Lorenzo Bianconi wrote:
> > Lorenzo Bianconi wrote:
> > > Update multi-buffer bit (mb) in xdp_buff to notify XDP/eBPF layer and
> > > XDP remote drivers if this is a "non-linear" XDP buffer. Access
> > > skb_shared_info only if xdp_buff mb is set in order to avoid possible
> > > cache-misses.
> > >
> > > Signed-off-by: Lorenzo Bianconi <lorenzo@xxxxxxxxxx>
> >
> > [...]
> >
> > > @@ -2320,8 +2325,12 @@ mvneta_swbm_build_skb(struct mvneta_port *pp, struct page_pool *pool,
> > > struct xdp_buff *xdp, u32 desc_status)
> > > {
> > > struct skb_shared_info *sinfo = xdp_get_shared_info_from_buff(xdp);
> > > - int i, num_frags = sinfo->nr_frags;
> > > struct sk_buff *skb;
> > > + u8 num_frags;
> > > + int i;
> > > +
> > > + if (unlikely(xdp_buff_is_mb(xdp)))
> > > + num_frags = sinfo->nr_frags;
> >
> > Doesn't really need a respin IMO, but rather an observation. Its not
> > obvious to me the unlikely/likely pair here is wanted. Seems it could
> > be relatively common for some applications sending jumbo frames.
> >
> > Maybe worth some experimenting in the future.
>
> Probably for mvneta it will not make any difference but in general I tried to
> avoid possible cache-misses here (accessing sinfo pointers). I will carry out
> some comparison to see if I can simplify the code.
Agree, I'll predict for mvneta it doesn't make a difference either way and
perhaps if you want to optimize small pkt benchmarks on a 100Gbps nic it would
show a win.
