kern/59043: thinkpad a475 xhci resume fails: resume timeout, USBSTS.RSS

[mrg%eterna23.net@localhost]

>Number:         59043
>Category:       kern
>Synopsis:       thinkpad a475 xhci resume fails: resume timeout, USBSTS.RSS
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Jan 31 06:40:00 +0000 2025
>Originator:     matthew green
>Release:        -current
>Organization:
bozotic softwar foundation
>Environment:
amd64
>Description:
on a lenovo thinkpad a475 (AMD PRO A12-9800B R7, 12 COMPUTE CORES 4C+8G),
most things resume fine after suspending (including amdgpu!), but xhci
fails, and most devices don't come back.

i think something recently has made this less bad, because before this
latest test i ran to file this PR, it would fail to resume and then it
would not do anything, but some things seem to come back now if not all
of the devices.  here's the dmesg from suspend to resume:

acpi0: entering state S3
Flushing disk caches: done
cpu0: searching errata for cpu revision 0x00660f51
ioapic0 reenabling
xhci0: Clear PME# now
ehci1: Clear PME# now
[drm] PCIE GART of 1024M enabled (table at 0x000000F400900000).
smu version 27.18.00
[drm] UVD initialized successfully.
[drm] VCE initialized successfully.
xhci0: resume timeout, USBSTS.RSS
acpivga0: connected output devices:
acpivga0:   0x0100: Ext. Monitor, head 0, bios detect
acpivga0:   0x0110 (acpiout0): LCD Panel, head 0, bios detect
acpivga0:   0x0200: TV, head 0
acpivga0:   0x0210: Unknown Output Device, head 0, bios detect
acpivga0:   0x0220: Unknown Output Device, head 0, bios detect
nvme0: for admin queue interrupting at msix3 vec 0
nvme0: for io queue 1 interrupting at msix3 vec 1 affinity to cpu0
nvme0: for io queue 2 interrupting at msix3 vec 2 affinity to cpu1
nvme0: for io queue 3 interrupting at msix3 vec 3 affinity to cpu2
nvme0: for io queue 4 interrupting at msix3 vec 4 affinity to cpu3
re0: link state DOWN (was UP)
video0: detached
uvideo0: detached
uvideo0: at uhub3 port 3 (addr 3) disconnected
uvideo0 at uhub3 port 3 configuration 1 interface 0: AzureWave (0x13d3) Integrated Camera (0x5619), rev 2.00/16.20, addr 3
video0 at uvideo0: AzureWave (0x13d3) Integrated Camera (0x5619), rev 2.00/16.20, addr 3
ugen0: detached
ugen0: at uhub4 port 1 (addr 5) disconnected
ubt0: detached
ubt0: at uhub4 port 2 (addr 6) disconnected
uhub4: detached
uhub4: at uhub3 port 4 (addr 4) disconnected
uhub4 at uhub3 port 4: Genesys Logic (0x05e3) USB2.0 Hub (0x0610), class 9/0, rev 2.00/60.52, addr 4
uhub4: multiple transaction translators
uhub4: 3 ports with 3 removable, self powered
xhci0: xhci_set_dequeue: endpoint 0x0: timed out
ugen0 at uhub4 port 1
ugen0: Validity Sensors, Inc. (0x138a) product 0097 (0x0097), rev 2.00/1.64, addr 5
ubt0 at uhub4 port 2
ubt0: Realtek (0x0bda) Bluetooth Radio (0xb023), rev 1.10/1.10, addr 6
re0: link state UP (was DOWN)
xhci0: xhci_set_dequeue: endpoint 0x0: timed out
xhci0: xhci_set_dequeue: endpoint 0x0: timed out
[ this xhci_set_dequeue message now repeats every 5 seconds or so. ]

i don't know if this matters, but the devices that now reattach
properly after suspend are the internal USB devices, and it is
the external ports that remain non-functional.

i can't explain why uhub3 doesn't detach above and allows both
uvideo0 and uhub4 to reattach to it, allowing ubt and ugen both
to attach.
>How-To-Repeat:
on a thinkpad a475 run:

   sysctl -w hw.acpi.sleep.state=3
>Fix: