On 8.1.2021 2.22, Kevin Locke wrote: > On a ThinkPad T430 (2342-CTO) with an Intel 7 Series USB xHCI Host > Controller (8086:1e31) running Linux 5.11-rc2 (or torvalds master > 71c061d24438, or 4.19.0-13-amd64 from Debian stable), I can provoke a > kernel Oops in code called from xhci_configure_endpoint by doing the > following (performed after booting into single-user mode): > > 1. Connect USB3 UAS disk drive (Seagate Backup Plus Hub - 1180:e823) > 2. When "Spinning up disk..." is logged by the kernel, initiate suspend > to memory (`echo mem >/sys/power/state`). > 3. Once suspended, disconnect the drive. > 4. Initiate wake then immediately plug and unplug a USB2 device (Yubikey > 1050:0406) in the same port repeatedly. > > This usually causes the system to freeze (unresponsive to Magic SysRq). > If not, repeat the steps. Relevant (decoded) dmesg (from EFI pstore): > > kernel BUG at lib/list_debug.c:54! This Panther Point xHCI was the only one that needed software bandwidth management. Looks like the bandwidth management list is left in some bad state, I'll take a look at it. ... > > Let me know if there's anything else I can do to help diagnose/debug. Thanks, I might need some help with testing patches -Mathias
