Re: [RFC PATCH] KVM: PPC: Book3S HV: add support for page faults in VM_IO|VM_PFNMAP vmas

[Paolo Bonzini <pbonzini@redhat.com>]

On 12/02/2018 15:38, Benjamin Herrenschmidt wrote:
> On Mon, 2018-02-12 at 12:01 +0100, Paolo Bonzini wrote:
>> On 12/02/2018 11:54, Benjamin Herrenschmidt wrote:
>>> On Mon, 2018-02-12 at 11:41 +0100, Paolo Bonzini wrote:
>>>> In fact I don't know about PPC but on x86 we actually do
>>>> write-protection except on the newest processors that have a dirty page
>>>> _log_ in addition to the bits.
>>>
>>> Can you explain ? I'm not sure I got what you mean here.
>>
>> On recent processors you can get an exit to the hypervisor every time N
>> dirty pages are marked dirty, and the processor also keeps a log of
>> _which_ pages are marked dirty.  Without the log, you'd have to walk the
>> entire radix tree to find the dirty pages, which is so expensive that
>> without the log we don't use dirty bits at all.
> 
> Oh interesting. How is that log "kept" ? Some kind of HW fifo ?

Yeah, forced vmexit after 512 dirty pages, plus a pointer to the last
written entry (kept in the hypervisor->processor communication block).
Pretty clever.

Paolo