ibmvnic: Add stat for tx direct vs tx batched

Allow tracking of packets sent with send_subcrq direct vs
indirect. `ethtool -S <dev>` will now provide a counter
of the number of uses of each xmit method. This metric will
be useful in performance debugging.

Signed-off-by: Nick Child <[email protected]>
Reviewed-by: Simon Horman <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Author: Nick Child

drivers/net/ethernet/ibm/ibmvnic.c

@@ -2310,7 +2310,7 @@ static void ibmvnic_tx_scrq_clean_buffer(struct ibmvnic_adapter *adapter,
 		tx_buff = &tx_pool->tx_buff[index];
 		adapter->netdev->stats.tx_packets--;
 		adapter->netdev->stats.tx_bytes -= tx_buff->skb->len;
-		adapter->tx_stats_buffers[queue_num].packets--;
+		adapter->tx_stats_buffers[queue_num].batched_packets--;
 		adapter->tx_stats_buffers[queue_num].bytes -=
 						tx_buff->skb->len;
 		dev_kfree_skb_any(tx_buff->skb);
@@ -2402,7 +2402,8 @@ static netdev_tx_t ibmvnic_xmit(struct sk_buff *skb, struct net_device *netdev)
 	unsigned int tx_map_failed = 0;
 	union sub_crq indir_arr[16];
 	unsigned int tx_dropped = 0;
-	unsigned int tx_packets = 0;
+	unsigned int tx_dpackets = 0;
+	unsigned int tx_bpackets = 0;
 	unsigned int tx_bytes = 0;
 	dma_addr_t data_dma_addr;
 	struct netdev_queue *txq;
@@ -2573,6 +2574,7 @@ static netdev_tx_t ibmvnic_xmit(struct sk_buff *skb, struct net_device *netdev)
 		if (lpar_rc != H_SUCCESS)
 			goto tx_err;
 
+		tx_dpackets++;
 		goto early_exit;
 	}
 
@@ -2601,14 +2603,15 @@ static netdev_tx_t ibmvnic_xmit(struct sk_buff *skb, struct net_device *netdev)
 			goto tx_err;
 	}
 
+	tx_bpackets++;
+
 early_exit:
 	if (atomic_add_return(num_entries, &tx_scrq->used)
 					>= adapter->req_tx_entries_per_subcrq) {
 		netdev_dbg(netdev, "Stopping queue %d\n", queue_num);
 		netif_stop_subqueue(netdev, queue_num);
 	}
 
-	tx_packets++;
 	tx_bytes += skb->len;
 	txq_trans_cond_update(txq);
 	ret = NETDEV_TX_OK;
@@ -2638,10 +2641,11 @@ static netdev_tx_t ibmvnic_xmit(struct sk_buff *skb, struct net_device *netdev)
 	rcu_read_unlock();
 	netdev->stats.tx_dropped += tx_dropped;
 	netdev->stats.tx_bytes += tx_bytes;
-	netdev->stats.tx_packets += tx_packets;
+	netdev->stats.tx_packets += tx_bpackets + tx_dpackets;
 	adapter->tx_send_failed += tx_send_failed;
 	adapter->tx_map_failed += tx_map_failed;
-	adapter->tx_stats_buffers[queue_num].packets += tx_packets;
+	adapter->tx_stats_buffers[queue_num].batched_packets += tx_bpackets;
+	adapter->tx_stats_buffers[queue_num].direct_packets += tx_dpackets;
 	adapter->tx_stats_buffers[queue_num].bytes += tx_bytes;
 	adapter->tx_stats_buffers[queue_num].dropped_packets += tx_dropped;
 
@@ -3806,7 +3810,10 @@ static void ibmvnic_get_strings(struct net_device *dev, u32 stringset, u8 *data)
 		memcpy(data, ibmvnic_stats[i].name, ETH_GSTRING_LEN);
 
 	for (i = 0; i < adapter->req_tx_queues; i++) {
-		snprintf(data, ETH_GSTRING_LEN, "tx%d_packets", i);
+		snprintf(data, ETH_GSTRING_LEN, "tx%d_batched_packets", i);
+		data += ETH_GSTRING_LEN;
+
+		snprintf(data, ETH_GSTRING_LEN, "tx%d_direct_packets", i);
 		data += ETH_GSTRING_LEN;
 
 		snprintf(data, ETH_GSTRING_LEN, "tx%d_bytes", i);
@@ -3871,7 +3878,9 @@ static void ibmvnic_get_ethtool_stats(struct net_device *dev,
 				      (adapter, ibmvnic_stats[i].offset));
 
 	for (j = 0; j < adapter->req_tx_queues; j++) {
-		data[i] = adapter->tx_stats_buffers[j].packets;
+		data[i] = adapter->tx_stats_buffers[j].batched_packets;
+		i++;
+		data[i] = adapter->tx_stats_buffers[j].direct_packets;
 		i++;
 		data[i] = adapter->tx_stats_buffers[j].bytes;
 		i++;

drivers/net/ethernet/ibm/ibmvnic.h

@@ -213,7 +213,8 @@ struct ibmvnic_statistics {
 
 #define NUM_TX_STATS 3
 struct ibmvnic_tx_queue_stats {
-	u64 packets;
+	u64 batched_packets;
+	u64 direct_packets;
 	u64 bytes;
 	u64 dropped_packets;
 };