[Bug 220906] New: Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[Bug 220906] New: Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

            Bug ID: 220906
           Summary: Regression: 4K display output fails on ThinkPad
                    Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1,
                    broken in 6.17.11)
           Product: Drivers
           Version: 2.5
          Hardware: AMD
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P3
         Component: Video(DRI - non Intel)
          Assignee: drivers_video-dri@kernel-bugs.osdl.org
          Reporter: danandrei@gmail.com
        Regression: No

Created attachment 309076
  --> https://bugzilla.kernel.org/attachment.cgi?id=309076&action=edit
dmesg boot output for 6.17.1 & 6.17.12

A regression in recent kernels prevents 4K output over a ThinkPad Thunderbolt 4
Dock when using amdgpu.

After upgrading to kernel 6.17.11‑300.fc43.x86_64 on a Lenovo T14s Gen 6 (AMD),
my 4K external monitor connected through a ThinkPad Thunderbolt 4 Dock no
longer works at 4K resolution. The display only shows output at lower
resolutions, and even those modes are effectively unusable.

This is a regression: the same hardware setup works correctly on kernel
6.17.1‑300.fc43.x86_64, including the version shipped with
Fedora‑KDE‑Desktop‑Live‑43‑1.6.x86_64.iso. Booting back into 6.17.1 restores
full 4K functionality.

The issue is fully reproducible by using the affected kernel versions with this
hardware combination. No out‑of‑tree modules are installed; this is a fresh
Fedora installation.

I also tested the latest Rawhide kernel (6.18 branch), and the problem persists
there as well.

When attempting to switch the display to 4K, the kernel logs repeatedly show
the following amdgpu messages:

```
journalctl -f | grep -i 'amdgpu'

Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:37 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:38 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:53 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
Dec 14 22:37:53 GOTO kernel: amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ
callback: link_index=5
```

I noticed that Bug 219748 (Pluggable UD‑4VPD dock resetting on AMD hardware)
shows the same repeated [drm] DMUB HPD RX IRQ callback messages. The hardware
is similar and also involves a USB‑C/Thunderbolt dock, so this may be related.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

Andrei (danandrei@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Regression|No                          |Yes

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

Andrei (danandrei@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Kernel Version|                            |6.17.11

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

Artem S. Tashkinov (aros@gmx.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |ANSWERED

--- Comment #1 from Artem S. Tashkinov (aros@gmx.com) ---
Please repost here: https://gitlab.freedesktop.org/drm/amd/-/issues

Bisection is advised or at the very least check kernel 6.18.2.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

--- Comment #2 from Andrei (danandrei@gmail.com) ---
Created attachment 309087
  --> https://bugzilla.kernel.org/attachment.cgi?id=309087&action=edit
dmesg_6.18.2-00.fc43.x86_64

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

--- Comment #3 from Andrei (danandrei@gmail.com) ---
Created attachment 309088
  --> https://bugzilla.kernel.org/attachment.cgi?id=309088&action=edit
dmesg_6.17.4-300.fc43.x86_64

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

--- Comment #4 from Andrei (danandrei@gmail.com) ---
Created attachment 309089
  --> https://bugzilla.kernel.org/attachment.cgi?id=309089&action=edit
dmesg_6.17.5-300.fc43.x86_64

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

--- Comment #5 from Andrei (danandrei@gmail.com) ---
Created attachment 309090
  --> https://bugzilla.kernel.org/attachment.cgi?id=309090&action=edit
dmesg_6.17.6-300.fc43.x86_64

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

Andrei (danandrei@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|ANSWERED                    |---

--- Comment #6 from Andrei (danandrei@gmail.com) ---
I also checked 6.18.2, it still exhibits the issue. I went through minor
versions between 6.17.1 and 6.17.11, and found the version where the issue was
first introduced.

6.17.4 - works
6.17.5 - works
6.17.6 - does not work

One extra note:

I see the amdgpu line that I originally thought associated the problem, also
appears in the dmesg output for working kernel versions.

For example, 6.17.4 (working):

[    6.602684] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5
[    6.612716] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5
[    6.634430] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5
[    6.639137] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5

6.17.5 (working):

[    6.468886] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5
[    6.473641] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5
[    6.479753] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5
[    6.492847] amdgpu 0000:c3:00.0: amdgpu: [drm] DMUB HPD RX IRQ callback:
link_index=5

The following may or may not be relevant, but here's something that appears
only in problematic versions, and is missing from working versions:

6.17.6:

[   18.911013] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 6935 (idx
101 out of bounds)
[   18.986136] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 7015 (idx
101 out of bounds)
[   19.061037] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 7095 (idx
101 out of bounds)
[   20.562028] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 6435 (idx
101 out of bounds)

6.17.12:

[   34.459370] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 6455 (idx
101 out of bounds)
[   34.534471] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 6535 (idx
101 out of bounds)
[   34.609497] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 6615 (idx
101 out of bounds)
[   34.684571] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 6695 (idx
101 out of bounds)

6.18.2:

[   14.236205] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 5180 (idx
101 out of bounds)
[   14.391215] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 5200 (idx
101 out of bounds)
[   14.546169] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 5220 (idx
101 out of bounds)
[   14.700826] ath12k_pci 0000:c2:00.0: chan info: invalid frequency 5240 (idx
101 out of bounds)

In working versions, ath12k_pci logging terminates with the following:

6.17.1:

[   38.914023] ath12k_pci 0000:c2:00.0: failed to enqueue rx buf: -28

6.17.4:

[   37.071153] ath12k_pci 0000:c2:00.0: failed to enqueue rx buf: -28

6.17.5:

[    6.586797] ath12k_pci 0000:c2:00.0: failed to enqueue rx buf: -28

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

--- Comment #7 from Andrei (danandrei@gmail.com) ---
I initially thought this was just an AMD driver bug, but after digging through
the logs across a few more versions, I think I’ve found a better lead.

In 6.17.6, where the regression started for the first time, I'm seeing these
bridge resource failures pop up for the first time:

pci 0000:02:00.0: bridge window [io  size 0x1000]: can't assign; no space
pci 0000:02:00.0: bridge window [io  size 0x1000]: failed to assign

Bridge 02:00.0 is the gateway for the USB4/Thunderbolt 4 controller on this
machine. In version 6.17.5, the kernel assigns it just fine and my 4K output
works perfectly. Starting with 6.17.6, that assignment fails, which lines up
exactly with when the display stops working.

Interestingly, the ath12k driver also starts throwing an "invalid ACPI DSM TAS
config size" error in this same version. It feels like 6.17.6 changed something
fundamental about how ACPI or PCI resource arbitration is handled, and now it's
just choking out the I/O space that the Thunderbolt subsystem needs to
function.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

Andrei (danandrei@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|Video(DRI - non Intel)      |PCI
     Kernel Version|6.17.11                     |6.17.6

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

--- Comment #8 from Artem S. Tashkinov (aros@gmx.com) ---
Please bisect.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220906] Regression: 4K display output fails on ThinkPad Thunderbolt 4 Dock with AMD GPU (Worked in 6.17.1, broken in 6.17.11)

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220906

--- Comment #9 from Andrei (danandrei@gmail.com) ---
Thanks, I attempted a bisection, but it appears that the 6.17.6 I built from
source doesn't exhibit the bug, whereas the build I got from Fedora's Koji
does.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.