[Bug 220069] [6.13.9] regression USB controller dies

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=220069

--- Comment #21 from Michał Pecio (michal.pecio@gmail.com) ---
(In reply to Claudio Wunder from comment #18)
> In the last 2 times these match times I've been playing games. Another
> symptom before the reset, is the system freezes for a few seconds, including
> audio, but then it unfreezes with the devices then disconnecting.
If my tricks fail to accelerate the failure you can always go back to gaming ;)

The freeze is a known (at least to me) bug. I believe it's caused by the driver
waiting for the command abort to complete with xhci->lock held, which blocks
the IRQ handler for a few seconds as it tries to acquire the same lock, which
apparently causes other random IRQs to get delayed as well. I have been putting
off looking into it properly since December, maybe it's time to alert
Mathias...

> The mount was done after I ran unbind/bind from `ssh`...
> But here's the zip anyways, maybe it helps.
> https://gist.github.com/ovflowd/0b0aa5c748683eca33909dc3ed7c66f7#file-
> debugfs-zip

This zip is OK, but it contains current data from normal operation, not a
failure. A few random transfers on some devices and it looks like the familiar
8-3 hub is periodically getting autosuspended, maybe that's what sometimes goes
wrong.

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

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