present at all times *Smile
For BSOD issues it is not recommended to turn off the pagefile, because the pagefile is used in the BSOD process for temporary storing the kernel memory before the system restarts. After the restart the system accesses the pagefile again to dump the kernel memory into an actual file (MEMORY.dmp).

 

Hi axe0,

Thank you for coming to my thread, love your work on the forum. *Smile

Regarding having page file set to system managed, yes, that's what I've always had it set to. What I always do when I install Windows is disable the "Automatically manage paging file size for all drives" option, and only turn it on for C: (my SSD and operating system drive), with "System managed" option.

Now, here's the issue. When I choose "Automatic memory dump" in "Startup and recovery", I get this exact warning:

"Windows might not be able to record details that could help identify system errors because your current paging file is disabled or less than 800 megabytes. Click OK to return to the Virtual Memory settings window, enable the paging file, and set the size to a value over 800 megabytes, or click Cancel to change your memory dump selection"

Obviously, that makes me wonder if it's able to properly create the dump files under System managed, because it's giving me a warning. Or is this normal?

 

Thank you *Smile


Could you make a screenshot of the settings (virtual memory and recovery) and the error please.

 

Okay, here are the screenshots:

Virtual Memory: Imgur: The most awesome images on the Internet

Startup and Recovery: Imgur: The most awesome images on the Internet

 

I would guess that the problem is the allocated memory, it is lower than the recommended size. Though I don't know if this really is the issue, because I've never seen it happening.

I would suggest to enable 'Automatically manage paging file size for all drives' while troubleshooting, when your system is stable again we can deal with the error message *Smile

 

Interesting. So, I enabled "Autoamtically manage paging file size for all drives" as you said and now I don't get that error anymore. I'm not sure I like this setting though, because from a performance standpoint having a page file only on the SSD is ideal as far as I know.

 

I haven't done the RAM test yet (since it requires a lot of time), but I can say that so far I haven't received another BSOD.

So, one question is: do I keep Driver Verifier on? Because it makes the experience a bit stuttery and I was wondering why I couldn't get 1 Gbps on Speedtest, but it was stuck around 300-600 Mbps with huge lag, until I turned it off of course. If it's still useful to keep it on, I will.

The other question is: how exactly do I do the RAM test? Do I do 1 stick at a time, or both?

Thank you once again, the both of you!

 

Driver verifier checks every checked action all checked drivers are doing, by doing this performance can be very sluggisly as probably mentioned.

How you do it is up to you, I would start with everything at once and if errors show switch to one per test.

 

OK so I got one BSOD just now while I tried to run OCCT (I want to run it so I can do a GPU memtest, easier to do first than the RAM test). The reason why I want to do a GPU memtest is because I've been getting this kind of graphic corruption: graphic corruption warhammer - YouTube

That makes me think it's a VRAM issue, and indeed on the Nvidia subreddit, that's what people assume is the issue as well.

Anyway, as soon as I tried to launch the program, I got the BSOD. Restarted, then ran program and it didn't BSOD this time. Odd. Here are the files.

In case it doesn't show what it said, during the error screen it said it was a "page fault in a non-paged area" and pointed to "cpuz x64" or something of the like.

 

