Re: [PATCH v1 01/14] iommu: Add iommu_get_unmanaged_domain helper

[Jason Gunthorpe <jgg@nvidia.com>]

On Wed, Mar 22, 2023 at 06:41:42PM +0100, Eric Auger wrote:
> > I would say this spec language is overly broad. If the FW knows the
> > reserved memory regions it creates are not sensitive to PCI layout
> > then it should not be forced to set this flag.
> 
> But do we have any guarantee the bus numbers can't change. I thought the
> guest was allowed to re-number at will? While further thinking at it,
> all RID ID mappings should be affected by this concern, I mean not only
> RID 2 RMRs? What do I miss?

Bus number changing is allowed, but qemu should not be sensitive to
this.

qemu always knows the current guest assigned bus number for the vPCI,
since it traps the bus number changes like anything else.

Thus when a STE is configured qemu has access to accurate data to
convert the vSID to the vPCI and vfio_device. Even if the bus numbers
change since boot.

> > We are forced to use RMR because of the hacky GIC ITS stuff.
> well we are not obliged to use RMRs. My first revisions did not use it
> and created a non direct S1 mapping. This is just a commodity that
> simplifies the integration and was nicely suggested by jean.

I undertand it is ARM's architectural preference..

Personally I would prefer the vGIC model include the ITS page itself
and that the guest put the ITS page into the S1 mapping in the usual
way. But we are a long way away from that..

Jason