1 files changed, 551 insertions, 276 deletions

diff --git a/drivers/net/wireless/mediatek/mt76/tx.c b/drivers/net/wireless/mediatek/mt76/tx.c
index fca38ea2441f..9ec6d0b53a84 100644
--- a/drivers/net/wireless/mediatek/mt76/tx.c
+++ b/drivers/net/wireless/mediatek/mt76/tx.c
@@ -1,79 +1,10 @@
-// SPDX-License-Identifier: ISC
+// SPDX-License-Identifier: BSD-3-Clause-Clear
 /*
  * Copyright (C) 2016 Felix Fietkau <nbd@nbd.name>
  */
 
 #include "mt76.h"
 
-static struct mt76_txwi_cache *
-mt76_alloc_txwi(struct mt76_dev *dev)
-{
-	struct mt76_txwi_cache *t;
-	dma_addr_t addr;
-	u8 *txwi;
-	int size;
-
-	size = L1_CACHE_ALIGN(dev->drv->txwi_size + sizeof(*t));
-	txwi = devm_kzalloc(dev->dev, size, GFP_ATOMIC);
-	if (!txwi)
-		return NULL;
-
-	addr = dma_map_single(dev->dev, txwi, dev->drv->txwi_size,
-			      DMA_TO_DEVICE);
-	t = (struct mt76_txwi_cache *)(txwi + dev->drv->txwi_size);
-	t->dma_addr = addr;
-
-	return t;
-}
-
-static struct mt76_txwi_cache *
-__mt76_get_txwi(struct mt76_dev *dev)
-{
-	struct mt76_txwi_cache *t = NULL;
-
-	spin_lock_bh(&dev->lock);
-	if (!list_empty(&dev->txwi_cache)) {
-		t = list_first_entry(&dev->txwi_cache, struct mt76_txwi_cache,
-				     list);
-		list_del(&t->list);
-	}
-	spin_unlock_bh(&dev->lock);
-
-	return t;
-}
-
-struct mt76_txwi_cache *
-mt76_get_txwi(struct mt76_dev *dev)
-{
-	struct mt76_txwi_cache *t = __mt76_get_txwi(dev);
-
-	if (t)
-		return t;
-
-	return mt76_alloc_txwi(dev);
-}
-
-void
-mt76_put_txwi(struct mt76_dev *dev, struct mt76_txwi_cache *t)
-{
-	if (!t)
-		return;
-
-	spin_lock_bh(&dev->lock);
-	list_add(&t->list, &dev->txwi_cache);
-	spin_unlock_bh(&dev->lock);
-}
-EXPORT_SYMBOL_GPL(mt76_put_txwi);
-
-void mt76_tx_free(struct mt76_dev *dev)
-{
-	struct mt76_txwi_cache *t;
-
-	while ((t = __mt76_get_txwi(dev)) != NULL)
-		dma_unmap_single(dev->dev, t->dma_addr, dev->drv->txwi_size,
-				 DMA_TO_DEVICE);
-}
-
 static int
 mt76_txq_get_qid(struct ieee80211_txq *txq)
 {
@@ -83,41 +14,74 @@ mt76_txq_get_qid(struct ieee80211_txq *txq)
 	return txq->ac;
 }
 
-static void
-mt76_check_agg_ssn(struct mt76_txq *mtxq, struct sk_buff *skb)
+void
+mt76_tx_check_agg_ssn(struct ieee80211_sta *sta, struct sk_buff *skb)
 {
 	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)skb->data;
+	struct ieee80211_txq *txq;
+	struct mt76_txq *mtxq;
+	u8 tid;
 
-	if (!ieee80211_is_data_qos(hdr->frame_control) ||
+	if (!sta || !ieee80211_is_data_qos(hdr->frame_control) ||
 	    !ieee80211_is_data_present(hdr->frame_control))
 		return;
 
+	tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK;
+	txq = sta->txq[tid];
+	mtxq = (struct mt76_txq *)txq->drv_priv;
+	if (!mtxq->aggr)
+		return;
+
 	mtxq->agg_ssn = le16_to_cpu(hdr->seq_ctrl) + 0x10;
 }
+EXPORT_SYMBOL_GPL(mt76_tx_check_agg_ssn);
 
 void
 mt76_tx_status_lock(struct mt76_dev *dev, struct sk_buff_head *list)
-		   __acquires(&dev->status_list.lock)
+		   __acquires(&dev->status_lock)
 {
 	__skb_queue_head_init(list);
-	spin_lock_bh(&dev->status_list.lock);
+	spin_lock_bh(&dev->status_lock);
 }
 EXPORT_SYMBOL_GPL(mt76_tx_status_lock);
 
 void
 mt76_tx_status_unlock(struct mt76_dev *dev, struct sk_buff_head *list)
-		      __releases(&dev->status_list.lock)
+		      __releases(&dev->status_lock)
 {
 	struct ieee80211_hw *hw;
 	struct sk_buff *skb;
 
-	spin_unlock_bh(&dev->status_list.lock);
+	spin_unlock_bh(&dev->status_lock);
 
+	rcu_read_lock();
 	while ((skb = __skb_dequeue(list)) != NULL) {
+		struct ieee80211_tx_status status = {
+			.skb = skb,
+			.info = IEEE80211_SKB_CB(skb),
+		};
+		struct ieee80211_rate_status rs = {};
+		struct mt76_tx_cb *cb = mt76_tx_skb_cb(skb);
+		struct mt76_wcid *wcid;
+
+		wcid = __mt76_wcid_ptr(dev, cb->wcid);
+		if (wcid) {
+			status.sta = wcid_to_sta(wcid);
+			if (status.sta && (wcid->rate.flags || wcid->rate.legacy)) {
+				rs.rate_idx = wcid->rate;
+				status.rates = &rs;
+				status.n_rates = 1;
+			} else {
+				status.n_rates = 0;
+			}
+		}
+
 		hw = mt76_tx_status_get_hw(dev, skb);
-		ieee80211_tx_status(hw, skb);
+		spin_lock_bh(&dev->rx_lock);
+		ieee80211_tx_status_ext(hw, &status);
+		spin_unlock_bh(&dev->rx_lock);
 	}
-
+	rcu_read_unlock();
 }
 EXPORT_SYMBOL_GPL(mt76_tx_status_unlock);
 
