On 09/08/2017 15:33, David Hildenbrand wrote:
> If I am not missing something important here, we can heavily simplify
> the kvm_slot code. Flatview will make sure that we don't have to deal
> with overlapping slots. E.g. when a memory section is resized, we are
> first notified about the removal and then about the new memory section.
>
> So basically, we can directly always map one memory section to one
> kvm slot (if the fixed up size is > 0).
>
> Only very briefly tested. Will do some more testing if we agree that this
> is the right thing to do.
Yes, it all looks very sane.
Paolo
> David Hildenbrand (6):
> kvm: require JOIN_MEMORY_REGIONS_WORKS
> kvm: factor out alignment of memory section
> kvm: use start + size for memory ranges
> kvm: we never have overlapping slots in kvm_set_phys_mem()
> kvm: kvm_log_start/stop are only called with known sections
> kvm: kvm_log_sync() is only called with known memory sections
>
> accel/kvm/kvm-all.c | 276 +++++++++++++++++-----------------------------------
> 1 file changed, 89 insertions(+), 187 deletions(-)
>