diff options
Diffstat (limited to 'results/classifier/deepseek-2/output/device/1998')
| -rw-r--r-- | results/classifier/deepseek-2/output/device/1998 | 28 |
1 files changed, 28 insertions, 0 deletions
diff --git a/results/classifier/deepseek-2/output/device/1998 b/results/classifier/deepseek-2/output/device/1998 new file mode 100644 index 000000000..7849f2057 --- /dev/null +++ b/results/classifier/deepseek-2/output/device/1998 @@ -0,0 +1,28 @@ + +acpihp does not work with some common guest kernels +Description of problem: +for pc-q35 6.1, 7.2, any guest kernel with `ACPI: Core revision` < 20230331, can not hot plug the nvidia GPUs. +So basically only guest kernel >= 6.5 can make it work so far. +But majority of server kernels are still at 4.18, 5.x. I wonder if it possible to be fixed? +I also don't know is this qemu bug? bios bug? or actually ACPIA's bug? + +journal -k report error like following: +``` +Nov 11 17:53:00 VMTEST kernel: pci 0000:08:00.0: BAR 0: no space for [mem size 0x01000000] +Nov 11 17:53:00 VMTEST kernel: pci 0000:08:00.0: BAR 0: failed to assign [mem size 0x01000000] +Nov 11 17:53:00 VMTEST kernel: pci 0000:08:00.0: BAR 6: assigned [mem 0x81800000-0x8187ffff pref] +Nov 11 17:53:00 VMTEST kernel: pci 0000:08:00.0: BAR 5: assigned [io 0xa000-0xa07f] +Nov 11 17:53:00 VMTEST kernel: nvidia 0000:08:00.0: enabling device (0000 -> 0003) +Nov 11 17:53:00 VMTEST kernel: NVRM: This PCI I/O region assigned to your NVIDIA device is invalid: + NVRM: BAR0 is 0M @ 0x0 (PCI:0000:08:00.0) +Nov 11 17:53:00 VMTEST kernel: nvidia: probe of 0000:08:00.0 failed with error -1 +``` +Steps to reproduce: +1. run the instance as I described above +2. in qemu monitor: device_add vfio-pci,host=0000:06:00.0,id=gpu0,bus=pci.8 +3. login to the vm console then nvidia-smi to see the failure + +workaround: +`ICH9-LPC.acpi-pci-hotplug-with-bridge-support=off` to disable the acpihp then pciehp can make it work. +Additional information: + |