@@ -135,11 +99,10 @@ __mt76_tx_status_skb_done(struct mt76_dev *dev, struct sk_buff *skb, u8 flags,
 	if ((flags & done) != done)
 		return;
 
-	__skb_unlink(skb, &dev->status_list);
-
 	/* Tx status can be unreliable. if it fails, mark the frame as ACKed */
-	if (flags & MT_TX_CB_TXS_FAILED) {
-		ieee80211_tx_info_clear_status(info);
+	if (flags & MT_TX_CB_TXS_FAILED &&
+	    (dev->drv->drv_flags & MT_DRV_IGNORE_TXS_FAILED)) {
+		info->status.rates[0].count = 0;
 		info->status.rates[0].idx = -1;
 		info->flags |= IEEE80211_TX_STAT_ACK;
 	}
@@ -159,35 +122,46 @@ int
 mt76_tx_status_skb_add(struct mt76_dev *dev, struct mt76_wcid *wcid,
 		       struct sk_buff *skb)
 {
+	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)skb->data;
 	struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
 	struct mt76_tx_cb *cb = mt76_tx_skb_cb(skb);
 	int pid;
 
-	if (!wcid)
+	memset(cb, 0, sizeof(*cb));
+
+	if (!wcid || !rcu_access_pointer(dev->wcid[wcid->idx]))
 		return MT_PACKET_ID_NO_ACK;
 
 	if (info->flags & IEEE80211_TX_CTL_NO_ACK)
 		return MT_PACKET_ID_NO_ACK;
 
 	if (!(info->flags & (IEEE80211_TX_CTL_REQ_TX_STATUS |
-			     IEEE80211_TX_CTL_RATE_CTRL_PROBE)))
+			     IEEE80211_TX_CTL_RATE_CTRL_PROBE))) {
+		if (mtk_wed_device_active(&dev->mmio.wed) &&
+		    ((info->flags & IEEE80211_TX_CTL_HW_80211_ENCAP) ||
+		     ieee80211_is_data(hdr->frame_control)))
+			return MT_PACKET_ID_WED;
+
 		return MT_PACKET_ID_NO_SKB;
+	}
 
-	spin_lock_bh(&dev->status_list.lock);
+	spin_lock_bh(&dev->status_lock);
 
-	memset(cb, 0, sizeof(*cb));
-	wcid->packet_id = (wcid->packet_id + 1) & MT_PACKET_ID_MASK;
-	if (wcid->packet_id == MT_PACKET_ID_NO_ACK ||
-	    wcid->packet_id == MT_PACKET_ID_NO_SKB)
-		wcid->packet_id = MT_PACKET_ID_FIRST;
+	pid = idr_alloc(&wcid->pktid, skb, MT_PACKET_ID_FIRST,
+			MT_PACKET_ID_MASK, GFP_ATOMIC);
+	if (pid < 0) {
+		pid = MT_PACKET_ID_NO_SKB;
+		goto out;
+	}
 
-	pid = wcid->packet_id;
 	cb->wcid = wcid->idx;
 	cb->pktid = pid;
-	cb->jiffies = jiffies;
 
-	__skb_queue_tail(&dev->status_list, skb);
-	spin_unlock_bh(&dev->status_list.lock);
+	if (list_empty(&wcid->list))
+		list_add_tail(&wcid->list, &dev->wcid_list);
+
+out:
+	spin_unlock_bh(&dev->status_lock);
 
 	return pid;
 }
@@ -197,139 +171,223 @@ struct sk_buff *
 mt76_tx_status_skb_get(struct mt76_dev *dev, struct mt76_wcid *wcid, int pktid,
 		       struct sk_buff_head *list)
 {
-	struct sk_buff *skb, *tmp;
+	struct sk_buff *skb;
+	int id;
 
-	skb_queue_walk_safe(&dev->status_list, skb, tmp) {
-		struct mt76_tx_cb *cb = mt76_tx_skb_cb(skb);
+	lockdep_assert_held(&dev->status_lock);
 
-		if (wcid && cb->wcid != wcid->idx)
-			continue;
+	skb = idr_remove(&wcid->pktid, pktid);
+	if (skb)
+		goto out;
 
-		if (cb->pktid == pktid)
-			return skb;
+	/* look for stale entries in the wcid idr queue */
+	idr_for_each_entry(&wcid->pktid, skb, id) {
+		struct mt76_tx_cb *cb = mt76_tx_skb_cb(skb);
 
-		if (pktid >= 0 && !time_after(jiffies, cb->jiffies +
-					      MT_TX_STATUS_SKB_TIMEOUT))
-			continue;
+		if (pktid >= 0) {
+			if (!(cb->flags & MT_TX_CB_DMA_DONE))
+				continue;
 
+			if (time_is_after_jiffies(cb->jiffies +
+						   MT_TX_STATUS_SKB_TIMEOUT))
+				continue;
+		}
+
+		/* It has been too long since DMA_DONE, time out this packet
+		 * and stop waiting for TXS callback.
+		 */
+		idr_remove(&wcid->pktid, cb->pktid);
 		__mt76_tx_status_skb_done(dev, skb, MT_TX_CB_TXS_FAILED |
 						    MT_TX_CB_TXS_DONE, list);
 	}
 
