Ryzen 7 8700G (Phoenix) + ASUS TUF B650-E: Sequoia 15.4.1 Recovery freezes at CoreAnalyticsHub

[Opp0815]

Hello AMD-OS-X community,

I am trying to boot macOS Sequoia 15.4.1 Recovery bare-metal, but it consistently freezes at or around the last visible message "Registered CoreAnalyticsHub functions" and Caps Lock becomes unresponsive. I am looking for Phoenix-specific advice rather than random OpenCore changes.

Hardware:

• AMD Ryzen 7 8700G Phoenix (8C/16T)
• ASUS TUF GAMING B650-E WIFI, BIOS 3223
• 64 GB DDR5
• PowerColor Radeon RX 5500 XT 8 GB (display connected directly), Radeon 780M iGPU disabled in BIOS
• OpenCore 1.0.8 DEBUG, SMBIOS MacPro7,1
• Linux Mint 22.3 works normally on this machine. macOS Recovery boots from a separate SanDisk USB drive; Linux SSD is untouched.

Verified:

• OpenCore loads and macOS kernel starts.
• Current AMD Vanilla kernel patches report Success in the OpenCore DEBUG log (including the 8-core cpuid_cores_per_package patch).
• Setting Booter/Quirks/DevirtualiseMmio=True fixed a prior boot.efi STOP 0x16 / no usable slide failure.
• Usual boot-args are '-v keepsyms=1'.

Tried with no change:

• Sequoia Recovery and Tahoe Recovery
• different USB 2/3 ports, safe mode (-x)
• dedicated Radeon RX 5500 XT instead of internal graphics
• Resizable BAR off, AMD PCI hotplug patch toggle
• several MMIO whitelist experiments
• cpus=1 (restored original boot args afterward)
• three targeted DSDT ACPI patches for conditional GPP5 device definitions UP00, RHUB, and XH00; OpenCore DEBUG log confirms each patch replaced 1 of 1, but the freeze stayed identical; patches subsequently reverted.

Questions:
1. Has anyone achieved a working bare-metal Sequoia or Tahoe boot on Ryzen 7 8700G/Phoenix with a dedicated AMD GPU?
2. Are known Phoenix-specific CPU/CPUID/kernel compatibility issues distinct from Ryzen 7000 desktop chips?
3. What additional targeted logs/tests would best distinguish Phoenix CPU incompatibility from motherboard-specific ACPI/MMIO problems?

I have a redacted config.plist and OpenCore DEBUG log available on request; original ACPI dumps can also be shared after review. The archived DEBUG log was taken during the cpus=1 test, whereas the restored config no longer has cpus=1. Please note this when comparing them.

Thank you for your help!

 

[Edhawk]

I've not read of any 'Phoenix' related issues on any of the Hackintosh sites I frequent. So I am not sure what you help if any you will receive.

As with any AM5 system, you will need to jump through some hoops to get a working system, but most if not all of the issues you will face have already been covered/highlighted in other B650E threads.

Does your OC debug log provide you with details of the MmioWhitelist entries you need to use in your config.plist?
Have you converted the entries from HEX to Decimal and added them to the config.plist?

As you are using an unsupported/incompatible AMD IGPU, I assume you have done one of these options to disable it.

1. Disabled in the Bios
2. Boot argument -wegnoigpu
3. Created a custom SSDT to disable the AMD IGPU

If not then you need to think about using one of the above methods.

Also you need the think about whether the ACPI patches you tested are required, even though your 'Freeze' issue wasn't resolved by using them. If your DSDT.aml shows the 3 x ACPI patches are required, then omitting them will be just adding to your woes.

Post whatever files and folders you feel comfortable sharing. We can only help if we see the full disclosure.

 

[Opp0815]

Hi Edhawk, many thanks for the thoughtful suggestions. I have now systematically reviewed my existing OpenCore 1.0.8 DEBUG log (from the unsuccessful cpus=1 test) against the restored, redacted config.plist.

MMIO findings:

• Booter/Quirks/DevirtualiseMmio = True.
• The two default MmioWhitelist entries in my config are both Enabled=False: 0xFED1C000 (4275159040) and 0xFF000000 (4278190080). They are not among the five runtime MMIO regions in this log.
• Exact logged regions (hex -> decimal, each has skip 0):

0xE0000000 -> 3758096384
0xF7000000 -> 4143972352
0xFEE00000 -> 4276092928
0xFEE01000 -> 4276097024
0x1080000000 -> 70866960384
- OpenCore reports 'MMIO devirt end, saved 935936 KB', 'Only 168/256 slide values are usable', 'Valid slides - 88-255'. The earlier boot.efi STOP 0x16 was resolved by DevirtualiseMmio=True, but macOS still hangs after kernel loading. Thus I cannot conclude from this alone which, if any, region must be whitelisted.

iGPU: Radeon 780M already disabled in ASUS BIOS; display physically on RX 5500 XT.

ACPI findings:

• Original DSDT has OperationRegion(CPVS, SystemMemory, 0x73068000, 0x100), within Linux-reported ACPI NVS. G000/G002/G004 control conditional PCIe/USB definitions under PCI0.GPP5 (GPP5, not the GPP7 shown in some generic examples).
• Our AML searches found unique intended 2-byte-package-length conditional patterns for UP00, RHUB and XH00 (offsets 0x0A727, 0x0B19C, 0x0B886). The OpenCore DEBUG log from that separate test explicitly says 'Patching DSDT ... replaced 1 of 1' for EACH of the three patches. The freeze remained identical, so they were reverted. I agree that this does not prove the patches are unnecessary; we have not fully validated the resulting namespace.

Could you advise which specific MMIO regions should be whitelisted (or how to identify required EFI runtime mappings on this board), and whether the GPP5 conditional DSDT patches should be retained together? I have a redacted OpenCore config and DEBUG log ready, and original ACPI dumps available locally. Please let me know the most useful ACPI tables/logs to attach, and I will check them for identifying data first.

Thanks again!

 

[Edhawk]

The normal practice for MmioWhitelist entries is to disable entries 1, 3, 4 & 5. Only the second entry is enabled. This works for the majority of users.

Retain the 3 x ACPI patches, you will see an error (G002 or similar) in the Verbose text if any of the 3 are not required.

Also make sure you have added apfs_aligned.efi to your /EFI/OC/Drivers folder with a corresponding entry in your config.plist.

The apfs_aligned.efi driver can be found in Acidanthera's OcBinaryData > Drivers folder in the Github repository linked below.

GitHub - acidanthera/OcBinaryData

Contribute to acidanthera/OcBinaryData development by creating an account on GitHub.

github.com