xdp: change ndo_xdp_xmit API to support bulking

This patch change the API for ndo_xdp_xmit to support bulking
xdp_frames.

When kernel is compiled with CONFIG_RETPOLINE, XDP sees a huge slowdown.
Most of the slowdown is caused by DMA API indirect function calls, but
also the net_device->ndo_xdp_xmit() call.

Benchmarked patch with CONFIG_RETPOLINE, using xdp_redirect_map with
single flow/core test (CPU E5-1650 v4 @ 3.60GHz), showed
performance improved:
 for driver ixgbe: 6,042,682 pps -> 6,853,768 pps = +811,086 pps
 for driver i40e : 6,187,169 pps -> 6,724,519 pps = +537,350 pps

With frames avail as a bulk inside the driver ndo_xdp_xmit call,
further optimizations are possible, like bulk DMA-mapping for TX.

Testing without CONFIG_RETPOLINE show the same performance for
physical NIC drivers.

The virtual NIC driver tun sees a huge performance boost, as it can
avoid doing per frame producer locking, but instead amortize the
locking cost over the bulk.

V2: Fix compile errors reported by kbuild test robot <lkp@intel.com>
V4: Isolated ndo, driver changes and callers.

Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

1 files changed, 18 insertions, 11 deletions

diff --git a/kernel/bpf/devmap.c b/kernel/bpf/devmap.c
index a9cd5c9..7790831 100644
--- a/kernel/bpf/devmap.c
+++ b/kernel/bpf/devmap.c
@@ -232,24 +232,31 @@ static int bq_xmit_all(struct bpf_dtab_netdev *obj,
 		prefetch(xdpf);
 	}
 
-	for (i = 0; i < bq->count; i++) {
-		struct xdp_frame *xdpf = bq->q[i];
-		int err;
-
-		err = dev->netdev_ops->ndo_xdp_xmit(dev, xdpf);
-		if (err) {
-			drops++;
-			xdp_return_frame_rx_napi(xdpf);
-		} else {
-			sent++;
-		}
+	sent = dev->netdev_ops->ndo_xdp_xmit(dev, bq->count, bq->q);
+	if (sent < 0) {
+		sent = 0;
+		goto error;
 	}
+	drops = bq->count - sent;
+out:
 	bq->count = 0;
 
 	trace_xdp_devmap_xmit(&obj->dtab->map, obj->bit,
 			      sent, drops, bq->dev_rx, dev);
 	bq->dev_rx = NULL;
 	return 0;
+error:
+	/* If ndo_xdp_xmit fails with an errno, no frames have been
+	 * xmit'ed and it's our responsibility to them free all.
+	 */
+	for (i = 0; i < bq->count; i++) {
+		struct xdp_frame *xdpf = bq->q[i];
+
+		/* RX path under NAPI protection, can return frames faster */
+		xdp_return_frame_rx_napi(xdpf);
+		drops++;
+	}
+	goto out;
 }
 
 /* __dev_map_flush is called from xdp_do_flush_map() which _must_ be signaled