[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 #67 from Francis DB (francisdb@gmail.com) ---
v4 (attachment 310447) results on the Framework Desktop (AMD Ryzen AI Max 300,
BIOS 03.05, xHCI 0000:c2:00.4): 146 clean s2idle cycles, zero controller deaths
(6 cycles at 300 s suspends, 140 at 60 s; baseline without the patch was ~32%
deaths per cycle, every run dead within 1-6 cycles). The link engages via the
SMU path: "amdgpu 0000:c2:00.0: [drm] USB controller 0000:c2:00.4 D0 power
state depends on 0000:c2:00.0". The v4 patch fixes the controller death on this
hardware.

On the full-system hang: after arming hung-task/soft/hardlockup panic (dump to
efi-pstore), I got 140 consecutive hang-free cycles on v4 with the rear ports
empty. Given I previously saw hangs within 1-13 cycles (twice on kernels
without the fix, once on v4 with USB devices attached to the affected
controller), it may be a rarer expression of the same resume race rather than
an independent bug. If it ever fires again the panic trace will be preserved
and I will attach 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.