[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

--- Comment #11 from Michał Pecio (michal.pecio@gmail.com) ---
Thanks.
This looks like we aren't getting IRQs or there is some problem with the IRQ
handler.

In dmesg (grep 0000:c4:00.4) we have the following:

// resuming the host controller and some root hub port manipulation
[  +0.190938] xhci_hcd 0000:c4:00.4: Setting command ring address to 0xffffe001
[  +0.002557] xhci_hcd 0000:c4:00.4: xhci_resume: starting usb1 port polling.
[  +0.000192] xhci_hcd 0000:c4:00.4: xhci_hub_status_data: stopping usb2 port
polling
[  +0.002367] xhci_hcd 0000:c4:00.4: xhci_hub_status_data: stopping usb1 port
polling
[  +0.000194] xhci_hcd 0000:c4:00.4: Get port status 1-1 read: 0xe63, return
0x507
[  +0.000007] xhci_hcd 0000:c4:00.4: Get port status 1-1 read: 0xe63, return
0x507
[  +0.000005] xhci_hcd 0000:c4:00.4: clear USB_PORT_FEAT_SUSPEND
[  +0.000001] xhci_hcd 0000:c4:00.4: PORTSC 0e63
[  +0.000008] xhci_hcd 0000:c4:00.4: Set port 1-1 link state, portsc: 0xe63,
write 0x10fe1
[  +0.009957] xhci_hcd 0000:c4:00.4: Get port status 2-1 read: 0x2a0, return
0x2a0
[  +0.006769] xhci_hcd 0000:c4:00.4: Set port 1-1 link state, portsc: 0xfe3,
write 0x10e01
[  +0.013286] xhci_hcd 0000:c4:00.4: Get port status 1-1 read: 0x400e03, return
0x40503
[  +0.000040] xhci_hcd 0000:c4:00.4: clear port1 suspend/resume change, portsc:
0xe03

// About 5 seconds later, a control URB is unlinked.
// This usually means timeout.
// It was probably an attempt to resume some USB device.
[  +4.026739] xhci_hcd 0000:c4:00.4: Cancel URB 000000003faffdb9, dev 1, ep
0x0, starting at offset 0xffff5960
[  +0.000065] xhci_hcd 0000:c4:00.4: // Ding dong!

// We try to stop this control EP, but no confirmation in 5 seconds.
[  +5.114893] xhci_hcd 0000:c4:00.4: Command timeout, USBSTS: 0x00000018 EINT
PCD
[  +0.000015] xhci_hcd 0000:c4:00.4: xHCI host not responding to stop endpoint
command

The command ring shows a single command at 0xffffe000 as expected:

 0 0x00000000ffffe000: Stop Ring Command: slot 1 sp 0 ep 1 flags C

The event ring shows completion of a transfer at 0xffff5980, which is probably
the last TRB of the URB starting at 0xffff5960, and completion of a command at
0xffffe000.

 0 0x00000000ffffd5f0: TRB 00000000ffff5980 status 'Success' len 0 slot 1 ep 1
type 'Transfer Event' flags e:C
 0 0x00000000ffffd600: TRB 00000000ffff5990 status 'Stopped - Length Invalid'
len 0 slot 1 ep 1 type 'Transfer Event' flags e:C
 0 0x00000000ffffd610: TRB 00000000ffffe000 status 'Success' len 0 slot 1 ep 0
type 'Command Completion Event' flags e:C

So it seems that the HW performed the control transfer and then stopped the
endpoint as requested, but we never learned about it.

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

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