MP 1,1-5,1 Manually Configured OpenCore on the Mac Pro

[h9826790]

What is the reason for these can you please elaborate?

Just one more way for us to ident the user is using which config.plist (when someone ask question).

Of course, we can ask them to check the OC version in terminal, but that won't help to ident if the user is using my config, or cdf's config, or his own config, etc.

By injecting different version when we publish new config, it will be easier for us to understand what the user did.

e.g. If someone complain he can't get DRM streaming works by using my 0.6.6 package, then I can ask him to check the BIOS version. If that's not 9144.0.6.6.0, then it means he isn't really running my package yet (e.g. he tried to build his own 0.6.6 config, but failed. Then try to use my package, but also failed to replace the existing OC). And I know straight away that I don't need to check my package, but just need to help him to install the newer OC package.

It actually helps me to save quite a bit of time, and provide more useful suggestion.

Some users actually mix use my package, cdf's config, and even some other OC sources on the net. Since this 9144.x.x.x.x should be good enough to provide the extra protection in short future. Therefore, I think we can use the numbers to help us in other area as well. But not just to avoid unwanted firmware update.

TBH, this may be only meaningful for cdf and me. All other OC users can use whatever number they want, and 9999.999.999.999.999 should provide the best protection.

 

[cdf]

From my own experience, it's actually better to ProcessorType=1281 in the config.

This is a sure bet. It corresponds to manually setting the Xeon processor type (0x0501). However, according to the OpenCore reference manual, "automatic value generation tries to provide most accurate value for the currently installed CPU. When this fails please make sure to create an issue and provide sysctl machdep.cpu and dmidecode output." So perhaps we should report the offending CPU.

 

[startergo]

This is a sure bet. It corresponds to manually setting the Xeon processor type (0x0501). However, according to the OpenCore reference manual, "automatic value generation tries to provide most accurate value for the currently installed CPU. When this fails please make sure to create an issue and provide sysctl machdep.cpu and dmidecode output." So perhaps we should report the offending CPU.

Does it fail without any spoofing ?

 

[cdf]

Does it fail without any spoofing ?

If you mean UpdateSMBIOS=false, then I don't believe so, because nothing is changed in the first place...

 

[cdf]

I wonder if this is a bug in Apple's implementation: according to the NVRAM variable, the original firmware-features value for the Mac Pro 5,1 is A1QMwA== (0xC00C5403), but according to the SMBIOS table, the value is AxQMwA== (0xC00C1403). A difference of one bit...

Regarding the SMBIOS properties: PlatformFeature and SmcVersion are not present in the original table, so it makes sense not to specify them. The failsafe for PlatformFeature is to not provide this value (although dmidecode indicates that it gets set 0) and the failsafe for SmcVersion is to ignore the value (confirmed with dmidecode).

 

[amstel78]

Greetings to all. I decided to install OpenCore v066 on my MP3,1. Everything appears to function fine except for two issues:

1. A setting (perhaps GOP related) in OpenCore is somehow overriding my 7950 with Mac BIOS meaning that it does not show the grey Apple loading screen at boot. All I get is a grey blank screen after selecting an OS to boot. After perhaps 10-15 seconds, the black Apple loading screen will appear followed by the login prompt.
2. Attempting to install Big Sur results in a hard reboot after selecting the Mac OS Installer in the OC boot picker. I started the installation from within patched Catalina (10.15.7 - DosDude1) using OC alternate with VMM flag enabled. This worked fine for me on my MP5,1. After initial Big Sur drive selection and preliminary install, Catalina reboots and brings me to RefindPlus/OC boot picker. I would select the Mac OS Installer option under OC alternate, which then results in a hard reboot.

I'm attaching a copy of my config.plist (both regular and alternate) for review. Grateful if someone could assist me with troubleshooting. Many thanks.

 

[Dayo]

Just on a hunch, try setting "ProvideConsoleGop" to "true" for Issue 1.

IIRC, while it is there to make MacPro5,1 show the OC bootpicker, I think it has the side effect of fixing the Stage 1 Loading Screen when broken on MacPro3,1.

 

[amstel78]

Just on a hunch, try setting "ProvideConsoleGop" to "true" for Issue 1.

IIRC, while it is there to make MacPro5,1 show the OC bootpicker, I think it has the side effect of fixing the Stage 1 Loading Screen when broken on MacPro3,1.

