Re: [RFC 01/12] ip: Introduce multi-address data structures for IPv4 and IPv6

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

[-- Attachment #1: Type: text/plain, Size: 1651 bytes --]

On Mon, Dec 15, 2025 at 08:58:32PM -0500, Jon Maloy wrote:
> 
> 
> On 2025-12-15 17:05, Jon Maloy wrote:
> > 
> > 
> > On 2025-12-15 04:40, David Gibson wrote:
> [...]
> > > > +struct ip4_addr_entry {
> > > > +    struct in_addr addr;
> > > > +    int prefix_len;
> > > > +    int permanent;
> > > 
> > > Might as well make these uint8_t and bool, respectively.  There will
> > > be some padding, but the overall structure will still be smaller.
> > > 
> > > Or, it might be worth considering replacing 'permanent' with a flags
> > > mask, in case we have future uses for it.
> > 
> > Agree with that. I'll make that change, and if we ver need more we can
> > replace the boolean with a bitmask.

Yes, that's what I was implying.  I think we will want some additional
flags to indicate the origin of the address (command line, scraped
from host, observed from guest).

> Provided we agree on keeping guest side subscriptions, we could mark
> the address entries "guest_side" and "host_side", hence giving only
> the owning side the right to remove/alter the entry.

I don't think that quite makes sense as is.  These are all guest side
addresses.  The main distinction is between addresses we're *telling*
the guest to use (currently 'addr') and guests we've observed the
guest to use (currently 'addr_seen').  Usually, those should have a
100% overlap, but we need to be clear on what the policy is when they
don't.

-- 
David Gibson (he or they)	| 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 --]