Il mer 21 mag 2025, 10:37 Paolo Bonzini <bonz...@gnu.org> ha scritto:
> > So... not sure what to do there. It seems like vm-memory is very close
> to
>
>> > being usable by QEMU, but maybe not completely. :(
>>
>> Is it possible or necessary for vm-memory to support overlap? Because I
>> feel that if it is possible, the problem might be simplified. (As a
>> beginner, I have yet to understand exactly how difficult it is.)
>>
>
> I don't think that's necessary. Just like in QEMU C code we have
> AddressSpace for DMA and MemoryRegion for hierarchy, in Rust code you have
> qemu_api::{AddressSpace,MemoryRegion}. FlatView, FlatRange,
> MemoryRegionSection are hidden in both cases, and users don't care much
> about which type implements GuestMemoryRegion because all they see is
> AddressSpace. Again, it's all hidden behind the prelude.
>
> The real problem is how hard it is to remove the references from the
> vm-memory API... Maybe not much.
>
Brain dump ahead! AddressSpaceDispatch is already storing
MemoryRegionSections. Therefore it should be possible to make
GuestMemory::R equal to MemoryRegionSection, or rather its Opaque wrapper.
Then one needs to implement Bytes<MemoryRegionAddress> in
MemoryRegionSection like
https://github.com/rust-vmm/vm-memory/blob/3f2fd80b11/src/mmap/mod.rs#L171,
use address_space_lookup_region() to implement find_region(), and (except
for IOMMU) everything should work.
To implement IOMMUs later, it's probably possible to call
flatview_translate() from the try_access() method, which is intended to be
internal but comes in handy here. try_access() is a bit complicated, but
we'll just have to copy some code from the default implementation at
https://github.com/rust-vmm/vm-memory/blob/main/src/guest_memory.rs, in
order to replace the call to find_region(). Notice how
address_space_lookup_region() returns MemoryRegionSection*, which
translates to Option<&Self::R>; while flatview_translate() returns a
MemoryRegionSection by value, that can become a local variable in
try_access() and everything is fine. Hopefully. The compiler will certainly
prove me wrong.
Another thing that will be needed later is support for MemTxAttrs. Maybe
squeeze those into the address, i.e. implement Bytes<(GuestAddress,
MemTxAttrs)> and Bytes<(MemoryRegionAddress, MemTxAttrs)>?
The dirty bitmap does need some changes in vm-memory, but it's pretty small
compared to the mess that I foreshadowed in order to change find_region().
Paolo
