I am not having any success passing through any of the built-in peripherals on my Epyc 7502 (SATA or USB).
When I start a VM that has the built-in SATA or USB as PCI pass-through devices, I get a sequence of...
... followed by the vm getting stuck and sometimes other PCI devices will end up stuck. In the end, I have to hard power cycle in order to recover.
I am able to pass-through other devices, such as an Adaptec controller which does not support FLR (FLReset-) and the motherboard's Intel gigE controllers and a Mellanox CX3 that do (FLReset+). So, there does not appear to be a problem with FLR, just with the AMD peripherals.
I'm running CentOS Linux release 8.1.1911 (Core) with kernel 4.18.0-147.8.1.el8_1.x86_64. Motherboard is an ASRock Rack EPYCD8 with BIOS 2.10.
Search results suggest that this is a known issue with no workaround other than a sketchy kernel patch that has been rejected.
Has anyone successfully used either the built-in USB or SATA controllers as pass-though devices?
When I start a VM that has the built-in SATA or USB as PCI pass-through devices, I get a sequence of...
Code:
vfio-pci 0000:45:00.0: not ready 1023ms after FLR; waiting
vfio-pci 0000:45:00.0: not ready 2047ms after FLR; waiting
...
vfio-pci 0000:45:00.0 not ready 65535ms after FLR; giving up
I am able to pass-through other devices, such as an Adaptec controller which does not support FLR (FLReset-) and the motherboard's Intel gigE controllers and a Mellanox CX3 that do (FLReset+). So, there does not appear to be a problem with FLR, just with the AMD peripherals.
I'm running CentOS Linux release 8.1.1911 (Core) with kernel 4.18.0-147.8.1.el8_1.x86_64. Motherboard is an ASRock Rack EPYCD8 with BIOS 2.10.
Search results suggest that this is a known issue with no workaround other than a sketchy kernel patch that has been rejected.
Has anyone successfully used either the built-in USB or SATA controllers as pass-though devices?