You are not logged in.
Laptop specs:
Acer Aspire E5-574
i3-6100U / Intel HD 520
8GB of ram
Insyde UEFI BIOS v1.18
Tested kernels: 5.5.5.arch-1, 5.5.6-18-tkg-bmq, 5.4.21-1 LTS, 4.19.102 LTS, and self compiled 5.5.6.
No matter what I do it seems like the GVT firmware always fails to load on my laptop.
➜ ~ dmesg | grep i915
[0.709301] i915 0000:00:02.0: vgaarb: deactivate vga console
[0.710055] i915 0000:00:02.0: Direct firmware load for i915/gvt/vid_0x8086_did_0x1916_rid_0x07.golden_hw_state failed with error -2
[0.732166] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[0.733503] [drm] Finished loading DMC firmware i915/skl_dmc_ver1_27.bin (v1.27)
[0.755662] [drm] Initialized i915 1.6.0 20191101 for 0000:00:02.0 on minor 0
[0.768226] fbcon: i915drmfb (fb0) is primary device
[0.790556] i915 0000:00:02.0: fb0: i915drmfb frame buffer device
[3.086957] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])The error :
[0.710055] i915 0000:00:02.0: Direct firmware load for i915/gvt/vid_0x8086_did_0x1916_rid_0x07.golden_hw_state failed with error -2I'm currently passing i915.enable_gvt=1 as a kernel parameter trough GRUB, passing it with a modprobe.d file results in the same error.
➜ ~ cat /etc/modprobe.d/i915.conf
options i915 enable_gvt=1
kvmgt vfio-iommu-type1 vfio-mdev➜ ~ cat /etc/mkinitcpio.conf
MODULES=(kvmgt vfio-iommu-type1 vfio-mdev i915)On my self compiled kernel I've also added the following to my .config
CONFIG_VFIO_MDEV_DEVICE=y
CONFIG_VFIO_MDEV=y
CONFIG_DRM_I915_GVT=y
CONFIG_DRM_I915_GVT_KVMGT=y
CONFIG_DRM_I915_GVT_XENGT=yMy cmdline:
➜ ~ cat /proc/cmdline
BOOT_IMAGE=/@root/boot/vmlinuz-linux-tkg-bmq-skylake root=UUID=2bda37a2-e146-46fa-a03a-f03126420f12 rw rootflags=subvol=@root resume=UUID=b86d43cc-d896-fa47-97ec-7c144e4a91f0 loglevel=3 nowatchdogd i915.modset=1 i915.enable_gvt=1 intel_pstate=passive Last edited by rwxr-xr-xis (2020-02-26 05:50:47)
Offline
That error doesn't actually preclude GVT-g from working. Do you have access to, for e.g., /sys/devices/pci0000\:00/0000\:00\:02.0/mdev_supported_types/*. If so GVTg is up and running TMK.
$ tree /sys/devices/pci0000:00/0000:00:02.0/mdev_supported_types/
├── i915-GVTg_V5_4
│ ├── available_instances
│ ├── create
│ ├── description
│ ├── device_api
│ └── devices
│ └── 247e4dd2-fa4a-41a7-b31f-e8a502710967 -> ../../../247e4dd2-fa4a-41a7-b31f-e8a502710967
└── i915-GVTg_V5_8
├── available_instances
├── create
├── description
├── device_api
└── devices
5 directories, 8 files
$ journalctl -b -k | grep -i 'error -2'
Feb 25 22:41:52 archlinux kernel: i915 0000:00:02.0: Direct firmware load for i915/gvt/vid_0x8086_did_0x191b_rid_0x06.golden_hw_state failed with error -2Last edited by CarbonChauvinist (2020-02-26 05:36:11)
"the wind-blown way, wanna win? don't play"
Offline
That error doesn't actually preclude GVT-g from working. Do you have access to, for e.g., /sys/devices/pci0000\:00/0000\:00\:02.0/mdev_supported_types/*. If so GVTg is up and running TMK.
Nope.
➜ ~ ls /sys/devices/
breakpoint cstate_pkg isa msr pnp0 system uncore_cbox_0 uprobe
cpu i915 kprobe pci0000:00 power tracepoint uncore_cbox_1 virtual
cstate_core intel_pt LNXSYSTM:00 platform software uncore_arb uncore_imc wakeup0➜ ~ ls /sys/devices/pci0000:00/0000:00:02.0
ari_enabled d3cold_allowed enable link numa_node resource4
boot_vga device firmware_node local_cpulist power revision
broken_parity_status dev_rescan graphics local_cpus remove rom
class devspec gvt_firmware max_link_speed reset subsystem
config dma_mask_bits i2c-0 max_link_width resource subsystem_device
consistent_dma_mask_bits driver i2c-1 modalias resource0 subsystem_vendor
current_link_speed driver_override i2c-2 msi_bus resource2 uevent
current_link_width drm irq msi_irqs resource2_wc vendor➜ ~ sudo ls /sys/devices/* | grep mdevLast edited by rwxr-xr-xis (2020-02-26 05:08:51)
Offline
Hmm. I'm not sure. I'm passing the following:
$ cat /proc/cmdline
initrd=\intel-ucode.img initrd=\initramfs-linux.img root=UUID=4d59661c-0a7e-41ce-8d7b-6cdb367f6e91 rw sysrq_always_enabled=1 quiet kvm.ignore_msrs=1 intel_iommu=on iommu=ptIs the
i915.modset=1
in your commandline a mistype or an actual typo? Do you have to pass the modeset? What happens if you don't do early KMS (i.e. take it out of your initramfs modules)?
