mainly because singlegpu is a nightmare to even attempt to support

V

Varian-Killbot•2/27/24, 2:34 PM

Yeah I was seeing that as I researched what I actually had to do

V

Varian-Killbot•2/27/24, 2:34 PM

No harm done yet thankfully

H

HikariKnightOP•2/27/24, 2:38 PM

sudo rm /etc/dracut.conf.d/vfio.conf
sudo touch /etc/bazzite/initramfs/rebuild
rpm-ostree kargs --delete-if-present="iommu" --delete-if-present="amd_iommu" --delete-if-present="intel_iommu" --delete-if-present="rd.driver.pre=vfio_pci" --delete-if-present="vfio_pci.disable_vga" --delete-if-present="vfio_pci.ids"

sudo rm /etc/dracut.conf.d/vfio.conf
sudo touch /etc/bazzite/initramfs/rebuild
rpm-ostree kargs --delete-if-present="iommu" --delete-if-present="amd_iommu" --delete-if-present="intel_iommu" --delete-if-present="rd.driver.pre=vfio_pci" --delete-if-present="vfio_pci.disable_vga" --delete-if-present="vfio_pci.ids"

sudo rm /etc/dracut.conf.d/vfio.conf
sudo touch /etc/bazzite/initramfs/rebuild
rpm-ostree kargs --delete-if-present="iommu" --delete-if-present="amd_iommu" --delete-if-present="intel_iommu" --delete-if-present="rd.driver.pre=vfio_pci" --delete-if-present="vfio_pci.disable_vga" --delete-if-present="vfio_pci.ids"

sudo rm /etc/dracut.conf.d/vfio.conf
sudo touch /etc/bazzite/initramfs/rebuild
rpm-ostree kargs --delete-if-present="iommu" --delete-if-present="amd_iommu" --delete-if-present="intel_iommu" --delete-if-present="rd.driver.pre=vfio_pci" --delete-if-present="vfio_pci.disable_vga" --delete-if-present="vfio_pci.ids"

i believe those are the commands you would run to undo

enable-vfio

enable-vfio

enable-vfio

enable-vfio

V

Varian-Killbot•2/27/24, 2:38 PM

I'll give it a shot!

H

HikariKnightOP•2/27/24, 2:39 PM

if you rely on the kargs for anything else, remove the ones you want to keep from the command

V

Varian-Killbot•2/27/24, 2:42 PM

Don't think so, just need go through this, restart, and see if it fixed anything. Very good chance that this is all coincidence and Factorio is just... angry. Even still, it worked last night, this was the only thing I changed, so I'm trying anyway.

H

HikariKnightOP•2/27/24, 2:42 PM

it is worth a shot

H

HikariKnightOP•2/27/24, 2:43 PM

i have had a handful of systems in the past that just refused to reboot normally after enabling iommu but everything else worked fine

H

HikariKnightOP•2/27/24, 2:43 PM

but those systems were also ancient

AAru oh, if you're trying to actually reproduce the flashing white screen issue, then...

J

jipiboily•2/27/24, 2:43 PM

could not repro, but here it is:

State: idle
Deployments:
● ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:248148043bcaf46be5a318e2d706350ce84a893cae296ff14fa247d87a7528ba
                  Version: 39.20240226.0 (2024-02-27T00:26:49Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"' 

  ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:38065b9645c29e4222926ef83c773b2ec4d900b713db2d8b12c3226d51ade84a
                  Version: 39.20240224.0 (2024-02-25T07:26:14Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"'

State: idle
Deployments:
● ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:248148043bcaf46be5a318e2d706350ce84a893cae296ff14fa247d87a7528ba
                  Version: 39.20240226.0 (2024-02-27T00:26:49Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"' 

  ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:38065b9645c29e4222926ef83c773b2ec4d900b713db2d8b12c3226d51ade84a
                  Version: 39.20240224.0 (2024-02-25T07:26:14Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"'

State: idle
Deployments:
● ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:248148043bcaf46be5a318e2d706350ce84a893cae296ff14fa247d87a7528ba
                  Version: 39.20240226.0 (2024-02-27T00:26:49Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"' 

  ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:38065b9645c29e4222926ef83c773b2ec4d900b713db2d8b12c3226d51ade84a
                  Version: 39.20240224.0 (2024-02-25T07:26:14Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"'

State: idle
Deployments:
● ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:248148043bcaf46be5a318e2d706350ce84a893cae296ff14fa247d87a7528ba
                  Version: 39.20240226.0 (2024-02-27T00:26:49Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"' 

  ostree-image-signed:docker://ghcr.io/ublue-os/bazzite-deck-gnome:stable
                   Digest: sha256:38065b9645c29e4222926ef83c773b2ec4d900b713db2d8b12c3226d51ade84a
                  Version: 39.20240224.0 (2024-02-25T07:26:14Z)
                Initramfs: '"-I /etc/crypttab /etc/modprobe.d/amdgpu.conf /etc/modprobe.d/deck-blacklist.conf"'

I'm using the gnome deck version, downloaded yesterday on Bazzite's website.

V

Varian-Killbot•2/27/24, 2:43 PM

Alright reboot time

J

jipiboily•2/27/24, 2:44 PM

Also, suspend doesn't seem to work well, so going to rollback to previous release

HHikariKnight mainly because singlegpu is a nightmare to even attempt to support

F

Frozen•2/27/24, 2:58 PM

Single GPU and dual GPU passthrough are very different beasts. I researched single GPU for a couple months before deciding to just save up for a used second GPU. So much easier.

Jjipiboily Also, suspend doesn't seem to work well, so going to rollback to previous releas...

