Question AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft) causing nvlddmkm.sys Video TDR BSOD?

[Deleted member 2900411]

I have been experiencing nvlddmkm.sys Video TDR Error BSOD crashes since mid-October. I have been keeping a detailed log of all events since November 01, which has allowed me to possibly narrow down the potential cause. This log is now 32 pages long, so I will try to summarize all of my observations and troubleshooting steps in this post.



Observations:

• In the 16 months since building my PC before this issue started, I experienced maybe 2-3 BSODs total of any kind, and my system was 99% reliable 24/7. (Including after I installed my GeForce 1050 TI in April, 2021.)
  
• The nvlddmkm.sys Video TDR BSODs started on or about 10-13-21, after a Windows update that updated the driver for "AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)". No other drivers were updated around this time aside from graphics driver.
  
• The crash happens randomly during any type of normal PC use (outside of gaming), or when idling.
  
• I can induce a guaranteed state of not crashing for 4-6 hours (outside of gaming) by cleanly shutting my system down through Windows, then powering back on as long as the only USB devices connected are my mouse and keyboard. (Restarting doesn't work for this, nor forced power button shutdowns, nor powering down during the POST screen after a BSOD auto-restart.)
  
• The crash occurs with no USB devices connected while idling overnight after a clean shutdown and restart, but the time until the first crash is increased to 10-13 hours.
  
• Once a crash occurs, more crashes will follow shortly and their frequency will increase until my PC is unusable. Performing a clean Windows shutdown resets this.
  
• If crashes are allowed to continue and increase in frequency without a clean shutdown, a nvlddmkm.sys or amdppm.sys DPC Watchdog Violation BSOD (or extremely rarely, a watchdog.sys VIDEO DXGKRNL LIVEDUMP error) may occur, and my PC may "hang" on a black screen after a BSOD automatic restart when trying to load Windows.
  
• The crash frequently occurs simultaneously with USB device input such as mouse movement, clicking, or a keyboard key press.
  
• The frequency of crash occurrence is directly affected by the quantity and type of USB devices connected outside of gaming. (More devices connected outside of gaming increases the crash frequency. Game controllers connected outside of gaming significantly increase crash frequency.)
  
• The crash has never happened while playing a game, no matter how long the session or how many USB devices of any type are connected.
  
• The crash has never happened in Safe Mode.

Troubleshooting steps taken (Performed one at a time after BSODs continued):

• Updated to the latest recommended graphics driver and software.
  
• Cleanly uninstalled all graphics drivers and software with Display Driver Uninstaller and installed the last driver version from before the BSOD issue started.
  
• Cleanly uninstalled all graphics drivers and software with Display Driver Uninstaller and installed the graphics driver version present when the BSOD issue started.
  
• Cleanly uninstalled all graphics drivers and software with Display Driver Uninstaller and installed the latest graphics driver again.
  
• Cleanly uninstalled all graphics drivers and software with Display Driver Uninstaller and tried only the driver and software Windows downloaded and installed.
  
• Cleanly uninstalled all graphics drivers and software with Display Driver Uninstaller and installed the latest graphics driver again.
  
• Kept Windows up-to-date.
  
• Cleaned all dust from my PC case and components.
  
• Confirmed all cable connections were secure.
  
• Ran System File Checker from admin account and got the following message: "Windows Resource Protection found corrupt files and successfully repaired them."
  
• Ran Windows Memory Diagnostic, no problems found.
  
• Turned off V-Sync.
  
• Confirmed "Link Power State" setting for PCI Express was turned off.
  
• Ran chkdsk / r c: from admin command prompt and allowed it to complete. Don't know how to find the results.
  
• Disabled "Windows Fast Startup".
  
• Performed soul-rending swear rampage.
  
• Cleanly uninstalled all graphics drivers and software with Display Driver Uninstaller and physically uninstalled my GeForce 1050 TI card.
  
• During two days of using the APU graphics of my AMD Ryzen 3200G with latest driver, not a single issue occurred.
  
• Ran a complete system scan with Malwarebytes, no malware found.
  
• Cleanly uninstalled all graphics drivers and software with Display Driver Uninstaller.
  
• Reinstalled the 1050 TI after cleaning connector pins. Made sure card was seated properly.
  
• Did not connect any USB devices normally connected except for mouse and keyboard.
  
• Installed latest driver for 1050 TI, but not the GeForce Experience software.
  
• Performed 3 hours of computer use without a crash, which would have certainly caused one in previous state.
  
• PC went additional 2 hours without a crash.
  
• Connected a USB game controller for remote PC use via keyboard and mouse emulation software, and a crash happened within one hour, followed by a return to a high frequency of crashes.
  
• Tried using only the front USB 2.0 ports.
  
• Tried using only the front USB 3.0 ports.
  
• Tried using only the rear USB 2.0 ports.
  
• Tried using only the rear USB 3.0 ports.
  
• Tried shutting down and restarting with no USB devices connected. Time before first crash was greatly increased.
  
• Manually searched through all Device Manager properties for device driver versions that had been recently updated.
  
• Noticed that the driver for "AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)" had been updated on 10-13-21. All other device drivers aside from graphics driver were at least 2 months older.
  
• Observed that the "Roll Back Driver" button was grayed-out and unclickable.
  
• Searched online for ways to roll back the "AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)" driver. Some forum posts suggested the driver may be included in the motherboard chipset drivers.
  
