9800X3D / ASUS TUF B850M-PLUS WIFI / RX 6900 XT - Tahoe recovery freezes before installer

P

plawpp

New member
AMD OS X Member
Oct 30, 2026
2
0
1
CPU:
9800x3d
Hi everyone,
I'm trying to install Tahoe on this PC, but I haven't managed to reach the recovery screen. OpenCore loads the recovery image and verbose boot starts, then it stops around CoreAnalyticsHub / early IOKit initialization. I left earlier attempts for five minutes with no change. A fresh test still stops at the same stage.
Specs:
  • CPU: Ryzen 7 9800X3D, 8 cores / 16 threads
  • Board: ASUS TUF GAMING B850M-PLUS WIFI, BIOS 1065 (original WIFI model, not WIFI7)
  • GPU: RX 6900 XT, device ID 1002:73BF; monitor connected by DisplayPort
  • RAM: 48 GB DDR5, EXPO 6000
  • SSDs: Samsung 990 EVO Plus 1 TB (macOS target), Lexar ARES PRO 1 TB, aigo P7000Z 2 TB
  • Tahoe 26.6.2 recovery, build 25G83 / Darwin 25.6.0, on a Kingston USB
  • OpenCore 1.0.8 DEBUG, MacPro7,1
Last checked BIOS settings: CSM off, Secure Boot off, Above 4G on, ReBAR on. ResizeAppleGpuBars is 0.
What I've tried so far:
  • AMD core-count patches set to 8, updated Tahoe PCI/TSC patches, and the AM5 PCI-enumeration patch enabled.
  • DevirtualiseMmio and SetupVirtualMap enabled, with F7000000 and E0000000 in the MMIO whitelist.
  • Updated Lilu to 1.7.3 and VirtualSMC to 1.3.9. Tried cpus=1 earlier without improvement; it is now removed.
  • Tried NootRX earlier. The current test has it disabled and uses the AGDP board-id patch without WhateverGreen.
  • Most recently, blocked IONVMeFamily and disabled NVMeFix for a USB-recovery-only test. The recording shows OpenCore reporting exclusion Success, but it still stops before recovery. All three SSDs remained physically connected.
Current boot-args:
-v keepsyms=1 debug=0x108 -lilubetaall io=0x80003f iokit_print_verbose_match_logs=1
The latest screen repeats "IOResources: family specific matching fails" and ends with a configuration-thread termination message. I don't know which thread is actually blocked.
Two things I should mention: USB is still using USBToolBox + UTBDefault; I haven't made a proper port map yet. The EFI also contains a locally modified DebugEnhancer and an early-console kernel patch added during troubleshooting. I haven't ruled out those changes affecting the result.
I've attached the EFI used for this test and a screenshot of the last screen. SMBIOS details are removed, and DebugEnhancer's debug symbols were stripped to remove local build paths. The original DSDT is at the top level of the ZIP for reference.
Could someone check whether there's an obvious problem in the ACPI/PCI setup? What would be the best way to get a useful trace at this point, or establish a cleaner baseline?
Thanks for taking a look.
 

Attachments

  • boot-stuck.jpg
    boot-stuck.jpg
    852.1 KB · Views: 4
  • EFI.zip
    EFI.zip
    7.7 MB · Views: 2
Sort by date Sort by votes
Boot using OC Debug version, so a boot log is generated. This will highlight any initial boot issues, such as ACPI errors plus it will show the MmioWhitelist entries you need to use to boot macOS.

 
Upvote 0 Downvote
Thanks Edhawk. I have captured three physical-machine boot logs using OpenCore 1.0.8 DEBUG with Target=67, AppleDebug and ApplePanic enabled. The attached ZIP contains sanitized logs and a short README.

Since my original post, I have replaced UTBDefault with a USBToolBox map created by testing the ports in Windows. I also completed the Tahoe installation onto the Samsung 990 EVO Plus through VMware using the physical SSD, but it still does not boot to the desktop on the real PC.

AppleMCEReporterDisabler is now enabled for MacPro7,1, and all three logs show successful prelinked injection. NVMeFix is enabled again, and the previous IONVMeFamily block has been removed.

The logs cover two installed-system boots and one KINGSTON USB recovery boot. All three stalled, and I manually restarted the PC.

All three show successful DSDT patch replacements, applicable AMD/PCI patches, and eight kext injections. They end at EXITBS:START and do not capture the later on-screen kernel output. No saved panic file was found.

The MMIO results are identical:

0xE0000000 0x10000 pages skip 1
0xF7000000 0x7E00 pages skip 1
0xFEE00000 0x1 page skip 0
0xFEE01000 0x11FF pages skip 0
0xC80000000 0x20200 pages skip 0
My local PE_init_kprintf diagnostic patch is Not Found in both installed-system boots, but matches once in recovery. The modified DebugEnhancer is still enabled, so this is not yet a clean baseline.

Could you review the logs and suggest the useful changes to test next?

Thank you for your help.
Edhawk said:
Boot using OC Debug version, so a boot log is generated. This will highlight any initial boot issues, such as ACPI errors plus it will show the MmioWhitelist entries you need to use to boot macOS.

Click to expand...
 

Attachments

Upvote 0 Downvote
Success with Sequoia on this motherboard was reported here:
henrikato

Thread 'Need help with a bare metal Sequoia build'

I have spent the past few days attempting to get the installer for macOS Sequoia 15.8.1 to boot on my setup:

The problem is no matter what I do, I never get...
I'm surprised that you went so far without SSDT-CPUR or equivalent.
So you may
  • Add SSDT-CPUR.
  • Try Sequoia first.
  • Enable all the ACPI patches (or try different combinations to find out which ones are necessary).
  • Remove the NVMe. I'm not sure if any of the three is macOS-friendly.
 
Upvote 0 Downvote
You must log in or register to reply here.
Top Bottom
✖
  AdBlock Detected
Sure, ad-blocking software does a great job at blocking ads, but it also blocks some useful and important features of our website. For the best possible site experience please take a moment to disable your AdBlocker.