Re: [PATCH bpf-next v2 00/16] Decouple skb metadata tracking from MAC header offset

On Mon, Jan 05, 2026 at 01:14 PM +01, Jakub Sitnicki wrote:

This series continues the effort to provide reliable access to xdp/skb
metadata from BPF context on the receive path.

FYI, I'm dropping the effort to make xdp/skb metadata survive L2
encap/decap, following the discussion here and in [1].

Instead, I'll work on an skb local storage backed by an skb extension,
similar to what we have for cgroups/sockets/etc.

This approach avoids patching skb_push/pull call sites and implicit
allocations on hot paths.
 
I'm happy to split out BPF_EMIT_CALL support for prologue/epilogue if
it's considered useful on its own.

[1] https://lore.kernel.org/r/20260110-skb-meta-fixup-skb_metadata_set-calls-v1-0-1047878ed1b0@cloudflare.com (local)

Thanks,
-jkbs