Drivers & Software

[BrKo123]

Yesterday I had my system BSOD out of the blue with an ATTEMPTED WRITE TO READONLY MEMORY stop code. It immediately bluescreened again after restarting with the same stopcode, before doing it again an hour later.

The memory dump shows ntkrnlmp.exe as the failing module.

Event Viewer shows for every BSOD an Event 41, Kernel-Power critical event.

Running a driver verifier has produced a minidump that shows amdgpio3.sys as a corrupted driver triggering BugCheck C4
The BSOD seems really inconsistent, having happened while watching an MKV file through VLC, 10 seconds after booting into the main screen and once while having two browsers open. I've tried updating and downgrading chipset drivers but due to the inconsistency of the BSOD I can't really verify if it worked.

I also got a PAGE FAULT IN NONPAGED AREA BSOD at one point right before restarting when I undid a driver update, but I cannot replicate that anymore, although it seems to match the dump produced by the driver verifier.

My RAM is a dual kit F5-6000J3040F16GA2-RS5K, which is not on the QVL but has passed a memtest86 stress test with and without XMP enabled. The crash also happened once with XMP disabled.

Nothing changes the driver verifier flagging amdgpio3.sys though, so I am still skeptical the issue was resolved.
Anybody here have any idea if this is at least a software, and not hardware issue? I don't have any spare parts on hand and I'd have to change the whole system to find the fault from what I've gathered.

If I need to upload the dumps or any other information if anybody can help, please let me know.

I am using an MSI B650 Tomahawk WIFI on bios beta version 7D75v164

[elstaci]

AMDGPIO3.sys sounds like it is part of AMD CHIPSET Driver set.

Use MS Install/Uninstaller to remove AMD GPIO Driver and reinstall it again or update your AMD CHIPSET driver package from AMD and see if that fixes your issue: https://www.amd.com/en/support/chipsets/amd-socket-am5/b650

[BrKo123]

That unfortunately did not work. amdgpio3 is still flagged by driver verifier. I already had the latest version of the chipset, although one thing of note is that the installer says the latest version is not installed for the Promontory GPIO driver, which seems to match the version shown by driver verifier (3.0.0.0), but it seems up to date.

[ThreeDee]

What are you running for a GPU? ..and power supply?

Update your BIOS to latest 7D75v16 (2023-05-29) 

IF still get BSOD's .. test system with XMP disabled

Check your temps/airflow

[BrKo123]

I am on the 7600's integrated graphics for now, with a Corsair RM850x 2021. Temps have been in high 40s every crash, I've never seen them go above 80 even without curve optimizer (I have curve optimizer off for now after the crashes, but one crash happened without curve optimizer)

I considered dirty power potentially being the issue but all the parts are brand new. I also just updated to the latest stable bios (the version you posted, which was not actually listed on MSI's website until yesterday for some reason). There have been no more BSODs since those last few on the first day (with XMP disabled for now) but driver validator still marks amdgpio3 as dangerous.

Hopefully it stays this way but due to that amdgpio3 validator thing I don't feel quite safe with my system rn. We'll see how things progress, I just hope it's not a hardware issue. With software I can at least take some comfort in a potential future fix.

[johnnyenglish]

I'm very inclined towards memory.

- I would try disabling memory context restore if you have it on enabled or auto.

- I would also try and run the system with just one memory stick, then the other for troubleshooting.

- You can try alternating between DOCP/EXPO/XMP profile 1 or 2, the difference is one will load just 4 main timings the other will load sub timings as well. There are reports that sometimes one works better than the other.

I had in my last rig some issues with my corsair kit, was on the QVL and memtest was 100% OK but it would crash in DX11 games. Had to increase CAS plus tRAS timings to maintain 3200 speed, also had to increase vSOC and work around VDD/Q voltage.

I would always follow QVL, its too risky when its not there.

PS: I would also be very careful with vSOC on 7000 series, don't go anywhere over 1.3v

Here's the same pointers (plus some more) in a video format.
Good Luck!

[elstaci]

Googling " ATTEMPTED WRITE TO READONLY MEMORY " found these two websites that gives good troubleshooting tips:

https://www.easeus.com/partition-manager-software/fix-attempted-write-to-readonly-memory-bsod-in-win... :

The following things could cause 0x000000BE error:

• Faulty or corrupt device driver
• Incompatible firmware software
• Corrupt system services

0x000000BE is Attempted Write To Readonly Memory BSOD error.

Workable Solutions	Step-by-step Troubleshooting
Fix 1. Rebuild MBR	Download and launch EaseUS Partition Master on a new PC which runs the same system as yours...Full steps
Fix 2. Enter Safe Mode	Restart your PC by pressing the Power button. Once the related options show up, hold the "Shift" key...Full steps
Fix 3. Check RAM	Press Win + R to open the Run dialog box. Type mdsched.exe in the blank area...Full steps
Fix 4. Run CHKDSK	Press the Windows key and type cmd. Right-click "Command Prompt"...Full steps
Other Useful Fixes	Update Device Driver > Uninstall Recent Software > Check for Windows 10 Update > Restore System...Full steps

also https://www.diskinternals.com/partition-recovery/attempted-to-write-to-read-only-memory/

Both does mention to check your RAM Memory. Since the RAM passed using MEMTEST86 all that would tell you was the RAM isn't defective BUT not whether it is compatible or not.

As mentioned in another reply just install one RAM Stick and see if the error occurs in DIMM Slot# 2. If it doesn't occur install a second RAM stick in DIMM # 4 and see if the error happens again.

This previous Microsoft Forum concerning amdgpio3.sys  Driver Verifier is very similar to your situation:

https://answers.microsoft.com/en-us/windows/forum/all/frequent-bsod-driver-verifier-suggests-it-migh...

Here is a couple of replies from MS Users from above thread: