News - Microsoft Submits Drivers for Paravirtualization

News - Microsoft Submits Drivers for Paravirtualization

[Nathan Eisenberg]

[-- Attachment #1.1: Type: text/plain, Size: 487 bytes --]

I don't normally news-post, cross-post, or even post to devel at all, but I thought this was quite interesting.

Apparently, Microsoft submitted 20,000 lines of device driver code for inclusion in the staging kernel.  Perhaps this is a step towards paravirtualized Windows guests in Xen?  That would be quite compelling.

http://www.networkworld.com/news/2009/072009-microsoft-linux-source-code.html

Best Regards
Nathan Eisenberg
Sr. Systems Administrator
Atlas Networks, LLC

[-- Attachment #1.2: Type: text/html, Size: 3131 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-devel] News - Microsoft Submits Drivers for Paravirtualization

[Thomas DuBuisson]

No, this seems to be a step in positioning Hyper-V to be the market
leading hypervisor by making Linux run as a guest on Hyper-V.  I don't
see this as any sort of good news from a Xen standpoint.  Perhaps I
missed something that made you think otherwise?

Thomas

On Mon, Jul 20, 2009 at 1:44 PM, Nathan
Eisenberg<nathan@atlasnetworks.us> wrote:
> I don’t normally news-post, cross-post, or even post to devel at all, but I
> thought this was quite interesting.
>
>
>
> Apparently, Microsoft submitted 20,000 lines of device driver code for
> inclusion in the staging kernel.  Perhaps this is a step towards
> paravirtualized Windows guests in Xen?  That would be quite compelling.
>
>
>
> http://www.networkworld.com/news/2009/072009-microsoft-linux-source-code.html
>
>
>
> Best Regards
>
> Nathan Eisenberg
>
> Sr. Systems Administrator
>
> Atlas Networks, LLC
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>
>

Re: News - Microsoft Submits Drivers for Paravirtualization

[Alan Cox]

On Mon, 20 Jul 2009 13:50:30 -0700
Thomas DuBuisson <thomas.dubuisson@gmail.com> wrote:

> No, this seems to be a step in positioning Hyper-V to be the market
> leading hypervisor by making Linux run as a guest on Hyper-V.  I don't
> see this as any sort of good news from a Xen standpoint.  Perhaps I
> missed something that made you think otherwise?

I assume the thinking is that from the Linux guest code you know what the
server is supposed to provide to pretend to be a hyper-v hypervisor to
windows. I'd bet there isn't sufficient info.

Re: News - Microsoft Submits Drivers for Paravirtualization

[Pasi Kärkkäinen]

On Mon, Jul 20, 2009 at 01:44:25PM -0700, Nathan Eisenberg wrote:
> I don't normally news-post, cross-post, or even post to devel at all, but I thought this was quite interesting.
> 
> Apparently, Microsoft submitted 20,000 lines of device driver code for inclusion in the staging kernel.  
>
> Perhaps this is a step towards paravirtualized Windows guests in Xen?  That would be quite compelling.
> 

Xen already supports "Enlightened" Win2008/Windows7 (and Vista?) when you
enable viridian=1 in the HVM domU cfg.

-- Pasi

Re: [Xen-devel] Re: Xen viridian support/features

[Pasi Kärkkäinen]

On Tue, Jul 21, 2009 at 10:23:26AM +0300, Pasi Kärkkäinen wrote:
> On Mon, Jul 20, 2009 at 01:44:25PM -0700, Nathan Eisenberg wrote:
> > I don't normally news-post, cross-post, or even post to devel at all, but I thought this was quite interesting.
> > 
> > Apparently, Microsoft submitted 20,000 lines of device driver code for inclusion in the staging kernel.  
> >
> > Perhaps this is a step towards paravirtualized Windows guests in Xen?  That would be quite compelling.
> > 
> 
> Xen already supports "Enlightened" Win2008/Windows7 (and Vista?) when you
> enable viridian=1 in the HVM domU cfg.
> 

Then again I'm not sure "how much" of viridian stuff Xen currently
supports.. 

