veth: fix data race in veth_get_ethtool_stats

In veth_get_ethtool_stats(), some statistics protected by
u64_stats_sync, are read and accumulated in ignorance of possible
u64_stats_fetch_retry() events. These statistics, peer_tq_xdp_xmit and
peer_tq_xdp_xmit_err, are already accumulated by veth_xdp_xmit(). Fix
this by reading them into a temporary buffer first.

Fixes: 5fe6e56776ba ("veth: rely on peer veth_rq for ndo_xdp_xmit accounting")
Signed-off-by: David Yang <[e-mail]>
Link: https://patch.msgid.link/[e-mail]
Signed-off-by: Jakub Kicinski <[e-mail]>

drivers/net/veth.c

···
 		const struct veth_rq_stats *rq_stats = &rcv_priv->rq[i].stats;
 		const void *base = (void *)&rq_stats->vs;
 		unsigned int start, tx_idx = idx;
+		u64 buf[VETH_TQ_STATS_LEN];
 		size_t offset;
 
-		tx_idx += (i % dev->real_num_tx_queues) * VETH_TQ_STATS_LEN;
 		do {
 			start = u64_stats_fetch_begin(&rq_stats->syncp);
 			for (j = 0; j < VETH_TQ_STATS_LEN; j++) {
 				offset = veth_tq_stats_desc[j].offset;
-				data[tx_idx + j] += *(u64 *)(base + offset);
+				buf[j] = *(u64 *)(base + offset);
 			}
 		} while (u64_stats_fetch_retry(&rq_stats->syncp, start));
+
+		tx_idx += (i % dev->real_num_tx_queues) * VETH_TQ_STATS_LEN;
+		for (j = 0; j < VETH_TQ_STATS_LEN; j++)
+			data[tx_idx + j] += buf[j];
 	}
 	pp_idx = idx + dev->real_num_tx_queues * VETH_TQ_STATS_LEN;