-	return NULL;
+out:
+	if (idr_is_empty(&wcid->pktid))
+		list_del_init(&wcid->list);
+
+	return skb;
 }
 EXPORT_SYMBOL_GPL(mt76_tx_status_skb_get);
 
 void
-mt76_tx_status_check(struct mt76_dev *dev, struct mt76_wcid *wcid, bool flush)
+mt76_tx_status_check(struct mt76_dev *dev, bool flush)
 {
+	struct mt76_wcid *wcid, *tmp;
 	struct sk_buff_head list;
 
 	mt76_tx_status_lock(dev, &list);
-	mt76_tx_status_skb_get(dev, wcid, flush ? -1 : 0, &list);
+	list_for_each_entry_safe(wcid, tmp, &dev->wcid_list, list)
+		mt76_tx_status_skb_get(dev, wcid, flush ? -1 : 0, &list);
 	mt76_tx_status_unlock(dev, &list);
 }
 EXPORT_SYMBOL_GPL(mt76_tx_status_check);
 
-void mt76_tx_complete_skb(struct mt76_dev *dev, struct sk_buff *skb)
+static void
+mt76_tx_check_non_aql(struct mt76_dev *dev, struct mt76_wcid *wcid,
+		      struct sk_buff *skb)
+{
+	struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
+	int pending;
+
+	if (!wcid || info->tx_time_est)
+		return;
+
+	pending = atomic_dec_return(&wcid->non_aql_packets);
+	if (pending < 0)
+		atomic_cmpxchg(&wcid->non_aql_packets, pending, 0);
+}
+
+void __mt76_tx_complete_skb(struct mt76_dev *dev, u16 wcid_idx, struct sk_buff *skb,
+			    struct list_head *free_list)
 {
+	struct mt76_tx_cb *cb = mt76_tx_skb_cb(skb);
+	struct ieee80211_tx_status status = {
+		.skb = skb,
+		.free_list = free_list,
+	};
+	struct mt76_wcid *wcid = NULL;
 	struct ieee80211_hw *hw;
 	struct sk_buff_head list;
 
-	if (!skb->prev) {
+	rcu_read_lock();
+
+	wcid = __mt76_wcid_ptr(dev, wcid_idx);
+	mt76_tx_check_non_aql(dev, wcid, skb);
+
+#ifdef CONFIG_NL80211_TESTMODE
+	if (mt76_is_testmode_skb(dev, skb, &hw)) {
+		struct mt76_phy *phy = hw->priv;
+
+		if (skb == phy->test.tx_skb)
+			phy->test.tx_done++;
+		if (phy->test.tx_queued == phy->test.tx_done)
+			wake_up(&dev->tx_wait);
+
+		dev_kfree_skb_any(skb);
+		goto out;
+	}
+#endif
+
+	if (cb->pktid < MT_PACKET_ID_FIRST) {
+		struct ieee80211_rate_status rs = {};
+
 		hw = mt76_tx_status_get_hw(dev, skb);
-		ieee80211_free_txskb(hw, skb);
-		return;
+		status.sta = wcid_to_sta(wcid);
+		if (status.sta && (wcid->rate.flags || wcid->rate.legacy)) {
+			rs.rate_idx = wcid->rate;
+			status.rates = &rs;
+			status.n_rates = 1;
+		}
+		spin_lock_bh(&dev->rx_lock);
+		ieee80211_tx_status_ext(hw, &status);
+		spin_unlock_bh(&dev->rx_lock);
+		goto out;
 	}
 
 	mt76_tx_status_lock(dev, &list);
+	cb->jiffies = jiffies;
 	__mt76_tx_status_skb_done(dev, skb, MT_TX_CB_DMA_DONE, &list);
 	mt76_tx_status_unlock(dev, &list);
+
+out:
+	rcu_read_unlock();
+}
+EXPORT_SYMBOL_GPL(__mt76_tx_complete_skb);
+
+static int
+__mt76_tx_queue_skb(struct mt76_phy *phy, int qid, struct sk_buff *skb,
+		    struct mt76_wcid *wcid, struct ieee80211_sta *sta,
+		    bool *stop)
+{
+	struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
+	struct mt76_queue *q = phy->q_tx[qid];
+	struct mt76_dev *dev = phy->dev;
+	bool non_aql;
+	int pending;
+	int idx;
+
+	non_aql = !info->tx_time_est;
+	idx = dev->queue_ops->tx_queue_skb(phy, q, qid, skb, wcid, sta);
+	if (idx < 0 || !sta)
+		return idx;
+
+	wcid = (struct mt76_wcid *)sta->drv_priv;
+	if (!wcid->sta)
+		return idx;
+
+	q->entry[idx].wcid = wcid->idx;
+
+	if (!non_aql)
+		return idx;
+
+	pending = atomic_inc_return(&wcid->non_aql_packets);
+	if (stop && pending >= MT_MAX_NON_AQL_PKT)
+		*stop = true;
+
+	return idx;
 }
-EXPORT_SYMBOL_GPL(mt76_tx_complete_skb);
 
 void
 mt76_tx(struct mt76_phy *phy, struct ieee80211_sta *sta,
 	struct mt76_wcid *wcid, struct sk_buff *skb)
 {
-	struct mt76_dev *dev = phy->dev;
 	struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
-	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)skb->data;
-	struct mt76_queue *q;
-	int qid = skb_get_queue_mapping(skb);
-	bool ext_phy = phy != &dev->phy;
+	struct ieee80211_hdr *hdr = (void *)skb->data;
+	struct sk_buff_head *head;
 
