Intel Volume Management Device

[jalavoui]

https://elixir.bootlin.com/linux/latest/source/drivers/pci/controller/vmd.c

 

Trying to start a new project

static const struct pci_device_id vmd_ids[] = {
	{PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_VMD_201D),
		.driver_data = VMD_FEAT_HAS_MEMBAR_SHADOW_VSCAP,},
	{PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_VMD_28C0),
		.driver_data = VMD_FEAT_HAS_MEMBAR_SHADOW |
				VMD_FEAT_HAS_BUS_RESTRICTIONS |
				VMD_FEAT_CAN_BYPASS_MSI_REMAP,},
	{PCI_VDEVICE(INTEL, 0x467f),
		.driver_data = VMD_FEATS_CLIENT,},
	{PCI_VDEVICE(INTEL, 0x4c3d),
		.driver_data = VMD_FEATS_CLIENT,},
	{PCI_VDEVICE(INTEL, 0xa77f),
		.driver_data = VMD_FEATS_CLIENT,},
	{PCI_VDEVICE(INTEL, 0x7d0b),
		.driver_data = VMD_FEATS_CLIENT,},
	{PCI_VDEVICE(INTEL, 0xad0b),
		.driver_data = VMD_FEATS_CLIENT,},
	{PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_VMD_9A0B),
		.driver_data = VMD_FEATS_CLIENT,},
	{0,}
};

https://github.com/Artoria2e5/SatSmartDriver_kext

https://cgit.freebsd.org/src/tree/sys/dev/vmd/vmd.c

Edited March 30, 2024 by jalavoui

[jalavoui]

meanwhile the workaround is disable it in bios and

 

the disk will be available as sata device on macos

[jalavoui]

starting code from freebsd/linux

 

todo: a lot

 

 

 

 

Edited August 5, 2024 by jalavoui

[jalavoui]

this is why the device as bugged 32 bits memory bars...

 

guess this afects many other devices that require BAR0 tobe 64 bits

on linux this is reg 0x10: [mem 0x6000000000-0x6001ffffff 64bit]

 

on os x...

 

 

 

 

 

gonna try hack this with some variables using ssdt injection

 

think might be a opencore option to solve this...

 

Edited March 16, 2025 by jalavoui

[jalavoui]

this looks like it is working (detection only)

 

 

guess a small but important step

 

 

 

 

the mac os "conversion" of bars to 32 bits makes this impossible so injecting linux/windows addresses allows driver to detect devices.

 

 

check ioregistry for values

 

 

 

Edited April 10, 2025 by jalavoui

[jalavoui]

mac os now as accepted the new devices in the pci table.

buses start at 0xe0

 

todo: correct interrupts and bridge setup

 

 

 

Edited May 4, 2025 by jalavoui

[jalavoui]

much better

 

 

 

IntelVMD.zip

Edited October 20, 2025 by jalavoui

[jalavoui]

can apple create a host bridge like linux ?

 

this is the result. the code logic changed a lot

 

 

 

IntelVMD.zip

 

Edited February 17 by jalavoui

[jalavoui]

IONVMeController does some acpi-path checks and if the device not on pci root.

 

it fails

 

great !

 

IONVMeController also  check for acpi-device var

 

 

if i use apple code to detect devices with

configOpParams cp = {.device = fBridgeDevice, .op = kConfigOpScan, .result = nullptr};

i get valid devices with all apple internal properties working

but

 

that doesn't help cause then i have 10% of things todo that make it impossible due to apple internal code

Edited February 14 by jalavoui