Re: [PATCH net-next v3 6/6] netlink: specs: rt-route: describe seg6-local attrs

[Hangbin Liu <hangbin.liu@linux.dev>]

On Thu, Oct 01, 2026 at 01:51:33AM +0000, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
> - [High] netlink: incorrect type: flags causes uAPI mismatch for seg6-local-flv-ops
> - [Low] netlink: yaml naming convention violation in seg6-local-flv-ops name-prefix
> - [Low] netlink: missing doc properties on new major yaml spec additions
> --
> 
> commit 1d8d137d71e90b2ea51b5a3ae23d03831ee06b0b
> Author: Hangbin Liu <liuhangbin@kylinos.cn>
> 
> netlink: specs: rt-route: describe seg6-local attrs
> 
> This commit adds SEG6 local actions enums and seg6-local-flv-ops flags to the
> rt-route Netlink specification. It also replaces binary attributes with nested
> seg6-local-bpf, seg6-local-cnt, and seg6-local-flv definitions.
> 
> > diff --git a/Documentation/netlink/specs/rt-route.yaml b/Documentation/netlink/specs/rt-route.yaml
> > index afec0375661f5..0260fca398d2c 100644
> > --- a/Documentation/netlink/specs/rt-route.yaml
> > +++ b/Documentation/netlink/specs/rt-route.yaml
> > @@ -116,6 +116,45 @@ definitions:
> >        - rpl
> >        - ioam6
> >        - xfrm
> > +  -
> > +    name: seg6-local-actions
> > +    type: enum
> 
> [Severity: Low]
> Does this new major property need a doc field? The Netlink subsystem
> guidelines recommend that each major property should carry a doc entry to
> explain its purpose.
> 
> [ ... ]
> 
> > +  -
> > +    name: seg6-local-flv-ops
> > +    type: flags
> > +    name-prefix: seg6_local_flv_op-
> 
> [Severity: High]
> Can using type: flags here cause a uAPI mismatch with the kernel?

Right, I forgot the change the type and name prefix.

> 
> The kernel uAPI defines these operations as sequential values (for example,
> SEG6_LOCAL_FLV_OP_USD is 3). When type: flags is used, the YNL code
> generator will emit bitmasks (1, 2, 4, 8), resulting in
> SEG6_LOCAL_FLV_OP_USD being incorrectly defined as 8 in the generated
> headers.
> 
> Would it be safer to use type: enum to ensure the generated values match
> the existing kernel constants?
> 
> [Severity: Low]
> Also, could the name-prefix use dashes instead of underscores? The naming
> convention for YAML specs specifies that names should contain dashes, as
> the code generator handles converting them to underscores for the C code.

Yes, I will fix it.

Thanks
Hangbin
> 
> [Severity: Low]
> Additionally, should seg6-local-flv-ops include a doc property?