xen.git plans

xen.git plans

[Jeremy Fitzhardinge]

Hi all,

I thought I'd give a quick update on my plans for xen.git for the next 
couple of months.

Firstly, I'll be taking ~3 weeks off over Christmas/NY, so nothing much 
will happen until I get back.

Current state:

xen/master:

    This seems to work fairly well for most people.  I merged blktap2
    support prematurely; it definitely isn't ready for use yet (it does,
    however, work enough to allow some enterprising person to debug, fix
    and send me patches, hint hint).

    But overall, this branch is getting a bit messy and tangled; see below.


xen/master-4.0:

    I'll make a branch off more-or-less current xen/master and add
    Xiantao's APIC API changes.  This branch will only work on Xen 4.0
    hypervisor (and 3.4 with backports), and will contain the "new" APIC
    setup code, which is what I intend to upstream.  So I'd like people
    to test this and report and regressions.

    Aside from that change, I'll keep it in sync with xen/master.


Future/WIP:

xen/next:

    At the moment I'm working on a fresh re-merge of all the Xen topic
    branches onto a new merge branch based on 2.6.32.  This is
    definitely Work-In-Progress (in fact, I haven't pushed my current
    state out yet).

    This idea is to make this a clean merge, containing the current best
    of all the topic branches, and ideally make it close to something we
    can upstream.  This will be Xen 4.0-only.

    When done, this will be the base for ongoing development.


xen/stable-2.6.32:

    Based on xen/next.  I'll only include bugfixes, stuff known to be
    stable, and merge in mainline stable point releases.  This should be
    a solid dom0 for Xen 4 users.

    I expect there'll be stable-2.6.X branches tracking future released
    upstream kernels.

xen/master:

    Main bleeding-edge development branch, based on xen/next re-merge.

    Policy remains the same as current xen/master: I won't merge
    absolutely broken things, and keep it generally useful, but new
    things will definitely be need of testing.


     J

Re: xen.git plans / 2.6.32 long-term stable kernel

[Pasi Kärkkäinen]

On Tue, Dec 22, 2009 at 12:59:29PM -0800, Jeremy Fitzhardinge wrote:
> Hi all,
>
> I thought I'd give a quick update on my plans for xen.git for the next  
> couple of months.
>
> Firstly, I'll be taking ~3 weeks off over Christmas/NY, so nothing much  
> will happen until I get back.
>

<snip>

>
> Future/WIP:
>
> xen/next:
>
>    At the moment I'm working on a fresh re-merge of all the Xen topic
>    branches onto a new merge branch based on 2.6.32.  This is
>    definitely Work-In-Progress (in fact, I haven't pushed my current
>    state out yet).
>
>    This idea is to make this a clean merge, containing the current best
>    of all the topic branches, and ideally make it close to something we
>    can upstream.  This will be Xen 4.0-only.
>
>    When done, this will be the base for ongoing development.
>
>
> xen/stable-2.6.32:
>
>    Based on xen/next.  I'll only include bugfixes, stuff known to be
>    stable, and merge in mainline stable point releases.  This should be
>    a solid dom0 for Xen 4 users.
>
>    I expect there'll be stable-2.6.X branches tracking future released
>    upstream kernels.
>

Hello,

It seems Linux 2.6.32 will be a long term supported stable release, 
ending up in multiple distros.. 

Distros that are probably going to use it:

	- Debian 6.0 ("Squeeze") 
	- Ubuntu 10.04 LTS ("Lucid Lynx")
	- RHEL6

Please correct me if I'm wrong :)

kernel.org developers are also going to make 2.6.32 a long-term stable release.
(announced yesterday).

The most interesting of those is Debian 6.0, since it looks like it'll ship pv_ops dom0 kernel..
but they need 2.6.32 based tree for that to happen :)

-- Pasi

Re: xen.git plans / 2.6.32 long-term stable kernel

[Jeremy Fitzhardinge]

On 01/20/2010 02:18 AM, Pasi Kärkkäinen wrote:
> It seems Linux 2.6.32 will be a long term supported stable release,
> ending up in multiple distros..
>
> Distros that are probably going to use it:
>
> 	- Debian 6.0 ("Squeeze")
> 	- Ubuntu 10.04 LTS ("Lucid Lynx")
> 	- RHEL6
>
> Please correct me if I'm wrong :)
>
> kernel.org developers are also going to make 2.6.32 a long-term stable release.
> (announced yesterday).
>    

Yep, that all works out nicely doesn't it?

