xdp: add tracepoint for devmap like cpumap have

Notice how this allow us get XDP statistic without affecting the XDP
performance, as tracepoint is no-longer activated on a per packet basis.

V5: Spotted by John Fastabend.
 Fix 'sent' also counted 'drops' in this patch, a later patch corrected
 this, but it was a mistake in this intermediate step.

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

include/linux/bpf.h

@@ -492,7 +492,8 @@ struct xdp_buff;
 struct bpf_dtab_netdev *__dev_map_lookup_elem(struct bpf_map *map, u32 key);
 void __dev_map_insert_ctx(struct bpf_map *map, u32 index);
 void __dev_map_flush(struct bpf_map *map);
-int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp);
+int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp,
+		    struct net_device *dev_rx);
 
 struct bpf_cpu_map_entry *__cpu_map_lookup_elem(struct bpf_map *map, u32 key);
 void __cpu_map_insert_ctx(struct bpf_map *map, u32 index);
@@ -579,7 +580,8 @@ struct xdp_buff;
 struct bpf_dtab_netdev;
 
 static inline
-int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp)
+int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp,
+		    struct net_device *dev_rx)
 {
 	return 0;
 }

include/trace/events/xdp.h

@@ -229,6 +229,45 @@ TRACE_EVENT(xdp_cpumap_enqueue,
 		  __entry->to_cpu)
 );
 
+TRACE_EVENT(xdp_devmap_xmit,
+
+	TP_PROTO(const struct bpf_map *map, u32 map_index,
+		 int sent, int drops,
+		 const struct net_device *from_dev,
+		 const struct net_device *to_dev),
+
+	TP_ARGS(map, map_index, sent, drops, from_dev, to_dev),
+
+	TP_STRUCT__entry(
+		__field(int, map_id)
+		__field(u32, act)
+		__field(u32, map_index)
+		__field(int, drops)
+		__field(int, sent)
+		__field(int, from_ifindex)
+		__field(int, to_ifindex)
+	),
+
+	TP_fast_assign(
+		__entry->map_id		= map->id;
+		__entry->act		= XDP_REDIRECT;
+		__entry->map_index	= map_index;
+		__entry->drops		= drops;
+		__entry->sent		= sent;
+		__entry->from_ifindex	= from_dev->ifindex;
+		__entry->to_ifindex	= to_dev->ifindex;
+	),
+
+	TP_printk("ndo_xdp_xmit"
+		  " map_id=%d map_index=%d action=%s"
+		  " sent=%d drops=%d"
+		  " from_ifindex=%d to_ifindex=%d",
+		  __entry->map_id, __entry->map_index,
+		  __print_symbolic(__entry->act, __XDP_ACT_SYM_TAB),
+		  __entry->sent, __entry->drops,
+		  __entry->from_ifindex, __entry->to_ifindex)
+);
+
 #endif /* _TRACE_XDP_H */
 
 #include <trace/define_trace.h>

kernel/bpf/devmap.c

@@ -58,6 +58,7 @@
 #define DEV_MAP_BULK_SIZE 16
 struct xdp_bulk_queue {
 	struct xdp_frame *q[DEV_MAP_BULK_SIZE];
+	struct net_device *dev_rx;
 	unsigned int count;
 };
 
@@ -219,6 +220,7 @@ static int bq_xmit_all(struct bpf_dtab_netdev *obj,
 			 struct xdp_bulk_queue *bq)
 {
 	struct net_device *dev = obj->dev;
+	int sent = 0, drops = 0;
 	int i;
 
 	if (unlikely(!bq->count))
@@ -235,11 +237,18 @@ static int bq_xmit_all(struct bpf_dtab_netdev *obj,
 		int err;
 
 		err = dev->netdev_ops->ndo_xdp_xmit(dev, xdpf);
-		if (err)
+		if (err) {
+			drops++;
 			xdp_return_frame(xdpf);
+		} else {
+			sent++;
+		}
 	}
 	bq->count = 0;
 
+	trace_xdp_devmap_xmit(&obj->dtab->map, obj->bit,
+			      sent, drops, bq->dev_rx, dev);
+	bq->dev_rx = NULL;
 	return 0;
 }
 
@@ -296,18 +305,28 @@ struct bpf_dtab_netdev *__dev_map_lookup_elem(struct bpf_map *map, u32 key)
 /* Runs under RCU-read-side, plus in softirq under NAPI protection.
  * Thus, safe percpu variable access.
  */
-static int bq_enqueue(struct bpf_dtab_netdev *obj, struct xdp_frame *xdpf)
+static int bq_enqueue(struct bpf_dtab_netdev *obj, struct xdp_frame *xdpf,
+		      struct net_device *dev_rx)
+
 {
 	struct xdp_bulk_queue *bq = this_cpu_ptr(obj->bulkq);
 
 	if (unlikely(bq->count == DEV_MAP_BULK_SIZE))
 		bq_xmit_all(obj, bq);
 
+	/* Ingress dev_rx will be the same for all xdp_frame's in
+	 * bulk_queue, because bq stored per-CPU and must be flushed
+	 * from net_device drivers NAPI func end.
+	 */
+	if (!bq->dev_rx)
+		bq->dev_rx = dev_rx;
+
 	bq->q[bq->count++] = xdpf;
 	return 0;
 }
 
-int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp)
+int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp,
+		    struct net_device *dev_rx)
 {
 	struct net_device *dev = dst->dev;
 	struct xdp_frame *xdpf;
@@ -319,7 +338,7 @@ int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp)
 	if (unlikely(!xdpf))
 		return -EOVERFLOW;
 
-	return bq_enqueue(dst, xdpf);
+	return bq_enqueue(dst, xdpf, dev_rx);
 }
 
 static void *dev_map_lookup_elem(struct bpf_map *map, void *key)

net/core/filter.c

@@ -3067,7 +3067,7 @@ static int __bpf_tx_xdp_map(struct net_device *dev_rx, void *fwd,
 	case BPF_MAP_TYPE_DEVMAP: {
 		struct bpf_dtab_netdev *dst = fwd;
 
-		err = dev_map_enqueue(dst, xdp);
+		err = dev_map_enqueue(dst, xdp, dev_rx);
 		if (err)
 			return err;
 		__dev_map_insert_ctx(map, index);