Manthony

Members
  • Posts

    3
  • Joined

  • Last visited

Everything posted by Manthony

  1. Thanks, I included both of those and it seems like pci=noats fixes the timeout issue without causing any other problems (iommu=soft didn't work on it's own and didn't seem to make a difference when set). With that out of the way, the next issue is that the kernel gets stuck at kernel: vfio-pci 0000:09:00.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=none afterwards nothing else happens and the VM doesn't start. spent some time on google today looking into this one, a fair number of amd VM users, apparently. I couldn't find any definitive solution though, at least not any that seemed applicable.
  2. Thanks for the tip, not having any issues saving the XLM document after adding that in, so at least that's working better. As for actually getting a VM working, it's somehow going much worse today than yesterday. I did see a brief blip of HDMI out, which was a new first and probably a good sign, but every time I start the VM unraid loses the flash drive after a few moments, the VM becomes unreachable, and I need to reboot everything. If I had to guess, either it's attempting to pass through all of the USB ports and losing the flash drive, or maybe it's just more bad igpu passthrough issues dropping the ball. Really baffling because it worked at least a couple of times yesterday, but today turning on the VM instantly breaks it, and I can't remote desktop at all. I've tried making a brand new VM following the instructions in this thread to a T, XML config is good, IOMMU groups are good, iGPU and audio are in their own groups without any USB controllers, so I don't know why the flash drive would be getting bumped off while the rest of the array stays 'on' EDIT, took a peek at the array logs when I kick off the VM. what stands out to me is that a bunch of things are timing out, but really I don't have a clue what this is about or how to fix it. Snippet ends when bunch of USBs disconnect which leads to all of the errors associated with the flash drive being gone. Edit 2: flash drive doesn't disconnect after I moved it to the front panel - not ideal for aesthetic reasons but maybe not a permanent change. I can now attempt to spin up the VM as much as I want without it impacting the array. Still consistently not taking off properly when using igpu passthrough though, logs continue to show many timeouts and remoting in is a no-go. Adding the server logs logs.txt VMlogs.txt
  3. Optiluca, after seeing you were successful I decided to try it myself. For flavor text, I've been snubbed this issue since 2018 with a raven ridge 2400 G, and I've long since given up on this ever actually working in a windows VM. Trying all of the steps you followed including your modifications, I'll admit that this is closer to "actually working" than it's ever been before, but still with quite a number of issues sadly. Something about the initial iGPU handoff is going buggy for me - I can start the VM cleanly, but as soon as I remote desktop into it, unraid loses connection to it's flash drive (I lose a good chunk of the webUI functionality, but the array doesn't crash). Within the VM, the vega graphics device is correctly showing but the display driver has a yellow alert next to it - Windows reports that the device has issues and the driver isn't properly installed (code 43). As a result of all this, I'm not able to get any HDMI out from the server itself. Really not sure what to try next, I can tell it's sort-of working because the remote desktop is full resolution and whatnot, but I have no idea how to iron out what's going on behind the scenes. I believe the driver issue is just a symptom of a bigger underlying problem. One thing I'm really curious about is the <qemu:override> section, I am also seeing the XML file seemingly revert that section when I reopen it, which makes me think there is something invalid there which is causing the document to not save properly. I also got the "hostdev0 not valid" issue when I used the code snippet in Tanne's guide, but using that code the XML at least saved properly... I'm still holding onto a flicker of hope that this can work, because there's no room in my server for a dedicated GPU and I would hate to have to give up/start over with a different processor after all this. But man, I really can't say I recommend using AMD APUs with unraid, considering all this hassle.