Intel 660p SSD BSOD "DRIVER_VERIFIER_DMA_VIOLATION" on Win10 latest KB4571756

[falken]

Hello, while I tested few NVMe SSDs inside TB3 boxes having DCH 987 driver (OWC Express 4M2, UNIBOS Portable SSD x4) with NUC10 kit, ONLY Intel 660p (fw:004C updated from 002C) fails almost always while detection/enabling it. Problem occurs only with latest MS nvmestor driver update in mentioned KB, so simply generic driver changes make him less reliable, but as other disks from different manufacturers are working well, bug is also in your drive. It was 2 days ago reported also to MS over FeedbackHub already. Please fix it working together with MS, tnx.

[BrusC_Intel · Accepted solution]

Hello, @falken.

Thank you very much for the feedback and all the documentation provided involving this error.

I can certainly submit this information and I'm glad to hear that you already reported this on the MS side as well.

The thread will be closed right now, but feel free to open a new one if necessary, this one will no longer be monitored.

Best regards,

Bruce C.

Intel Customer Support Technician

A Contingent Worker at Intel

[SteinOiestad]

FYI

I got the same problem BSOD "VERIFIER_DMA_VIOLATION" as soon as I connected OWC Express 4M2 containing one Intel 660p NVMe. This was on Intel NUC Frost Canyon i7 - BIOS just updated. Windows 10 latest installed. Intel Driver package OK.

As soon as I tried to connect the OWC Express 4M2 Thunderbolt cable the computer rebooted (BSOD as described) immediately. I will try to remove the Intel 660p from the OWC express box (luckily now sw raid yet) and see if the problem persists.

[SteinOiestad]

FYI - the OWC M42 worked OK when I used only a Crucial 1TB NVMe.

I have both Crucial, Kingston, Samsung and Intel NVMe's, so I will test and see what works and what will not... strange....

[falken]

Hi, I spent few days with it, it is simply latest stornvme driver with latest 660p firmware only which is failing. 6 other SSDs works there without problem even with new stornvme, but there is definitelly something bad in new driver too. The same happens in single slot UNIBOS PCIe 16x4 enclosure. Posted file "!! -- OWC Express 4M2 TB3, testy.zip" contains timestamped history of AS SSD results on working disks. I also later enabled only "Native OS AHCI" in BIOS and left ASPM disabled there. Since then this OWC  box is working for me fine, having four different SSD inside mounted as stripped NTFS volume "R:". As I will need to process few hundreds of gigybyte ZIP files with my custom software, performance of this all seems to be OK, but now I have here also OWC "Mercury Helios 3" (not new 3S) enclosure and ASUS "Hyper M.2 x16 Card V2", but I expect it will be the same. Thunderbolt3 transparently exposed PCIe just works, but probably as there are potential security issues related to DMA (see thunderspy), new windows "Kernel DMA Protection" feature probably clashes sometimes with drivers/firmware bugs for such externally exposed bus yet. (I had no chance to test new stornvme in any internal M.2 slot. It may be OK too)