Success with recent Xen pv_ops dom0 kernel (2.6.29-rc8)

Success with recent Xen pv_ops dom0 kernel (2.6.29-rc8)

[Pasi Kärkkäinen]

Hello!

In addition to all the nagging about bugs I decided to report some positive
stuff too :) 

My current pv_ops dom0 testbox seems to work pretty well. It's Linux 2.6.29-rc8
based, built last saturday (2009-03-21). I'm running 32bit PAE pv_ops dom0 kernel 
on some older Intel P4 machine, which doesn't have HVM (Intel VT) support. 
So I can't test HVM guests.

Things that work based on my testing: 
- "xm info"
- "xm list"
- "xm create" <- PV domU, LVM volume as disk, bridged network, VNC console
- "xm destroy"
- "virt-install" to install new guests
- "virt-manager" to manage and install new guests
- "virsh" to manage guests

I successfully ran and installed CentOS 5.3 domU, Fedora 10 domU, and Fedora 11 (rawhide) domU.
I'm able to run multiple PV guests at the same time.. No crashes or anything weird based on 
a couple of hours of testing.

Things that are still broken:
- dom0 VGA text console doesn't work, only serial console works
- CONFIG_HIGHPTE=y crashes the dom0 kernel sooner or later, so I'm running CONFIG_HIGHPTE=n

Otherwise my setup consists of:
- Fedora 10 distribution on dom0, latest updates installed
- libvirt 0.6.1 and related packages from f10 updates-testing (earlier versions had a lot of problems)
- F11/rawhide Xen 3.3.1-9 packages rebuilt for F10
- New enough pv_ops dom0 kernel so it has /sys/hypervisor support for libvirt/virt-install/virt-manager

Missing features:
- blktap (tap:aio). Any plans for this? 
- PV domU PCI-passthrough? 

Congratulations and thanks to Jeremy (and others) for the hard work so far! 

-- Pasi

Re: Success with recent Xen pv_ops dom0 kernel (2.6.29-rc8)

[Pasi Kärkkäinen]

On Wed, Mar 25, 2009 at 05:34:06PM +0200, Pasi Kärkkäinen wrote:
> Hello!
> 
> In addition to all the nagging about bugs I decided to report some positive
> stuff too :) 
> 
> My current pv_ops dom0 testbox seems to work pretty well. It's Linux 2.6.29-rc8
> based, built last saturday (2009-03-21). I'm running 32bit PAE pv_ops dom0 kernel 
> on some older Intel P4 machine, which doesn't have HVM (Intel VT) support. 
> So I can't test HVM guests.
> 
> Things that work based on my testing: 
> - "xm info"
> - "xm list"
> - "xm create" <- PV domU, LVM volume as disk, bridged network, VNC console
> - "xm destroy"

- "xm reboot"
- "xm shutdown"
- "xm console" 

Those do work aswell, just forgot about them.

-- Pasi

Re: Success with recent Xen pv_ops dom0 kernel (2.6.29-rc8)

[Boris Derzhavets]

[-- Attachment #1.1: Type: text/plain, Size: 2082 bytes --]

>So I can't test HVM guests 

Lucky you.

Boris.
--- On Wed, 3/25/09, Pasi Kärkkäinen <pasik@iki.fi> wrote:
From: Pasi Kärkkäinen <pasik@iki.fi>
Subject: [Xen-devel] Success with recent Xen pv_ops dom0 kernel (2.6.29-rc8)
To: xen-devel@lists.xensource.com
Date: Wednesday, March 25, 2009, 11:34 AM

Hello!

In addition to all the nagging about bugs I decided to report some positive
stuff too :) 

My current pv_ops dom0 testbox seems to work pretty well. It's Linux
2.6.29-rc8
based, built last saturday (2009-03-21). I'm running 32bit PAE pv_ops dom0
kernel 
on some older Intel P4 machine, which doesn't have HVM (Intel VT) support. 
So I can't test HVM guests.

Things that work based on my testing: 
- "xm info"
- "xm list"
- "xm create" <- PV domU, LVM volume as disk, bridged network, VNC
console
- "xm destroy"
- "virt-install" to install new guests
- "virt-manager" to manage and install new guests
- "virsh" to manage guests

I successfully ran and installed CentOS 5.3 domU, Fedora 10 domU, and Fedora 11
(rawhide) domU.
I'm able to run multiple PV guests at the same time.. No crashes or
anything weird based on 
a couple of hours of testing.

Things that are still broken:
- dom0 VGA text console doesn't work, only serial console works
- CONFIG_HIGHPTE=y crashes the dom0 kernel sooner or later, so I'm running
CONFIG_HIGHPTE=n

Otherwise my setup consists of:
- Fedora 10 distribution on dom0, latest updates installed
- libvirt 0.6.1 and related packages from f10 updates-testing (earlier versions
had a lot of problems)
- F11/rawhide Xen 3.3.1-9 packages rebuilt for F10
- New enough pv_ops dom0 kernel so it has /sys/hypervisor support for
libvirt/virt-install/virt-manager

Missing features:
- blktap (tap:aio). Any plans for this? 
- PV domU PCI-passthrough? 

Congratulations and thanks to Jeremy (and others) for the hard work so far! 

-- Pasi

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



      

[-- Attachment #1.2: Type: text/html, Size: 2483 bytes --]

[-- Attachment #2: Type: text/plain, Size: 138 bytes --]

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

Re: Success with recent Xen pv_ops dom0 kernel (2.6.29-rc8)

[Jeremy Fitzhardinge]

Pasi Kärkkäinen wrote:
> Things that are still broken:
> - dom0 VGA text console doesn't work, only serial console works
>   

OK, Gerd identified a mapping problem that probably explains the lack of 
VGA.  I haven't had a chance to cook up a fix for it yet.

> - CONFIG_HIGHPTE=y crashes the dom0 kernel sooner or later, so I'm running CONFIG_HIGHPTE=n
>   

It's possible this is related.  Well, it would be nice ;)

> Otherwise my setup consists of:
> - Fedora 10 distribution on dom0, latest updates installed
> - libvirt 0.6.1 and related packages from f10 updates-testing (earlier versions had a lot of problems)
> - F11/rawhide Xen 3.3.1-9 packages rebuilt for F10
> - New enough pv_ops dom0 kernel so it has /sys/hypervisor support for libvirt/virt-install/virt-manager
>
> Missing features:
> - blktap (tap:aio). Any plans for this? 
> - PV domU PCI-passthrough? 
>   

Yes, both are planned.  blktap is on a fairly short list (waiting for 
blktap2 to land, rather than waste time on blktap1).

I think a lot of the tricky bits PCI passthrough are already done in the 
form of the rest of the dom0 pci changes, but I haven't looked at 
pciback in detail yet.

    J