Re: [PATCH net-next 4/6] netlink: specs: rt-route: describe lwtunnel IP options

On Fri, Sep 18, 2026 at 06:36:58AM +0000, [email protected] wrote:
> Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
> - [Medium] netlink: specs: missing byte-order property for lwtunnel-ip-opt-erspan index
> - [Low] netlink: specs: missing max-len validation for lwtunnel-ip-opt-geneve data

Thanks, I will fix these.

> --
> 
> commit 15d568f88486d1cb50af0dbb912eabab2c37d852
> Author: Hangbin Liu <[email protected]>
> 
> netlink: specs: rt-route: describe lwtunnel IP options
> 
> This commit replaces binary options in lwtunnel-ip and lwtunnel-ip6 with
> a nested lwtunnel-ip-opts set. It also adds attribute sets for geneve, vxlan,
> and erspan IP options to align with the kernel's definitions in lwtunnel.h.
> 
> > diff --git a/Documentation/netlink/specs/rt-route.yaml b/Documentation/netlink/specs/rt-route.yaml
> > index 0e236da140dc5..6030d7cb8d0d1 100644
> > --- a/Documentation/netlink/specs/rt-route.yaml
> > +++ b/Documentation/netlink/specs/rt-route.yaml
> [ ... ]
> > @@ -555,6 +557,59 @@ attribute-sets:
> >        -
> >          name: link
> >          type: u32
> > +  -
> > +    name: lwtunnel-ip-opts
> > +    name-prefix: lwtunnel-ip-opts-
> > +    attributes:
> > +      -
> > +        name: geneve
> > +        type: nest
> > +        nested-attributes: lwtunnel-ip-opt-geneve
> > +      -
> > +        name: vxlan
> > +        type: nest
> > +        nested-attributes: lwtunnel-ip-opt-vxlan
> > +      -
> > +        name: erspan
> > +        type: nest
> > +        nested-attributes: lwtunnel-ip-opt-erspan
> > +  -
> > +    name: lwtunnel-ip-opt-geneve
> > +    name-prefix: lwtunnel-ip-opt-geneve-
> > +    attributes:
> > +      -
> > +        name: class
> > +        type: u16
> > +        byte-order: big-endian
> > +      -
> > +        name: type
> > +        type: u8
> > +      -
> > +        name: data
> > +        type: binary
> 
> [Severity: Low]
> Does this binary attribute need a max-len validation check?
> 
> The kernel explicitly enforces a maximum length of 127 bytes for this
> attribute in its Netlink policy:
> 
> net/ipv4/ip_tunnel_core.c:geneve_opt_policy[] {
>     [LWTUNNEL_IP_OPT_GENEVE_DATA] = { .type = NLA_BINARY, .len = 127 },
> }
> 
> Without a corresponding max-len constraint in the spec, auto-generated
> tools will not catch oversized attributes locally before transmitting
> them to the kernel.
> 
> > +  -
> > +    name: lwtunnel-ip-opt-vxlan
> > +    name-prefix: lwtunnel-ip-opt-vxlan-
> > +    attributes:
> > +      -
> > +        name: gbp
> > +        type: u32
> > +  -
> > +    name: lwtunnel-ip-opt-erspan
> > +    name-prefix: lwtunnel-ip-opt-erspan-
> > +    attributes:
> > +      -
> > +        name: ver
> > +        type: u8
> > +      -
> > +        name: index
> > +        type: u32
> 
> [Severity: Medium]
> Will this result in an incorrect ERSPAN index on little-endian architectures?
> 
> The kernel parses this attribute using nla_get_be32():
> 
> net/ipv4/ip_tunnel_core.c:ip_tun_parse_opts_erspan() {
>     ...
>     attr = tb[LWTUNNEL_IP_OPT_ERSPAN_INDEX];
>     md->u.index = nla_get_be32(attr);
>     ...
> }
> 
> Since the spec defines index as a u32 without the byte-order: big-endian
> property, auto-generated tools will send the value in host byte order.
> When run on a little-endian system, the kernel would then read a
> byte-swapped, incorrect value, which would break the tunnel configuration.
> 
> > +      -
> > +        name: dir
> > +        type: u8
> > +      -
> > +        name: hwid
> > +        type: u8
> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=4