-	if (WARN_ON(qid >= MT_TXQ_PSD)) {
-		qid = MT_TXQ_BE;
-		skb_set_queue_mapping(skb, qid);
+	if (mt76_testmode_enabled(phy)) {
+		ieee80211_free_txskb(phy->hw, skb);
+		return;
 	}
 
-	if (!(wcid->tx_info & MT_WCID_TX_INFO_SET))
+	if (WARN_ON(skb_get_queue_mapping(skb) >= MT_TXQ_PSD))
+		skb_set_queue_mapping(skb, MT_TXQ_BE);
+
+	if (wcid && !(wcid->tx_info & MT_WCID_TX_INFO_SET))
 		ieee80211_get_tx_rates(info->control.vif, sta, skb,
 				       info->control.rates, 1);
 
-	if (sta && ieee80211_is_data_qos(hdr->frame_control)) {
-		struct ieee80211_txq *txq;
-		struct mt76_txq *mtxq;
-		u8 tid;
-
-		tid = skb->priority & IEEE80211_QOS_CTL_TID_MASK;
-		txq = sta->txq[tid];
-		mtxq = (struct mt76_txq *)txq->drv_priv;
-
-		if (mtxq->aggr)
-			mt76_check_agg_ssn(mtxq, skb);
-	}
+	info->hw_queue |= FIELD_PREP(MT_TX_HW_QUEUE_PHY, phy->band_idx);
 
-	if (ext_phy)
-		info->hw_queue |= MT_TX_HW_QUEUE_EXT_PHY;
+	if ((info->flags & IEEE80211_TX_CTL_TX_OFFCHAN) ||
+	    ((info->control.flags & IEEE80211_TX_CTRL_DONT_USE_RATE_MASK) &&
+	     ieee80211_is_probe_req(hdr->frame_control)))
+		head = &wcid->tx_offchannel;
+	else
+		head = &wcid->tx_pending;
 
-	q = dev->q_tx[qid].q;
+	spin_lock_bh(&head->lock);
+	__skb_queue_tail(head, skb);
+	spin_unlock_bh(&head->lock);
 
-	spin_lock_bh(&q->lock);
-	dev->queue_ops->tx_queue_skb(dev, qid, skb, wcid, sta);
-	dev->queue_ops->kick(dev, q);
-
-	if (q->queued > q->ndesc - 8 && !q->stopped) {
-		ieee80211_stop_queue(phy->hw, skb_get_queue_mapping(skb));
-		q->stopped = true;
-	}
+	spin_lock_bh(&phy->tx_lock);
+	if (list_empty(&wcid->tx_list))
+		list_add_tail(&wcid->tx_list, &phy->tx_list);
+	spin_unlock_bh(&phy->tx_lock);
 
-	spin_unlock_bh(&q->lock);
+	mt76_worker_schedule(&phy->dev->tx_worker);
 }
 EXPORT_SYMBOL_GPL(mt76_tx);
 
 static struct sk_buff *
-mt76_txq_dequeue(struct mt76_phy *phy, struct mt76_txq *mtxq, bool ps)
+mt76_txq_dequeue(struct mt76_phy *phy, struct mt76_txq *mtxq)
 {
 	struct ieee80211_txq *txq = mtxq_to_txq(mtxq);
 	struct ieee80211_tx_info *info;
-	bool ext_phy = phy != &phy->dev->phy;
 	struct sk_buff *skb;
 
-	skb = skb_dequeue(&mtxq->retry_q);
-	if (skb) {
-		u8 tid = skb->priority & IEEE80211_QOS_CTL_TID_MASK;
-
-		if (ps && skb_queue_empty(&mtxq->retry_q))
-			ieee80211_sta_set_buffered(txq->sta, tid, false);
-
-		return skb;
-	}
-
 	skb = ieee80211_tx_dequeue(phy->hw, txq);
 	if (!skb)
 		return NULL;
 
 	info = IEEE80211_SKB_CB(skb);
-	if (ext_phy)
-		info->hw_queue |= MT_TX_HW_QUEUE_EXT_PHY;
+	info->hw_queue |= FIELD_PREP(MT_TX_HW_QUEUE_PHY, phy->band_idx);
 
 	return skb;
 }
 
 static void
-mt76_queue_ps_skb(struct mt76_dev *dev, struct ieee80211_sta *sta,
+mt76_queue_ps_skb(struct mt76_phy *phy, struct ieee80211_sta *sta,
 		  struct sk_buff *skb, bool last)
 {
 	struct mt76_wcid *wcid = (struct mt76_wcid *)sta->drv_priv;
@@ -341,7 +399,7 @@ mt76_queue_ps_skb(struct mt76_dev *dev, struct ieee80211_sta *sta,
 			       IEEE80211_TX_CTL_REQ_TX_STATUS;
 
 	mt76_skb_set_moredata(skb, !last);
-	dev->queue_ops->tx_queue_skb(dev, MT_TXQ_PSD, skb, wcid, sta);
+	__mt76_tx_queue_skb(phy, MT_TXQ_PSD, skb, wcid, sta, NULL);
 }
 
 void
@@ -353,7 +411,7 @@ mt76_release_buffered_frames(struct ieee80211_hw *hw, struct ieee80211_sta *sta,
 	struct mt76_phy *phy = hw->priv;
 	struct mt76_dev *dev = phy->dev;
 	struct sk_buff *last_skb = NULL;
-	struct mt76_queue *hwq = dev->q_tx[MT_TXQ_PSD].q;
+	struct mt76_queue *hwq = phy->q_tx[MT_TXQ_PSD];
 	int i;
 
 	spin_lock_bh(&hwq->lock);
@@ -366,23 +424,20 @@ mt76_release_buffered_frames(struct ieee80211_hw *hw, struct ieee80211_sta *sta,
 			continue;
 
 		do {
-			skb = mt76_txq_dequeue(phy, mtxq, true);
+			skb = mt76_txq_dequeue(phy, mtxq);
 			if (!skb)
 				break;
 
