Re: [Qemu-devel] [PATCH 5/7] memory: Allow replay of IOMMU mapping notifications

[Paolo Bonzini <pbonzini@redhat.com>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256



On 25/09/2015 07:24, David Gibson wrote:
>> When memory_listener_register() replays mappings, it does so on 
>> an rcu copy of the flatview for each AddressSpace.  Here we
>> don't seem to have anything protecting against concurrency... do
>> we need to worry about that?
> 
> I was assuming that the IOMMU mappings are protected by the BQL. I
> _think_ that's the case (for every IOMMU we have so far), but I'm 
> not really sure how to be sure.

Yes, even in listener_add_address_space there's no real need to use
address_space_get_flatview because updates to both the memory maps and
the MemoryListener list are protected by the BQL.  It could just read
as->current_map directly.

listener_add_address_space plays it a bit safe because QEMU doesn't
(yet?) have stuff like rcu_dereference_check.  It's sad that we'll
have to reinvent so much debugging stuff from Linux...

Paolo
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBCAAGBQJWBS9AAAoJEL/70l94x66D5l0H/AxF9flRUf6PaNO6RLpd0N8y
z17BXqKuGIIevkGtF66xCRNDTRgfHKlugKkSQjaQIlGlh2k/nKfozcFERiE6/unv
QmUS3+c3ryLMhOC4VidU90Krq1ZsXLLNE1Z81aSQwD4Y0LXEp3hFuL1F/K6tDUtY
Vzk2KJu/bEaeDhJa4UglBOBLGWIYNRvokIFc2TdoMjdBvKfygLHfjL3rroQOle4U
40/mBO+3J58cyfCBYha0U9DxVTvceaFubj3M72W3ajenXZ15lq75B0M9GCNNujHE
oaP8ehQ5wByyiGGmcZkStEXU/ups7J6cMWWjpTak3PqO0O5DVo+j5P4QbvJE57Q=
=gQRk
-----END PGP SIGNATURE-----