On Fri, 22 May 2020 11:44:04 -0600
David Ahern <dsahern@xxxxxxxxx> wrote:
> On 5/22/20 10:11 AM, Lorenzo Bianconi wrote:
> > @@ -259,28 +270,64 @@ static int cpu_map_kthread_run(void *data)
> > * kthread CPU pinned. Lockless access to ptr_ring
> > * consume side valid as no-resize allowed of queue.
> > */
> > - n = ptr_ring_consume_batched(rcpu->queue, frames, CPUMAP_BATCH);
> > + n = ptr_ring_consume_batched(rcpu->queue, xdp_frames,
> > + CPUMAP_BATCH);
> >
> > + rcu_read_lock();
> > +
> > + prog = READ_ONCE(rcpu->prog);
> > for (i = 0; i < n; i++) {
> > - void *f = frames[i];
> > + void *f = xdp_frames[i];
> > struct page *page = virt_to_page(f);
> > + struct xdp_frame *xdpf;
> > + struct xdp_buff xdp;
> > + u32 act;
> >
> > /* Bring struct page memory area to curr CPU. Read by
> > * build_skb_around via page_is_pfmemalloc(), and when
> > * freed written by page_frag_free call.
> > */
> > prefetchw(page);
> > + if (!prog) {
> > + frames[nframes++] = xdp_frames[i];
> > + continue;
> > + }
> > +
> > + xdpf = f;
> > + xdp.data_hard_start = xdpf->data - xdpf->headroom;
> > + xdp.data = xdpf->data;
> > + xdp.data_end = xdpf->data + xdpf->len;
> > + xdp.data_meta = xdpf->data - xdpf->metasize;
> > + xdp.frame_sz = xdpf->frame_sz;
> > + /* TODO: rxq */
> > +
> > + act = bpf_prog_run_xdp(prog, &xdp);
>
> Why not run the program in cpu_map_enqueue before converting from
> xdp_buff to xdp_frame?
Because we want to run the XDP-prog on the remote-CPU.
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer
