PCI connection issues Common problems with PCI passthrough Troubleshooting tips
PCI connection issues Common problems with PCI passthrough Troubleshooting tips
I’m weighing my options carefully. Based on what I understand, it looks like the Troubleshooting forum is more focused on hardware problems, so I’ll try posting here instead. My setup includes Intel integrated graphics and an NVIDIA GTX 1060 with a 3GB slot. The integrated graphics are linked via DVI to my monitor, while my dedicated graphics use HDMI. I’m attempting to run Linux on the integrated graphics and use QEMU’s PCI passthrough for Windows on the dedicated card. Currently, I have just one problem: when booting, the display vanishes if the HDMI cable is removed, but works fine with DVI. I’ve already resolved this issue in a previous post. The second concern involves a guest VM running Windows 10 via QEMU and libvirt. If both DVI and HDMI are connected and the host is set to integrated graphics, the screen flickers on and off, often showing a black screen. When only DVI is used, it works without flickering, and the Linux desktop appears normal. I’m unsure how to resolve this without dual-booting. Any advice would be greatly appreciated.
The monitor issues could stem from outdated firmware possibly? Switching between two inputs might seem simple, but the monitors' firmware might be outdated or damaged—consider updating it. Another potential problem is Nvidia's absence of official PCIe support. I've noticed GeForce cards function reasonably well with KVM, though some virtualisation tools fail because of driver incompatibility; the issue often lies in the drivers' refusal to pass through. It’s usually fixable.
You're facing an issue with your monitor firmware compatibility while using KVM via QEMU. It seems you can switch between DVI and HDMI, but there might be limitations specific to this model (IO-DATA DIOS-MF271XDB). Consider checking the manufacturer's documentation or support channels for detailed guidance.
I thought you were using KVM with QEMU because it seemed to work, but the flickering screen could be due to a PCIe passthrough issue. Since I’m not entirely confident, I’ve had good experiences with older FirePro or AMD cards when using passthrough without Linux. You might want to try routing the iGPU through and check if the problem persists, or look for another AMD card to test. It’s strange how it flickers, so it’s hard to pinpoint.
It might seem like a simple question, but I'm curious: have you looked up the NVIDIA card's position in the IOMMU group, and are you connecting all the required devices within that group?
It's not an obvious inquiry; I didn't provide much detail on the post, probably. Anyway, yes—I've looked into the IOMMU groups. IOMMU Group 0 at 00:00.0 Host bridge [0600]: Intel Xeon E3-1200 v2/Ivy Bridge DRAM Controller (rev 09). IOMMU Group 1 at 00:01:00 PCI bridge [0604]: Intel Xeon E3-1200 v2/3rd Gen Core PCI Express Root Port (rev 09). IOMMU Group 1 at 01:00:00 VGA compatible chip [0300]: NVIDIA GP106 [GeForce GTX 1060 3GB] (rev a1). IOMMU Group 1 at 01:00:1 Audio device [0403]: NVIDIA GP106 High Definition Audio Controller [10de:10f1]. IOMMU Group 2 at 00:02:00 VGA compatible chip [0300]: Intel Xeon E3-1200 v2/3rd Gen Core PCI Express Root Port (rev 09). IOMMU Group 3 at 00:14:00 USB controller [0c03]: Intel 7 Series/C210 Chipset USB xHCI Host Controller (rev 04). IOMMU Group 4 at 00:19:00 Ethernet controller [0200]: Intel Gigabit Network Connection (Lewisville) (rev 04). IOMMU Group 5 at 00:1a:0 USB controller [0c03]: Intel 7 Series/C216 Chipset USB Enhanced Host Controller #2 (rev 04). IOMMU Group 6 at 00:1b:0 Audio device [0403]: Intel 7 Series/C216 Chipset High Definition Audio Controller (rev 04). IOMMU Group 7 at 00:1d:0 USB controller [0c03]: Intel 7 Series/C216 Chipset USB Enhanced Host Controller #1 (rev 04). IOMMU Group 8 at 00:1e:0 PCI bridge [0604]: Intel 82801 PCI Bridge (rev a4). IOMMU Group 9 at 00:1f:0 ISA bridge [0601]: C216 Chipset USB LPC Controller (rev 04). IOMMU Group 9 at 00:1f:2 SATA controller [0106]: 6-port SATA Controller (AHCI) [8086:1e53]. IOMMU Group 9 at 00:1f:3 SMBus [0c05]: SMBus Controller [8086:1e22]. The NVIDIA card resides in IOMMU Group 1. I'm employing vfio-pci drivers on 10de:1c02 and 10de:10f1 (for VGA & HDMI audio), and routing both to the VM. The only component not using those drivers and not being sent to the VM is the PCI bridge, which seems essential for host access. EDIT: Also confirmed; both utilize vfio-pci drivers.**
I managed to resolve the initial issue by including the video=efifb:off parameter to direct console text to my iGPU. I also created an Xorg configuration file at /etc/X11/xorg.conf.d/20-intel.conf to tell Xorg to utilize the iGPU. Simply adding that section worked perfectly. Now I can boot with the dGPU's HDMI connected and display Linux output through the iGPU.