macOS Tahoe 26 on Unsupported Macs Discussion

[Oxygen-X1]

If you tell me which graphics card you’re using and whether you’re working with Hackintosh tools (Clover/OpenCore, WhateverGreen, Platform IDs), I can help you determine if and how QE/CI can be enabled.

Thanks for the effort but I don't do Hackintosh. The AMD Radeon, will do a fine job when the Devs have everything in place, I will wait for that, in the meantime for everyday use T26 is fine for my needs, I even find it to be very quick in Firefox, I do not use this computer for work. I did find out that launchpad has been replace with "apps", it opens but then quickly disappears, that does not bother me because I never use it, the finder menu bar is fine.

Thanks for your offer to help me figure out if QE/Ci works on my nVidia (Kepler chipset).
I already know that it doesn't!! work, just like your AMD Radeon not work.

And your AMD Radeon will work fine once the developers have everything set up,
or they'll never get it done! I'm still waiting for that... Me too.

And I don't use a Hackintosh! You should check out my signature.

 

[HippoInWindow20]

Note: feel free to correct me on anything here, this is just from my own testing and investigation.

I've done some testing with Tahoe on a 2018 MBP (MacBookPro15,1). OCLP is not an option, since the 2018 was never unsupported before and OCLP hasn't been updated yet. So I went with OpenCore and some very basic spoofing. My goal is spoof to 16,1 and see what happens.

RECOVERY
My first attempt was recovery. Booting into recovery directly through OpenCore is not possible. Instead, you have to manually attempt to boot it without OC, which will initiate an iBridge update. After iBridge updates, Tahoe Recovery can be booted through OpenCore with no issues on 15,1. WiFi, external display, SSD, etc. all work fine.

INSTALLATION
Then there is a huge blockade: macOS Installer. Since I did not hide my T2, the installer checks for this multiple times during installation:

1. When you click Continue on the first page
2. After extracting SharedSupport.dmg
3. After extracting the payload .aea file
4. After extracting the actual BaseSystem
5. During first reboot phase

These errors occur due to a personalization step in the installer. When it sees the T2, it checks the plists to see if any extra work needs to be done during installation to accommodate for it (eg. update iBridge). By manually extracting SharedSupport and changing all references to the MBP16,2 chip (J152FAP) to the 15,1 one (J680AP), the first two checks are circumvented.

The third check is a bit different, though. The aea file is signed, so you cannot modify it beforehand. This can be circumvented, but requires precise timing. You have to manually patch the plists after the installer extracts them to temporary directories, but before the installer reads them. From testing, you have about 10-15 seconds to do this.

I have not been able to get past the last two checks. They seem to be in plist files deep into the installer and I have not found a way to fix this.

BOOTING FULL OS
After fumbling with the installer for way too long, I decided to just use another Mac to preinstall the OS on an external drive. macOS supports booting from external media, so this is fine. Unfortunately, this was unsuccessful. The system attempts to boot but fails while searching for the root device. More specifically, the APFS driver cannot verify the system image. I have no idea why this happens.

On Macs without T2, the same external SSD works fine for booting Tahoe (even on 9,1!). But it seems that T2 Macs file system verification works a bit differently. I think the OS expects a root hash from UEFI, but OpenCore does not provide one. The APFS driver receives null data, and subsequently panics with the following message:

panic(cpu 0 caller 0xffffff801f441676): "The global payload bytes pointer is NULL\n" @apfs_vfsops.c:2736

MY EXPECTATION
In Recovery, ioreg properly shows all hardware. It also seems to have attached all drivers. Considering that, I fully expect Tahoe to be possible on this Mac. The APFS issues seem like they are solvable, as long as it is possible to provide the data the OS expects on boot for filesystem verification.

Okay that explains why I didn't got it to work on a 2019 13-inch (MacbookPro15,4). Looks like the APFS Driver panic is T2-related.

panic(cpu 0 caller 0xffffff80096489ce): "The global payload bytes pointer is NULL\n" @apfs_vfsops.c:2738

 

[Portiere]

After 2 days I installed tahoe on my mac mini 2012. The only problem is there is not metal lib pkg for it. How to have it? Thanks a lot

 

[iMac-Oldschool]

After 2 days I installed tahoe on my mac mini 2012. The only problem is there is not metal lib pkg for it. How to have it? Thanks a lot

You have a simple choice: join the dev team and write it yourself, or wait for somebody else to write it for you.

