Hi
I have been recently having nvme passthrough problems. I have two samsung 990 in a bifurcation card (asus hyper)
These are passed through to one of my virtual machines, but recently I have been getting
When restarting (commonly) or sometimes when running. See Thread.
In a bid to fix this; I tried setting kernel parameters
However booting back in; I was unable to start VM, and was getting some weird kernel panic errors. I was also unable to kill the process or machine, resulting in me having to flip the power switch. This would happen even without passing the NVMEs
Currently; I have cloned the VM - This seems to give me proper errors and the start and stop commands are able to function and be killed. The error message has now changed to this;
Does anyone know what this means? Why I am suddenly unable to pass one through ? I noticed that now I have cloned the machine, I am able to pass the other one through seemingly without issue (currently)
I have been recently having nvme passthrough problems. I have two samsung 990 in a bifurcation card (asus hyper)
These are passed through to one of my virtual machines, but recently I have been getting
Code:
Unable to change power state from D0 to D3hot, device inaccessibleIn a bid to fix this; I tried setting kernel parameters
Code:
nvme_core.default_ps_max_latency_us=0 pcie_aspm=offHowever booting back in; I was unable to start VM, and was getting some weird kernel panic errors. I was also unable to kill the process or machine, resulting in me having to flip the power switch. This would happen even without passing the NVMEs
Currently; I have cloned the VM - This seems to give me proper errors and the start and stop commands are able to function and be killed. The error message has now changed to this;
Code:
() kvm: ../hw/pci/pci.c:1633: pci_irq_handler: Assertion `0 <= irq_num && irq_num < PCI_NUM_PINS' failed.Does anyone know what this means? Why I am suddenly unable to pass one through ? I noticed that now I have cloned the machine, I am able to pass the other one through seemingly without issue (currently)
