[SOLVED] Dxgkrnl.sys BSOD

[andeekaii]

Hi all

 

Bit of an unusual one here and the steps taken that put me in this situation are quite unusual, but maybe someone else has faced something similar.

 

Steps Taken

• Used sdelete to remove (zero out) free space on my Windows 10 (Raw) vdisk
• Converted from raw to qcow2 to reduce space allocation and make my backups faster
• Added a second cache drive in 'Single Device' mode (not raid 1)
• Created a new qcow2 vdisk specifically to store my gaming library (stored on cache)
• Mounted said qcow2 vdisk to my Windows 10 VM as a second drive so that it's local and all the game launchers can install games to it without complaining that it's a network drive
• Installed/move my game library to gaming vdisk

 

Problem

• Three times in a row a game I was playing would blue screen at pretty much the same place (the end of a level). The bluescreen code was Dxgkrnl.sys. Previous to this we had 24 hours gameplay without a single issue.
• Attempted to redownload the game in case it was corrupted, but every time I tried to redownload it to the gaming qcow2 vdisk, it would reset the entire VM after a few minutes. This happened several times in a row and only seemed to happen when downloading to the vdisk.

 

Debugging

• Ran the Windows VM with a different GPU. Same issue
• Reinstalled the GPU drivers. Same issue.
• Installed a fresh windows 10 VM. Only installed GPU drivers. Couldn't even download the game because it kept resetting the VM.

 

Thoughts

• I am fairly confident it isn't a GPU issue.
• I know it isn't a corrupt Windows installation as it happens on a fresh install.
• I haven't made any hardware changes other than installing the new m.2 nvme drive for the gaming vdisk.
• The fact that it happens even when downloading makes me think it's something to do with either my original or my new nvme drive.
   

Next Steps

• I have formatted my cache drive completely and gone back to a single cache drive (the original drive that I know was working). If this still blue screen then the only other variable left is using qcow2 instead of raw. Has anyone had issues with qcow2 with either windows or gaming?
• Is there anything special about setting the size on a qcow2 disk? If I set the size to 100G to begin with, and then in the future I increase it through the GUI and expand the partition in Windows, is this ok?

 

Probably no use uploading a log at the moment because I've been busy fixing a different issue (see my other post), but if it happens again on the fresh cache pool then I will attach in another post.

 

Interested to here any suggestions. I'm aware that it's quite an unusual error - previously my server was stable for over a year so it's really unusual for me to experience so many crashes in such quick succession.

 

Andy

 

 

Edited June 27, 2021 by andeekaii

[andeekaii]

Update: Even after formatting the cache drive, it still happens. It now happens every time I shut down the VM - loads saying please wait and then blue screens (too fast to catch the error message).

 

Something is seriously wrong but I don't know what. I'm going to reseat the GPUs and the drives and try again

[Squid]

Update BIOS?

[andeekaii]

Update:

Just wanted to come back to this and let everyone know what the outcome was. 

 

However it happened, my VM vdisks got corrupted during the steps taken in my first post. They still booted into Windows fine and ran fine most of the time but certain things caused the system to crash.

 

Since having created a new vdisk (for VM and for my Gaming Library) and installed a fresh copy of windows, everything is now resolved and back to normal. No more crashes.

 

A little anti-climatic but it was easier to just re-install Windows than chase my tail around trying to find exactly where the corruption was.

[andeekaii]

Further Update:

I encountered the same issue again on my completely fresh install and managed to isolate the cause this time.

The crashing began after installing the latest MSI Afterburner 4.6.3 beta. As soon as I rolled back to 4.6 the BSODs immediately stopped.

 

Maybe this can help people searching for this issue in the future...

 

Andy

[Bullerwins]

On 6/30/2021 at 7:41 PM, andeekaii said:

Further Update:

I encountered the same issue again on my completely fresh install and managed to isolate the cause this time.

The crashing began after installing the latest MSI Afterburner 4.6.3 beta. As soon as I rolled back to 4.6 the BSODs immediately stopped.

 

Maybe this can help people searching for this issue in the future...

 

Andy

Just wanted to save that you're a lifesaver, for some reason it worked fine for everything for months. But now installing the Trezor Bridge software I was getting a BSOD. Downgrading to 4.6.3 (now has a stable version) worked, i had the 4.6.3 beta too.