Maybe someone else has more knowledge about this? 

-- Pasi

Re: [Xen-devel] Re: Xen viridian support/features

[Pasi Kärkkäinen]

On Tue, Jul 21, 2009 at 08:15:33PM +0300, Pasi Kärkkäinen wrote:
> On Tue, Jul 21, 2009 at 10:23:26AM +0300, Pasi Kärkkäinen wrote:
> > On Mon, Jul 20, 2009 at 01:44:25PM -0700, Nathan Eisenberg wrote:
> > > I don't normally news-post, cross-post, or even post to devel at all, but I thought this was quite interesting.
> > > 
> > > Apparently, Microsoft submitted 20,000 lines of device driver code for inclusion in the staging kernel.  
> > >
> > > Perhaps this is a step towards paravirtualized Windows guests in Xen?  That would be quite compelling.
> > > 
> > 
> > Xen already supports "Enlightened" Win2008/Windows7 (and Vista?) when you
> > enable viridian=1 in the HVM domU cfg.
> > 
> 
> Then again I'm not sure "how much" of viridian stuff Xen currently
> supports.. 
> 
> Maybe someone else has more knowledge about this? 
> 

CCing Keir.. maybe he has some thoughts about this.

-- Pasi

Re: Re: [Xen-users] Xen viridian support/features

[Tim Deegan]

At 18:15 +0100 on 21 Jul (1248200133), Pasi K?rkk?inen wrote:
> Then again I'm not sure "how much" of viridian stuff Xen currently
> supports.. 

At the moment we provide MSR access to APIC registers, and the hypercall
to yield the CPU when waiting for a spinlock.  We looked at a few others
(e.g. address-space-change hypercall) but found the benefits, if any,
were negligible.

The most important part of the interface is that we can use it to tell
Windows not to expect timer interrupt delivery to be as even across all
CPUs as it would be on bare metal.  This gets rid of annoying STOP 0x101
bluescreens in multi-vcpu Windows from Vista on.

Cheers,

Tim.

-- 
Tim Deegan <Tim.Deegan@citrix.com>
Principal Software Engineer, Citrix Systems (R&D) Ltd.
[Company #02300071, SL9 0DZ, UK.]

Re: Re: [Xen-users] Xen viridian support/features

[Pasi Kärkkäinen]

On Thu, Jul 23, 2009 at 03:11:01PM +0100, Tim Deegan wrote:
> At 18:15 +0100 on 21 Jul (1248200133), Pasi K?rkk?inen wrote:
> > Then again I'm not sure "how much" of viridian stuff Xen currently
> > supports.. 
> 
> At the moment we provide MSR access to APIC registers, and the hypercall
> to yield the CPU when waiting for a spinlock.  We looked at a few others
> (e.g. address-space-change hypercall) but found the benefits, if any,
> were negligible.
> 
> The most important part of the interface is that we can use it to tell
> Windows not to expect timer interrupt delivery to be as even across all
> CPUs as it would be on bare metal.  This gets rid of annoying STOP 0x101
> bluescreens in multi-vcpu Windows from Vista on.
> 

Thanks for the reply!

Any plans to write support for paravirtual viridian disk/network backends?
(I assume they have stuff like that implemented in hyper-v).

-- Pasi

Re: Re: [Xen-users] Xen viridian support/features

[Tim Deegan]

At 15:19 +0100 on 23 Jul (1248362393), Pasi K?rkk?inen wrote:
> Any plans to write support for paravirtual viridian disk/network backends?
> (I assume they have stuff like that implemented in hyper-v).

I don't think anyone's doing that, but there is a choice of Xen PV
drivers for Windows that will do the job.  Are the PV driver interfaces
even public?  I don't remember seeing them in the hyper-v spec, though I
haven't read the latest version.

Cheers,

Tim.

-- 
Tim Deegan <Tim.Deegan@citrix.com>
Principal Software Engineer, Citrix Systems (R&D) Ltd.
[Company #02300071, SL9 0DZ, UK.]