-			if (mtxq->aggr)
-				mt76_check_agg_ssn(mtxq, skb);
-
 			nframes--;
 			if (last_skb)
-				mt76_queue_ps_skb(dev, sta, last_skb, false);
+				mt76_queue_ps_skb(phy, sta, last_skb, false);
 
 			last_skb = skb;
 		} while (nframes);
 	}
 
 	if (last_skb) {
-		mt76_queue_ps_skb(dev, sta, last_skb, true);
+		mt76_queue_ps_skb(phy, sta, last_skb, true);
 		dev->queue_ops->kick(dev, hwq);
 	} else {
 		ieee80211_sta_eosp(sta);
@@ -392,27 +447,33 @@ mt76_release_buffered_frames(struct ieee80211_hw *hw, struct ieee80211_sta *sta,
 }
 EXPORT_SYMBOL_GPL(mt76_release_buffered_frames);
 
+static bool
+mt76_txq_stopped(struct mt76_queue *q)
+{
+	return q->stopped || q->blocked ||
+	       q->queued + MT_TXQ_FREE_THR >= q->ndesc;
+}
+
 static int
-mt76_txq_send_burst(struct mt76_phy *phy, struct mt76_sw_queue *sq,
-		    struct mt76_txq *mtxq)
+mt76_txq_send_burst(struct mt76_phy *phy, struct mt76_queue *q,
+		    struct mt76_txq *mtxq, struct mt76_wcid *wcid)
 {
 	struct mt76_dev *dev = phy->dev;
 	struct ieee80211_txq *txq = mtxq_to_txq(mtxq);
 	enum mt76_txq_id qid = mt76_txq_get_qid(txq);
-	struct mt76_wcid *wcid = mtxq->wcid;
-	struct mt76_queue *hwq = sq->q;
 	struct ieee80211_tx_info *info;
 	struct sk_buff *skb;
-	int n_frames = 1, limit;
-	struct ieee80211_tx_rate tx_rate;
-	bool ampdu;
-	bool probe;
+	int n_frames = 1;
+	bool stop = false;
 	int idx;
 
 	if (test_bit(MT_WCID_FLAG_PS, &wcid->flags))
 		return 0;
 
-	skb = mt76_txq_dequeue(phy, mtxq, false);
+	if (atomic_read(&wcid->non_aql_packets) >= MT_MAX_NON_AQL_PKT)
+		return 0;
+
+	skb = mt76_txq_dequeue(phy, mtxq);
 	if (!skb)
 		return 0;
 
@@ -420,62 +481,41 @@ mt76_txq_send_burst(struct mt76_phy *phy, struct mt76_sw_queue *sq,
 	if (!(wcid->tx_info & MT_WCID_TX_INFO_SET))
 		ieee80211_get_tx_rates(txq->vif, txq->sta, skb,
 				       info->control.rates, 1);
-	tx_rate = info->control.rates[0];
-
-	probe = (info->flags & IEEE80211_TX_CTL_RATE_CTRL_PROBE);
-	ampdu = IEEE80211_SKB_CB(skb)->flags & IEEE80211_TX_CTL_AMPDU;
-	limit = ampdu ? 16 : 3;
-
-	if (ampdu)
-		mt76_check_agg_ssn(mtxq, skb);
-
-	idx = dev->queue_ops->tx_queue_skb(dev, qid, skb, wcid, txq->sta);
 
+	spin_lock(&q->lock);
+	idx = __mt76_tx_queue_skb(phy, qid, skb, wcid, txq->sta, &stop);
+	spin_unlock(&q->lock);
 	if (idx < 0)
 		return idx;
 
 	do {
-		bool cur_ampdu;
-
-		if (probe)
+		if (test_bit(MT76_RESET, &phy->state) || phy->offchannel)
 			break;
 
-		if (test_bit(MT76_RESET, &phy->state))
-			return -EBUSY;
+		if (stop || mt76_txq_stopped(q))
+			break;
 
-		skb = mt76_txq_dequeue(phy, mtxq, false);
+		skb = mt76_txq_dequeue(phy, mtxq);
 		if (!skb)
 			break;
 
 		info = IEEE80211_SKB_CB(skb);
-		cur_ampdu = info->flags & IEEE80211_TX_CTL_AMPDU;
-
-		if (ampdu != cur_ampdu ||
-		    (info->flags & IEEE80211_TX_CTL_RATE_CTRL_PROBE)) {
-			skb_queue_tail(&mtxq->retry_q, skb);
-			break;
-		}
-
-		info->control.rates[0] = tx_rate;
-
-		if (cur_ampdu)
-			mt76_check_agg_ssn(mtxq, skb);
+		if (!(wcid->tx_info & MT_WCID_TX_INFO_SET))
+			ieee80211_get_tx_rates(txq->vif, txq->sta, skb,
+					       info->control.rates, 1);
 
-		idx = dev->queue_ops->tx_queue_skb(dev, qid, skb, wcid,
-						   txq->sta);
+		spin_lock(&q->lock);
+		idx = __mt76_tx_queue_skb(phy, qid, skb, wcid, txq->sta, &stop);
+		spin_unlock(&q->lock);
 		if (idx < 0)
-			return idx;
+			break;
 
 		n_frames++;
-	} while (n_frames < limit);
-
-	if (!probe) {
-		hwq->entry[idx].qid = sq - dev->q_tx;
-		hwq->entry[idx].schedule = true;
-		sq->swq_queued++;
-	}
+	} while (1);
 
