Contribute
Register

Fix for failure NVIDIA Web Driver on High Sierra, black screen, panics.

Status
Not open for further replies.
i do all the steps by #1, and install the 378.10.10.10.15.120 driver.
but aslo blank screen.

but, i connected by remote desktop and changed the display setting, then the screen is ok.
so, to blank screen uses, perhaps you should connect the remote destop and change the display setting.

Check clover configuration.
 
This post addresses very specific issue which seems to be common problem world-wide - black screen on infinite reboot loop after installing NVIDIA Web Drivers.

View attachment 284837

You have installed High Sierra on your comp or upgraded Sierra into High Sierra and everything looks nice expect you month have proper graphics with your NVIDIA graphics card. So your next step is to install NVIDIA Web drivers and instead of having nice and smooth picture you got black screen or infinite reboot loop. If you upgraded from Sierra and have updated Sierra several times you might scratch your head, what has happened now, why previous methods doesn't work.

I assume that you read this topic because you already got into trouble and doesn't know how to solve your problem with NVIDIA web drivers.

How to get back access to the High Sierra GUI?

This is your primary goal once you lost access to your Desktop. According to various web recourses an easiest option is to remove NVDAStartupWeb.kext from /Library/Extension. If you have installed High Sierra properly you should have access to the Recovery Mode. So reboot you comp into Recovery mode and launch Termina and remove NVDAStartupWeb.kext from /Library/Extension. Replace HDD with name of your High Sierra startup disk.

Code:
rm -R /Volumes/HDD/Library/Extensions/NVDAStartupWeb.kext
touch /Volumes/HDD/Library/Extensions/
reboot

or commands below to clean completely NVIDIA Web Drivers.
Code:
cd /Volumes/HDD/System/Library/Extensions/
rm -R GeF*.*
touch /Volumes/HDD/System/Library/Extensions/
cd /Volumes/HDD/Library/Extensions/
rm -R GeF*.* NVDA*.*
touch /Volumes/HDD/Library/Extensions/
reboot

Do some cleanup once you gain access to your Desktop

Once you got access back to your Desktop you have to do some cleanup. Mount your EFI partition and remove all these kexts:
  • Lilu.kext
  • NVWebDriverLibValFix.kext
  • NvidiaGraphicsFixup.kext
Check folders /System/Library/Extensions/ and /Library/Extensions/ for these kexts and remove these. Also check /Volumes/EFI/EFI/CLOVER/kexts/Other etc for these kexts and remove these! Also remove FakeSMC_GPUSensors.kext etc.

Uninstall completely NVIDIA Web Drivers.

Choose proper location for your kexts

I recommend to install your hackintosh kexts into /Volumes/EFI/EFI/CLOVER/kexts/Other, which makes things a lot easier on High Sierra.

Install Clover driver EmuVariableUefi if not installed
Add EmuVariableUefi.efi into /EFI/CLOVER/Drivers64UEFI. With CLover Configuratior you can easily to do it from panel Install Drivers. On section "DRIVERS UEFI 64 BIT" should be button EmuVariableUefi grey. If not, then press it and EmuVariableUefi will be installed.

Download and install new NVIDIA Web Drivers fix kexts
Usage of Lilu.kext, NVWebDriverLibValFix.kext and NvidiaGraphicsFixup.kext together can cause your HS to loop reboot as the latest NvidiaGraphicsFixup.kext already has built in the NVWebDriverLibValFix.kext!

Use only the latest:
Do not use NVWebDriverLibValFix.kext with them!

Reboot

Enable SIP temporarily
To avoid error during installation of NVIDIA Web Drivers you might choose to enable SIP temporarily. Edit Colver configuration file, set SIP related setting CsrActiveConfig to 0x00. Reboot.

Install NVIDIA Web driver

Download proper version of NVIIDA web drivers for your High Sierra.
Once you have got proper version of NVIDIA web drivers follow this procedure:
  • Mount EFI partition of your Clover
  • Edit Colver configuration file, set SIP related setting CsrActiveConfig to 0x67
  • Do not reboot yet!
  • Install Nvidia Web Driver and reboot as installer demands.

