Xen users - Jul 2011 - Re: [Xen-devel] Failure to create HVM DomU at Xen 4.1 ( kernel 3.0.0-5-generic) Ubuntu 11.10 (alpha 2)

According to several threads, such as mine last month - ''[Xen-users] 
Working with Fedora 15 & systemd'' - and  ''[Xen-users]
Problems
with HVM after upgrade from 4.0.1 to 4.1.1'', and a private
communication with
t.wagner in ''[Xen-users] XEN-4.1.1 and linux kernel 3.0-rc5'',
and finally
''[Xen-users] Re: Trouble starting HVM domU with Linux 3.0.0 and Xen
4.1.1'',

hvm & xen 4.1.x don''t mix, either with kernel 3.0.0, or pvops
2.6.32 (which
was working with xen 4.0.2 prior to upgrading from fedora 14 to f15, and xen 
4.1. In the last thread mentioned above, the solution was to upgrade to xen 
4.2 unstable. My equally effective solution was to downgrade to xen 4.0.2, 
since fedora rawhide doesn''t have 4.2 yet.

Sorry for the top post - I''m not subscribed, but I thought the summary
was
worth posting.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

hi jim,

i just posted a thread severa hours ago about xen 4.2 unstable, 2.6.39.3 and
problem creating hvm guest''s. I would be interested if you have more
details
why there are problems. my problem actually is that my guest''s cant
access
hw resources from dom0!

thanks walter.

-----Original Message-----
From: xen-users-bounces@lists.xensource.com
[mailto:xen-users-bounces@lists.xensource.com] On Behalf Of jim burns
Sent: Sonntag, 17. Juli 2011 21:47
To: bderzhavets@yahoo.com; xen-users@lists.xensource.com
Subject: [Xen-users] Re: [Xen-devel] Failure to create HVM DomU at Xen 4.1 (
kernel 3.0.0-5-generic) Ubuntu 11.10 (alpha 2)

According to several threads, such as mine last month - ''[Xen-users]
Working
with Fedora 15 & systemd'' - and  ''[Xen-users] Problems
with HVM after
upgrade from 4.0.1 to 4.1.1'', and a private communication with t.wagner
in
''[Xen-users] XEN-4.1.1 and linux kernel 3.0-rc5'', and finally
''[Xen-users]
Re: Trouble starting HVM domU with Linux 3.0.0 and Xen 4.1.1'',

hvm & xen 4.1.x don''t mix, either with kernel 3.0.0, or pvops
2.6.32 (which
was working with xen 4.0.2 prior to upgrading from fedora 14 to f15, and xen
4.1. In the last thread mentioned above, the solution was to upgrade to xen
4.2 unstable. My equally effective solution was to downgrade to xen 4.0.2,
since fedora rawhide doesn''t have 4.2 yet.

Sorry for the top post - I''m not subscribed, but I thought the summary
was
worth posting.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Sun July 17 2011 3:52:48 PM you wrote:
> hi jim,
> 
> i just posted a thread several hours ago about xen 4.2 unstable, 2.6.39.3
> and problem creating hvm guest''s. I would be interested if you
have more
> details why there are problems. my problem actually is that my
guest''s
> cant access hw resources from dom0!
This looks like a very different problem. Let''s see what jumps out at
me:

from dmesg: (looks like dom0 to me)
> [    0.000000] Kernel command line: dummy root=UUID=6d870fbe-f4a5-4aa6-
bc2f-502f81344e7b ro quiet console=tty0 nomodeset

Ah yes, grub*2*, sigh!
> [    0.041309] ACPI: EC: Look up EC in DSDT
> [    0.042064] ACPI Error: Field [CDW3] at 96 exceeds Buffer [NULL] size 64
> (bits) (20110316/dsopcode-236)
> [    0.042081] ACPI Error: Method parse/execution failed [\_SB_._OSC] (Node
ffff8800319119c0), AE_AML_BUFFER_LIMIT
(20110316/psparse-536)
> [    0.046996] ACPI: Interpreter enabled
> [    0.046996] ACPI: (supports S0 S5)
> [    0.047028] ACPI: Using PIC for interrupt routing
This bothers me that your dom0 can''t initialize acpi w/o errors. Maybe
that''s
why you chose ''acpi=0'' in your hvm config. (Btw, use hda/hdc
in your disk=
line, not xvda/xvdc, though xen *should* translate for you.)
> [    0.263779] Failed to setup GSI :1, err_code:-22
Lot of dom0 errors of this form. Any hardware problems on dom0 itself?