• Downloaded previous and latest chipset driver installers for my motherboard.
  
• Installed previous chipset drivers and software. "AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)" driver version did not change when viewed in Device Manager. Assumed installer package did not overwrite a newer driver with an old one.
  
• Installed newest chipset drivers and software.
  
• Upgraded BIOS version from F50 to F54d (latest).

None of these steps have stopped the nvlddmkm.sys BSOD crash from occurring at least once a day under the most stable system state I can thus far induce. I have searched for days online with Google and Bing, using every manor of wording and keyword combination I can think of and can't find anyone having this exact problem. I also can't find out how to replace/roll back the driver for the "AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)". Any help or suggestions would be greatly appreciated.



System Specs:



OS: Windows 10 Home 64-bit Version 21H1 Build 19043.1348

Motherboard: Gigabyte A320M-S2H Rev 1.1 BIOS Version F54d

RAM: HyperX Fury 16GB DDR4 2666 MHz DIMM

Processor: AMD Ryzen 3 3200G With Radeon Vega Graphics

System HDD: Seagate BarraCuda 3.5" 2TB 7200 RPM (ST2000DM008-2FR102)

Storage HDD: WD Red Pro NAS 3.5" 4TB 7200 RPM (WD4003FFBX-68MU3N0)

Dedicated Graphics Card: Nvidia GeForce 1050 TI 4GB Dual Slot Single Fan

Power Supply: Corsair CV550 550 Watt PSU

Internal Optical Drive: LG WH14NS40 Blu-ray Writer

Case: Rosewill Line-M Micro-ATX

USB Mouse: Teknet 6-Button Wired USB Mouse

USB Keyboard: Verbatim Slimline Corded USB Keyboard

 

[kerberos_20]

you cant rollback non oem drivers
that usb driver is from windows update, youd have to uninstall windows updates to roll back

but ummm..your issue is a bit unique
what gpu and usb have in common is that they use PCI interrupts
can you open device manager, on menu bar, click on view -> resources by type
then open interrupt request (IRQ)
find your GPU there and say if it has positive or negative number there
same will aply for your USB if positive or negative
example of positive number:
(PCI) x00000001 (1)
example of negative number:
(PCI) x FFFFFFFF (-1)

 

[Deleted member 2900411]

you cant rollback non oem drivers
that usb driver is from windows update, youd have to uninstall windows updates to roll back

but ummm..your issue is a bit unique
what gpu and usb have in common is that they use PCI interrupts
can you open device manager, on menu bar, click on view -> resources by type
then open interrupt request (IRQ)
find your GPU there and say if it has positive or negative number there
same will aply for your USB if positive or negative
example of positive number:
(PCI) x00000001 (1)
example of negative number:
(PCI) x FFFFFFFF (-1)

Here are the IRQ values for my graphics card and everything below it:

(PCI) 0x00000010 (24) NVIDIA GeForce GTX 1050 Ti
(PCI) 0X00000019 (25) High Definition Audio Controller
(PCI) 0X00000036 (54) High Definition Audio Controller
(PCI) 0xFFFFFFE9 (-23) Realtek PCIe GbE Family Controller
(PCI) 0xFFFFFFEA (-22) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFEB (-21) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFEC (-20) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFED (-19) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFEE (-18) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFEF (-17) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF0 (-16) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF1 (-15) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF2 (-14) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF3 (-13) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF4 (-12) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF5 (-11) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF6 (-10) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF7 (-9) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF8 (-8) AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)
(PCI) 0xFFFFFFF9 (-7) AMD PSP 10.0 Device
(PCI) 0xFFFFFFFA (-6) AMD PSP 10.0 Device
(PCI) 0xFFFFFFFB (-5) Standard SATA AHCI Controller
(PCI) 0xFFFFFFFC (-4) PCI Express Root Port
(PCI) 0xFFFFFFFD (-3) PCI Express Root Port
(PCI) 0xFFFFFFFE (-2) PCI Express Root Port

 

[Deleted member 2900411]

The nvlddmkm.sys issue has been solved by user "Greybear" on the Nvidia GeForce Forum. He had me run the 3DMark benchmark "FireStrike" while logging with HWiNFO64. This generated a log that I shared with Greybear. Greybear examined the log and noticed that my CPU's Infinity Fabric clock frequency (FCLK) was varying, indicating (to him) a defective CPU. I contacted AMD about it, and they replaced my CPU under warranty.

While the new CPU appears to have the exact same FCLK variation as the old one, the practical results indicate that the CPU was definitely defective, if not for the reason originally suspected. The crashes had been a daily occurrence for eight months before replacing the CPU. It has been 16 days now without a crash, despite my best efforts to induce one. I'm satisfied that a defective CPU was the cause.

 

[Deleted member 2900411]

The issue was not solved by replacing the CPU. The nvlddmkm.sys crash returned after about a month on the first replacement CPU. I RMAed the first replacement CPU, and received a second replacement. The FCLK variation was present with this (overall third) Ryzen 3 3200G as well. Greybear and I began to suspect that something else in my system was making these CPUs behave this way. He suggested that it has been known for Ryzen systems to become unstable when using only a single RAM module. I then purchased a 16GB (2x8GB) kit and installed it. After enabling XMP in the BIOS, the FCLK is now rock-steady at 1,600Mhz. If the varying FCLK frequency was causing the nvlddmkm.sys crash, then it may well be solved now. Only more time will tell for sure.