Search results

Sorry for the shitty follow up on the initial post. Recap of the initial post and problem description, this to verify for you if what you attempt is feasible or necessary. It is a bit more elaborate than another post i made on this recently. You report to have passthrough success for the SSD...

I just noticed this as was of checking passthrough is actually functional. At least, i expect this is a good indication it is at least initialized as expected (with the iommu) dmesg | grep -i passthrough [ 0.198326] iommu: Default domain type: Passthrough (set via kernel command line)

The output here looks really good. Check again with dmesg | grep -e IOMMU -e iommu -e vfio -e Intel -e bug -e Bug -e BAR

check here https://www.kernel.org/doc/html/v5.4/admin-guide/kernel-parameters.html i have see eek, i have not looked into what it mean, may work for your setup Use this line, you cannot simply copy stuff :) iommu=pt intel_iommu=on video=vesafb:eek:ff,efifb:eek:ff pci=nobar,noats pcie_pme=nomsi...

Yeah, sorry, typo, my bad,indeed, notintxmask=1 Not sure what you are trying to do really. At this moment i think all the configuration you have is working as expected and should not change much. The option i do think you can work with are check: dmesg | grep -e IOMMU -e iommu -e vfio -e...

in case it may be of any relevance, here are the boot paramaters used at the time. iommu=pt amd_iommu=forced_isolation noats nobar video=efifb:off,vesafb:off hugepagesz=2M hugepages=1375 default_hugepagesz=1G hugepagesz=1G hugepages=8 text the GP occured on two different days Jun 6 06:09:00...

Important to know is the VM itself must also be configured appropriately. In my experience you must reinstall the VM sometimes for the PCI passthrough to catch on, or reboot the server. CPU: host Chipset: q35 When you add a PCI device it is a matter of experimenting with the settings...

iGPU 00:02.0 VGA compatible controller [0300]: Intel Corporation HD Graphics 5500 [8086:1616] (rev 09) GPU 04:00.0 3D controller [0302]: NVIDIA Corporation GM108M [GeForce 940M] [10de:1347] (rev a2) Audio PCI 00:03.0 Audio device [0403]: Intel Corporation Broadwell-U Audio Controller...

not sure what i was doing, the GP was found while viewing dmesg output just now pveproxy also crashed ASK ERROR: command '/usr/bin/termproxy 5900 --path /vms/1001 --perm VM.Console -- /usr/bin/dtach -A /var/run/dtach/vzctlconsole1001 -r winch -z lxc-console -n 1001 -e -1' failed: exit code 4

Hmm. Looks good. My suggestion is to gradually increase the number of vfio-pci ids you configured, that are a lot of ids. Try with one GPU and one other device first. Verify if all vfio-pci ids are correct, maybe also try nointmask=1 ? Make sure to have IOMMU set to enabled in the BIOS, not...

can you post the output for cat /proc/cmdline also, i just realised i must have made a mistake pci=nobar,noats seems more appropriate than just nobar

Any chances this kernel patch could be released as a hot-patch such as with Ksplice Kgraft Kpatch Livepatch KernelCare ?

iommu=pt intel_iommu=on nobar pci=noats video=vesafb:off,efifb:off" pcie_acs_override=downstream,multifunction text nomodeset What i personally add to vfio-pci as module option is disable_vga=1 disable_idle_d3=1 nointmask=1 wrt. the error, choose the 64 bit .msi file not the 32 bit .exe

do you mind sharing the output for: dmesg | grep -e IOMMU -e iommu -e vfio -e Intel -e bug -e Bug kernel boot params you could try: pci=noats nomodeset Also, did you use the https://fedorapeople.org/groups/virt/virtio-win/direct-downloads/archive-virtio/virtio-win-0.1.171-1/ or...

fixed: modified the vm.conf file hostpci to match the current pci

1. took a snapshot 2. upgraded bios 3. bios changed network interface id and pci id assigned 4. restore of snapshot fails because the old pci id cannot be found how to fix ?

Holding wood, for now it appears GPU passthrough is working and stable. Tested with CPU = qemu64, kvm64 Sharing cmdline appends here, AMD+AMD specific but shoud be easily customisable. iommu=pt amd_iommu=forced_isolation nobar video=efifb:off text nomodeset rcu_nocbs=0-15 amdgpu.runpm=0...

This seems a legit fix to most issues, just suffering from a VM which loses vncproxy when switching the console, thi scan be fixed with: systemctl restart pveproxy , not really how i hoped to fix it but hey. kernel boot parameters to test: nobar pci=noats amdgpu.pm=0 nomodeset include...

Filed bug report https://bugzilla.kernel.org/show_bug.cgi?id=208091 because of https://patchwork.kernel.org/patch/7670771/

well, i think the PCI-e passthrough is finally working. THE (not so) miracle solution is adding nomodeset to the kernel boot parameter string one stupifying observation is thestrings for teh PCI devices change on reboot, so 00:nn.0 and 00:nn.1 beome both 00:nn The GPU is now definitely...