Bug 2266062 - AMD dGPU no longer available
Summary: AMD dGPU no longer available
Keywords:
Status: CLOSED COMPLETED
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: 39
Hardware: x86_64
OS: Linux
unspecified
high
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2024-02-26 14:17 UTC by sub
Modified: 2024-03-04 10:36 UTC (History)
18 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2024-03-04 10:36:00 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
dmesg.txt (138.79 KB, text/plain)
2024-02-26 14:17 UTC, sub
no flags Details
dmesg (270.13 KB, text/plain)
2024-02-28 22:37 UTC, Manu Reis
no flags Details

Description sub 2024-02-26 14:17:32 UTC
Created attachment 2018946 [details]
dmesg.txt

1. Please describe the problem:

After updating to kernel version 6.7.5, my AMD dGPU (AMD Radeon RX 6600) is no longer seen by my system. When performing [lspci | grep VGA] in the Terminal, it is seen, as well as in my System Details panel in Settings, but not from any other application that uses a Discrete Graphics card. As a result, my integrated graphics card is constantly working.

2. What is the Version-Release number of the kernel:

6.7.5 (and 6.7.6)

3. Did it work previously in Fedora? If so, what kernel version did the issue
   *first* appear?  Old kernels are available for download at
   https://koji.fedoraproject.org/koji/packageinfo?packageID=8 :

It did work properly on F39 with kernel version 6.7.4 (I rolled back to 6.7.4 and my AMD dGPU appears and works as intended)

4. Can you reproduce this issue? If so, please provide the steps to reproduce
   the issue below:

The problem appears as soon as I boot into Fedora 39 on kernel version 6.7.5 and onwards.

5. Does this problem occur with the latest Rawhide kernel? To install the
   Rawhide kernel, run ``sudo dnf install fedora-repos-rawhide`` followed by
   ``sudo dnf update --enablerepo=rawhide kernel``:

It still occurs on the latest Rawhide kernel. (6.8.0)

6. Are you running any modules that not shipped with directly Fedora's kernel?:

No.

7. Please attach the kernel logs. You can get the complete kernel log
   for a boot with ``journalctl --no-hostname -k > dmesg.txt``. If the
   issue occurred on a previous boot, use the journalctl ``-b`` flag.

Attaching below.

Comment 1 Christopher Klooz 2024-02-27 11:25:12 UTC
There is an ask.fedora ticket of the user about this: https://discussion.fedoraproject.org/t/cannot-suspend-computer-amd-dgpu-does-not-work-after-kernel-update-6-7-5/106362  
  
The user seems to have installed 6.7.5 using the normal dnf updates, and additionally tested 6.7.6 through the bodhi update [1] that is still in testing. They report that 6.7.4 always works properly, while 6.7.5 and 6.7.6 always have the reported problems.  
  
The user experiences two problems on 6.7.5 & 6.7.6 for which they wrote two bug reports: the other is BZ#2266057

[1] https://bodhi.fedoraproject.org/updates/FEDORA-2024-d16d94b00d

Comment 2 Manu Reis 2024-02-28 22:37:07 UTC
Created attachment 2019343 [details]
dmesg

Motherboard: Asrock B650e Taichi
BIOSv: 2.03
CPU: AMD Ryzen 7800x3D
RAM: DDR5 64GB
iGPU: AMD Radeon Graphics (RADV RAPHAEL_MENDOCINO)
dGPU: AMD Radeon RX 6600 XT (RADV NAVI23)

Comment 3 Manu Reis 2024-02-28 22:38:17 UTC
1. Please describe the problem:

I'm using Fedora 39 Silverblue and from kernel version 6.7.5 up to 6.7.6-200 my system won't fully recognize the dedicated GPU (AMD Radeon RX 6600 XT RADV NAVI23). As mention before, 'lspci' will show that the correct modules were loaded. However 'vulkaninfo' throws this warning: 
ERROR: [../src/amd/vulkan/radv_physical_device.c:1740] Code 0 : Could not open device /dev/dri/renderD128: Invalid argument (VK_ERROR_INCOMPATIBLE_DRIVER).

Related to this, the system won't enter sleep mode, and just power down any USB device attached but keep power to fans and leds.

2. What is the Version-Release number of the kernel:

Version 6.7.5 up to 6.7.6-200

3. Did it work previously in Fedora? If so, what kernel version did the issue
   *first* appear?  Old kernels are available for download at
   https://koji.fedoraproject.org/koji/packageinfo?packageID=8 :

It did work properly on F39 Silverblue with kernel version 6.7.4. 

Also, I've found that if plug the monitor directly to the dGPU (RX6600) this will fix the issue on kernels 6.7.5 up to 6.7.6-200. The system will suspend and wake without issues and 'vulkaninfo' will show both cards and the LLVM device without warnings.
However gnome will have issues, specifically window animations when I use the overview and switch from app to app. If I keep switching the animation will smooth out, but after stopping for a few seconds the issue will return. In general the system feels “choppy” but not sluggish. Sound through HDMI won't work despite having loaded the correct kernel modules.

4. Can you reproduce this issue? If so, please provide the steps to reproduce
   the issue below:

The problem appears as soon as I boot into Fedora Silverblue 39 on kernel version 6.7.5 and onwards on a system with two GPUs, both AMD. In my case: One is an integrated GPU from a Ryzen 7800x3D, the other being a 6600XT.

5. Does this problem occur with the latest Rawhide kernel? To install the
   Rawhide kernel, run ``sudo dnf install fedora-repos-rawhide`` followed by
   ``sudo dnf update --enablerepo=rawhide kernel``:

I did not test any rawhide kernel

6. Are you running any modules that not shipped with directly Fedora's kernel?:

No.

7. Please attach the kernel logs. You can get the complete kernel log
   for a boot with ``journalctl --no-hostname -k > dmesg.txt``. If the
   issue occurred on a previous boot, use the journalctl ``-b`` flag.

Attaching below.

Comment 4 sub 2024-03-04 10:35:32 UTC
Reporting back after updating to kernel 6.7.7.

This update fixes the missing AMD dGPU bug, as well as the irregular suspend behavior bug I reported alongside this one:

https://bugzilla.redhat.com/show_bug.cgi?id=2266057

Thank you for your hard work.


Note You need to log in before you can comment on or make changes to this bug.