Polishing + installing CUDA
Install other kexts which might be relevant for your system.

If you want to use CUDA you might need to patch it to work with current Nvidia drivers, as CUDA Framework is prepared to read only specific version of the GPU Drivers, and with the last update at the moment, it doesn't recognise the installed version.
  • Check the driver version from CUDA preference panel (for example GPU Driver Version: 10.25.28 378.10.10.10.15.117)
  • Go to /Library/Frameworks/CUDA.framework/Versions/A/Libraries
  • Find closest dylib match to your driver version, on my case for 378.10.10.10.15.117 and duplicate it. Rename duplicated file, on my case libcuda_378.05.05.25_mercury copy.dylib into libcuda_378.10.10.10_mercury.dylib.
  • Copy /Library/Frameworks/CUDA.framework/Versions/A/Resources/Info.plist on yourDdesktop
  • Add related lines at the end of NVDAResmanVersions section
Code:
<key>378.10.10.10</key>
<string>378.10.10.10</string>
  • Replace /Library/Frameworks/CUDA.framework/Versions/A/Resources/Info.plist with an updated file.
  • run these commands on terminal:
Code:
sudo kextunload -b com.nvidia.CUDA
sudo kextload -b com.nvidia.CUDA

Voila, you should have CUDA working now.

Hardware Acceleration Issues
If macOS High Sierra is logging you out while you’re using Chrome, try opening up the Chrome web browser and going to the settings. At the bottom of the screen click ‘Advanced’, then turn off hardware acceleration.

It’s VERY IMPORTANT also, for many users having the forbidden sign when rebooting, to clear de pram/nvram. This can be done using this command in the console:

nvram -c

It helps also clearing the kext cache using touch and kextcache -u

I hope this helps, cause it saved my day :)
 
It’s VERY IMPORTANT also, for many users having the forbidden sign when rebooting, to clear de pram/nvram. This can be done using this command in the console:

nvram -c

It helps also clearing the kext cache using touch and kextcache -u

I hope this helps, cause it saved my day :)

Please can you be more specific and explain exactly what and why?
 
Unfortunately this didn't fix the problem for me. I already tried it several times with different SMBIOS and a couple of clean installs. I just managed to use Screen Sharing and noticed something.
It seems that it tries to connect to the monitor which would be similar for the chosen Model in the SMBIOS.



Edit: It seems that this is not everytime the case. For the 17,1 SMBIOS it uses a 23" Built-in-Display (Wouldn't fit for the real Monitor nor the iMac).
 
Hi,
When I set CsrActiveConfig to 0x00 my hackintosh crash immediately when booting,
 

Attachments

  • image.jpg
    image.jpg
    2.1 MB · Views: 401
Thanks!! This fixed my problems after trying other fixes in the other main driver thread without success.
 
Please can you be more specific and explain exactly what and why?
Enabling SIP will prevent unsigned kexts being included in the kernel cache. For example, a user who only has FakeSMC installed to /L/E enables SIP and installs the web driver. The cache is invalidated and/or updated as part of the installation process, excluding unsigned kexts due to SIP being turned on. The next boot, FakeSMC is not in the cache, whatever CSR flags have now been set, and the user is unable to reach the login screen. Same applies to other unsigned kexts not being injected by the bootloader.
 
Ugg, I can’t win with Nvidia to save my life.

I tried all of these steps but still am in a reboot loop.

Any other suggestions?

*** update ***
@Raouws had some additional ideas that, combined with this post, finally made everything work.

https://www.tonymacx86.com/threads/macos-high-sierra-10-13-supplemental-update.233897/page-48

Thanks much! I was ready to jump off a bridge
Hi.
I have same videocard as you and have a problem running Davinci on CUDA. Want to ask you - did you try run it by any chance?
 
Last edited:
Status
Not open for further replies.
Back
Top