Re: [PATCH RFC 03/20] ctlv: Add the root family specification and generated files

[Jiri Pirko <jiri@resnulli.us>]

Tue, Oct 20, 2026 at 01:29:45AM +0200, jgg@nvidia.com wrote:
>> [ ... 238 lines skipped ... ]
>> --- /dev/null
>> +++ b/Documentation/ctlv/families.yaml
>> @@ -0,0 +1,13 @@
>> +# SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
>> +%YAML 1.2
>> +---
>> +# Append-only registry of CTLV family names, indexed by family ID: the entry at
>> +# index N has family ID N, up to 65535. The root family owns ID 0 and heads the
>> +# list, and every other entry comes after the entry of the family it inherits.
>> +# An entry may be marked deprecated, but never reordered, removed, renamed or
>> +# reused, its position being its wire identity. Every entry must have a
>> +# specification in specs/<name>.yaml, and is added by the change that adds the
>> +# implementation. Fixtures under tools/ctlv/tests/ are not registered.
>> +
>> +families:
>> +  - ctlv
>
>I see where you are coming from with this, but it is going to be
>problematic from a global kernel perspective. Given CTLV is imagined
>as a general tool for any subsystem we cannot control the tree that
>things merge through. It will not be like netdev where there is one
>authoritative tree with a linear history.
>
>So if Bob and Jane maintainers both add their own subsystems by the
>time it gets to Linus the IDs will get scrambled, and bisection will
>be lost. 
>
>Either changes to this file must be strictly done in a single tree
>with shared branches (which is going to be a bit hard process wise)
>
>Or we need to put the ID #s here and have some simple process to
>coordinate them during the development process. Since new families are
>probably rare it shouldn't be a big deal

Exactly. Collision is not very likely. Explicit ID or implicit, scramble
would still happen. And in that case it is resolvable.