Re: [RFC net-next 1/6] net: add kfree_skb_for_sk function

[Eric Dumazet]

On Fri, May 31, 2024 at 9:05 PM Yan Zhai <yan@xxxxxxxxxxxxxx> wrote:
>

> I don't quite follow it. I think this specific commit using skb->cb is
> unnecessary so I am going to re-work it. As you initially mentioned,
> maybe I should just extend kfree_skb tracepoint. I saw a similar
> change dd1b527831a3("net: add location to trace_consume_skb()"), is it
> something I might follow, or do you specifically mean changes like
> this can annoy stable teams?
>

I do not think trace points arguments are put in stone.

If they were, I would nack the addition of new tracepoints, to prevent
ossification.