Il 11/09/2013 00:02, Alex Williamson ha scritto:
>>> > > I would hope that when you do the below
>>> > > memory_region_register_iommu_notifier() the callback gets a replay of
>>> > > the current state of the iommu like we do for a memory region when we
>>> > > register the listener that gets us here. Is that not the case?
>> >
>> >
>> > From what I see, it just adds a notifier:
>> >
>> > void memory_region_register_iommu_notifier(MemoryRegion *mr, Notifier *n)
>> > {
>> > notifier_list_add(&mr->iommu_notify, n);
>> > }
> Ok, that's too bad. I guess your comment is justified then.
>
That can be fixed by adding a new element ("replay") to the
MemoryRegionIOMMUOps. It can be a follow-up.
Paolo