Thanks. I tried changing ProvideConsoleGop to "true" in config.plist. Unfortunately no effect; still blank gray screen until black Apple loading sequence.

That said, what does DirectGopRendering do? It's currently set to "false." Setting it to "true" has no effect either - at least on my flashed 7950.

 

[Dayo]

I tried changing ProvideConsoleGop to "true" in config.plist. Unfortunately no effect;

Raise an issue on the OC Github project for insights.

@startergo suggested something earlier. No harm in trying that first.

What does DirectGopRendering do?

Something about GOP for some GPUs ... Refer to the OC Manual for information about params.

 

[TECK]

@cdf what is the FirmwareFeatures fix @h9826790 is referring to? Failsafe values are:

'PlatformInfo': {
    'PlatformNVRAM': {
        'FirmwareFeatures': Data(''),
        'FirmwareFeaturesMask': Data('')
    }
}

Should we also leave the KernelArch to Auto? It should always pick the x86_64 value, I ran it on Auto for a while.

 

[cdf]

• A setting (perhaps GOP related) in OpenCore is somehow overriding my 7950 with Mac BIOS meaning that it does not show the grey Apple loading screen at boot. All I get is a grey blank screen after selecting an OS to boot. After perhaps 10-15 seconds, the black Apple loading screen will appear followed by the login prompt.

So the boot picker is working, but the first stage loading screen isn't, correct? To help diagnose where the gray blank screen is coming from, try setting DefaultBackgroundColor to AAAAAA== and see if that changes anything.

• Attempting to install Big Sur results in a hard reboot after selecting the Mac OS Installer in the OC boot picker. I started the installation from within patched Catalina (10.15.7 - DosDude1) using OC alternate with VMM flag enabled. This worked fine for me on my MP5,1. After initial Big Sur drive selection and preliminary install, Catalina reboots and brings me to RefindPlus/OC boot picker. I would select the Mac OS Installer option under OC alternate, which then results in a hard reboot.

This happen can happen if the selected disk is not APFS formatted.

 

[cdf]

@cdf what is the FirmwareFeatures fix @h9826790 is referring to? Failsafe values are:

'PlatformInfo': {
    'PlatformNVRAM': {
        'FirmwareFeatures': Data(''),
        'FirmwareFeaturesMask': Data('')
    }
}

Should we also leave the KernelArch to Auto? It should always pick the x86_64 value, I ran it on Auto for a while.

The fix concerns the SMBIOS values. It seems that installing Big Sur from Catalina still needs the firmware features. I've put them back in the guide. In fact, the failsafes for FirmwareFeatures and FirmwareFeaturesMask are 0, which effectively changes the original values (0xC00C1403 and 0xFF1FFF3F). I'm currently taking a closer look at the SMBIOS values with failsafes different than the original values. As for KernelArch, it shouldn't really matter. If you want, we can use Auto. But we should probably stick with x86_64 for consistency with the samples provided by Acidanthera.

 

[amstel78]

So the boot picker is working, but the first stage loading screen isn't, correct? To help diagnose where the gray blank screen is coming from, try setting DefaultBackgroundColor to AAAAAA== and see if that changes anything.


This happen can happen if the selected disk is not APFS formatted.

Thanks. Yes, boot picker functions but 1st stage loading screen remains a blank gray. My 7950 is flashed so shows the boot screen natively if not using OC. Will try changing background color as suggested and report back.

As for issue number 2, the disk is APFS formatted. That said, my MP3,1 has had its boot ROM flashed to boot APFS drives natively. Could that be throwing something off with OC?

 

[cdf]

Could that be throwing something off with OC?

I wouldn't think so if you're already successfully booting Catalina through OC.

 

[TECK]

If you want, we can use Auto.

I applied the Firmware Features, let's use the failsafe Auto, one less setting to fiddle with. See #77, VMM flag should stay on?

 

[amstel78]

I wouldn't think so if you're already successfully booting Catalina through OC.

Ok. Well, I tried setting DefaultBackgroundColor to AAAAAA== and that didn't do anything. First stage is still blank.

 

[cdf]

Ok. Well, I tried setting DefaultBackgroundColor to AAAAAA== and that didn't do anything. First stage is still blank.

Still blank, but still gray?

 

[cdf]

let's use the failsafe Auto, one less setting to fiddle with.

See my edit to the post.

 

[TECK]

See my edit to the post.

