SOLVED 12th gen iGPU passthrough a possibility?

[Mosq1to]

Still have the same issue :S

what do you get when you run:

sudo dmesg | grep i915??

i get:
[ 0.017219] Kernel command line: BOOT_IMAGE=/ROOT/22.12.3@/boot/vmlinuz-5.15.107+truenas root=ZFS=boot-pool/ROOT/22.12.3 ro libata.allow_tpm=1 amd_iommu=on iommu=pt kvm_amd.npt=1 kvm_amd.avic=1 intel_iommu=on zfsforce=1 nvme_core.multipath=N i915.force_probe=4c9a
[ 9.746534] i915 0000:00:02.0: [drm] VT-d active for gfx access
[ 9.746563] i915 0000:00:02.0: vgaarb: deactivate vga console
[ 9.772790] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=none:owns=io+mem
[ 9.876754] i915 0000:00:02.0: [drm] Finished loading DMC firmware i915/rkl_dmc_ver2_03.bin (v2.3)
[ 9.882046] [drm] Initialized i915 1.6.0 20201103 for 0000:00:02.0 on minor 1
[ 9.884577] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])
[ 9.899722] fbcon: i915drmfb (fb1) is primary device
[ 11.041914] i915 0000:00:02.0: [drm] fb1: i915drmfb frame buffer device

and could you also post your container config?

steps:
1. sudo docker ps

2. get container ID for plex container and run

3. sudo docker inspect "containderID"

curious to se if you have both devices like me:
"Devices": [
{
"PathOnHost": "/dev/dri/card1",
"PathInContainer": "/dev/dri/card1",
"CgroupPermissions": "rw"
},
{
"PathOnHost": "/dev/dri/renderD128",
"PathInContainer": "/dev/dri/renderD128",
"CgroupPermissions": "rw"
}

thx in advance

 

[omeganot]

what do you get when you run:

sudo dmesg | grep i915??

Code:

[   15.034046] i915 0000:00:02.0: [drm] VT-d active for gfx access
[   15.058624] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=none:owns=io+mem
[   15.060276] i915 0000:00:02.0: Direct firmware load for i915/rkl_dmc_ver2_03.bin failed with error -2
[   15.060289] i915 0000:00:02.0: [drm] Failed to load DMC firmware i915/rkl_dmc_ver2_03.bin. Disabling runtime power management.
[   15.060289] i915 0000:00:02.0: [drm] DMC firmware homepage: https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/tree/i915
[   15.730899] i915 0000:00:02.0: [drm] failed to retrieve link info, disabling eDP
[   15.854308] [drm] Initialized i915 1.6.0 20201103 for 0000:00:02.0 on minor 1
[   15.857940] i915 0000:00:02.0: [drm] Cannot find any crtc or sizes
[   15.858099] i915 0000:00:02.0: [drm] Cannot find any crtc or sizes

Not sure why it's not loading, but once again, I can do full HW transcodes, that I just checked again immediately on content I know is h265 that I want to force to h264.

and could you also post your container config?

In entirety? No. But when it comes to devices:

Code:

            "Devices": [
                {
                    "PathOnHost": "/dev/dri/card1",
                    "PathInContainer": "/dev/dri/card1",
                    "CgroupPermissions": "rw"
                },
                {
                    "PathOnHost": "/dev/dri/renderD128",
                    "PathInContainer": "/dev/dri/renderD128",
                    "CgroupPermissions": "rw"
                }
            ],

Is your iGPU isolated in the advanced system settings? (/ui/system/advanced)
Mine isn't. Not sure if I could make it be isolated anyway, but it's not.

I have "Enable GPU support" checked in the Kubernetes settings, and in the Plex container I allocated #5 just to pick the last one:

 

[eldxmgw]

This Scale passthrough issue is well known and documentated by dozens of users since years.
But it isn't limited to GPU related devices. In general all PCIe devices suffer from this misbehavior.
And despite to opened jira tickets, feedbacks, backlogs aso. iX Systems seems to have no interest at all to fix that and rather leave the users alone.
Cause there're more unfixed than fixed community threads they speak for themselves.
My problem is i can't passthough my SAS PCIe card: https://www.truenas.com/community/t...ding-a-sas-hba-pci-passthrough-device.115874/

 

[Davvo]

Cause there're more unfixed than fixed community threads they speak for themselves.

That's because people don't know they can set a thread's status as fixed... or just don't care.
In your thread you did not post the required informations, I wouldn't be that assertive in your criticism for iX.

 

[eldxmgw]

That's because people don't know they can set a thread's status as fixed... or just don't care.
In your thread you did not post the required informations, I wouldn't be that assertive in your criticism for iX.

Really? Here's your proof: https://ixsystems.atlassian.net/jira/software/c/projects/NAS/issues/NAS-127338
And btw. i did so in the thread you're refering to also.

 

[Davvo]

And btw. i did so in the thread you're refering to also.

At the time my post was made your thread contained no hardware list at all.

 

[eldxmgw]

At the time my post was made your thread contained no hardware list at all.

That was not my point.
Even more critics, you told me aren't correct in terms of iX.

 

[Davvo]

That was not my point.
Even more critics, you told me aren't correct in terms of iX.

I'm not sure I understand what you want to convey. I replied to a specific statement and pointed out you were not following the forum's rules in your own thread while writing about the state of the community. Then, as you wrote that you did provide the required informations, I pointed out it was after my post in this thread.
Have I misunderstood your point?

 

[eldxmgw]

Seems so.

Key point was that my initial statement was:
".... and despite to opened jira tickets, feedbacks, backlogs aso. iX Systems seems to have no interest at all to fix that and rather leave the users alone. Cause there're more unfixed than fixed community threads they speak for themselves...."

You answered:
".... that's because people don't know they can set a thread's status as fixed... or just don't care. ... // ... I wouldn't be that assertive in your criticism for iX."

The proof of my initial critics can be seen today in my closed jira ticket.

 

[Davvo]

You answered:
".... that's because people don't know they can set a thread's status as fixed... or just don't care. ... // ... I wouldn't be that assertive in your criticism for iX."

You do not understand my posts.

 

[eldxmgw]

You do not understand my posts.

Your reality don't has to be my one.