Re: [PATCH v6 1/7] xen/mm: Refactor claim deduction for later functional changes

[Jan Beulich <jbeulich@xxxxxxxx>]

On 22.04.2026 16:35, Bernhard Kaindl wrote:
> 
> Jan Beulich wrote:
> 
>> As to the rename to deduct_global_claims(): With that, wouldn't its 2nd
>> parameter then better also change to "deduction"? Furthermore, "global" in
>> the name is ambiguous: It may mean "not per-node", but it may also mean
>> "not per-domain". The v4 name didn't have such an issue.
> 
> I see the point about 'global'. While the struct domain argument provides
> context, the term is indeed too generic.
> 
> In this logic, 'global' was intended to refer to the opposite of
> node-specific claims. While the legacy codebase uses 'no-node' or
> NUMA_NO_NODE as a sentinel, I find those terms ambiguous for this new
> logic; they often imply a "missing" assignment rather than an explicit
> "any-node" policy.
> 
> 'Node-agnostic' better describes the architectural intent: that these
> claims are valid regardless of node locality. To resolve the ambiguity,
> I will replace 'global' with 'node-agnostic' across the board (design
> documentation, code, and tests).
> 
> Specifically, I will:
> 
>     Rename the function to deduct_node_agnostic_claims().

Or deduct_host_claims() (short for deduct_host_wide_claims())?

Jan

>     Update the second parameter to 'deduction' for consistency.
> 
>     Update the function comment to explicitly define the scope.
> 
> /* Deduct from the domain's node-agnostic (non-per-node) claims pool */
> static unsigned long deduct_node_agnostic_claims(struct domain *d,
>                                                  unsigned long deduction)
> 
> While 'global' was shorter, 'node-agnostic' is much more descriptive of
> the actual logic and the node-locality (or lack thereof) being addressed,
> avoiding the potential confusion associated with 'no-node'.
> 
> For consistency, the corresponding renames across the series will be:
> 
> +--------------------------------+---------------------------------------+
> | Original Term                  | New Term                              |
> +--------------------------------+---------------------------------------+
> | "global claims"                | "node-agnostic claims"                |
> | d->global_claims               | d->node_agnostic_claims               |
> | XEN_DOMCTL_CLAIM_MEMORY_GLOBAL | XEN_DOMCTL_CLAIM_MEMORY_NODE_AGNOSTIC |
> | deduct_global_claims()         | deduct_node_agnostic_claims()         |
> +--------------------------------+---------------------------------------+
> 
> I believe these changes result in less ambiguous language universally.
> Jan, could you Ack these renames for the next version of the series?
> 
> Bernhard
>