Re: Re: [Xen-users] Xen viridian support/features

[Jeremy Fitzhardinge]

On 07/23/09 07:27, Tim Deegan wrote:
> At 15:19 +0100 on 23 Jul (1248362393), Pasi K?rkk?inen wrote:
>   
>> Any plans to write support for paravirtual viridian disk/network backends?
>> (I assume they have stuff like that implemented in hyper-v).
>>     
>
> I don't think anyone's doing that, but there is a choice of Xen PV
> drivers for Windows that will do the job.  Are the PV driver interfaces
> even public?  I don't remember seeing them in the hyper-v spec, though I
> haven't read the latest version.
>   

They were implicitly published, at least to some extent, in the form of
the big HyperV PV driver dump for Linux.  But I don't think those
drivers are very featureful yet, though they claim to be planning to
release more updates.

    J

Re: [Xen-devel] Re: Xen viridian support/features

[Florian Manschwetus]

[-- Attachment #1.1: Type: text/plain, Size: 1701 bytes --]

Uhm, just to get clear in this topic,
the great work of James is not that useful as it supposed in this 
thread. It leaks reliable support for exactly those Windows Versions 
with best Hyper-V integrations.
So there is the choice to exploit Hyper-V guest integration services or 
absolutely ugly hacking around in a win2008 x64 guest policies with a 
real great chance to get the guest perfectly and unrecoverable meshed up.
So it would be great at least to have things like shutdown notification 
out of the box, as it will be with an viridian interface implementation.

Again if xen provides a feature using viridian API win2008 x64 guests 
would use it without ANY modification.

Florian

Am 23.07.2009 23:28, schrieb Jeremy Fitzhardinge:
> On 07/23/09 07:27, Tim Deegan wrote:
>> At 15:19 +0100 on 23 Jul (1248362393), Pasi K?rkk?inen wrote:
>>
>>> Any plans to write support for paravirtual viridian disk/network backends?
>>> (I assume they have stuff like that implemented in hyper-v).
>>>
>> I don't think anyone's doing that, but there is a choice of Xen PV
>> drivers for Windows that will do the job.  Are the PV driver interfaces
>> even public?  I don't remember seeing them in the hyper-v spec, though I
>> haven't read the latest version.
>>
>
> They were implicitly published, at least to some extent, in the form of
> the big HyperV PV driver dump for Linux.  But I don't think those
> drivers are very featureful yet, though they claim to be planning to
> release more updates.
>
>      J
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>



[-- Attachment #1.2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 3686 bytes --]

[-- Attachment #2: Type: text/plain, Size: 137 bytes --]

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

Re: Re: [Xen-users] Xen viridian support/features

[Andrew Lyon]

On Fri, Jul 24, 2009 at 3:40 PM, Florian
Manschwetus<florianmanschwetus@gmx.de> wrote:
> Uhm, just to get clear in this topic,
> the great work of James is not that useful as it supposed in this thread. It
> leaks reliable support for exactly those Windows Versions with best Hyper-V
> integrations.
> So there is the choice to exploit Hyper-V guest integration services or
> absolutely ugly hacking around in a win2008 x64 guest policies with a real
> great chance to get the guest perfectly and unrecoverable meshed up.
> So it would be great at least to have things like shutdown notification out
> of the box, as it will be with an viridian interface implementation

When you say "shutdown notification" I assume you mean dom0 telling a
hvm guest to shutdown , in which case "xm trigger <domid> power" seems
to do the job with no additional drivers or services required, and
should work on most operating systems.

Perhaps I misunderstood..

