[Critical Bug] FanControl crashes completely on Nvidia driver restart

[gashtal]

Latest version of FanControl V260 crashes completely if Nvidia's driver is restarted. If a hardware or software issues causes the driver to hang, the driver can restart automatically as a recovery mechanism, which then crashes FanControl. This can turn into a very critical situation if the user is also controlling GPU fan speeds using FanControl, which will stop working after driver restart, and can result in GPU overheating if the driver hangs and restarts while playing a game.

Steps to reproduce:

• Install FanControl
• Enable LHM plugin for GPU and disable Nvidia's plugins
• Restart the driver using restart64.exe from CRU (https://www.monitortests.com/forum/Thread-Custom-Resolution-Utility-CRU)
• FanControl will crash completely

Crash log in Event Viewer:

Application: FanControl.EXE
CoreCLR Version: 10.0.125.57005
.NET Version: 10.0.1
Description: The process was terminated due to an unhandled exception.
Stack:
   at LibreHardwareMonitor.Interop.NvidiaML.NvmlDeviceGetPowerUsage(NvmlDevice)
   at LibreHardwareMonitor.Hardware.Gpu.NvidiaGpu.Update()
   at LibreHardwareMonitor.Hardware.Hardware.Accept(LibreHardwareMonitor.Hardware.IVisitor)
   at FanControl.Domain.BackendProviders.LHM.LHMBackendProvider.Update()
   at FanControl.Domain.ComputerAccessLayer.Update()
   at FanControl.Domain.ApplicationClock.DoActions()
   at System.Timers.Timer.MyTimerCallback(System.Object)
   at System.Threading.ExecutionContext.RunInternal(System.Threading.ExecutionContext, System.Threading.ContextCallback, System.Object)
   at System.Threading.TimerQueue.FireNextTimers()
   at System.Threading.ThreadPoolWorkQueue.Dispatch()
   at System.Threading.PortableThreadPool+WorkerThread.WorkerThreadStart()
   at System.Threading.Thread.StartCallback()

Faulting application name: FanControl.EXE, version: 260.0.0.0, time stamp: 0x69760000
Faulting module name: coreclr.dll, version: 10.0.125.57005, time stamp: 0x691ddc6e
Exception code: 0xc0000005
Fault offset: 0x0000000000367c3a
Faulting process id: 0x309C
Faulting application start time: 0x1DCA7A5995BC968
Faulting application path: C:\Program Files (x86)\FanControl\FanControl.EXE
Faulting module path: C:\Program Files\dotnet\shared\Microsoft.NETCore.App\10.0.1\coreclr.dll
Report Id: 7eb2a6e4-ceba-4b17-b354-3f87f2d46321
Faulting package full name: 
Faulting package-relative application ID: 

This crash is not recorded in FanControl's log.txt at all.

OS: Windows 11 25H2
GPU: Nvidia RTX 5090

P.S. I was previously using Nvidia's plugin and FanControl did not crash during driver restarts back then, but generated an error about the plugin, and stopped monitoring GPU temps. I am not sure if this crash started after I switched to the LHM plugin, or something has changed in FanControl itself.

[Rem0o]

at LibreHardwareMonitor.Interop.NvidiaML.NvmlDeviceGetPowerUsage(NvmlDevice)

The bug needs to be fixed/reported on there.

[gashtal]

Sure, I will report this to LHM's repo. In the meantime, can some provision be added to FanControl to handle LHM crashes gracefully and not crash?

[Rem0o]

It's a native exception, I can't catch it on the .NET side. This whole code block is already in a try catch.

[gashtal]

@Rem0o The unhandled exception has been fixed in LHM now (LibreHardwareMonitor/LibreHardwareMonitor#2249), but now an issues similar to what previously happened with NVAPI at #3553 is happening after the driver restart, where FanControl no longer reads the GPU sensors (and fans curves that depend on those sensors do not get applied anymore), and detects a new set of GPU fan sensors on every driver restart, which can only be fixed with a manual refresh of FanControl. Doesn't the auto recover mechanism that was implemented to fix #3553 apply here?

[Rem0o]

It should apply. Any nvapiwrapper unhandled exception should trigger a refresh.

[gashtal]

This is with LHM, not NVAPI, and it doesn't seem the refresh is happening on my side. Is it possible you could test this on your environment with the modified version provided by @Blacktempel and confirm if the auto refresh works or not?

[Rem0o]

I don't have a Nvidia card on hand right now, so I can't test it. Howerver if it interfaces with IHardwareChanged, FanControl should pick it up.