[Bug 221073] xHCI host controller dies on resume from s2idle on AMD Strix Halo [1022:1587]

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=221073

jase_harley@protonmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jase_harley@protonmail.com

--- Comment #43 from jase_harley@protonmail.com ---
I can reproduce a very similar issue on a Lenovo ThinkPad T14 Gen 6 AMD / Ryzen
AI 7 PRO 350.

I do not currently have access to the device, so this is from notes/memory
rather than freshly collected logs.

Hardware:
- Lenovo ThinkPad T14 Gen 6 AMD
- CPU: AMD Ryzen AI 7 PRO 350
- BIOS: 1.13
- Internal camera: Chicony USB camera 04f2:b840

Distributions/kernels where observed:
- Ubuntu 24.04 LTS
- Ubuntu 26.04
- Fedora 43
- Fedora 44
- Kernel range observed: approximately 6.17.0 through 7.0.8

Symptoms:
- Internal Chicony USB camera 04f2:b840 disappears after suspend/resume.
- /dev/video* disappears when the failure occurs.
- Logs have shown USB resume failure / PM resume errors.
- The visible symptom is the webcam disappearing, but the failure appears to be
at the AMD xHCI / USB controller layer rather than a camera-specific issue.
- The machine has also had hard freezes/lockups around power-management events.

Workaround observations:
- Unbind/rebind of the affected xHCI controller can restore the camera.
- Forcing the AMD USB/xHCI PCI controllers to power/control=on before suspend
appears to have prevented the hard freezes so far.
- However, power/control=on has not fully prevented the internal webcam from
disappearing after resume.

Affected controllers seen on this machine include:
- 0000:c4:00.4
- 0000:c6:00.0
- 0000:c6:00.3
- 0000:c6:00.4

This may suggest two related behaviours:
1. runtime PM / autosuspend seems involved in the more severe freeze/lockup
behaviour;
2. the remaining webcam disappearance may still match this bug’s suspected xHCI
resume / interrupt / MSI-X issue.

I have not yet tested xhci_hcd.quirks=0x40. I can provide logs and test
xhci_hcd.quirks=0x40 if that would help.

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

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