     J

Re: xen.git plans / 2.6.32 long-term stable kernel

[Pasi Kärkkäinen]

On Wed, Jan 20, 2010 at 10:40:02AM -0800, Jeremy Fitzhardinge wrote:
> On 01/20/2010 02:18 AM, Pasi Kärkkäinen wrote:
>> It seems Linux 2.6.32 will be a long term supported stable release,
>> ending up in multiple distros..
>>
>> Distros that are probably going to use it:
>>
>> 	- Debian 6.0 ("Squeeze")
>> 	- Ubuntu 10.04 LTS ("Lucid Lynx")
>> 	- RHEL6
>>
>> Please correct me if I'm wrong :)
>>
>> kernel.org developers are also going to make 2.6.32 a long-term stable release.
>> (announced yesterday).
>>    
>
> Yep, that all works out nicely doesn't it?
>

Indeed it does. 

-- Pasi

Re: xen.git plans / 2.6.32 long-term stable kernel

[Josip Rodin]

On Wed, Jan 20, 2010 at 10:40:02AM -0800, Jeremy Fitzhardinge wrote:
> On 01/20/2010 02:18 AM, Pasi Kärkkäinen wrote:
>> It seems Linux 2.6.32 will be a long term supported stable release,
>> ending up in multiple distros..
>>
>> Distros that are probably going to use it:
>>
>> 	- Debian 6.0 ("Squeeze")
>> 	- Ubuntu 10.04 LTS ("Lucid Lynx")
>> 	- RHEL6
>>
>> Please correct me if I'm wrong :)
>>
>> kernel.org developers are also going to make 2.6.32 a long-term stable release.
>> (announced yesterday).
>
> Yep, that all works out nicely doesn't it?

Can you please update xen.git to have a branch based off
stable/linux-2.6.32.y.git ? Right now it's still a stable .31.6 and mainline
.32-rc5 in there, so we're missing a couple of months worth of bug fixes.

Since this kernel is going to enter so many stable distribution releases,
it would really be a good idea to get that exact combination widely tested
'in the wild'.

I'm asking particularly because the current Debian dom0 kernel (based on
.18 patches forward-ported onto .26) turned out to have a major regression
issue described in http://bugs.debian.org/534880 that didn't even get
reported until months after the release. Perhaps this particular bug is
specific to the old branch, but then the new paravirt_ops branch may have
some other, newer bugs of its own, so it makes sense to encourage widespread
testing as soon as possible.

My own 2.6.31.6+paravirt_ops dom0 test machine is still working great after
several months, so I'm optimistic. Thanks for all the hard work that made it
happen.

-- 
     2. That which causes joy or happiness.

Re: xen.git plans / 2.6.32 long-term stable kernel

[Sassy Natan]

[-- Attachment #1.1: Type: text/plain, Size: 1555 bytes --]

Hi Pasi/Jeremy

So Just let me understand, do u plan to ship Xen version 4 with linux 2.6.32
including pv_ops dom0 kernel?

I wonder when the Linux group will decide to include xen dom0 support in the
mainline?

Can you provide some more feedback on this?

Does Version 4 will include PV-USB?

If Debian intend to ship version 6 with 2.6.32 kernel based, I guess I
should send my patches to the dev group since I have build my own debian
package with xen 3.4.2 + Linux Kernel 2.6.31.5 + Xen dom0 support (old
style) + l7filter support.

The only bug I found in this version is that xen restore make some troubles.

Thanks
Sassy



On Wed, Jan 20, 2010 at 8:55 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote:

> On Wed, Jan 20, 2010 at 10:40:02AM -0800, Jeremy Fitzhardinge wrote:
> > On 01/20/2010 02:18 AM, Pasi Kärkkäinen wrote:
> >> It seems Linux 2.6.32 will be a long term supported stable release,
> >> ending up in multiple distros..
> >>
> >> Distros that are probably going to use it:
> >>
> >>      - Debian 6.0 ("Squeeze")
> >>      - Ubuntu 10.04 LTS ("Lucid Lynx")
> >>      - RHEL6
> >>
> >> Please correct me if I'm wrong :)
> >>
> >> kernel.org developers are also going to make 2.6.32 a long-term stable
> release.
> >> (announced yesterday).
> >>
> >
> > Yep, that all works out nicely doesn't it?
> >
>
> Indeed it does.
>
> -- Pasi
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>

[-- Attachment #1.2: Type: text/html, Size: 2401 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: xen.git plans / 2.6.32 long-term stable kernel

[Sassy Natan]

[-- Attachment #1.1: Type: text/plain, Size: 2193 bytes --]

Did u find kernel 2.6.31.6+paravirt_ops dom0 and Xen 3.4.2 I guess stable
enough?
I did found some issues with domain restore, but indeed this seems like the
best virtualization systems.

