Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
That is bizarre... I don't experience this on my system at all. Nothing has changed in the way that I build/sign the executables for quite a while. Is anyone else getting this issue too?
Yes, on my MBP5,2 I also get the ad-hoc-signed messages, regarding CoreDisplay only.
I already happens right after OTA and post-install patch with 1.3.0, including the legacy video card patch, see post #8203.

Then I replaced the CoreDisplay+CoreDisplayOriginal in
/System/Library/Frameworks/CoreDisplay.framework/Versions/A/
and SkyLight+SkyLightOriginal in
/System/Library/PrivateFrameworks/SkyLight.framework/Versions/A/
with the versions provided in your post #8213.
No change, the boot process still hangs at the same place.
Photos taken at the time when the boot stalls are attached (left and right part of screen). Looks the same when using safe mode.

Looks like a change in 10.15.4 causes this as it happens already with the versions used in patcher 1.3.0 ?
 

Attachments

  • IMG_3699.JPG
    IMG_3699.JPG
    591.7 KB · Views: 601
  • IMG_3701.JPG
    IMG_3701.JPG
    449.7 KB · Views: 321
Last edited:
  • Like
Reactions: TimothyR734
Yes, on my MBP5,2 I also get the ad-hoc-signed messages, regarding CoreDisplay only.
I already happens right after OTA and post-install patch with 1.3.0, including the legacy video card patch, see post #8203.

Then I replaced the CoreDisplay+CoreDisplayOriginal in
/System/Library/Frameworks/CoreDisplay.framework/Versions/A/
and SkyLight+SkyLightOriginal in
/System/Library/PrivateFrameworks/SkyLight.framework/Versions/A/
with the versions provided in your post #8213.
No change, the boot process still hangs at the same place.
Photos taken at the time when the boot stalls are attached (left and right part of screen). Looks the same when using safe mode.

Looks like a change in 10.15.4 causes this as it happens already with the versions used in patcher 1.3.0 ?

Apparently apple have enforced their AMFI signature security checks, I'd exclude that OTA swu script has done something wrong for a simple reason, if you use only these stock untouched from Catalina 10.15.3 or 10.15.4 :

CoreDisplay.framework , OpenGL.framework, SkyLight.framework, GPUSupport.framework, the 10.15.4 system will still boot correctly to the GUI mode (without video acceleration) from an unsupported mac.

Meanwhile you could try this from a recovery or installer terminal:
nvram boot-args="-no_compat_check -v amfi_get_out_of_my_way=0x1"
 
Apparently apple have enforced their AMFI signature security checks, I'd exclude that OTA swu script has done something wrong for a simple reason, if you use only these stock untouched from Catalina 10.15.3 or 10.15.4 :

CoreDisplay.framework , OpenGL.framework, SkyLight.framework, GPUSupport.framework, the 10.15.4 system will still boot correctly to the GUI mode (without video acceleration) from an unsupported mac.

Meanwhile you could try this from a recovery or installer terminal:
nvram boot-args="-no_compat_check -v amfi_get_out_of_my_way=0x1"
I just set the boot-args as you said.
The boot process still hangs with the same adhoc-signed message for CoreDisplay...
There may be more AMFI options ?
 
I just set the boot-args as you said.
The boot process still hangs with the same adhoc-signed message for CoreDisplay...
There may be more AMFI options ?

There are, but I don't know much about AMFI, I was just reading this useful article:


Anyway the CoreDisplay and SkyLight's Resources subfolders are also important, for example the SkyLight's one contains the WindowServer, not sure if @ASentientBot using the latest one compiled by apple on February 3 or the Resources from Mojave.
 
Last edited:
Out of interest, if you've used this tool, DM me with your thoughts, version, and device. I'm interested to see how spoofing a supported Mac model performs and effects firmware updates (which if installed onto the wrong device, would probably kill said device)
I did as the installer stated, made the usb drive, rebooted into it.
it just stays in the boot screen.
 
I looked through the thread and I didn’t see exactly what I was looking for. I apologize if this has been answered before. I am looking at buying a 2012 Mac Pro and I am wondering what in addition to either Juliens or DOSdudes installer might be necessary to get Catalina running.

have heard conflicting stories

Running a 2010 Mac mini on Catalina just fine.
 