Likely from CPU-Z, the driver is very old.
You may want to update CPU-Z to the newest version.
Code: Child-SP RetAddr : Args to Child : Call Site ffffdb00`7438f2b8 fffff802`2eb64490 : 00000000`00000050 ffffaf0f`00000018 00000000`00000000 ffffdb00`7438f5b0 : nt!KeBugCheckEx ffffdb00`7438f2c0 fffff802`2ea7974a : 00000000`00000000 00000000`00000000 ffffdb00`7438f5b0 ffffdb00`7438f420 : nt! ?? ::FNODOBFM::`string'+0x9200 ffffdb00`7438f3b0 fffff802`2eb556fc : 00000000`00000000 00000000`00000000 00000000`00000000 ffffa800`0631fd70 : nt!MmAccessFault+0x9ca ffffdb00`7438f5b0 fffff802`6511217b : 00000000`20646156 fffff802`2ec4d6db ffffaf0f`a8aae080 00000000`00000003 : nt!KiPageFault+0x13c (TrapFrame @ ffffdb00`7438f5b0) ffffdb00`7438f740 00000000`20646156 : fffff802`2ec4d6db ffffaf0f`a8aae080 00000000`00000003 fffff802`2ed26400 : cpuz_x64+0x217b ffffdb00`7438f748 fffff802`2ec4d6db : ffffaf0f`a8aae080 00000000`00000003 fffff802`2ed26400 ffffaf0f`a7c05000 : 0x20646156 ffffdb00`7438f750 ffffaf0f`aee05ee0 : 00000000`00000001 ffffaf0f`00000000 ffffaf0f`a8aae080 00000000`00000024 : nt!ExFreePoolWithTag+0x34b ffffdb00`7438f830 00000000`00000001 : ffffaf0f`00000000 ffffaf0f`a8aae080 00000000`00000024 00000000`00000000 : 0xffffaf0f`aee05ee0 ffffdb00`7438f838 ffffaf0f`00000000 : ffffaf0f`a8aae080 00000000`00000024 00000000`00000000 ffffaf0f`a7d27080 : 0x1 ffffdb00`7438f840 ffffaf0f`a8aae080 : 00000000`00000024 00000000`00000000 ffffaf0f`a7d27080 ffffaf0f`ad4377e0 : 0xffffaf0f`00000000 ffffdb00`7438f848 00000000`00000024 : 00000000`00000000 ffffaf0f`a7d27080 ffffaf0f`ad4377e0 00000000`00000000 : 0xffffaf0f`a8aae080 ffffdb00`7438f850 00000000`00000000 : ffffaf0f`a7d27080 ffffaf0f`ad4377e0 00000000`00000000 00000000`00000001 : 0x24 Bugcheck code 00000050 Arguments ffffaf0f`00000018 00000000`00000000 fffff802`6511217b 00000000`00000002 Debug session time: Mon Sep 12 19:18:48.985 2016 (UTC + 2:00) System Uptime: 0 days 7:37:16.641 0: kd> lmvm cpuz_x64 Browse full module list start end module name fffff802`65110000 fffff802`65118000 cpuz_x64 T (no symbols) Loaded symbol image file: cpuz_x64.sys Image path: \??\C:\Users\RaduPC\AppData\Local\Temp\cpuz131\cpuz_x64.sys Image name: cpuz_x64.sys Browse all global symbols functions data Timestamp: Tue Dec 02 14:51:19 2008 (49353D57) CheckSum: 000138BD ImageSize: 00008000 Translations: 0000.04b0 0000.04e4 0409.04b0 0409.04e4[/quote]

 

My CPU-Z is 1.74.0 (latest is 1.77.0). So, it's not that old I think.

I think OCCT comes with built-in CPU-Z perhaps, and the version I downloaded (3.1.0) may be using an older one? I need 3.1.0 because the latest one doesn't seem to have memtest support, and 3.1.0 is the last one that does.

I'm guessing that's what happened there.

 

The driver mentioned is from 2008, thats old.

Testing the VRAM is possible with Furmark.

Run Furmark tutorial



Note Run Furmark for around 30 minutes



Warning Your GPU temperatures will rise quickly while Furmark is running. Keep a keen eye on them and abort the test if temperatures rise too high
When finished, make a photo/screenshot of Furmark and post it.

In Furmark you can find a tool called 'GPU-Z' (if you want to install GPU-Z is up to you), when GPU-Z is open click on the 'Sensors' tab, on this tab you can monitor

• the frequency of the GPU,
• the GPU memory clock,
• the temperature,
• the power,
• the load on the GPU,
• Etc.

I would like you to make a screenshot of GPU-Z after you have either cancelled Furmark for any reason or Furmark is finished, post this screenshot in your next reply with the result of Furmark.

 

I'll do a Furmark GPU test, then I'll do the RAM test probably tomorrow or the day after.

Indeed, the driver is old but it's not the installed CPU-Z. Again, it must be from the CPU-Z that comes with OCCT (it's bundled I think, that's how you get CPU monitoring in the program). It's more obvious because it's in a folder called "cpuz131", and I never ever installed that version. At least, that's how I think it is.