The following reply was made to PR port-xen/35557; it has been noted by GNATS.

From: "Thomas Feddersen" <thomas-feddersen@t-online.de>
To: <gnats-bugs@NetBSD.org>
Cc: 
Subject: RE: port-xen/35557: Xen 3.0.3 crashes when attempting to mount a usb-drive
Date: Wed, 7 Feb 2007 14:25:17 +0100

 > -----Original Message-----
 > From: Manuel Bouyer [mailto:bouyer@antioche.eu.org]
 > Sent: Wednesday, February 07, 2007 12:25 PM
 > To: port-xen-maintainer@netbsd.org; gnats-admin@netbsd.org;=20
 > netbsd-bugs@netbsd.org; thomas.feddersen@t-online.de
 > Subject: Re: port-xen/35557: Xen 3.0.3 crashes when=20
 > attempting to mount a usb-drive
 >=20
 > =20
 >  The message is missing :)
 >  also, does the same happens with a plain i386 kernel ?
 > =20
 
 Somehow the webinterface ate part of my pr, here it comes again:
 
 >Description:
 when I plug a usb-drive in, the system=20
 - hangs immediately when in use a 2,5" drive
 - recognizes the 3,5" drive (WD2500JB), shows the=20
 disklabel correctly but when I try to mount it,=20
 # mount /dev/sd0e /usb
 the following pair of messages (green letters) keeps scolling over=20
 the console:
 	.
 	.
 	.
 xennetback_ifstart: no mc_pages (0)
 xennetback: can't get new mcl pages
 	.
 	.
 	.
 When I invoke DDB and execute
 db> ps
 The relevant process is denoted with a ">":
 PID=3D29 PPID=3D0 PGRP=3D0 UID=3D0 S=3D2 FLAGS=3D0x20200 LWPS=3D1 =
 COMMAND=3Dpagedaemon WAIT=3D
 
 
 The only way-out I know is to reboot:
 db> reboot
 
 Then I see a lot of green messages:
 rtk0: unable to allocate Rx mbuf
 	.
 	.
 	.
 xennetback_ifstart: no mc_pages (0)
 xennetback: can't get new mcl pages
 xennetback_ifstart: no mc_pages (0)
 xennetback: can't get new mcl pages
 pagedaemon: deadlock detected
 panic: pagedaemon deadlock
 
 and the system drops into debugger
 db> reboot
 
 Then the system reboots okay.=20
 =20
 >How-To-Repeat:
 Hardware: Intel DG965WHMKR MoBo + Core2Duo E6300 + 1GB KINGSTON =
 (PC2-4200
 CL4 240) + 6 SATA drives and one ATAPI DVD. The system boots from wd0a =
 and
 dumps on wd0b, wd1a-wd5a are configured as RAID5 (raidframe).
 
 Set up a system like above, boot it, plug in a usb drive. =20
 
 I tried AHCI and IDE settings for the SATA disks, both with the same
 results.
 
 This morning I saw the same phenomenon (abovementioned pair of messages
 scolling over console) but without operator interaction, the system had
 crashed sometime during the night. On a frozen terminal window I could =
 see
 that the (one and only) unprivileged domain=20
 -->
 kernel =3D "/usr/pkg/etc/xen/kernels/netbsd-XEN3_DOMU.gz" #Ver. Jan 23
 memory =3D 256
 name =3D "netbsd.kde"
 vif =3D [ 'mac=3D00:16:3e:00:01:13, bridge=3Dbridge0' ]
 disk =3D [ 'file:/raid/VirtualMachines/netbsd.kde,0x1,w' ]
 extra =3D ""
 <--
 had been active making /compiling KDE3 at the time.=20
 I invoked DDB but the system couldn't reboot. So I had to hardware reset =
 it.
 
 After restart (i386) I found nothing pertinent in the /var/log/messages =
 nor
 in the /var/log/xen/xend.log.=20
 
 I haven't restarted the Xen system yet, but I would be glad to perform
 certain tests and record messages if required.=20
 
 >Fix:
 The problem does not occur when I use a NetBSD 4.99.9 (GENERIC.MPACPI)
 kernel.=20
 (NetBSD 4 beta2 garbles the message buffer, cf. pr 35507)
 
 Kind Regards
  --
 Dipl.-Ing. Thomas Feddersen
 Beratender Ingenieur
 engineering for pleasure and profit