Andy
.
>
> Again if xen provides a feature using viridian API win2008 x64 guests would
> use it without ANY modification.
>
> Florian
>
> Am 23.07.2009 23:28, schrieb Jeremy Fitzhardinge:
>>
>> On 07/23/09 07:27, Tim Deegan wrote:
>>>
>>> At 15:19 +0100 on 23 Jul (1248362393), Pasi K?rkk?inen wrote:
>>>
>>>> Any plans to write support for paravirtual viridian disk/network
>>>> backends?
>>>> (I assume they have stuff like that implemented in hyper-v).
>>>>
>>> I don't think anyone's doing that, but there is a choice of Xen PV
>>> drivers for Windows that will do the job.  Are the PV driver interfaces
>>> even public?  I don't remember seeing them in the hyper-v spec, though I
>>> haven't read the latest version.
>>>
>>
>> They were implicitly published, at least to some extent, in the form of
>> the big HyperV PV driver dump for Linux.  But I don't think those
>> drivers are very featureful yet, though they claim to be planning to
>> release more updates.
>>
>>     J
>>
>> _______________________________________________
>> 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
>

Re: Re: [Xen-users] Xen viridian support/features

[Florian Manschwetus]

[-- Attachment #1.1: Type: text/plain, Size: 2696 bytes --]

Am 24.07.2009 21:31, schrieb Andrew Lyon:
> On Fri, Jul 24, 2009 at 3:40 PM, Florian
> Manschwetus<florianmanschwetus@gmx.de>  wrote:
>> Uhm, just to get clear in this topic,
>> the great work of James is not that useful as it supposed in this thread. It
>> leaks reliable support for exactly those Windows Versions with best Hyper-V
>> integrations.
>> So there is the choice to exploit Hyper-V guest integration services or
>> absolutely ugly hacking around in a win2008 x64 guest policies with a real
>> great chance to get the guest perfectly and unrecoverable meshed up.
>> So it would be great at least to have things like shutdown notification out
>> of the box, as it will be with an viridian interface implementation
>
> When you say "shutdown notification" I assume you mean dom0 telling a
> hvm guest to shutdown , in which case "xm trigger<domid>  power" seems
> to do the job with no additional drivers or services required, and
> should work on most operating systems.
>
> Perhaps I misunderstood..
Hm no it is fine but when I shutdown dom0 the windows system just get 
killed (xm shutdown is issued afaik). Furthermore this won't include a 
back channel (does it work?) and not every system surely shutsdown when 
powerbutton is pressed, maybe there must be a bit configured...

Florian
>
> Andy
> .
>> Again if xen provides a feature using viridian API win2008 x64 guests would
>> use it without ANY modification.
>>
>> Florian
>>
>> Am 23.07.2009 23:28, schrieb Jeremy Fitzhardinge:
>>> On 07/23/09 07:27, Tim Deegan wrote:
>>>> At 15:19 +0100 on 23 Jul (1248362393), Pasi K?rkk?inen wrote:
>>>>
>>>>> Any plans to write support for paravirtual viridian disk/network
>>>>> backends?
>>>>> (I assume they have stuff like that implemented in hyper-v).
>>>>>
>>>> I don't think anyone's doing that, but there is a choice of Xen PV
>>>> drivers for Windows that will do the job.  Are the PV driver interfaces
>>>> even public?  I don't remember seeing them in the hyper-v spec, though I
>>>> haven't read the latest version.
>>>>
>>> They were implicitly published, at least to some extent, in the form of
>>> the big HyperV PV driver dump for Linux.  But I don't think those
>>> drivers are very featureful yet, though they claim to be planning to
>>> release more updates.
>>>
>>>      J
>>>
>>> _______________________________________________
>>> 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
>>
>



[-- Attachment #1.2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 3686 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: Re: [Xen-users] Xen viridian support/features

[Andrew Lyon]