To be clear, the GTX 750M is Kepler, and the others (750 and 750 Ti) are Maxwell. Other 700 series cards might be Firmi, Kepler, or Maxwell.

Maxwell GPUs require the Nvidia web driver, so they won't work beyond High Sierra.

@joevt does that mean that Kepler Nvidia cards (even pc ones) will work in Catalina? Is there a list of which chip architecture can be run in Catalina without needing the Nvidia Web Driver?

Could something like an NVIDIA GeForce GTX 1060 SC work (which uses the Pascal architecture)? or what about the NVIDIA cards that use Turing architecture?
 
Last edited:
  • Like
Reactions: TimothyR734
it looks like there is a big problem with AMFI in the DP1. I once looked at the link to AMFI and read through it, so much can be changed by Apple.

I come after installation level 2 and do not run the patches (graphics) with dosdude1 patcher 1.3.0 to the graphical desktop.

Also not with exchanging the corresponding files that @ASentientBot has attached, plus OpenGL and GPUSupport.
No chance to see the graphical desktop on my Medion Akoya P8610 laptop (hack book),
No problems on my native hackintosh 1 and 2.
 
Hi,

I've managed to install Catalina on my beloved friend Mac Mini Late 2009 (macmini3,1).
I have an ASUS BT400 Bluetooth dongle, but I couldn't manage to make AirDrop work. I understand that I need a patched driver. I search the thread and found instructions and files for various Macs but no for the macmini3,1.

Could someone help me? Im feeling like I'm hitting a wall!

Thanks in advance
 
Damn, you guys are right about AMFI. I totally missed the change because I had already set amfi_get_out_of_my_way for some other things I was testing.

That said, it seems less like a change to AMFI itself and more that binaries like WindowServer are now using Library Validation: code signature in (/System/Library/PrivateFrameworks/SkyLight.framework/Versions/A/SkyLight) not valid for use in process using Library Validation: mapped file has no Team ID and is not a platform binary (signed with custom identity or adhoc?)

The solutions thus seem: the boot flag (easiest method, can add to Preboot's com.apple.Boot.plist), DisableLibraryValidation.kext, or maybe re-signing every binary in the OS to remove Library Validation. (I don't see any reason to do this, but I remember it working for, eg. Xcode when I modified a supporting library in ~DP2.) Patching or disabling amfid itself (as is done by jailbreaks) is also an option, but not one that I can see being necessary for a while if ever.

Thanks to @jackluke, @Pinarek, @hvds and others for catching this -- looks like @dosdude1 will definitely need to add this to the patcher.

For anyone stuck with a non-booting system, run nvram boot-args=amfi_get_out_of_my_way=1; reboot from single-user mode.

Sorry for missing this (major!) change.
 
Last edited:
@joevt does that mean that Kepler Nvidia cards (even pc ones) will work in Catalina? Is there a list of which chip architecture can be run in Catalina without needing the Nvidia Web Driver?

Could something like an NVIDIA GeForce GTX 1060 SC work (which uses the Pascal architecture)? or what about the NVIDIA cards that use Turing architecture?
Kepler: yes (metal) (older cards such as Firmi, maybe? - legacy).
Maxwell, Pascal, Turing: No.
You can use Maxwell and Pascal in High Sierra. After that, only Kepler.
 
Damn, you guys are right about AMFI. I totally missed the change because I had already set amfi_get_out_of_my_way for some other things I was testing.

That said, it seems less like a change to AMFI itself and more that binaries like WindowServer are now using Library Validation: code signature in (/System/Library/PrivateFrameworks/SkyLight.framework/Versions/A/SkyLight) not valid for use in process using Library Validation: mapped file has no Team ID and is not a platform binary (signed with custom identity or adhoc?)

The solutions thus seem: the boot flag (easiest method, can add to Preboot's com.apple.Boot.plist), DisableLibraryValidation.kext, or maybe re-signing every binary in the OS to remove Library Validation. (I don't see any reason to do this, but I remember it working for, eg. Xcode when I modified a supporting library in ~DP2.) Patching or disabling amfid itself (as is done by jailbreaks) is also an option, but not one that I can see being necessary for a while if ever.

Thanks to @jackluke, @Pinarek, @hvds and others for catching this -- looks like @dosdude1 will definitely need to add this to the patcher.

For anyone stuck with a non-booting system, run nvram boot-args=amfi_get_out_of_my_way=1; reboot from single-user mode.

