@olivierlambert said in Google Coral TPU PCIe Passthrough Woes:

Can you try an on older kernel in your VM just to be sure? (Eg Debian 10 guest with default bundled kernel)

I'm just now getting back to this.

Might the problem be related to this issue?

"Unfortunately the device in question violates PCI specification by mapping PBA, MSI-X vector table, and other registers into same 4KB page (PBA is at 0x46068, VT at 0x46800, but there is a bunch of other registers in 0x46XXX range)."

https://github.com/google-coral/edgetpu/issues/343#issuecomment-1287251821

dakota created this issue in google-coral/edgetpu

open Apex failing with error -110 (No /dev/apex_0) #343

Update: I put the server back to the exact configuration it was in, rebooted, and nothing showed up in xl pci-assignable-list. There must have been some configuration... somewhere... that something was looking for that was related to the exact locations of those 4 USB controllers.

TL;DR, all is well, thanks for the help!

I am attempting to remove a PCIe card that was formerly hidden/passed through (it 4 USB controller card) causes xcp-ng to fail to boot from my M.2 boot drive (can't find the root drive).

Prior to removing the card, all 4 controllers were hidden from dom0. I removed the pass through from the two VMs that used a controller each, but I forgot to clear the dom0 pciback.hide parameter before. I put the card back in, booted up, cleared dom0 pciback.hide, and rebooted, thinking that would solve the problem - but I still get the same error to boot.

I have confirmed that neither of the two VMs that had PCIe controllers passed through no longer have them set in the other-config (MRW) parameters. I've also confirmed that the controllers are not in /boot/efi/EFI/xenserver/grub.cfg.

But, strangely, 3 of the 4 USB controllers that were hidden do still show up with xl pci-assignable-list - and I still can't boot without the card.

What am I missing?