$ grep ^MODULES /etc/mkinitcpio.conf
MODULES=(kvmgt vfio vfio-iommu-type1 vfio-mdevDo you have VT-d enabled in your system's firmware?
journalctl -b -k | grep i915
Feb 25 22:41:52 archlinux kernel: i915 0000:00:02.0: VT-d active for gfx access
Feb 25 22:41:52 archlinux kernel: i915 0000:00:02.0: vgaarb: deactivate vga console
Feb 25 22:41:52 archlinux kernel: i915 0000:00:02.0: Direct firmware load for i915/gvt/vid_0x8086_did_0x191b_rid_0x06.golden_hw_state failed with error -2
Feb 25 22:41:52 archlinux kernel: i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=io+mem
Feb 25 22:41:52 archlinux kernel: [drm] Finished loading DMC firmware i915/skl_dmc_ver1_27.bin (v1.27)
Feb 25 22:41:52 archlinux kernel: [drm] Initialized i915 1.6.0 20191101 for 0000:00:02.0 on minor 0
Feb 25 22:41:52 archlinux kernel: fbcon: i915drmfb (fb0) is primary device
Feb 25 22:41:52 archlinux kernel: i915 0000:00:02.0: fb0: i915drmfb frame buffer device
Feb 25 22:41:52 archlinux kernel: i915 0000:00:02.0: MDEV: Registered
Feb 25 22:41:53 lap kernel: snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])Last edited by CarbonChauvinist (2020-02-26 05:28:00)
"the wind-blown way, wanna win? don't play"
Offline
Hmm. I'm not sure. I'm passing the following:
$ cat /proc/cmdline initrd=\intel-ucode.img initrd=\initramfs-linux.img root=UUID=4d59661c-0a7e-41ce-8d7b-6cdb367f6e91 rw sysrq_always_enabled=1 quiet kvm.ignore_msrs=1 intel_iommu=on iommu=ptIs the
i915.modset=1
in your commandline a mistype or an actual typo? Do you have to pass the modeset? What happens if you don't do early KMS (i.e. take it out of your initramfs modules)?
$ grep ^MODULES /etc/mkinitcpio.conf MODULES=(kvmgt vfio vfio-iommu-type1 vfio-mdev
i915.modset=1 was a typo from me copying the post :c, I put it there trying to get VT-d to work, anyway, with no modules on mkinitcpio.conf and passing kvm.ignore_msrs=1 intel_iommu=on iommu=pt i915.enable_gvt=1, VT-d/g seems to be working! (Although haven't tested in a vm yet).
Thanks a bunch for helping me out!
~ dmesg | grep i915
[ 0.000000] Command line: BOOT_IMAGE=/@root/boot/vmlinuz-linux root=UUID=2bda37a2-e146-46fa-a03a-f03126420f12 rw rootflags=subvol=@root resume=UUID=b86d43cc-d896-fa47-97ec-7c144e4a91f0 loglevel=3 nowatchdogd kvm.ignore_msrs=1 intel_iommu=on iommu=pt i915.fastboot=1 i915.enable_guc=0 i915.enable_gvt=1 i915.enable_fbc=1 i915.enable_psr=1 i915.enable_dc=4
[ 0.161986] Kernel command line: BOOT_IMAGE=/@root/boot/vmlinuz-linux root=UUID=2bda37a2-e146-46fa-a03a-f03126420f12 rw rootflags=subvol=@root resume=UUID=b86d43cc-d896-fa47-97ec-7c144e4a91f0 loglevel=3 nowatchdogd kvm.ignore_msrs=1 intel_iommu=on iommu=pt i915.fastboot=1 i915.enable_guc=0 i915.enable_gvt=1 i915.enable_fbc=1 i915.enable_psr=1 i915.enable_dc=4
[ 0.967787] i915 0000:00:02.0: VT-d active for gfx access
[ 0.967901] i915 0000:00:02.0: vgaarb: deactivate vga console
[ 0.968730] i915 0000:00:02.0: Direct firmware load for i915/gvt/vid_0x8086_did_0x1916_rid_0x07.golden_hw_state failed with error -2
[ 0.999476] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[ 0.999853] [drm] Finished loading DMC firmware i915/skl_dmc_ver1_27.bin (v1.27)
[ 1.010721] [drm] Initialized i915 1.6.0 20191101 for 0000:00:02.0 on minor 0
[ 1.026158] fbcon: i915drmfb (fb0) is primary device
[ 1.060988] i915 0000:00:02.0: fb0: i915drmfb frame buffer device
[ 1.087278] i915 0000:00:02.0: MDEV: Registered
[ 4.701995] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])➜ ~ ls /sys/devices/pci0000:00/0000:00:02.0/mdev_supported_types/
i915-GVTg_V5_4 i915-GVTg_V5_8Last edited by rwxr-xr-xis (2020-02-26 05:49:17)
Offline
Nice, please edit your first post's title and prepend with [SOLVED] (if you feel it is in fact resolved).
Last edited by CarbonChauvinist (2020-02-26 05:33:27)
"the wind-blown way, wanna win? don't play"
Offline