-	dev->queue_ops->kick(dev, hwq);
+	spin_lock(&q->lock);
+	dev->queue_ops->kick(dev, q);
+	spin_unlock(&q->lock);
 
 	return n_frames;
 }
@@ -484,32 +524,34 @@ static int
 mt76_txq_schedule_list(struct mt76_phy *phy, enum mt76_txq_id qid)
 {
 	struct mt76_dev *dev = phy->dev;
-	struct mt76_sw_queue *sq = &dev->q_tx[qid];
-	struct mt76_queue *hwq = sq->q;
 	struct ieee80211_txq *txq;
 	struct mt76_txq *mtxq;
 	struct mt76_wcid *wcid;
+	struct mt76_queue *q;
 	int ret = 0;
 
-	spin_lock_bh(&hwq->lock);
 	while (1) {
-		if (sq->swq_queued >= 4)
-			break;
-
-		if (test_bit(MT76_RESET, &phy->state)) {
-			ret = -EBUSY;
-			break;
-		}
+		int n_frames = 0;
 
 		txq = ieee80211_next_txq(phy->hw, qid);
 		if (!txq)
 			break;
 
 		mtxq = (struct mt76_txq *)txq->drv_priv;
-		wcid = mtxq->wcid;
-		if (wcid && test_bit(MT_WCID_FLAG_PS, &wcid->flags))
+		wcid = __mt76_wcid_ptr(dev, mtxq->wcid);
+		if (!wcid || test_bit(MT_WCID_FLAG_PS, &wcid->flags))
+			continue;
+
+		phy = mt76_dev_phy(dev, wcid->phy_idx);
+		if (test_bit(MT76_RESET, &phy->state) || phy->offchannel)
 			continue;
 
+		q = phy->q_tx[qid];
+		if (dev->queue_ops->tx_cleanup &&
+		    q->queued + 2 * MT_TXQ_FREE_THR >= q->ndesc) {
+			dev->queue_ops->tx_cleanup(dev, q, false);
+		}
+
 		if (mtxq->send_bar && mtxq->aggr) {
 			struct ieee80211_txq *txq = mtxq_to_txq(mtxq);
 			struct ieee80211_sta *sta = txq->sta;
@@ -518,32 +560,31 @@ mt76_txq_schedule_list(struct mt76_phy *phy, enum mt76_txq_id qid)
 			u8 tid = txq->tid;
 
 			mtxq->send_bar = false;
-			spin_unlock_bh(&hwq->lock);
 			ieee80211_send_bar(vif, sta->addr, tid, agg_ssn);
-			spin_lock_bh(&hwq->lock);
 		}
 
-		ret += mt76_txq_send_burst(phy, sq, mtxq);
-		ieee80211_return_txq(phy->hw, txq,
-				     !skb_queue_empty(&mtxq->retry_q));
+		if (!mt76_txq_stopped(q))
+			n_frames = mt76_txq_send_burst(phy, q, mtxq, wcid);
+
+		ieee80211_return_txq(phy->hw, txq, false);
+
+		if (unlikely(n_frames < 0))
+			return n_frames;
+
+		ret += n_frames;
 	}
-	spin_unlock_bh(&hwq->lock);
 
 	return ret;
 }
 
 void mt76_txq_schedule(struct mt76_phy *phy, enum mt76_txq_id qid)
 {
-	struct mt76_dev *dev = phy->dev;
-	struct mt76_sw_queue *sq = &dev->q_tx[qid];
 	int len;
 
 	if (qid >= 4)
 		return;
 
-	if (sq->swq_queued >= 4)
-		return;
-
+	local_bh_disable();
 	rcu_read_lock();
 
 	do {
@@ -553,28 +594,142 @@ void mt76_txq_schedule(struct mt76_phy *phy, enum mt76_txq_id qid)
 	} while (len > 0);
 
 	rcu_read_unlock();
+	local_bh_enable();
 }
 EXPORT_SYMBOL_GPL(mt76_txq_schedule);
 
+static int
+mt76_txq_schedule_pending_wcid(struct mt76_phy *phy, struct mt76_wcid *wcid,
+			       struct sk_buff_head *head)
+{
+	struct mt76_dev *dev = phy->dev;
+	struct ieee80211_sta *sta;
+	struct mt76_queue *q;
+	struct sk_buff *skb;
+	int ret = 0;
+
+	spin_lock(&head->lock);
+	while ((skb = skb_peek(head)) != NULL) {
+		struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)skb->data;
+		struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
+		int qid = skb_get_queue_mapping(skb);
+
+		if ((dev->drv->drv_flags & MT_DRV_HW_MGMT_TXQ) &&
+		    !(info->flags & IEEE80211_TX_CTL_HW_80211_ENCAP) &&
+		    !ieee80211_is_data(hdr->frame_control) &&
+		    (!ieee80211_is_bufferable_mmpdu(skb) ||
+		     ieee80211_is_deauth(hdr->frame_control) ||
+		     head == &wcid->tx_offchannel))
+			qid = MT_TXQ_PSD;
+
+		q = phy->q_tx[qid];
+		if (mt76_txq_stopped(q) || test_bit(MT76_RESET, &phy->state)) {
+			ret = -1;
+			break;
+		}
+
+		__skb_unlink(skb, head);
+		spin_unlock(&head->lock);
+
+		sta = wcid_to_sta(wcid);
+		spin_lock(&q->lock);
+		__mt76_tx_queue_skb(phy, qid, skb, wcid, sta, NULL);
+		dev->queue_ops->kick(dev, q);
+		spin_unlock(&q->lock);
+
+		spin_lock(&head->lock);
+	}
+	spin_unlock(&head->lock);
+
+	return ret;
+}
+
+static void mt76_txq_schedule_pending(struct mt76_phy *phy)
+{
+	LIST_HEAD(tx_list);
+	int ret = 0;
+
+	if (list_empty(&phy->tx_list))
+		return;
+
+	local_bh_disable();
+	rcu_read_lock();
+
+	spin_lock(&phy->tx_lock);
+	list_splice_init(&phy->tx_list, &tx_list);
+	while (!list_empty(&tx_list)) {
+		struct mt76_wcid *wcid;
+
+		wcid = list_first_entry(&tx_list, struct mt76_wcid, tx_list);
+		list_del_init(&wcid->tx_list);
+
+		spin_unlock(&phy->tx_lock);
+		if (ret >= 0)
+			ret = mt76_txq_schedule_pending_wcid(phy, wcid, &wcid->tx_offchannel);
+		if (ret >= 0 && !phy->offchannel)
+			ret = mt76_txq_schedule_pending_wcid(phy, wcid, &wcid->tx_pending);
+		spin_lock(&phy->tx_lock);
+
+		if (!skb_queue_empty(&wcid->tx_pending) &&
+		    !skb_queue_empty(&wcid->tx_offchannel) &&
+		    list_empty(&wcid->tx_list))
+			list_add_tail(&wcid->tx_list, &phy->tx_list);
+	}
+	spin_unlock(&phy->tx_lock);
+
+	rcu_read_unlock();
+	local_bh_enable();
+}
+
 void mt76_txq_schedule_all(struct mt76_phy *phy)
 {
+	struct mt76_phy *main_phy = &phy->dev->phy;
 	int i;
 