Sorry for missing this (major!) change.
It's looking like DisableLibraryValidation.kext may be the best option in this circumstance... This will also be an issue for other framework patches, such as the NightShift Patch in CoreBrightness.
 
Damn, you guys are right about AMFI. I totally missed the change because I had already set amfi_get_out_of_my_way for some other things I was testing.

That said, it seems less like a change to AMFI itself and more that binaries like WindowServer are now using Library Validation: code signature in (/System/Library/PrivateFrameworks/SkyLight.framework/Versions/A/SkyLight) not valid for use in process using Library Validation: mapped file has no Team ID and is not a platform binary (signed with custom identity or adhoc?)

The solutions thus seem: the boot flag (easiest method, can add to Preboot's com.apple.Boot.plist), DisableLibraryValidation.kext, or maybe re-signing every binary in the OS to remove Library Validation. (I don't see any reason to do this, but I remember it working for, eg. Xcode when I modified a supporting library in ~DP2.) Patching or disabling amfid itself (as is done by jailbreaks) is also an option, but not one that I can see being necessary for a while if ever.

Thanks to @jackluke, @Pinarek, @hvds and others for catching this -- looks like @dosdude1 will definitely need to add this to the patcher.

For anyone stuck with a non-booting system, run nvram boot-args=amfi_get_out_of_my_way=1; reboot from single-user mode.

Sorry for missing this (major!) change.

After set nvram for AMFI verbose boot continue normally, but in my case before loginUI I get a WindowServer crash because I don't have more the 10.15.4 right version (the one compiled by apple on 3 February 2020), so I am re-downloading the entire .4 beta 1 update to retrieve those CoreDisplay/SkyLight Resources folders.

Done, but still getting a WindowServer and loginwindow pid crashes right before the loginUI.

@ASentientBot the SkyLight's Resources WindowServer executable that you are using has an apple stock Date Modified 3 February 2020 ?

If so, then I was using the right Resources subfolders, so probably something else has mismatched on my .4 beta 1 installation. Maybe some my other mistake of manual patching, but I repeat the weird fact is that if I put back the 4 stock Metal framework I can boot correctly again to the loginUI (without video acceleration).

edit:
Same behavior with a dualGPUs unsupported mac, with all post-patches and keeping stock 4 Metal Frameworks (CoreDisplay, OpenGL, SkyLight and GPUSupport), 10.15.4 beta 1 still boot correctly to the GUI.
Instead using your wrappers get WindowServer and loginwindow crashes and consequently no GUI.

Please @ASentientBot check the Resources folder Date modified files (picture attached) if match yours.
 

Attachments

  • DualGPU 4beta1 with 4 Metal Frameworks.png
    DualGPU 4beta1 with 4 Metal Frameworks.png
    1.3 MB · Views: 254
Last edited:
It's looking like DisableLibraryValidation.kext may be the best option in this circumstance... This will also be an issue for other framework patches, such as the NightShift Patch in CoreBrightness.

If you roll out another Catalina Patcher, it would be nice if the installation of the MouSSE kext were limited to hardware with AMD cards. Currently, it gets installed on systems with GTX680 and fouls up the stability of sleep on those machines.
 
  • Like
Reactions: TimothyR734
Anyone know how to get v.1.3.0 "Install to this machine" installations to complete? When attempting to install 10.15.3 onto a Mini3,1 APFS SSD which has 10.15.1 running great, the first '6 minutes remaining' ran but after the mandated auto-reboot, the installer fails and leaves a 8.5GB ƒ "macOS Install Data" on top of the target "Data" drive upon rebooting. Only since v.1.3.0 have installer issues cropped-up. v.1.2.3 was perfect as were all the Mojave patchers. High Sierra was fine, too.

Screen Shot 2020-02-07 at 9.36.20 AM.pngScreen Shot 2020-02-07 at 9.37.49 AM.png
 
Anyone know how to get v.1.3.0 "Install to this machine" installations to complete? When attempting to install 10.15.3 onto a Mini3,1 APFS SSD which has 10.15.1 running great, the first '6 minutes remaining' ran but after the mandated auto-reboot, the installer fails and leaves a 8.5GB ƒ "macOS Install Data" on top of the target "Data" drive upon rebooting. Only since v.1.3.0 have installer issues cropped-up. v.1.2.3 was perfect as were all the Mojave patchers. High Sierra was fine, too.

