[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

bugzilla@logical.ink changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bugzilla@logical.ink

--- Comment #39 from bugzilla@logical.ink ---
Also experiencing this bug on Framework 13 AMD Ryzen AI 300 using Fedora. 

### System Information

| Field | Value |
|---|---|
| Machine | Framework Laptop 13 (AMD Ryzen AI 300 Series) |
| CPU | AMD Ryzen AI 9 HX 370 (Hawk Point, 24 threads) |
| GPU | AMD Radeon 890M (integrated) |
| BIOS | FRANMGCP09 v03.05 (2025-10-30) |
| RAM | 64 GB DDR5 |
| Kernel (at time of event) | 6.18.12-100.fc42.x86_64 |
| Kernel (current) | 6.18.16-100.fc42.x86_64 |
| Distribution | Fedora 42 |
| xHCI controller | 0000:c1:00.4 |
| xHCI hcc params | 0x0118ffc5 |
| xHCI version | 0x120 (USB 3.1) |
| xHCI quirks at init | 0x0000000200000010 |

The affected USB device is `usb 1-1`: a full-speed internal device (device
number 2, bus 1) which corresponds to the Framework laptop's internal input
device hub (trackpad, keyboard controller, fingerprint reader).

Relevant Logs:
**Failing resume (44-minute sleep, lid closed):**
```
Mar 13 22:40:43 kernel: usb 1-1: reset full-speed USB device number 2 using
xhci_hcd
Mar 13 22:40:43 kernel: PM: suspend entry (s2idle)

[~44 minutes elapse]

Mar 13 23:25:33 kernel: PM: suspend devices took 0.162 seconds
Mar 13 23:25:33 kernel: PM: resume devices took 0.285 seconds
Mar 13 23:25:33 systemd-logind: Lid opened.
Mar 13 23:25:33 kernel: PM: suspend exit
                                                          <-- usb 1-1 reset
ABSENT
Mar 13 23:26:13 kernel: PM: suspend entry (s2idle)        <-- next suspend,
still no usb 1-1 reset
Mar 13 23:26:18 kernel: PM: suspend exit
                                                          <-- usb 1-1 reset
still ABSENT
Mar 13 23:27:09 kernel: PM: suspend entry (s2idle)
Mar 13 23:28:56 kernel: PM: suspend exit
                                                          <-- usb 1-1 reset
still ABSENT
[reboot required to restore input]
```

**xHCI controller initialization (boot):**
```
kernel: xhci_hcd 0000:c1:00.4: xHCI Host Controller
kernel: xhci_hcd 0000:c1:00.4: new USB bus registered, assigned bus number 1
kernel: xhci_hcd 0000:c1:00.4: hcc params 0x0118ffc5 hci version 0x120 quirks
0x0000000200000010
```

Additional Context:
- The xHCI controller at `0000:c1:00.4` already has quirks `0x0000000200000010`
applied at init. The `XHCI_RESET_ON_RESUME` bit (0x80) is not present and may
need to be added to the AMD platform quirk table for this hardware.
- `bluetoothd` logs `Controller resume with wake event 0x0` on all resumes
including the failing one, indicating the Bluetooth/wakeup path is not the
differentiating factor.
- No kernel error or warning is emitted at the time of the failed USB reinit —
the failure is silent.

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

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