[patch 0/4] KVM/ia64 SN2 RTC backed ITC emulation

[patch 0/4] KVM/ia64 SN2 RTC backed ITC emulation

[jes]

This set of patches switches KVM to use the SN2 RTC clock source as
backing for the ITC emulation, as the ITC isn't synchronized on SN2.

The RTC is a memory mapped register which needs to be mapped into the
address space of the VMM module, and also accessed from the kernel. I
could try and split the changes into seperate files, but I honestly
thing it would cause more clutter than doing it this way.

With this change I can boot 64 vcpus on ia64 at stable boot times,
with the ITC backed ITC emulation, the boot would regularly hang for
minutes because time would suddenly go backwards.

Let me know what you think - this patch set is against Avi's tree as
of earlier today.

Cheers,
Jes

RE: [patch 0/4] KVM/ia64 SN2 RTC backed ITC emulation

[Zhang, Xiantao]

Jes
    Thanks for your work!  Okay, the patchset is good to me! 

Ack-ed: Xiantao Zhang <xiantao.zhang@intel.com>
  
Avi, 
	Could you help to queue it for next merge window ?
Xiantao



Jes@sgi.com wrote:
> This set of patches switches KVM to use the SN2 RTC clock source as
> backing for the ITC emulation, as the ITC isn't synchronized on SN2.
> 
> The RTC is a memory mapped register which needs to be mapped into the
> address space of the VMM module, and also accessed from the kernel. I
> could try and split the changes into seperate files, but I honestly
> thing it would cause more clutter than doing it this way.
> 
> With this change I can boot 64 vcpus on ia64 at stable boot times,
> with the ITC backed ITC emulation, the boot would regularly hang for
> minutes because time would suddenly go backwards.
> 
> Let me know what you think - this patch set is against Avi's tree as
> of earlier today.
> 
> Cheers,
> Jes

Re: [patch 0/4] KVM/ia64 SN2 RTC backed ITC emulation

[Avi Kivity]

jes@sgi.com wrote:
> This set of patches switches KVM to use the SN2 RTC clock source as
> backing for the ITC emulation, as the ITC isn't synchronized on SN2.
>
> The RTC is a memory mapped register which needs to be mapped into the
> address space of the VMM module, and also accessed from the kernel. I
> could try and split the changes into seperate files, but I honestly
> thing it would cause more clutter than doing it this way.
>
> With this change I can boot 64 vcpus on ia64 at stable boot times,
> with the ITC backed ITC emulation, the boot would regularly hang for
> minutes because time would suddenly go backwards.
>
> Let me know what you think - this patch set is against Avi's tree as
> of earlier today.
>   

Applied, thanks.

-- 
error compiling committee.c: too many arguments to function