Pkg xen devel - Jun 2011 - Bug#630175: 32 bit hvm windows xp pro guests crash on 64 bit amd host

Package: xen-hypervisor-4.0-amd64
Version: 4.0.1-2
Severity: normal



-- System Information:
Debian Release: 6.0.1
  APT prefers stable-updates
  APT policy: (500, 'stable-updates'), (500, 'stable')
Architecture: amd64 (x86_64)

Kernel: Linux 2.6.32-5-xen-amd64 (SMP w/4 CPU cores)
Locale: LANG=en_CA.UTF-8, LC_CTYPE=en_CA.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

xen-hypervisor-4.0-amd64 depends on no packages.

Versions of packages xen-hypervisor-4.0-amd64 recommends:
ii  xen-utils-4.0                 4.0.1-2    XEN administrative tools

Versions of packages xen-hypervisor-4.0-amd64 suggests:
pn  xen-docs-4.0                  <none>     (no description available)

-- no debconf information



after updrade from lenny to squeeze my windows xp pro guests began
repeatedly crashing. i am still unclear as to what is needed to
crash
the guests but it seems somewhat predictable. the less work the
guest
is doing the less frequent the crashes are.

one interesting thing to note which may or may not be related, is
that
/sometimes/, maybe once every 3 or 4 shutdowns, a shutdown of the xp
guest                                             will result in the
ssh connection to the host
being severed. it returns shortly afterward (within 30 - 120
seconds)
but usually results in a timeout and i then have to reestablish the
connection. this also happened in lenny.

things i have tried:

- memtest86: no errors
- upgrade xen to testing: the lockups became less frequent but still
occur several times/hour under use
- moved installation to different hardware completely: no
improvement
- tryed disk: phy, file: no help

in the end i am testing with completely new server hardware. on
it i have a stock
squeeze install with almost nothing but xen installed. i created
a new
lvm volume and tried to install xp pro into it from scratch. the
installer itself
crashes predictably at about the same progress point. i also
tried
installing onto an image filesystem image using disk: file and
exactly the same
thing happens.

here is the config i am using:


import os, re
arch = os.uname()[4]

kernel = "/usr/lib/xen-default/boot/hvmloader"
builder='hvm'

memory = 2048
maxmem = 4096
shadow_memory = 20
name = "brown3"
vif = [ 'type=ioemu, bridge=eth0' ]
disk = [ 'phy:/dev/mapper/lime-brown3,ioemu:xvda,w',
'phy:/dev/sr0,ioemu:xvdc:cdrom,r' ]
device_model = '/usr/lib/xen-default/bin/qemu-dm'
boot="dc"
sdl=0
vnc=1
vnclisten="0.0.0.0"
vncconsole=1
vncdisplay=4

vncpasswd='xxxxxxxx'
stdvga=0
serial='pty'
usbdevice='tablet'
localtime=1

on_poweroff = 'destroy'
on_reboot   = 'restart'
on_crash    = 'preserve'



i used the "on_crash = preserve" to try to capture what was
happening on
the dying host. on the install crash i have witnessed 2
different
messages on crash:
"an invalid parameter was passed to a service or function"
and
"not enough virtual memory or paging file quota is available to
complete
the specified operation"

shadow_memory was increased from 8 to 20 for testing purposed
but made
no improvement. anyway, as i understand it HAP is enabled on
this
machine, and overrides shadow_memory anyway.

both host machines are amd64, although not identical:
AMD Phenom(tm) II X4 970
AMD Athlon(tm) II X4 630
both have 8G ram


thank you very much for any clarification you can provide me
with.

sincerely,
proctor