Lenovo Legion 5 15ACH6H (82JU), Ryzen 5800H — Sonoma Recovery stalls/reboots after OpenCore

[wojko]

​

Hi everyone,

I'm trying to boot macOS Sonoma 14.6.1 Recovery on a Lenovo Legion 5 15ACH6H (type 82JU) with OpenCore. I have not been able to reach the Recovery UI.

I'm looking for a way to isolate the failing stage, rather than a ready-made EFI or a series of random configuration changes.

Hardware and environment​

• Laptop: Lenovo Legion 5 15ACH6H / 82JU
• BIOS: GKCN65WW
• CPU: AMD Ryzen 7 5800H (8 cores / 16 threads)
• Graphics: AMD Cezanne iGPU + NVIDIA RTX 3060 Laptop dGPU (I do not expect macOS support for the NVIDIA GPU)
• RAM: 16 GB
• Internal NVMe: WD PC SN730 and Crucial P3 CT1000P3SSD8
• Boot media: External SanDisk USB with Sonoma 14.6.1 (23G93) Recovery
• Bootloader: OpenCore 1.0.8 RELEASE
• Kexts / patches: AMD Vanilla kernel patches, Lilu, VirtualSMC, ForgedInvariant and NootedRed
• Host OS: Fedora Linux, running normally from the internal WD drive

Symptoms​

These observations come from different attempts, not necessarily the same boot:

• Verbose output sometimes reaches IONVMeController AppleNVMe Assert failed and NVMe Request Pool messages, after which the screen stops progressing.
• Other attempts show ACPI/CPU initialization messages, including AE_NOT_FOUND</span> involving _SB.PCI0.PB2, TSCSyncer and, in another capture, CoreAnalyticsHub.
• Some attempts restart automatically. Others leave a frozen-looking verbose screen.
• No confirmed kernel panic backtrace has been captured.
• Recovery has never opened.

I cannot yet prove which last visible message is actually fatal.

What has already been checked​

1. OpenCore and AMD kernel patches

• ocvalidate passes.
• ProvideCurrentCpuInfo=True.
• The 8-core AMD patch is configured correctly for Ryzen 7 5800H.
• A read-only comparison of the 27 local kernel-patch entries with a pinned AMD Vanilla revision found 23 exact matches.
• The two apparently missing reference patches have local equivalents with matching 10/10 core fields, but different version-range comments.

This is static validation, not proof that XNU executed successfully.

2. OpenCore debug logs

In one historical OpenCore log:

• Kernel patcher: 14 Success, 0 non-success
• Successful 64-bit XNU hook
• EXITBS:START

Other logs also reach EXITBS:START but do not contain detailed patcher markers.

I cannot confidently map these logs to individual physical boot attempts.

3. NVMe diagnostics

Both internal NVMe drives report:

• critical_warning=0
• media_errors=0

This does not prove macOS compatibility or rule out a driver-related issue.

4. Previous experiments

I have already tried:

• NVMeFix enabled and disabled
• A temporary Crucial P3 class-code spoof
• A dGPU-off SSDT experiment

None resulted in a verified Recovery boot. The experimental changes were rolled back.

5. TSC observations

Fedora has recorded TSC synchronization problems with an HPET fallback, but there is no evidence yet that this causes the macOS failure.

My questions​

1. On this Ryzen 5800H / Lenovo Legion platform, what would be the most useful single next diagnostic to distinguish an early XNU panic, NVMe initialization issue, ACPI/TSC problem, or a display-only hang?
2. Is there a reliable way to obtain evidence after ExitBootServices when no accessible hardware serial/UART debug port has been confirmed?
3. Are there known Lenovo Legion 5 15ACH6H / 82JU-specific firmware or kernel issues that fit these mixed symptoms? What precise evidence should I collect before changing the EFI again?

I can provide redacted screenshots and relevant OpenCore log excerpts if needed. I will remove serial numbers, MLB, ROM, UUID, device-unique identifiers and unrelated private values.

I would prefer not to publish a complete, unreviewed EFI.

Thank you for any targeted diagnostic suggestions.

Attached is a sanitized EFI archive for diagnostic review only. The original config.plist and device-specific identifiers have been removed. This archive is not intended for booting.