From ''qemu-dm'':
> qemu: ignoring not-understood drive `xvda''
Reinforces my comment above. Everything else looks ok here.

From xen-hotplug:
> RTNETLINK answers: Operation not supported
> /etc/xen/scripts/xen-network-common.sh: line 131: brctl: command not found
> /etc/xen/scripts/xen-network-common.sh: line 131: brctl: command not found
Ah ha! Nothing is going to work if you don''t have bridge utils
installed.
Also, make sure ''brctl'' is in the PATH.

From your xl log:

Each ''xl create'' ends in a lot of
''int13_harddisk:'' errors. Possibly also from
the xvda/hda mixup. Also, your server02 config says
''boot=''cd'', but your xl
log says:
> Boot from Hard Disk failed: could not read the boot disk
and then tries to boot from the cd. Don''t know why you didn''t
see anything
from the cd, tho''. (You had the cd inserted, right?)

Scratch that - I just saw you''re accessing the .iso file on /var/xen
with a
''phy:'' prefix. It should be ''file:''. (Or
some variation of tap: or tap2:.
However, since 3.0.0 doesn''t have blktap.ko, I doubt that 2.6.39.3
does,
unless Debian added it themselves.)

I''m going to assume that some of your dom0 dmesg errors are because xen
4.2 is
still ''unstable''. I haven''t worked with 4.2 yet.

Hope this helped. Good luck.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Sun July 17 2011 3:47:16 PM you wrote:
> According to several threads, such as mine last month -
''[Xen-users]
> Working with Fedora 15 & systemd'' - and  ''[Xen-users]
Problems
> with HVM after upgrade from 4.0.1 to 4.1.1'', and a private
communication
> with t.wagner in ''[Xen-users] XEN-4.1.1 and linux kernel
3.0-rc5'', and
> finally ''[Xen-users] Re: Trouble starting HVM domU with Linux
3.0.0 and
> Xen 4.1.1'',
> 
> hvm & xen 4.1.x don''t mix, either with kernel 3.0.0, or pvops
2.6.32 (which
> was working with xen 4.0.2 prior to upgrading from fedora 14 to f15, and
> xen 4.1. In the last thread mentioned above, the solution was to upgrade
> to xen 4.2 unstable. My equally effective solution was to downgrade to xen
> 4.0.2, since fedora rawhide doesn''t have 4.2 yet.
> 
> Sorry for the top post - I''m not subscribed, but I thought the
summary was
> worth posting.
On 7/18, in the main fork of this thread, Konrad proposed another solution:
> I believe the xl command expects you to use ''vfb'' option
instead of
> the vnc and vncdisplay for 4.1
> So:
> vfb = [ ''vnc=1, vnclisten=0.0.0.0,vncunused=1'']
I can confirm that this works. In fact, my winxp config has both the 
individual variables, and a vfb= line, and both xl and xm can happily create 
my domu. This worked under xen 4.0.2, and then I reinstalled xen 4.1.1 from 
fedora rawhide (where I got 3.0.0 dom0 from), and xm/xl create still work. 
Strangely, tho'', I''m still getting the error in
qemu-dm-winxp.log:
> xen be: console-0: xen be: console-0: initialise() failed
xm works as well as it always did on fedora 14/xen 4.0.2, with only the 
addition of a vfb= line. This is with 3.0.0, or pvops 2.6.32 (myoung),
altho''
the unoptimized drivers in 3.0.0 make 2.6.32 the clear favorite.

xl has several problems. ''localtime=1'' seems to be ignored,
and a vncviewer is
not auto-launched with vncviewer=1 or vncconsole=1. either as a separate 
variable, on in the vfb= line.

However, the major problem is xl puts the tap interface on the old style 
xenbr0 bridge, instead of the new style eth0 bridge. (Yes, I still have xend 
running - I want both systems to operate in parallel.) And the vif interface 
is created w/o an inet6 addr, so even if I manually move tap and vif to the 
eth0 bridge with brctl, my domu still comes up with no network connectivity.

Anybody found a solution to this? It probably only affects hvm domus.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users