 

[Portiere]

Thanks a lot.

 

[iMac-Oldschool]

Thanks a lot.

Veuillez pardonner la réponse inutile. Beaucoup d'entre nous attendent une solution. La première version de Tahoe ne s'est pas installée; mais maintenant nous pouvons au moins faire les premiers pas.

 

[hvds]

You have a simple choice: join the dev team and write it yourself, or wait for somebody else to write it for you.

...recently tried it myself for my MBP11,1, exploring https://github.com/dortania/MetallibSupportPkg. It was a short go (one full day) and it formally worked and produced a metallib pkg. Using it in root patching worked to some extent but several instances of metallib were missing, because during patching them in the MetallibSupportPkg failures occurred so I skipped them. Resulting system of course didn't boot.
This was an interesting exercise and I admired the ingenuity in this piece of OCLP software. But getting it to really, not only formally work was far beyond my capabilities.

 

[Neon Ball]

After 2 days I installed tahoe on my mac mini 2012. The only problem is there is not metal lib pkg for it. How to have it? Thanks a lot

...recently tried it myself for my MBP11,1, exploring https://github.com/dortania/MetallibSupportPkg. It was a short go (one full day) and it formally worked and produced a metallib pkg. Using it in root patching worked to some extent but several instances of metallib were missing, because during patching them in the MetallibSupportPkg failures occurred so I skipped them. Resulting system of course didn't boot.
This was an interesting exercise and I admired the ingenuity in this piece of OCLP software. But getting it to really, not only formally work was far beyond my capabilities.

The MetallibSupportPkg is just a support package like it says, it supplements the root patches to patch some issues with built-in applications and such to work with the patched drivers and frameworks. Full Tahoe graphics patches aren't ready or public yet, so the package has no use without them.

 

[Portiere]

thanks a lot to imacoldschool - hvds and neon ball for their answers. I will wait OCLP and continuing using Sequoia.

 

[hvds]

The MetallibSupportPkg is just a support package like it says, it supplements the root patches to patch some issues with built-in applications and such to work with the patched drivers and frameworks. Full Tahoe graphics patches aren't ready or public yet, so the package has no use without them.

Thank you for explaining, always interesting and much appreciated.

 

[ronton3]

Can someone please verify if this or is not the latest 3.0.0 nightly build. Please note that out of respect for the Dev's there are no clickable links shown.

 

[macpro_mid2014]

View attachment 2534408

Can someone please verify if this or is not the latest 3.0.0 nightly build. Please note that out of respect for the Dev's there are no clickable links shown.

This is the last modification to OCLP 3.0.0 beta:

Artifacts​

Produced during runtime

Name	Size	Digest
AutoPkg-Assets.pkg	709 MB	sha256:1ad26a405b0ed4d9b010980c8ab0db5e529473cfcb226a3c49dce1aa5fcc6750
OpenCore-Patcher-Uninstaller.pkg	156 KB	sha256:5e2f8f5b5667a66c04b60d03e85aea42a71662a5d9576a466c06f628af0e5aa5
OpenCore-Patcher.pkg	709 MB	sha256:a98df2d4ce41d70347f02bd2ad645d5217399668185e91e039db7237efa925c9

 

[ronton3]

OpenCore-Patcher.pkg 709 MB sha256:a98df2d4ce41d70347f02bd2ad645d5217399668185e91e039db7237efa925c9

Thank You for replying, your number is much different than mine. I have just been refreshing the page thinking that when a new update was available it would change, now I finally know that is not the case. I hope now I can get back to the right place. I don't want to impose but help would be appreciated. Thanks again.

Edit: I have blundered into a way to access the latest thanks to your help. For some reason my T26 test volume had not been enabling Btooth the other mouse or keyboards, I could access the desktop but that was it. Now using the latest nightly and installing it from another volume on the T26 volume the mice and keyboards are working again. What a relief, now I can continue my fun experiments. I am impressed by how snappy Firefox is onTahoe apps are slower, of course I have not transferred any previous info to the volume except Firefox(synced) and mail.

Edit: T26b5 is available: I am going to let things settle down before downloading. The new feedback assistant which I have never used before, contains loads of info about updates etc.

 

[Dimvol]

A1707 2017. Full support for HD630 and Baffin Radeon 555 Pro has been discovered in the public beta.

