mac80211: expose txq queue depth and size to drivers

This will allow drivers to make more educated
decisions whether to defer transmission or not.

Relying on wake_tx_queue() call count implicitly
was not possible because it could be called
without queued frame count actually changing on
software tx aggregation start/stop code paths.

It was also not possible to know how long
byte-wise queue was without dequeueing.
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

include/net/mac80211.h

@@ -5596,4 +5596,19 @@ void ieee80211_unreserve_tid(struct ieee80211_sta *sta, u8 tid);
  */
 struct sk_buff *ieee80211_tx_dequeue(struct ieee80211_hw *hw,
 				     struct ieee80211_txq *txq);
+
+/**
+ * ieee80211_txq_get_depth - get pending frame/byte count of given txq
+ *
+ * The values are not guaranteed to be coherent with regard to each other, i.e.
+ * txq state can change half-way of this function and the caller may end up
+ * with "new" frame_cnt and "old" byte_cnt or vice-versa.
+ *
+ * @txq: pointer obtained from station or virtual interface
+ * @frame_cnt: pointer to store frame count
+ * @byte_cnt: pointer to store byte count
+ */
+void ieee80211_txq_get_depth(struct ieee80211_txq *txq,
+			     unsigned long *frame_cnt,
+			     unsigned long *byte_cnt);
 #endif /* MAC80211_H */

net/mac80211/ieee80211_i.h

@@ -804,6 +804,7 @@ enum txq_info_flags {
 struct txq_info {
 	struct sk_buff_head queue;
 	unsigned long flags;
+	unsigned long byte_cnt;
 
 	/* keep last! */
 	struct ieee80211_txq txq;

net/mac80211/iface.c

@@ -979,6 +979,7 @@ static void ieee80211_do_stop(struct ieee80211_sub_if_data *sdata,
 
 		spin_lock_bh(&txqi->queue.lock);
 		ieee80211_purge_tx_queue(&local->hw, &txqi->queue);
+		txqi->byte_cnt = 0;
 		spin_unlock_bh(&txqi->queue.lock);
 
 		atomic_set(&sdata->txqs_len[txqi->txq.ac], 0);

net/mac80211/sta_info.c

@@ -116,6 +116,7 @@ static void __cleanup_single_sta(struct sta_info *sta)
 
 			ieee80211_purge_tx_queue(&local->hw, &txqi->queue);
 			atomic_sub(n, &sdata->txqs_len[txqi->txq.ac]);
+			txqi->byte_cnt = 0;
 		}
 	}
 

net/mac80211/tx.c

@@ -1270,7 +1270,11 @@ static void ieee80211_drv_tx(struct ieee80211_local *local,
 	if (atomic_read(&sdata->txqs_len[ac]) >= local->hw.txq_ac_max_pending)
 		netif_stop_subqueue(sdata->dev, ac);
 
-	skb_queue_tail(&txqi->queue, skb);
+	spin_lock_bh(&txqi->queue.lock);
+	txqi->byte_cnt += skb->len;
+	__skb_queue_tail(&txqi->queue, skb);
+	spin_unlock_bh(&txqi->queue.lock);
+
 	drv_wake_tx_queue(local, txqi);
 
 	return;
@@ -1298,6 +1302,8 @@ struct sk_buff *ieee80211_tx_dequeue(struct ieee80211_hw *hw,
 	if (!skb)
 		goto out;
 
+	txqi->byte_cnt -= skb->len;
+
 	atomic_dec(&sdata->txqs_len[ac]);
 	if (__netif_subqueue_stopped(sdata->dev, ac))
 		ieee80211_propagate_queue_wake(local, sdata->vif.hw_queue[ac]);

net/mac80211/util.c

@@ -3368,3 +3368,17 @@ void ieee80211_init_tx_queue(struct ieee80211_sub_if_data *sdata,
 		txqi->txq.ac = IEEE80211_AC_BE;
 	}
 }
+
+void ieee80211_txq_get_depth(struct ieee80211_txq *txq,
+			     unsigned long *frame_cnt,
+			     unsigned long *byte_cnt)
+{
+	struct txq_info *txqi = to_txq_info(txq);
+
+	if (frame_cnt)
+		*frame_cnt = txqi->queue.qlen;
+
+	if (byte_cnt)
+		*byte_cnt = txqi->byte_cnt;
+}
+EXPORT_SYMBOL(ieee80211_txq_get_depth);