View attachment 893019View attachment 893020
From the the Catalina Patcher I would set the macsOS Install Data as the startup disk to finish install Catalina or pressing the option key on your keyboard while booting and selecting that one
 
  • Like
Reactions: K two and jackluke
Anyone know how to get v.1.3.0 "Install to this machine" installations to complete? When attempting to install 10.15.3 onto a Mini3,1 APFS SSD which has 10.15.1 running great, the first '6 minutes remaining' ran but after the mandated auto-reboot, the installer fails and leaves a 8.5GB ƒ "macOS Install Data" on top of the target "Data" drive upon rebooting. Only since v.1.3.0 have installer issues cropped-up. v.1.2.3 was perfect as were all the Mojave patchers. High Sierra was fine, too.

View attachment 893019View attachment 893020

The "macOS Install Data" if I don't get wrong is only a mount point on the "APFS Data Volume" contained as a simple stage2 installer folder, that apple creates to install the APFS system update (otherwise they can't overwrite system core files) , from Catalina this mode is a similar concept as OTA updates perfom.
 
  • Like
Reactions: K two
Great news (for my old MBP4,1 anyway) - got the 10.15.4 beta 1 up and running with NVidia GT full acceleration! And discovered that Handoff actually works better than ever, displaying running active Apps on the iPhone in the leftmost section of the dock. Never had that on my old machines yet ;-)

Bildschirmfoto 2020-02-07 um 17.52.34.jpg


Here´s what I did (thanks to the previous findings @jackluke @ASentientBot et al !):
- changed to the correct SWupdatecatalog to get the 10.15.4 beta 1
- downloaded it via system settings panel
- used the catalinaswufix.command (still v2, I admit ;-) / newer editions should work fine as well)
- installed the 10.5.4 beta 1 by rebooting
- after installer reboot, booted USB patch stick (v1.3.0 of dosdude1´s installer), did all 4,1 specific patches
- booted into single user mode, set amfi_get_out_of_my_way=0x1 in nvram boot-args
- /sbin/mount -uw / to get write access to the S/L/E folder
- delete NVDAstartup.kext
- exit-boot to login
- use kext utility to install DisableLibraryValidation.kext
- rebuild caches & reboot
At this point I reached the login to Finder, but of course still without acceletion due to missing NVDAstartup.
- re-installed NVDAstartup.kext and again after cache rebuild did a reboot
Voilá -> all running fine with accel again!
(Somehow, the patcher cannot bring accel back in one go, instead window service seems to crash before login then - that´s why I did the detour above).
 
Last edited:
Great news (for my old MBP4,1 anyway) - got the 10.15.4 beta 1 up and running with NVidia GT full acceleration! And discovered that Handoff actually works better than ever, displaying running active Apps on the iPhone in the leftmost section of the dock. Never had that on my old machines yet ;-)

View attachment 893031

Here´s what I did (thanks to the previous findings @jackluke @ASentientBot et al !):
- changed to the correct SWupdatecatalog to get the 10.15.4 beta 1
- downloaded it via system settings panel
- used the catalinaswufix.command (still v2, I admit ;-) / newer editions should work fine as well)
- installed the 10.5.4 beta 1 by rebooting
- after installer reboot, booted USB patch stick (v1.3.0 of dosdude1´s installer), did all 4,1 specific patches
- booted into single user mode, set amfi_get_out_of_my_way=0x1 in nvram boot-args
- /sbin/mount -uw / to get write access to the S/L/E folder
- delete NVDAstartup.kext
- exit-boot to login
- use kext utility to install DisableLibraryValidation.kext
- rebuild caches & reboot
At this point I reached the login to Finder, but of course still without acceletion due to missing NVDAstartup.
- re-installed NVDAstartup.kext and again after cache rebuild did a reboot
Viola -> all running fine with accel again!
(Somehow, the patcher cannot bring accel back in one go, instead window service seems to crash before login then - that´s why I did the detour above).

Well done!

I don't know if you also noticed that from this OTA .4 beta 1 wave apple added a new stage after the Catalina logo OTA installing, there is a new phase with apple logo and remaining time (takes some minutes also with an SSD), and this should let to complete without applying any post-install patches, because it's some kind of new apple OTA post-scripts phase.

