On Tue, 3 Feb 2026 11:53:28 +0100 Larysa Zaremba wrote: > Many ethernet drivers report xdp Rx queue frag size as being the same as > DMA write size. However, the only user of this field, namely > bpf_xdp_frags_increase_tail(), clearly expects a truesize.
truesize may not be the most fortunate term. truesize is how much memory we charge to the skb (for memory management) which may relate to frag size in strange ways for strange HW :S I don't have a great term, but I have a feeling truesize already confuses driver developers. Note two our xdp.py tests covers tail adjust, I wonder if it would have caught this for you. Looking at Intel's results the XDP tests don't run due to missing BTF. Could you work with Adrian to address that?
