Re: [Qemu-devel] Re: [PATCH 3/3] Add KVM support to QEMU

[Avi Kivity <avi@redhat.com>]

Anthony Liguori wrote:
> Avi Kivity wrote:
>> Anthony Liguori wrote:
>>> Another place "hook" is updating a slot's dirty bitmap.  Right now, 
>>> with my patchset we don't have live migration or the VGA RAM 
>>> optimization.  There's nothing about the VGA RAM optimization that 
>>> wouldn't work for QEMU.  I'm not sure that it really is an 
>>> optimization in the context of TCG, but I certainly don't think it's 
>>> any worse.  The only thing you really need is to query the KVM dirty 
>>> bitmap when it comes time to enable start over querying the VGA 
>>> dirty bits.
>>
>> I don't understand this.  The VGA optimization really is qemu's, the 
>> kvm modifications only cater to the different way of getting the 
>> dirty bits.
>
> Right.  I'm just not sure that it's going to be as much of an 
> optimization for TCG as it is for KVM.

When qemu is loaded certainly this is lost in the noise.  But when 
idling (and as a bonus, the screen doesn't change much), tcg and kvm 
will benefit equally.

>
> Right now, in QEMU, code looks like this:
>
> for (i = 0; i < addr; i += TARGET_PAGE_SIZE) {
>   if (cpu_p_m_g_dirty(i, DIRTY_FLAG))  {
>       cpu_p_m_r_dirty(i, i + TARGET_PAGE_SIZE, DIRTY_FLAG);
>
>   // do something with dirty memory
> }
>
> All we need to do is add another cpu_physical_memory_sync_dirty(i, i + 
> REGION_SIZE, DIRTY_FLAG); that would go at the start of this.  For 
> QEMU, this is a nop since dirty bits are updated as soon as they are 
> reset.  For KVM, this would update the entire set of dirty bits for 
> the given memory region.

You mean merge the kvm bitmap into the qemu bitmap?  That's what it does 
now, no?

> We also need something to enable dirty tracking for a particular 
> region.  We already have something for migration, we could perhaps 
> extend that API (cpu_p_m_s_dirty_tracking).

I don't think qemu would benefit much from per-region tracking, but 
maybe I'm wrong.

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