That said, just one question, your current SkyLight.framework/Versions/A/Resources/ path of Date modified files match with this I posted (I mean especially the WindowServer exec) ?

https://forums.macrumors.com/thread...ge-330?post=28200653#lg=post-28199827&slide=0
 
Last edited:
  • Like
Reactions: TimothyR734
...
That said, just one question, your current SkyLight.framework/Versions/A/Resources/ path of Date modified files match with this I posted (I mean especially the WindowServer exec) ?
Seems to be the same mod date, see screenshot:
Bildschirmfoto 2020-02-07 um 18.49.44.jpg


And yes, there is some additional install stage that you noticed. But my impression was that I´ve seen this since some 10.15.3 (or a bit earlier) beta back then..!?
[automerge]1581098263[/automerge]
Unfortunately, Music.App does not work any more with 10.15.4 beta 1 and only crashes each time started:

Dyld Error Message:
dependent dylib '@executable_path/../Frameworks/libgnsdk_dsp.3.06.1.dylib' not found for '/System/Applications/Music.app/Contents/MacOS/Music', tried but didn't find: '@executable_path/../Frameworks/libgnsdk_dsp.3.06.1.dylib' '/usr/lib/libgnsdk_dsp.3.06.1.dylib'

Tested more programs, ahem, "Apps":
These work without hickups: Mail, Pages, Numbers, Calendar, Stocks, TV, Messages, Safari, M$Office 2019, Preview, ...
 
Last edited:
Seems to be the same mod date, see screenshot:
View attachment 893044

And yes, there is some additional install stage that you noticed. But my impression was that I´ve seen this since some 10.15.3 (or a bit earlier) beta back then..!?

Yes, they match, infact the 3 February 2020 is when apple developers compiled their exec before packaging for 4beta1 release.

This confirm that I have a different issue and ASentientBot wrappers worked correctly.
 
Thanks @ASentientBot for your 10.15.4 beta1 wrappers, my issue was due to some mismatched replacement or permissions, FYI the Display.prefpane is correctly showed. Only night shift tab not shown after patched CoreBrightness with pikeralpha method, but the NotificationCenter button worked and warm colours are night-shifted properly.

@Larsvonhier I used the catalinaswufix3.command so I'd say that Pinarek OTA update method is still valid, thanks for your tips.
 

Attachments

  • Screenshot 2020-02-07 at 20.00.39.png
    Screenshot 2020-02-07 at 20.00.39.png
    951.1 KB · Views: 184
  • 10154b1 nightshift CoreBrightness.zip
    346.8 KB · Views: 113
  • catalinaswufix3amfi.zip
    1.7 KB · Views: 122
Last edited:
  • Like
Reactions: ASentientBot
From the the Catalina Patcher I would set the macsOS Install Data as the startup disk to finish install Catalina or pressing the option key on your keyboard while booting and selecting that one

TimothyR734 and jackluke -

• Just now, made a fresh v.1.3.0 USB 10.15.3 installer. Came back "damaged."

• The macsOS Install Data ƒ does not appear in any start-up manager under any circumstances.

But thanks for the suggestions.
 
  • Like
Reactions: jackluke
TimothyR734 and jackluke -

• Just now, made a fresh v.1.3.0 USB 10.15.3 installer. Came back "damaged."

• The macsOS Install Data ƒ does not appear in any start-up manager under any circumstances.

But thanks for the suggestions.

Because if the stage2 installer failed, the "macOS Install Data" is automatically deleted at next auto-reboot, so I guess you should retry a new usb thumb drive or this "damaged" is due to some AMFI check.
In case it's some kind of security check, try this before install the next Catalina major build:
Code:
sudo nvram boot-args="-no_compat_check amfi_get_out_of_my_way=0x1 amfi_allow_any_signature=0x1 cs_enforcement_disable=0x1"

I learned something from here: https://pikeralpha.wordpress.com/2014/06/11/yosemite-dp1-adds-applemobilefileintegrity-kext-to-os-x/
and here: https://www.theiphonewiki.com/wiki/AppleMobileFileIntegrity
 
Last edited:
  • Like
Reactions: hvds and K two
When I upgrade (e.g., from 10.15.1 to 10.15.3), can I install to this machine or do I need to make a USB and install from that?
In addition, can I use macOSUpdCombo dmg or is "install macOS Catalina" app required?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.