J

jipiboily•2/27/24, 2:59 PM

rolling back did fix the suspend issue, so there's that, at least

T

Tayler•2/27/24, 3:01 PM

Is anyone else here running VRR-patched Gnome with an AMD GPU? I think I might be narrowing in on a bug but it would be nice if someone else could test and verify, too

FFrozen Single GPU and dual GPU passthrough are *very* different beasts. I researched si...

H

HikariKnightOP•2/27/24, 3:01 PM

i do both single and dual gpu, i also develop 2 tools to specifically help with gpu passthrough (mainly dual but one of the tools is useful for both)

testers are few and far between (especially since i dont use the main tool myself anymore since most my stuff is now ostree based)

TTayler Is anyone else here running VRR-patched Gnome with an AMD GPU? I think I might b...

T

Tayler•2/27/24, 3:05 PM

And to clarify a bit more, I think VRR does not function properly when running games with high GPU utilization unless you disable KMS modifiers by setting MUTTER_DEBUG_SEND_KMS_MODIFIERS=0

T

Tayler•2/27/24, 3:05 PM

I believe in both Xorg and Wayland but most noticably in Wayland.

V

Varian-Killbot•2/27/24, 3:08 PM

Unfortunately, removing the args was definitely not the fix, but I expected that anyway. Regardless, good to remove args I won't actually be using lol

T

Tayler•2/27/24, 3:11 PM

I've been trying to figure out what's going on with VRR on Wayland for a while now and the other day I finally put together this theory from several different issue tickets in different window managers. Many of these WMs and DEs just use VRRTest or something to test that VRR works, but the issue only pops up when you're running an actual game. If you start up VRRTest, then start a demanding game on another workspace and switch back to VRRTest, you should see that VRRTest starts to stutter and skip. When I disabled KMS Modifiers, VRRTest no longer stuttered and skipped--it's also noticable in game of course, but checking VRRTest with high GPU load is a more consistent and obvious way to see if VRR is still working.

T

Tayler•2/27/24, 3:11 PM

I'm not really sure where to put this information--since it's happening in every WM I've tried, I think it's probably an AMD bug? But if disabling modifiers fixes it, then it's also something that WM/DEs could work around, so...

T

Tayler•2/27/24, 3:14 PM

And also something bazzite could work around, if it's found this really is affecting other AMD users

TTayler I've been trying to figure out what's going on with VRR on Wayland for a while n...

J

Jan•2/27/24, 3:16 PM

so what you are saying is that the GPU can't keep up the frametime and so starts stuttering when expected to spew out as many frames as possible

J

Jan•2/27/24, 3:17 PM

I dont think this is a bug unless you can verify this on games that aren't very demanding

T

Tayler•2/27/24, 3:25 PM

It appears to be a frametime issue, yeah. But the point of VRR is that your frames are synced to the refresh rate, which is only useful when you're playing a demanding game that can't hit your max refresh rate, right? So if VRR "works" for low GPU usage situations but not high GPU usage ones... isn't it pretty much not working?

T

Tayler•2/27/24, 3:26 PM

I did find a ticket from I think a year ago, I can find it in a sec, where previously Gnome disabled modifiers on AMD because of some issue with xwayland. This wasn't VRR related but it at least shows there's a history of this being an issue for functionality I guess

T

Tayler•2/27/24, 3:29 PM

This is the revert for that, looks like modifiers were preventing direct scanout for DRI3 clients in xwayland: https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3030

T

Tayler•2/27/24, 3:34 PM

Also seems this was an issue people found 3 years ago: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4308

T

Tayler•2/27/24, 3:38 PM

Lots of discussion in this issue as well on sway, some of the testing people did in here seems to indicate disabling direct scanout can fix it as well: https://github.com/swaywm/sway/issues/5601

TTayler I've been trying to figure out what's going on with VRR on Wayland for a while n...

V

Virulent•2/27/24, 3:47 PM

Did you change the power profile to 3d_fullscreen?

V

Virulent•2/27/24, 3:47 PM

Because that's a known issue with AMD and has been for like two years

V

Virulent•2/27/24, 3:47 PM

You need to change the power profile to 3d_fullscreen

T

Tayler•2/27/24, 3:47 PM

Yes, I've had that fix in place for a long time

T

Tayler•2/27/24, 3:48 PM

For a long time I didn't think the issue was with VRR itself so I was running down other known issues like that power profile one

V

Virulent•2/27/24, 3:48 PM

Ah so it is a gnome-vrr bug?

T

Tayler•2/27/24, 3:49 PM

That's what I'm not sure about. It could be? But it seems there's similar issues in sway and hyprland

T

Tayler•2/27/24, 3:50 PM

and also I'm not completely confident that it's a generalized issue for everyone with amd on wayland, that's what I'm trying to figure out. it could be some specific setting or something

V

Virulent•2/27/24, 3:50 PM

I can try it out on plasma

J

j0rge•2/27/24, 3:56 PM

J

j0rge•2/27/24, 3:56 PM

bazzite is on GoL this morning!

O

OptimumSlinky•2/27/24, 3:56 PM

@Kyle Gospo Fresh boot, still getting that update error in game mode, so let me know what logs you’d like

Jj0rge bazzite is on GoL this morning!

D

dnkmmr•2/27/24, 4:07 PM

GoL?

J

j0rge•2/27/24, 4:07 PM

https://www.gamingonlinux.com/2024/02/pc-and-handheld-gaming-with-linux-levels-up-with-bazzite-v230/

GamingOnLinux

PC and handheld gaming with Linux levels up with Bazzite v2.3.0

Bazzite version 2.3.0 has been released, and it should make installing it a whole lot easier across different devices from PCs to handhelds. The developers think it's the next generation of Linux Gaming.