+	mt76_txq_schedule_pending(phy);
+
+	if (phy != main_phy && phy->hw == main_phy->hw)
+		return;
+
 	for (i = 0; i <= MT_TXQ_BK; i++)
 		mt76_txq_schedule(phy, i);
 }
 EXPORT_SYMBOL_GPL(mt76_txq_schedule_all);
 
-void mt76_tx_tasklet(unsigned long data)
+void mt76_tx_worker_run(struct mt76_dev *dev)
 {
-	struct mt76_dev *dev = (struct mt76_dev *)data;
+	struct mt76_phy *phy;
+	int i;
 
 	mt76_txq_schedule_all(&dev->phy);
-	if (dev->phy2)
-		mt76_txq_schedule_all(dev->phy2);
+	for (i = 0; i < ARRAY_SIZE(dev->phys); i++) {
+		phy = dev->phys[i];
+		if (!phy)
+			continue;
+
+		mt76_txq_schedule_all(phy);
+	}
+
+#ifdef CONFIG_NL80211_TESTMODE
+	for (i = 0; i < ARRAY_SIZE(dev->phys); i++) {
+		phy = dev->phys[i];
+		if (!phy || !phy->test.tx_pending)
+			continue;
+
+		mt76_testmode_tx_pending(phy);
+	}
+#endif
 }
+EXPORT_SYMBOL_GPL(mt76_tx_worker_run);
 
-void mt76_stop_tx_queues(struct mt76_dev *dev, struct ieee80211_sta *sta,
+void mt76_tx_worker(struct mt76_worker *w)
+{
+	struct mt76_dev *dev = container_of(w, struct mt76_dev, tx_worker);
+
+	mt76_tx_worker_run(dev);
+}
+
+void mt76_stop_tx_queues(struct mt76_phy *phy, struct ieee80211_sta *sta,
 			 bool send_bar)
 {
 	int i;
@@ -587,8 +742,8 @@ void mt76_stop_tx_queues(struct mt76_dev *dev, struct ieee80211_sta *sta,
 		if (!txq)
 			continue;
 
+		hwq = phy->q_tx[mt76_txq_get_qid(txq)];
 		mtxq = (struct mt76_txq *)txq->drv_priv;
-		hwq = mtxq->swq->q;
 
 		spin_lock_bh(&hwq->lock);
 		mtxq->send_bar = mtxq->aggr && send_bar;
@@ -602,53 +757,173 @@ void mt76_wake_tx_queue(struct ieee80211_hw *hw, struct ieee80211_txq *txq)
 	struct mt76_phy *phy = hw->priv;
 	struct mt76_dev *dev = phy->dev;
 
-	if (!test_bit(MT76_STATE_RUNNING, &phy->state))
-		return;
-
-	tasklet_schedule(&dev->tx_tasklet);
+	mt76_worker_schedule(&dev->tx_worker);
 }
 EXPORT_SYMBOL_GPL(mt76_wake_tx_queue);
 
-void mt76_txq_remove(struct mt76_dev *dev, struct ieee80211_txq *txq)
+u8 mt76_ac_to_hwq(u8 ac)
 {
-	struct ieee80211_hw *hw;
-	struct mt76_txq *mtxq;
-	struct sk_buff *skb;
+	static const u8 wmm_queue_map[] = {
+		[IEEE80211_AC_BE] = 0,
+		[IEEE80211_AC_BK] = 1,
+		[IEEE80211_AC_VI] = 2,
+		[IEEE80211_AC_VO] = 3,
+	};
+
+	if (WARN_ON(ac >= IEEE80211_NUM_ACS))
+		return 0;
+
+	return wmm_queue_map[ac];
+}
+EXPORT_SYMBOL_GPL(mt76_ac_to_hwq);
 
