Re: vlan and bpf

[Greg Troxel <gdt%lexort.com@localhost>]

Roy Marples <roy%marples.name@localhost> writes:

>  ---- On Fri, 15 Dec 2023 18:47:54 +0000  Mouse  wrote --- 
>  > Again, I would expect to find it in the packet, bytes [14] and [15]
>  > containing the tag and priority, but, depending on the use case, "need"
>  > may or may not be appropriate.
>  > 
>  > If you're stuck with hardware that insists on stripping VLAN tags, IMO
>  > the only sane thing is to synthesize the stripped 4 octets for
>  > presentation via bpf.
>
> Sane or not, you don't see VLAN tags in Linux BPF.
> They move ancillary data such as that to a negative offset (which is invalid in BPF syntax)
> https://github.com/torvalds/linux/blob/7acac4b3196caee5e21fb5ea53f8bc124e6a16fc/include/uapi/linux/filter.h#L60
>
> They have more there than just VLAN, so if we feel the urge to do anything more, it might be something to consider.
> I'm not advocating for one or the other, just noting it as a datapoint.

Hmm; that doesn't quite make sense to me.   What yamt@ did in netbsd-10
looks correct.