On Fri, Jul 24, 2009 at 8:56 PM, Florian
Manschwetus<florianmanschwetus@gmx.de> wrote:
> Am 24.07.2009 21:31, schrieb Andrew Lyon:
>>
>> On Fri, Jul 24, 2009 at 3:40 PM, Florian
>> Manschwetus<florianmanschwetus@gmx.de>  wrote:
>>>
>>> Uhm, just to get clear in this topic,
>>> the great work of James is not that useful as it supposed in this thread.
>>> It
>>> leaks reliable support for exactly those Windows Versions with best
>>> Hyper-V
>>> integrations.
>>> So there is the choice to exploit Hyper-V guest integration services or
>>> absolutely ugly hacking around in a win2008 x64 guest policies with a
>>> real
>>> great chance to get the guest perfectly and unrecoverable meshed up.
>>> So it would be great at least to have things like shutdown notification
>>> out
>>> of the box, as it will be with an viridian interface implementation
>>
>> When you say "shutdown notification" I assume you mean dom0 telling a
>> hvm guest to shutdown , in which case "xm trigger<domid>  power" seems
>> to do the job with no additional drivers or services required, and
>> should work on most operating systems.
>>
>> Perhaps I misunderstood..
>
> Hm no it is fine but when I shutdown dom0 the windows system just get killed
> (xm shutdown is issued afaik). Furthermore this won't include a back channel

Yes thats right when the system shuts down xm shutdown is called, I
don't know what happens if a guest fails to shutdown, myself I'd
probably want it to be xm destroyed after a timeout as thats probably
what I'd end up doing anyway, I could easily modify the script for you
so that after trying xm shutdown it sends xm trigger <domid> power to
any remaining domains and if any still fail to shutdown I could make
it destroy them.

Its something I looked into and could do in few minutes, I just didn't
bother because all of our servers have remote kvm and 99% of reboots
are supervised, but let me know if you'd like me to make a patch for
you.

Andy


> (does it work?) and not every system surely shutsdown when powerbutton is
> pressed, maybe there must be a bit configured...
>
> Florian
>>
>> Andy
>> .
>>>
>>> Again if xen provides a feature using viridian API win2008 x64 guests
>>> would
>>> use it without ANY modification.
>>>
>>> Florian
>>>
>>> Am 23.07.2009 23:28, schrieb Jeremy Fitzhardinge:
>>>>
>>>> On 07/23/09 07:27, Tim Deegan wrote:
>>>>>
>>>>> At 15:19 +0100 on 23 Jul (1248362393), Pasi K?rkk?inen wrote:
>>>>>
>>>>>> Any plans to write support for paravirtual viridian disk/network
>>>>>> backends?
>>>>>> (I assume they have stuff like that implemented in hyper-v).
>>>>>>
>>>>> I don't think anyone's doing that, but there is a choice of Xen PV
>>>>> drivers for Windows that will do the job.  Are the PV driver interfaces
>>>>> even public?  I don't remember seeing them in the hyper-v spec, though
>>>>> I
>>>>> haven't read the latest version.
>>>>>
>>>> They were implicitly published, at least to some extent, in the form of
>>>> the big HyperV PV driver dump for Linux.  But I don't think those
>>>> drivers are very featureful yet, though they claim to be planning to
>>>> release more updates.
>>>>
>>>>     J
>>>>
>>>> _______________________________________________
>>>> 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
>>>
>>
>
>
>

Re: Re: [Xen-users] Xen viridian support/features

[Florian Manschwetus]

[-- Attachment #1.1: Type: text/plain, Size: 3773 bytes --]

Am 24.07.2009 23:29, schrieb Andrew Lyon:
> On Fri, Jul 24, 2009 at 8:56 PM, Florian
> Manschwetus<florianmanschwetus@gmx.de>  wrote:
>> Am 24.07.2009 21:31, schrieb Andrew Lyon:
>>> On Fri, Jul 24, 2009 at 3:40 PM, Florian
>>> Manschwetus<florianmanschwetus@gmx.de>    wrote:
>>>> Uhm, just to get clear in this topic,
>>>> the great work of James is not that useful as it supposed in this thread.
>>>> It
>>>> leaks reliable support for exactly those Windows Versions with best
>>>> Hyper-V
>>>> integrations.
>>>> So there is the choice to exploit Hyper-V guest integration services or
>>>> absolutely ugly hacking around in a win2008 x64 guest policies with a
>>>> real
>>>> great chance to get the guest perfectly and unrecoverable meshed up.
>>>> So it would be great at least to have things like shutdown notification
>>>> out
>>>> of the box, as it will be with an viridian interface implementation
>>> When you say "shutdown notification" I assume you mean dom0 telling a
>>> hvm guest to shutdown , in which case "xm trigger<domid>    power" seems
>>> to do the job with no additional drivers or services required, and
>>> should work on most operating systems.
>>>
>>> Perhaps I misunderstood..
>> Hm no it is fine but when I shutdown dom0 the windows system just get killed
>> (xm shutdown is issued afaik). Furthermore this won't include a back channel
>
> Yes thats right when the system shuts down xm shutdown is called, I
> don't know what happens if a guest fails to shutdown, myself I'd
> probably want it to be xm destroyed after a timeout as thats probably
> what I'd end up doing anyway, I could easily modify the script for you
> so that after trying xm shutdown it sends xm trigger<domid>  power to
> any remaining domains and if any still fail to shutdown I could make
> it destroy them.
>
> Its something I looked into and could do in few minutes, I just didn't
> bother because all of our servers have remote kvm and 99% of reboots
> are supervised, but let me know if you'd like me to make a patch for
> you.
>
> Andy
Would be great until a more sophisticated way is realized.