Thanks
Sassy



On Mon, Jan 25, 2010 at 3:38 PM, Josip Rodin <joy@entuzijast.net> wrote:

> On Wed, Jan 20, 2010 at 10:40:02AM -0800, Jeremy Fitzhardinge wrote:
> > On 01/20/2010 02:18 AM, Pasi Kärkkäinen wrote:
> >> It seems Linux 2.6.32 will be a long term supported stable release,
> >> ending up in multiple distros..
> >>
> >> Distros that are probably going to use it:
> >>
> >>      - Debian 6.0 ("Squeeze")
> >>      - Ubuntu 10.04 LTS ("Lucid Lynx")
> >>      - RHEL6
> >>
> >> Please correct me if I'm wrong :)
> >>
> >> kernel.org developers are also going to make 2.6.32 a long-term stable
> release.
> >> (announced yesterday).
> >
> > Yep, that all works out nicely doesn't it?
>
> Can you please update xen.git to have a branch based off
> stable/linux-2.6.32.y.git ? Right now it's still a stable .31.6 and
> mainline
> .32-rc5 in there, so we're missing a couple of months worth of bug fixes.
>
> Since this kernel is going to enter so many stable distribution releases,
> it would really be a good idea to get that exact combination widely tested
> 'in the wild'.
>
> I'm asking particularly because the current Debian dom0 kernel (based on
> .18 patches forward-ported onto .26) turned out to have a major regression
> issue described in http://bugs.debian.org/534880 that didn't even get
> reported until months after the release. Perhaps this particular bug is
> specific to the old branch, but then the new paravirt_ops branch may have
> some other, newer bugs of its own, so it makes sense to encourage
> widespread
> testing as soon as possible.
>
> My own 2.6.31.6+paravirt_ops dom0 test machine is still working great after
> several months, so I'm optimistic. Thanks for all the hard work that made
> it
> happen.
>
> --
>     2. That which causes joy or happiness.
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>

[-- Attachment #1.2: Type: text/html, Size: 3022 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: xen.git plans / 2.6.32 long-term stable kernel

[Josip Rodin]

On Mon, Jan 25, 2010 at 10:37:41PM +0200, Sassy Natan wrote:
> I wonder when the Linux group will decide to include xen dom0 support in
> the mainline?

Based on Jeremy's November presentation, that's not actually that important,
because the dom0 paravirt_ops support doesn't have a pressing need to
qualify for mainline Linux inclusion standards, while there is a practical
matter of it needing to become a qualified replacement for the old 2.6.18
kernel Xen patch. That patch has continued to be forward-ported to newer
versions and shipped with distribution kernels, which is what most users
use, but that process is becoming untenable as the kernel versions continue
to diverge.

-- 
     2. That which causes joy or happiness.

Re: xen.git plans / 2.6.32 long-term stable kernel

[Pasi Kärkkäinen]

On Wed, Jan 27, 2010 at 12:32:00PM +0100, Josip Rodin wrote:
> On Mon, Jan 25, 2010 at 10:37:41PM +0200, Sassy Natan wrote:
> > I wonder when the Linux group will decide to include xen dom0 support in
> > the mainline?
> 
> Based on Jeremy's November presentation, that's not actually that important,
> because the dom0 paravirt_ops support doesn't have a pressing need to
> qualify for mainline Linux inclusion standards, while there is a practical
> matter of it needing to become a qualified replacement for the old 2.6.18
> kernel Xen patch. That patch has continued to be forward-ported to newer
> versions and shipped with distribution kernels, which is what most users
> use, but that process is becoming untenable as the kernel versions continue
> to diverge.
> 

Yesterday I changed http://wiki.xensource.com/xenwiki/XenDom0Kernels to hilight 
that pv_ops dom0 is the kernel people should be testing and using nowadays.

-- Pasi