Contribute
Register

Installation of 10.14.4 Update Causes KP

Status
Not open for further replies.
Why monitor resolution U2718Q 5K?
4k is written on the site

5120*2880 is the render resolution for macOS, 3840*2160 is the actual display resolution.

So in reality, items on my display would have the size of 2560*1440, but rendered in 2x, which is 5120*2880, and have the clarity of 3840*2160.

This is very common on Apple devices (including iOS and macOS devices), for example, my MacBook Pro 13 inch 2016 have physical resolution of 2560*1600,
395390

but the default render resolution is 2x of 1440*900. So it renders on resolution of 2880*1800, items have the size of 1440*900,
395391

and have clarity of 2560*1600.

In this way, GPU would need to fill more pixel per frame, but items on screen would look a lot smoother.

It's very similar to super sampling:
 
Deleted the device entry and platform ID and checked inject Intel, but still hangs on boot at the apple logo. I have attached my clover folder, please take a look and see if I missed something.
 
Deleted the device entry and platform ID and checked inject Intel, but still hangs on boot at the apple logo. I have attached my clover folder, please take a look and see if I missed something.

Your CLOVER directory has tons of old stuff.

I'm not sure if you're using drivers64 at all, and your drivers64UEFI has so many efi's that I don't think you need them all. Also config.plist clearly has leftover things like the -shikivga, and you have tons of .aml's too (do you actually need all that?).

I really would recommend you delete all that and reinstall CLOVER clean and choose only the stuff you know you need.
 
I really would recommend you delete all that and reinstall CLOVER clean and choose only the stuff you know you need.

If the Unibeast-built USB drive boots your system mostly OK, you can even use the Clover directory on that as a base. Especially if you've gone through a bunch of various troubleshooting steps to get your machine up and running in the first place or have been running the same installation for a long time, you may have a bunch of stuff laying around in your EFI partition and /L/E that you don't actually need.

I took the opportunity to strip my config back to the barest essentials that did not result in any lost functionality: a couple DSDT renames(EC0->EC, SAT0->SATA), SSDT->Generate->PluginType for CPU PM, audio layout injection for AppleALC, inject Intel, a couple things in KernelAndKextPatches, and SMBIOS information -- nothing else.
 
Update: I was able to get it working with WEG 1.2.7. The only change I made to my config was disabling the following:

change HDAS to HDEF
change HECI to IMEI
change GFX0 to IGPU
change PEGP to GFX0

According to the WEG documentation it automatically takes care of these renames so they should be disabled. After disabling I was able to upgrade to WEG 1.2.6 and boot cleanly, so I then tried again with WEG 1.2.7 and was able to boot cleanly. I'm not sure why these renames caused issues on my system in 10.14.4 but it's fully functional now.
 
I'm not sure why these renames caused issues on my system in 10.14.4 but it's fully functional now.
That’s great to hear, would you mind looking at my earlier post and see if you think my issue is the same? Screenshots posted.

 
That’s great to hear, would you mind looking at my earlier post and see if you think my issue is the same? Screenshots posted.

Left you a note on your thread. TLDR; Try compiling WEG 1.2.8 and see if it fixes things for you.
 
Also, I am tracking an update committed to WEG that purports to fix KP on Kaby Lake. See https://github.com/acidanthera/WhateverGreen/commit/0d22f8e1f5dd23d8e8b2a22e8d05849602b7e9d9
Ooh, good find, that seems like it might be what I'm running into as the panic is in the KBL kext and happens immediately as the OS tries to init graphics (black screen before login window). Hopefully with 1.2.8 I can boot with Intel injection off and have the boot loader do even less :D

Specifically, this comment is promising:

Code:
    // In 10.14.4b1 the return condition changed from (flags & 7U) != 0 to (flags & 1U) != 0. The change makes good sense to me, but
    // it results in crashes on 10.14.4b1 on many ASUS Z170 and Z270 boards with connector-less IGPU framebuffer.
 
Status
Not open for further replies.
Back
Top