panic: kernel diagnostic assertion "mii != NULL" failed

[Jaap Boender]

Hi all,

since 9.99.19, whenever I try to activate an urndis device, I get a
kernel panic with the following message:

kernel diagnostic assertion "mii != NULL" failed: file
"/usr/src/sys/dev/usb/usbnet.c" line 1191

The backtrace is:
_KERNEL_OPT_NARCNET() at 0
?() at ffff989b51e1f418
vpanic() at vpanic+0x181
kern_assert() at kern_assert+0x48
usbnet_tick_task() at usbnet_tick_task+0x20e
usb_task_thread() at usb_task_thread+0x47

The last kernel that does not have the problem is 9.99.18; the problem
occurs from 9.99.19, up to and including a -current I checked out a few
hours ago.

The problem occurs when, for example, trying to run dhcpcd -MnL urndis0.

This is on a NetBSD/amd64 system; kernel configuration is GENERIC with
nonexistent devices (on that particular machine) removed.

best

Jaap

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Alexander Nasonov]

Post by Jaap Boender
Hi all,
since 9.99.19, whenever I try to activate an urndis device, I get a kernel
kernel diagnostic assertion "mii != NULL" failed: file
"/usr/src/sys/dev/usb/usbnet.c" line 1191

kern/54762.

[Jaap Boender]

Post by Alexander Nasonov

Post by Jaap Boender
Hi all,
since 9.99.19, whenever I try to activate an urndis device, I get a kernel
kernel diagnostic assertion "mii != NULL" failed: file
"/usr/src/sys/dev/usb/usbnet.c" line 1191

kern/54762.

Yup - I can definitely confirm that 9.99.18 did not have that bug, and
that 9.99.19 did. There's a typo in the original bug report (mii, not
mi) so I guess that's why Google didn't pick it up...

best

Jaap


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Alexander Nasonov]

Post by Alexander Nasonov

Post by Jaap Boender
Hi all,
since 9.99.19, whenever I try to activate an urndis device, I get a kernel
kernel diagnostic assertion "mii != NULL" failed: file
"/usr/src/sys/dev/usb/usbnet.c" line 1191

kern/54762.

Yup - I can definitely confirm that 9.99.18 did not have that bug, and that
9.99.19 did. There's a typo in the original bug report (mii, not mi) so I
guess that's why Google didn't pick it up...

My working kernel is a very early 9.99.19:

NetBSD neva 9.99.19 NetBSD 9.99.19 (GENERIC) #0: Sun Dec 1 20:56:49 GMT
2019
***@lonet:/scratch/alnsn/netbsd-current/obj/sys/arch/amd64/compile/GENERIC
amd64

So, the bug was introduced (or triggered) by some change between these
two version bumps:

$ cvs log sys/sys/param.h
...
revision 1.621
date: 2019-12-06 21:07:07 +0000; author: ad; state: Exp; lines: +2
-2; commitid: 4eFCDlNIb8C7QGNB;
NetBSD 9.99.20 cpu_info/cpu_data changes for topology
----------------------------
revision 1.620
date: 2019-12-01 16:36:25 +0000; author: ad; state: Exp; lines: +2
-2; commitid: v6SbsZvvByg4v1NB;
NetBSD 9.99.19 - many kernel data structure changes

--
Alex

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de