Florian

>
>
>> (does it work?) and not every system surely shutsdown when powerbutton is
>> pressed, maybe there must be a bit configured...
>>
>> Florian
>>> Andy
>>> .
>>>> Again if xen provides a feature using viridian API win2008 x64 guests
>>>> would
>>>> use it without ANY modification.
>>>>
>>>> Florian
>>>>
>>>> Am 23.07.2009 23:28, schrieb Jeremy Fitzhardinge:
>>>>> On 07/23/09 07:27, Tim Deegan wrote:
>>>>>> At 15:19 +0100 on 23 Jul (1248362393), Pasi K?rkk?inen wrote:
>>>>>>
>>>>>>> Any plans to write support for paravirtual viridian disk/network
>>>>>>> backends?
>>>>>>> (I assume they have stuff like that implemented in hyper-v).
>>>>>>>
>>>>>> I don't think anyone's doing that, but there is a choice of Xen PV
>>>>>> drivers for Windows that will do the job.  Are the PV driver interfaces
>>>>>> even public?  I don't remember seeing them in the hyper-v spec, though
>>>>>> I
>>>>>> haven't read the latest version.
>>>>>>
>>>>> They were implicitly published, at least to some extent, in the form of
>>>>> the big HyperV PV driver dump for Linux.  But I don't think those
>>>>> drivers are very featureful yet, though they claim to be planning to
>>>>> release more updates.
>>>>>
>>>>>      J
>>>>>
>>>>> _______________________________________________
>>>>> 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
>>>>
>>
>>
>



[-- Attachment #1.2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 3686 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: Re: [Xen-users] Xen viridian support/features

[Florian Manschwetus]

[-- Attachment #1.1: Type: text/plain, Size: 1264 bytes --]

Am 23.07.2009 16:19, schrieb Pasi Kärkkäinen:
> On Thu, Jul 23, 2009 at 03:11:01PM +0100, Tim Deegan wrote:
>> At 18:15 +0100 on 21 Jul (1248200133), Pasi K?rkk?inen wrote:
>>> Then again I'm not sure "how much" of viridian stuff Xen currently
>>> supports..
>> At the moment we provide MSR access to APIC registers, and the hypercall
>> to yield the CPU when waiting for a spinlock.  We looked at a few others
>> (e.g. address-space-change hypercall) but found the benefits, if any,
>> were negligible.
>>
>> The most important part of the interface is that we can use it to tell
>> Windows not to expect timer interrupt delivery to be as even across all
>> CPUs as it would be on bare metal.  This gets rid of annoying STOP 0x101
>> bluescreens in multi-vcpu Windows from Vista on.
>>
>
> Thanks for the reply!
>
> Any plans to write support for paravirtual viridian disk/network backends?
> (I assume they have stuff like that implemented in hyper-v).
>
> -- Pasi
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>
Take a shorter target, reboot notification using hyper-v guest 
integration stuff would be a great win.

Florian


[-- Attachment #1.2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 3686 bytes --]

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

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