No fixes or patches. Works as is.
Radeon began to use a different port, for this reason it did not work on the internal display. After fixing it works. There is no black screen when starting the Photo application and the internal display does not go out when connecting an external monitor. Radeon passes tests (for example, Novabench).

There is no more good news. The screen brightness is not adjustable. Black screen after waking up. Probably T1 will work conditionally: system accept the user password, but not accept the administrator password.
Support for previously unsupported video accelerators may be temporary or erroneous.

 

[ronton3]

T26b5/3.0.0n installed OTA 3.01GB, it took a restart for everything to come up but all is working as before, some small differences, the apps that replaced launchpad still quickly disappears, but it can be opened from the finder. Still no patches from OCLP. There was a mail update. Firefox works very well. Safari looks very different but I don't use it.

 

[Dimvol]

Brightness ajustment appeared. Now everything is fine with the video system.

 

[deeveedee]

Smooth OTA upgrade from Tahoe Beta 4 -> Beta 5 on my MBP6,2. OTA upgrade downloaded the full 16.7GB installer in my case. Still patiently waiting for OCLP non-metal post-install patches. At this time, my bar is low, so I'm just impressed that the install completes and I can login.

Note that I did need to decrypt my Tahoe volume after applying the Beta 4 -> Beta 5 update (I believe this is because the update was applied with the full installer and not an incremental installer) by following the steps here.

 

[thegdn]

looks good... no wifi... yet...

 

[thegdn]

looks good with no patching... yet...

 

[amaze1499]

Can someone get back to me about performance comparison Taheo vs Sequoia already? A feedback along the lines "Better, same, worse" would be appreciated!

 

[sidj]

Looks like OCLP development is either slowed or stopped.

Last commit on GitHub was 3 weeks ago. Last year around same time I was running Sequoia on my 2017 MBP.


Any insider information, if the work is going on in background?

 

[amaze1499]

Looks like OCLP development is either slowed or stopped.

Last commit on GitHub was 3 weeks ago. Last year around same time I was running Sequoia on my 2017 MBP.


Any insider information, if the work is going on in background?

There is some work going on. A nightly build is in development. Check out the nextMacOS branch.

 

[hvds]

Looks like OCLP development is either slowed or stopped.

Last commit on GitHub was 3 weeks ago. Last year around same time I was running Sequoia on my 2017 MBP.


Any insider information, if the work is going on in background?

No insider information - but the last commit to the main branch (Sequoia) was 11hrs ago, to macos-next (Tahoe) as you said.
Booting (not patching) seems to work for most systems, as for my MBP11,1 and MBP5,2, and that is already something. Patching takes longer development, but at least for my mentioned machines, I can do partial patching with a slightly source modified OCLP 3.0.0n.
As Apple is removing Intel components progressively from Tahoe, I imagine root patching to become even more difficult than for Sequoia. So that will take time, fingers crossed and all the best to the developers.

 

[ronton3]

Can someone get back to me about performance comparison Taheo vs Sequoia already? A feedback along the lines "Better, same, worse" would be appreciated!

Many things do not work others do, for instance on my Imac 2015, 17.1 the Tahoe background will briefly appear and then retreat, but I can use a color and it stays in place, same with the screen saver, I use Flurry it has a muted response but is present, the clock works but it is in Cupertino time because location services do not work, Firefox is quick, but apps from Apple are slower to open and activate sleep does not work, at the same time mail and entry level things do. Much information is in Feedback assistant. This is still a testing and experimenting phase, best on a simple separate volume you do not mind deleting, at the same time I am logged into my Apple acct. It can't really be compared to Seq as a work source, thankfully the Devs are still working.

 

[Oxygen-X1]

Many things do not work others do, for instance on my Imac 2015, 17.1 the Tahoe background will briefly appear and then retreat, but I can use a color and it stays in place, same with the screen saver, I use Flurry it has a muted response but is present, the clock works but it is in Cupertino time because location services do not work, Firefox is quick, but apps from Apple are slower to open and activate sleep does not work, at the same time mail and entry level things do. Much information is in Feedback assistant. This is still a testing and experimenting phase, best on a simple separate volume you do not mind deleting, at the same time I am logged into my Apple acct. It can't really be compared to Seq as a work source, thankfully the Devs are still working.

You think:thankfully the Devs are still working.....If that's the case ???