AppleIGC doesnt work on Tahoe Developer Beta 9

[Kampot-vkusna]

MB: Asus Rog Strix B850-G Gaming Wifi
Intel i226-V

I have already tried to turn off and on DisableIoMapper with DisableIoMapperMapping, I tried boot-arg e1000=0, but it doesn't help. The problem is that there is no access to the Internet, in the settings it is displayed first as Self-Assigned IP, then as Connected, but the configuration is incorrect. I will also add that when trying to disconnect or connect the Internet cable, the system freezes

[jzrodriguez98]

Did you disable VT-D in the BIOS? Keeping VT-D disabled and the boot-arg are the only things I have right now. I had DMAR.aml but got rid off it and it's still working for me

[Kampot-vkusna]

VT-D is an exclusive intel platform, right? Do you mean SVM?

[jzrodriguez98]

Is your MB an AMD one? Mine is Intel so it does have VT-D.

[Kampot-vkusna]

yes, am5 platform

[jzrodriguez98]

@SongXiaoXi Have you verified what changed in Tahoe that is causing AppleIGC kext to fail when loading Tahoe with AppleVTD enabled (VT-D enabled in BIOS for Intel CPUs)? AppleIGC was working in Sequoia with VT-D enabled.

[SongXiaoXi]

@jzrodriguez98 Thanks for filing this. The behavior appears tied to AppleVTD/DMAR, which is outside this NIC driver’s scope, so I don’t have actionable guidance on ACPI/DMAR or BIOS/firmware issues.

[jzrodriguez98]

@SongXiaoXi about two months ago mieze added AppleVTD support in Tahoe for the Ethernet kexts she developed like IntelMausi and Realtek8111 because the kexts had issues with VTD enabled in Tahoe, same as AppleIGC. Have you looked at what Mieze did? Is it something that you can replicate here?

[SongXiaoXi]

@jzrodriguez98 Thanks for the suggestion. I did look at Mieze’s AppleVTD-related changes before making a decision.

From what I can see, AppleVTD support there was not added as a small isolated patch. It came with broader changes to buffer/resource allocation, which suggests this is not something I can safely port as a narrow feature addition in AppleIGC.

For AppleIGC, supporting AppleVTD would likely require touching DMA/resource-management assumptions more broadly, and I do not want to merge a partial implementation that may behave unpredictably on some systems.

On top of that, the testing cost would be fairly high, and I do not have matching hardware for proper validation. So AppleVTD support is not planned for AppleIGC at this time.