On 14.03.22 23:20, Daniel Borkmann wrote:
On 3/14/22 11:16 PM, Toke Høiland-Jørgensen wrote:
Felix Fietkau <nbd@xxxxxxxx> writes:
On 14.03.22 21:39, Jesper D. Brouer wrote:
(Cc. BPF list and other XDP maintainers)
On 14/03/2022 11.22, Felix Fietkau wrote:
Most ethernet drivers allocate a packet headroom of NET_SKB_PAD. Since it is
rounded up to L1 cache size, it ends up being at least 64 bytes on the most
common platforms.
On most ethernet drivers, having a guaranteed headroom of 256 bytes for XDP
adds an extra forced pskb_expand_head call when enabling SKB XDP, which can
be quite expensive.
Many XDP programs need only very little headroom, so it can be beneficial
to have a way to opt-out of the 256 bytes headroom requirement.
IMHO 64 bytes is too small.
We are using this area for struct xdp_frame and also for metadata
(XDP-hints). This will limit us from growing this structures for
the sake of generic-XDP.
I'm fine with reducting this to 192 bytes, as most Intel drivers
have this headroom, and have defacto established that this is
a valid XDP headroom, even for native-XDP.
We could go a small as two cachelines 128 bytes, as if xdp_frame
and metadata grows above a cache-line (64 bytes) each, then we have
done something wrong (performance wise).
Here's some background on why I chose 64 bytes: I'm currently
implementing a userspace + xdp program to act as generic fastpath to
speed network bridging.
Any reason this can't run in the TC ingress hook instead? Generic XDP is
a bit of an odd duck, and I'm not a huge fan of special-casing it this
way...
+1, would have been fine with generic reduction to just down to 192 bytes
(though not less than that), but 64 is a bit too little. Also curious on
why not tc ingress instead?
I chose XDP because of bpf_redirect_map, which doesn't seem to be
available to tc ingress classifier programs.
When I started writing the code, I didn't know that generic XDP
performance would be bad on pretty much any ethernet/WLAN driver that
wasn't updated to support it.
- Felix