-	if (!txq)
+int mt76_skb_adjust_pad(struct sk_buff *skb, int pad)
+{
+	struct sk_buff *iter, *last = skb;
+
+	/* First packet of a A-MSDU burst keeps track of the whole burst
+	 * length, need to update length of it and the last packet.
+	 */
+	skb_walk_frags(skb, iter) {
+		last = iter;
+		if (!iter->next) {
+			skb->data_len += pad;
+			skb->len += pad;
+			break;
+		}
+	}
+
+	if (skb_pad(last, pad))
+		return -ENOMEM;
+
+	__skb_put(last, pad);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mt76_skb_adjust_pad);
+
+void mt76_queue_tx_complete(struct mt76_dev *dev, struct mt76_queue *q,
+			    struct mt76_queue_entry *e)
+{
+	if (e->skb)
+		dev->drv->tx_complete_skb(dev, e);
+
+	spin_lock_bh(&q->lock);
+	q->tail = (q->tail + 1) % q->ndesc;
+	q->queued--;
+	spin_unlock_bh(&q->lock);
+}
+EXPORT_SYMBOL_GPL(mt76_queue_tx_complete);
+
+void __mt76_set_tx_blocked(struct mt76_dev *dev, bool blocked)
+{
+	struct mt76_phy *phy = &dev->phy;
+	struct mt76_queue *q = phy->q_tx[0];
+
+	if (blocked == q->blocked)
 		return;
 
-	mtxq = (struct mt76_txq *)txq->drv_priv;
+	q->blocked = blocked;
 
-	while ((skb = skb_dequeue(&mtxq->retry_q)) != NULL) {
-		hw = mt76_tx_status_get_hw(dev, skb);
-		ieee80211_free_txskb(hw, skb);
+	phy = dev->phys[MT_BAND1];
+	if (phy) {
+		q = phy->q_tx[0];
+		q->blocked = blocked;
 	}
+	phy = dev->phys[MT_BAND2];
+	if (phy) {
+		q = phy->q_tx[0];
+		q->blocked = blocked;
+	}
+
+	if (!blocked)
+		mt76_worker_schedule(&dev->tx_worker);
 }
-EXPORT_SYMBOL_GPL(mt76_txq_remove);
+EXPORT_SYMBOL_GPL(__mt76_set_tx_blocked);
 
-void mt76_txq_init(struct mt76_dev *dev, struct ieee80211_txq *txq)
+int mt76_token_consume(struct mt76_dev *dev, struct mt76_txwi_cache **ptxwi)
 {
-	struct mt76_txq *mtxq = (struct mt76_txq *)txq->drv_priv;
+	int token;
 
-	skb_queue_head_init(&mtxq->retry_q);
+	spin_lock_bh(&dev->token_lock);
 
-	mtxq->swq = &dev->q_tx[mt76_txq_get_qid(txq)];
+	token = idr_alloc(&dev->token, *ptxwi, dev->token_start,
+			  dev->token_start + dev->token_size,
+			  GFP_ATOMIC);
+	if (token >= dev->token_start)
+		dev->token_count++;
+
+#ifdef CONFIG_NET_MEDIATEK_SOC_WED
+	if (mtk_wed_device_active(&dev->mmio.wed) &&
+	    token >= dev->mmio.wed.wlan.token_start)
+		dev->wed_token_count++;
+#endif
+
+	if (dev->token_count >= dev->token_size - MT76_TOKEN_FREE_THR)
+		__mt76_set_tx_blocked(dev, true);
+
+	spin_unlock_bh(&dev->token_lock);
+
+	return token;
 }
-EXPORT_SYMBOL_GPL(mt76_txq_init);
+EXPORT_SYMBOL_GPL(mt76_token_consume);
 
-u8 mt76_ac_to_hwq(u8 ac)
+int mt76_rx_token_consume(struct mt76_dev *dev, void *ptr,
+			  struct mt76_txwi_cache *t, dma_addr_t phys)
 {
-	static const u8 wmm_queue_map[] = {
-		[IEEE80211_AC_BE] = 0,
-		[IEEE80211_AC_BK] = 1,
-		[IEEE80211_AC_VI] = 2,
-		[IEEE80211_AC_VO] = 3,
-	};
+	int token;
+
+	spin_lock_bh(&dev->rx_token_lock);
+	token = idr_alloc(&dev->rx_token, t, 0, dev->rx_token_size,
+			  GFP_ATOMIC);
+	if (token >= 0) {
+		t->ptr = ptr;
+		t->dma_addr = phys;
+	}
+	spin_unlock_bh(&dev->rx_token_lock);
 
-	if (WARN_ON(ac >= IEEE80211_NUM_ACS))
-		return 0;
+	return token;
+}
+EXPORT_SYMBOL_GPL(mt76_rx_token_consume);
 
-	return wmm_queue_map[ac];
+struct mt76_txwi_cache *
+mt76_token_release(struct mt76_dev *dev, int token, bool *wake)
+{
+	struct mt76_txwi_cache *txwi;
+
+	spin_lock_bh(&dev->token_lock);
+
+	txwi = idr_remove(&dev->token, token);
+	if (txwi) {
+		dev->token_count--;
+
+#ifdef CONFIG_NET_MEDIATEK_SOC_WED
+		if (mtk_wed_device_active(&dev->mmio.wed) &&
+		    token >= dev->mmio.wed.wlan.token_start &&
+		    --dev->wed_token_count == 0)
+			wake_up(&dev->tx_wait);
+#endif
+	}
+
+	if (dev->token_count < dev->token_size - MT76_TOKEN_FREE_THR &&
+	    dev->phy.q_tx[0]->blocked)
+		*wake = true;
+
+	spin_unlock_bh(&dev->token_lock);
+
+	return txwi;
 }
-EXPORT_SYMBOL_GPL(mt76_ac_to_hwq);
+EXPORT_SYMBOL_GPL(mt76_token_release);
+
+struct mt76_txwi_cache *
+mt76_rx_token_release(struct mt76_dev *dev, int token)
+{
+	struct mt76_txwi_cache *t;
+
+	spin_lock_bh(&dev->rx_token_lock);
+	t = idr_remove(&dev->rx_token, token);
+	spin_unlock_bh(&dev->rx_token_lock);
+
+	return t;
+}
+EXPORT_SYMBOL_GPL(mt76_rx_token_release);