Re: [RFC 13/20] iommu: Extend iommu_at[de]tach_device() for multiple devices group

[David Gibson <david@gibson.dropbear.id.au>]

[-- Attachment #1: Type: text/plain, Size: 1311 bytes --]

On Mon, Oct 25, 2021 at 08:36:02PM -0300, Jason Gunthorpe wrote:
> On Tue, Oct 26, 2021 at 12:16:43AM +1100, David Gibson wrote:
> > If you attach devices A and B (both in group X) to IOAS 1, then detach
> > device A, what happens?  Do you detach both devices?  Or do you have a
> > counter so you have to detach as many time as you attached?
> 
> I would refcount it since that is the only thing that makes semantic
> sense with the device centric model.

Yes, I definitely think that's the better option here.  This does
still leave (at least) one weird edge case where the group structure
can "leak" into the awareness of code that otherwise doesn't care,
though it's definitely less nasty that the ones I mentioned before:

If an app wants to move a bunch of devices from one IOAS to another,
it can do it either:

A)
	for each dev:
		detach dev from IOAS
	for each dev:
		attach dev to new IOAS

or B)

	for each dev:
		detach dev from IOAS
		attach dev to new IOAS

With only singleton groups they're pretty much equivalent, but with
multiple devices in a group, (B) will fail.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]