The goal is to have the examples/setup.py identical to your settings, Let me check VMM flag, anything else that changed and I missed? Edit, for now I'll leave VMM flag on, if you want me to change it to off, let me know.

 

[amstel78]

Still blank, but still gray?

Yes, still blank but gray.

To recap:

1. Soft bless and reboot
2. Select standard OC instance
3. Select Catalina
4. Boot picker disappears. Screen remains gray for about 10-15 seconds
5. Second stage (black screen/Apple logo with progress bar) appears
6. Login Window

Disabling OC, MP3,1 reboots with grey Apple 1st stage loading screen. Very weird indeed. I tried different GPU ports just to rule that out. It makes no difference whether connected to HDMI or miniDP. Once OC starts though, Catalina runs fine.

 

[TECK]

But we should probably stick with x86_64 for consistency with the samples provided by Acidanthera.

I just saw your edit, will add it back. Fixed on #78

 

[cdf]

Yes, still blank but gray.

To recap:

1. Soft bless and reboot
2. Select standard OC instance
3. Select Catalina
4. Boot picker disappears. Screen remains gray for about 10-15 seconds
5. Second stage (black screen/Apple logo with progress bar) appears
6. Login Window

Disabling OC, MP3,1 reboots with grey Apple 1st stage loading screen. Very weird indeed. I tried different GPU ports just to rule that out. It makes no difference whether connected to HDMI or miniDP. Once OC starts though, Catalina runs fine.

The gray background is from the boot picker, which is in text mode, so perhaps the text-to-graphics transition is broken. To verify this, try booting in verbose mode to see if the blank screen actually gets filled with text.

 

[Dayo]

I tried changing ProvideConsoleGop to "true" in config.plist. Unfortunately no effect;

Thought about it and yes, would not work for HD 7950 which doesn't use GOP to start with.
Unless when you said it was flashed, you meant it was flashed with GOP.
Attach your debug logs (including from RP).

try booting in verbose mode

You can set this from RP by "uncommenting" set_mac_boot_args "-v" in your overrides config file.

On Issue 2, are you sure your HD 7950 supports Big Sur in the first place? You might want to check on the GPU Compatibility or similar threads as it is a rather old GPU to expect Big Sur support from. Maybe ask the specific question in a new thread.

 

[Bmju]

Yes, still blank but gray.

To recap:

1. Soft bless and reboot
2. Select standard OC instance
3. Select Catalina
4. Boot picker disappears. Screen remains gray for about 10-15 seconds
5. Second stage (black screen/Apple logo with progress bar) appears
6. Login Window

Disabling OC, MP3,1 reboots with grey Apple 1st stage loading screen. Very weird indeed. I tried different GPU ports just to rule that out. It makes no difference whether connected to HDMI or miniDP. Once OC starts though, Catalina runs fine.

Just a random thought, but you haven't manually enabled any extra debugging in OC? Could be slowing down the boot process?

 

[amstel78]

Thought about it and yes, would not work for HD 7950 which doesn't use GOP to start with.
Unless when you said it was flashed, you meant it was flashed with GOP.
Attach your debug logs (including from RP).


You can set this from RP by "uncommenting" set_mac_boot_args "-v" in your overrides config file.

On Issue 2, are you sure your HD 7950 supports Big Sur in the first place? You might want to check on the GPU Compatibility or similar threads as it is a rather old GPU to expect Big Sur support from. Maybe ask the specific question in a new thread.

Hi Dayo. No, the 7950 is flashed with a Mac EFI bios.

As for BS support, it should be. According to this link, the 7750 is fully compatible using AMD7000Controller.kext. in principle, the 7950 being slightly newer, should also work using the same kext.

AMD Radeon HD 7750: supported in macOS Mojave / Catalina / Big Sur / Monterey

Final Notes : Mac OS Ventura has dropped support for HD 7750. It has been a great run folks running mac os on this amd gpu. We have really pushed this budget gpu as far as we could! It remains my best ever budget gpu for mac os to date! Dear Members, Mac OS Catalina I am pleased again to confirm ...

osxlatitude.com

Specifically regarding item 2, spoke to @freqrider who told me that BS, if on the same drive must be in it's own container. In my case, I only had it in a separate volume in the same container of a single SSD also containing Catalina. Perhaps that is the cause of the hard reboot?

Anyway, I will post the debug logs and test while in verbose mode later today.