The following reply was made to PR kern/50491; it has been noted by GNATS.
From: Michael Plass <mfpnb%plass-family.net@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc:
Subject: Re: kern/50491: unkillable wait in usbd_transfer while using usmsc0 on raspberry pi 2
Date: Tue, 8 Dec 2015 19:27:02 -0800
On Dec 7, 2015, at 9:50 PM, Michael Plass wrote:
> I'm going to try it using a kernel without LOCKDEBUG
It looks good! No more unkillable lockups.
Maybe related, maybe not:
I did have a dropped tcp connection (ip4, I think). Also, I had a ipv6 =
connection
transferring from the pi to my laptop (inetd/chargen on the pi), and =
that transfer stalled - no data moving, but the connection didn't drop. =
This happened twice. The first time I killed the transfer, the other =
one is still in the stuck state.
The kernel and user stack traces are:
db{0}> bt/a b5ce74a0
trace: pid 4567 lid 1 at 0xb4e2fcdc
0xb4e2fcdc: netbsd:mi_switch+0x10
0xb4e2fd0c: netbsd:sleepq_block+0xb4
0xb4e2fd44: netbsd:cv_timedwait_sig+0x108
0xb4e2fee4: netbsd:kevent1+0x2cc
0xb4e2ff0c: netbsd:sys___kevent50+0x38
0xb4e2ff7c: netbsd:syscall+0x88
0xb4e2ffac: netbsd:swi_handler+0x98
db{0}> cont