Subject: kern/10440: Occasional panics from kue
To: None <gnats-bugs@gnats.netbsd.org>
From: Ben Harris <bjh21@cam.ac.uk>
List: netbsd-bugs
Date: 06/25/2000 09:06:15
>Number:         10440
>Category:       kern
>Synopsis:       Occasional panics from kue
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Jun 25 09:07:02 PDT 2000
>Closed-Date:
>Last-Modified:
>Originator:     Ben Harris
>Release:        1.4X
>Organization:
University of Cambridge
>Environment:
NetBSD cromarty 1.4X NetBSD 1.4X (CROMARTY) #9: Wed May  3 23:49:51 BST 2000
     bjh21@cromarty:/usr/src/sys/arch/macppc/compile/CROMARTY macppc

>Description:
Occasionally, usually under load, the "kue" driver seems to cause a
panic.  This doesn't happen to me very often, and I've only just set up my
system to automatically core-dump and reboot on panics, so I don't have a
core from it yet.  I do have a bunch of log messages from just before a
panic though:

Jun 18 13:57:05 cromarty /netbsd: kue0: watchdog timeout
Jun 18 13:57:05 cromarty /netbsd: usb_block_allocmem: in interrupt context, size=4096
Jun 18 13:57:05 cromarty /netbsd: usb_block_allocmem: in interrupt context, size=4096
Jun 18 13:57:05 cromarty /netbsd: usbd_transfer_cb: pipe==0, xfer=0xe007b980
Jun 18 13:57:05 cromarty last message repeated 2 times
Jun 18 13:57:05 cromarty /netbsd: usb_transfer_complete: bad dequeue 0xe0091400 != 0x0
[...]
Jun 18 14:53:49 cromarty uservd[29370]: user ceb28 -> root : prod-kue0
[...]
Jun 18 16:31:03 cromarty /netbsd: kue0: watchdog timeout
Jun 18 16:31:04 cromarty /netbsd: usbd_transfer_cb: pipe==0, xfer=0xe007b980
Jun 18 16:31:04 cromarty last message repeated 2 times
Jun 18 16:31:04 cromarty /netbsd: usb_transfer_complete: bad dequeue 0xe0091400 != 0x0
Jun 18 16:35:35 cromarty /netbsd: kue0: watchdog timeout
Jun 18 16:35:35 cromarty /netbsd: usbd_transfer_cb: pipe==0, xfer=0xe007b980
Jun 18 16:35:35 cromarty last message repeated 2 times
Jun 18 16:35:35 cromarty /netbsd: usb_transfer_complete: bad dequeue 0xe0091400 != 0x0

At this point, I think the system panicked, though problems with the
console make it hard to be sure.

The "uservd" message indicates a user running the script which does
"ifconfig kue0 down; ifconfig kue0 up", in order to deal with kue0 hanging
up as described in PR#10337.

I'll append more information if I manage to get a more useful panic.

>How-To-Repeat:

Not known.  Perhaps just repeatedly ifconfigging a kue down and up again
will do it.

-- 
Ben Harris
Unix Support, University of Cambridge Computing Service.
E-mail: bjh21@cam.ac.uk  Tel: +44 (0)1223 334728  Fax: +44 (0)1223 334679


>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted: