Merge branch 'for-linville' of git://github.com/kvalo/ath

13 files changed, 638 insertions, 657 deletions

diff --git a/drivers/net/wireless/ath/ath10k/ce.c b/drivers/net/wireless/ath/ath10k/ce.c
index d44d618b05f9..a79499c82350 100644
--- a/drivers/net/wireless/ath/ath10k/ce.c
+++ b/drivers/net/wireless/ath/ath10k/ce.c
@@ -266,12 +266,12 @@ static inline void ath10k_ce_engine_int_status_clear(struct ath10k *ar,
  * ath10k_ce_sendlist_send.
  * The caller takes responsibility for any needed locking.
  */
-static int ath10k_ce_send_nolock(struct ath10k_ce_pipe *ce_state,
-				 void *per_transfer_context,
-				 u32 buffer,
-				 unsigned int nbytes,
-				 unsigned int transfer_id,
-				 unsigned int flags)
+int ath10k_ce_send_nolock(struct ath10k_ce_pipe *ce_state,
+			  void *per_transfer_context,
+			  u32 buffer,
+			  unsigned int nbytes,
+			  unsigned int transfer_id,
+			  unsigned int flags)
 {
 	struct ath10k *ar = ce_state->ar;
 	struct ath10k_ce_ring *src_ring = ce_state->src_ring;
@@ -1067,9 +1067,9 @@ struct ath10k_ce_pipe *ath10k_ce_init(struct ath10k *ar,
 	 *
 	 * For the lack of a better place do the check here.
 	 */
-	BUILD_BUG_ON(TARGET_NUM_MSDU_DESC >
+	BUILD_BUG_ON(2*TARGET_NUM_MSDU_DESC >
 		     (CE_HTT_H2T_MSG_SRC_NENTRIES - 1));
-	BUILD_BUG_ON(TARGET_10X_NUM_MSDU_DESC >
+	BUILD_BUG_ON(2*TARGET_10X_NUM_MSDU_DESC >
 		     (CE_HTT_H2T_MSG_SRC_NENTRIES - 1));
 
 	ret = ath10k_pci_wake(ar);
diff --git a/drivers/net/wireless/ath/ath10k/ce.h b/drivers/net/wireless/ath/ath10k/ce.h
index 67dbde6a5c74..8eb7f99ed992 100644
--- a/drivers/net/wireless/ath/ath10k/ce.h
+++ b/drivers/net/wireless/ath/ath10k/ce.h
@@ -23,7 +23,7 @@
 
 /* Maximum number of Copy Engine's supported */
 #define CE_COUNT_MAX 8
-#define CE_HTT_H2T_MSG_SRC_NENTRIES 2048
+#define CE_HTT_H2T_MSG_SRC_NENTRIES 4096
 
 /* Descriptor rings must be aligned to this boundary */
 #define CE_DESC_RING_ALIGN	8
@@ -152,6 +152,13 @@ int ath10k_ce_send(struct ath10k_ce_pipe *ce_state,
 		   unsigned int transfer_id,
 		   unsigned int flags);
 
+int ath10k_ce_send_nolock(struct ath10k_ce_pipe *ce_state,
+			  void *per_transfer_context,
+			  u32 buffer,
+			  unsigned int nbytes,
+			  unsigned int transfer_id,
+			  unsigned int flags);
+
 void ath10k_ce_send_cb_register(struct ath10k_ce_pipe *ce_state,
 				void (*send_cb)(struct ath10k_ce_pipe *),
 				int disable_interrupts);
diff --git a/drivers/net/wireless/ath/ath10k/core.h b/drivers/net/wireless/ath/ath10k/core.h
index 1fc26fe057e8..0e71979d837c 100644
--- a/drivers/net/wireless/ath/ath10k/core.h
+++ b/drivers/net/wireless/ath/ath10k/core.h
@@ -62,16 +62,13 @@ struct ath10k;
 
 struct ath10k_skb_cb {
 	dma_addr_t paddr;
-	bool is_mapped;
-	bool is_aborted;
 	u8 vdev_id;
 
 	struct {
 		u8 tid;
 		bool is_offchan;
-
-		u8 frag_len;
-		u8 pad_len;
+		struct ath10k_htt_txbuf *txbuf;
+		u32 txbuf_paddr;
 	} __packed htt;
 
 	struct {
@@ -87,32 +84,6 @@ static inline struct ath10k_skb_cb *ATH10K_SKB_CB(struct sk_buff *skb)
 	return (struct ath10k_skb_cb *)&IEEE80211_SKB_CB(skb)->driver_data;
 }
 
-static inline int ath10k_skb_map(struct device *dev, struct sk_buff *skb)
-{
-	if (ATH10K_SKB_CB(skb)->is_mapped)
-		return -EINVAL;
-
-	ATH10K_SKB_CB(skb)->paddr = dma_map_single(dev, skb->data, skb->len,
-						   DMA_TO_DEVICE);
-
-	if (unlikely(dma_mapping_error(dev, ATH10K_SKB_CB(skb)->paddr)))
-		return -EIO;
-
-	ATH10K_SKB_CB(skb)->is_mapped = true;
-	return 0;
-}
-
-static inline int ath10k_skb_unmap(struct device *dev, struct sk_buff *skb)
-{
-	if (!ATH10K_SKB_CB(skb)->is_mapped)
-		return -EINVAL;
-
-	dma_unmap_single(dev, ATH10K_SKB_CB(skb)->paddr, skb->len,
-			 DMA_TO_DEVICE);
-	ATH10K_SKB_CB(skb)->is_mapped = false;
-	return 0;
-}
-
 static inline u32 host_interest_item_address(u32 item_offset)
 {
 	return QCA988X_HOST_INTEREST_ADDRESS + item_offset;
@@ -288,6 +259,7 @@ struct ath10k_vif {
 
 	u8 fixed_rate;
 	u8 fixed_nss;
+	u8 force_sgi;
 };
 
 struct ath10k_vif_iter {
diff --git a/drivers/net/wireless/ath/ath10k/hif.h b/drivers/net/wireless/ath/ath10k/hif.h
index dcdea68bcc0a..2ac7beacddca 100644
--- a/drivers/net/wireless/ath/ath10k/hif.h
+++ b/drivers/net/wireless/ath/ath10k/hif.h
@@ -21,6 +21,14 @@
 #include <linux/kernel.h>
 #include "core.h"
 
+struct ath10k_hif_sg_item {
+	u16 transfer_id;
+	void *transfer_context; /* NULL = tx completion callback not called */
+	void *vaddr; /* for debugging mostly */
+	u32 paddr;
+	u16 len;
+};
+
 struct ath10k_hif_cb {
 	int (*tx_completion)(struct ath10k *ar,
 			     struct sk_buff *wbuf,
@@ -31,11 +39,9 @@ struct ath10k_hif_cb {
 };
 
 struct ath10k_hif_ops {
-	/* Send the head of a buffer to HIF for transmission to the target. */
-	int (*send_head)(struct ath10k *ar, u8 pipe_id,
-			 unsigned int transfer_id,
-			 unsigned int nbytes,
-			 struct sk_buff *buf);
+	/* send a scatter-gather list to the target */
+	int (*tx_sg)(struct ath10k *ar, u8 pipe_id,
+		     struct ath10k_hif_sg_item *items, int n_items);
 
 	/*
 	 * API to handle HIF-specific BMI message exchanges, this API is
@@ -86,12 +92,11 @@ struct ath10k_hif_ops {
 };
 
 
-static inline int ath10k_hif_send_head(struct ath10k *ar, u8 pipe_id,
-				       unsigned int transfer_id,
-				       unsigned int nbytes,
-				       struct sk_buff *buf)
+static inline int ath10k_hif_tx_sg(struct ath10k *ar, u8 pipe_id,
+				   struct ath10k_hif_sg_item *items,
+				   int n_items)
 {
-	return ar->hif.ops->send_head(ar, pipe_id, transfer_id, nbytes, buf);
+	return ar->hif.ops->tx_sg(ar, pipe_id, items, n_items);
 }
 
 static inline int ath10k_hif_exchange_bmi_msg(struct ath10k *ar,
diff --git a/drivers/net/wireless/ath/ath10k/htc.c b/drivers/net/wireless/ath/ath10k/htc.c
index edc57ab505c8..7f1bccd3597f 100644
--- a/drivers/net/wireless/ath/ath10k/htc.c
+++ b/drivers/net/wireless/ath/ath10k/htc.c
@@ -63,7 +63,9 @@ static struct sk_buff *ath10k_htc_build_tx_ctrl_skb(void *ar)
 static inline void ath10k_htc_restore_tx_skb(struct ath10k_htc *htc,
 					     struct sk_buff *skb)
 {
-	ath10k_skb_unmap(htc->ar->dev, skb);
+	struct ath10k_skb_cb *skb_cb = ATH10K_SKB_CB(skb);
+
+	dma_unmap_single(htc->ar->dev, skb_cb->paddr, skb->len, DMA_TO_DEVICE);
 	skb_pull(skb, sizeof(struct ath10k_htc_hdr));
 }
 
@@ -122,6 +124,9 @@ int ath10k_htc_send(struct ath10k_htc *htc,
 		    struct sk_buff *skb)
 {
 	struct ath10k_htc_ep *ep = &htc->endpoint[eid];
+	struct ath10k_skb_cb *skb_cb = ATH10K_SKB_CB(skb);
+	struct ath10k_hif_sg_item sg_item;
+	struct device *dev = htc->ar->dev;
 	int credits = 0;
 	int ret;
 
@@ -157,19 +162,25 @@ int ath10k_htc_send(struct ath10k_htc *htc,
 
 	ath10k_htc_prepare_tx_skb(ep, skb);
 
-	ret = ath10k_skb_map(htc->ar->dev, skb);
+	skb_cb->paddr = dma_map_single(dev, skb->data, skb->len, DMA_TO_DEVICE);
+	ret = dma_mapping_error(dev, skb_cb->paddr);
 	if (ret)
 		goto err_credits;
 
-	ret = ath10k_hif_send_head(htc->ar, ep->ul_pipe_id, ep->eid,
-				   skb->len, skb);
+	sg_item.transfer_id = ep->eid;
+	sg_item.transfer_context = skb;
+	sg_item.vaddr = skb->data;
+	sg_item.paddr = skb_cb->paddr;
+	sg_item.len = skb->len;
+
+	ret = ath10k_hif_tx_sg(htc->ar, ep->ul_pipe_id, &sg_item, 1);
 	if (ret)
 		goto err_unmap;
 
 	return 0;
 
 err_unmap:
-	ath10k_skb_unmap(htc->ar->dev, skb);
+	dma_unmap_single(dev, skb_cb->paddr, skb->len, DMA_TO_DEVICE);
 err_credits:
 	if (ep->tx_credit_flow_enabled) {
 		spin_lock_bh(&htc->tx_lock);
@@ -191,10 +202,8 @@ static int ath10k_htc_tx_completion_handler(struct ath10k *ar,
 	struct ath10k_htc *htc = &ar->htc;
 	struct ath10k_htc_ep *ep = &htc->endpoint[eid];
 
-	if (!skb) {
-		ath10k_warn("invalid sk_buff completion - NULL pointer. firmware crashed?\n");
+	if (WARN_ON_ONCE(!skb))
 		return 0;
-	}
 
 	ath10k_htc_notify_tx_completion(ep, skb);
 	/* the skb now belongs to the completion handler */
diff --git a/drivers/net/wireless/ath/ath10k/htt.h b/drivers/net/wireless/ath/ath10k/htt.h
index b93ae355bc08..654867fc1ae7 100644
--- a/drivers/net/wireless/ath/ath10k/htt.h
+++ b/drivers/net/wireless/ath/ath10k/htt.h
@@ -20,6 +20,7 @@
 
 #include <linux/bug.h>
 #include <linux/interrupt.h>
+#include <linux/dmapool.h>
 
 #include "htc.h"
 #include "rx_desc.h"
@@ -1181,11 +1182,20 @@ struct htt_rx_info {
 		u32 info1;
 		u32 info2;
 	} rate;
+
+	u32 tsf;
 	bool fcs_err;
 	bool amsdu_more;
 	bool mic_err;
 };
 
+struct ath10k_htt_txbuf {
+	struct htt_data_tx_desc_frag frags[2];
+	struct ath10k_htc_hdr htc_hdr;
+	struct htt_cmd_hdr cmd_hdr;
+	struct htt_data_tx_desc cmd_tx;
+} __packed;
+
 struct ath10k_htt {
 	struct ath10k *ar;
 	enum ath10k_htc_ep_id eid;
@@ -1267,11 +1277,18 @@ struct ath10k_htt {
 	struct sk_buff **pending_tx;
 	unsigned long *used_msdu_ids; /* bitmap */
 	wait_queue_head_t empty_tx_wq;
+	struct dma_pool *tx_pool;
 
 	/* set if host-fw communication goes haywire
 	 * used to avoid further failures */
 	bool rx_confused;
 	struct tasklet_struct rx_replenish_task;
+
+	/* This is used to group tx/rx completions separately and process them
+	 * in batches to reduce cache stalls */
+	struct tasklet_struct txrx_compl_task;
+	struct sk_buff_head tx_compl_q;
+	struct sk_buff_head rx_compl_q;
 };
 
 #define RX_HTT_HDR_STATUS_LEN 64
@@ -1343,4 +1360,5 @@ int ath10k_htt_tx_alloc_msdu_id(struct ath10k_htt *htt);
 void ath10k_htt_tx_free_msdu_id(struct ath10k_htt *htt, u16 msdu_id);
 int ath10k_htt_mgmt_tx(struct ath10k_htt *htt, struct sk_buff *);
 int ath10k_htt_tx(struct ath10k_htt *htt, struct sk_buff *);
+
 #endif
diff --git a/drivers/net/wireless/ath/ath10k/htt_rx.c b/drivers/net/wireless/ath/ath10k/htt_rx.c
index 4767c24bf819..cdcbe2de95f9 100644
--- a/drivers/net/wireless/ath/ath10k/htt_rx.c
+++ b/drivers/net/wireless/ath/ath10k/htt_rx.c
@@ -43,7 +43,7 @@
 
 
 static int ath10k_htt_rx_get_csum_state(struct sk_buff *skb);
-
+static void ath10k_htt_txrx_compl_task(unsigned long ptr);
 
 static int ath10k_htt_rx_ring_size(struct ath10k_htt *htt)
 {
@@ -225,18 +225,16 @@ static void ath10k_htt_rx_ring_refill_retry(unsigned long arg)
 	ath10k_htt_rx_msdu_buff_replenish(htt);
 }
 
-static unsigned ath10k_htt_rx_ring_elems(struct ath10k_htt *htt)
-{
-	return (__le32_to_cpu(*htt->rx_ring.alloc_idx.vaddr) -
-		htt->rx_ring.sw_rd_idx.msdu_payld) & htt->rx_ring.size_mask;
-}
-
 void ath10k_htt_rx_detach(struct ath10k_htt *htt)
 {
 	int sw_rd_idx = htt->rx_ring.sw_rd_idx.msdu_payld;
 
 	del_timer_sync(&htt->rx_ring.refill_retry_timer);
 	tasklet_kill(&htt->rx_replenish_task);
+	tasklet_kill(&htt->txrx_compl_task);
+
+	skb_queue_purge(&htt->tx_compl_q);
+	skb_queue_purge(&htt->rx_compl_q);
 
 	while (sw_rd_idx != __le32_to_cpu(*(htt->rx_ring.alloc_idx.vaddr))) {
 		struct sk_buff *skb =
@@ -270,10 +268,12 @@ static inline struct sk_buff *ath10k_htt_rx_netbuf_pop(struct ath10k_htt *htt)
 	int idx;
 	struct sk_buff *msdu;
 
-	spin_lock_bh(&htt->rx_ring.lock);
+	lockdep_assert_held(&htt->rx_ring.lock);
 
-	if (ath10k_htt_rx_ring_elems(htt) == 0)
-		ath10k_warn("htt rx ring is empty!\n");
+	if (htt->rx_ring.fill_cnt == 0) {
+		ath10k_warn("tried to pop sk_buff from an empty rx ring\n");
+		return NULL;
+	}
 
 	idx = htt->rx_ring.sw_rd_idx.msdu_payld;
 	msdu = htt->rx_ring.netbufs_ring[idx];
@@ -283,7 +283,6 @@ static inline struct sk_buff *ath10k_htt_rx_netbuf_pop(struct ath10k_htt *htt)
 	htt->rx_ring.sw_rd_idx.msdu_payld = idx;
 	htt->rx_ring.fill_cnt--;
 
-	spin_unlock_bh(&htt->rx_ring.lock);
 	return msdu;
 }
 
@@ -307,8 +306,7 @@ static int ath10k_htt_rx_amsdu_pop(struct ath10k_htt *htt,
 	struct sk_buff *msdu;
 	struct htt_rx_desc *rx_desc;
 
-	if (ath10k_htt_rx_ring_elems(htt) == 0)
-		ath10k_warn("htt rx ring is empty!\n");
+	lockdep_assert_held(&htt->rx_ring.lock);
 
 	if (htt->rx_confused) {
 		ath10k_warn("htt is confused. refusing rx\n");
@@ -400,6 +398,7 @@ static int ath10k_htt_rx_amsdu_pop(struct ath10k_htt *htt,
 		msdu_len = MS(__le32_to_cpu(rx_desc->msdu_start.info0),
 			      RX_MSDU_START_INFO0_MSDU_LENGTH);
 		msdu_chained = rx_desc->frag_info.ring2_more_count;
+		msdu_chaining = msdu_chained;
 
 		if (msdu_len_invalid)
 			msdu_len = 0;
@@ -427,7 +426,6 @@ static int ath10k_htt_rx_amsdu_pop(struct ath10k_htt *htt,
 
 			msdu->next = next;
 			msdu = next;
-			msdu_chaining = 1;
 		}
 
 		last_msdu = __le32_to_cpu(rx_desc->msdu_end.info0) &
@@ -529,6 +527,12 @@ int ath10k_htt_rx_attach(struct ath10k_htt *htt)
 	tasklet_init(&htt->rx_replenish_task, ath10k_htt_rx_replenish_task,
 		     (unsigned long)htt);
 
+	skb_queue_head_init(&htt->tx_compl_q);
+	skb_queue_head_init(&htt->rx_compl_q);
+
+	tasklet_init(&htt->txrx_compl_task, ath10k_htt_txrx_compl_task,
+		     (unsigned long)htt);
+
 	ath10k_dbg(ATH10K_DBG_BOOT, "htt rx ring size %d fill_level %d\n",
 		   htt->rx_ring.size, htt->rx_ring.fill_level);
 	return 0;
@@ -632,6 +636,12 @@ struct amsdu_subframe_hdr {
 	__be16 len;
 } __packed;
 
+static int ath10k_htt_rx_nwifi_hdrlen(struct ieee80211_hdr *hdr)
+{
+	/* nwifi header is padded to 4 bytes. this fixes 4addr rx */
+	return round_up(ieee80211_hdrlen(hdr->frame_control), 4);
+}
+
 static void ath10k_htt_rx_amsdu(struct ath10k_htt *htt,
 				struct htt_rx_info *info)
 {
@@ -681,7 +691,7 @@ static void ath10k_htt_rx_amsdu(struct ath10k_htt *htt,
 		case RX_MSDU_DECAP_NATIVE_WIFI:
 			/* pull decapped header and copy DA */
 			hdr = (struct ieee80211_hdr *)skb->data;
-			hdr_len = ieee80211_hdrlen(hdr->frame_control);
+			hdr_len = ath10k_htt_rx_nwifi_hdrlen(hdr);
 			memcpy(addr, ieee80211_get_DA(hdr), ETH_ALEN);
 			skb_pull(skb, hdr_len);
 
@@ -768,7 +778,7 @@ static void ath10k_htt_rx_msdu(struct ath10k_htt *htt, struct htt_rx_info *info)
 	case RX_MSDU_DECAP_NATIVE_WIFI:
 		/* Pull decapped header */
 		hdr = (struct ieee80211_hdr *)skb->data;
-		hdr_len = ieee80211_hdrlen(hdr->frame_control);
+		hdr_len = ath10k_htt_rx_nwifi_hdrlen(hdr);
 		skb_pull(skb, hdr_len);
 
 		/* Push original header */
@@ -846,6 +856,20 @@ static bool ath10k_htt_rx_has_mic_err(struct sk_buff *skb)
 	return false;
 }
 
+static bool ath10k_htt_rx_is_mgmt(struct sk_buff *skb)
+{
+	struct htt_rx_desc *rxd;
+	u32 flags;
+
+	rxd = (void *)skb->data - sizeof(*rxd);
+	flags = __le32_to_cpu(rxd->attention.flags);
+
+	if (flags & RX_ATTENTION_FLAGS_MGMT_TYPE)
+		return true;
+
+	return false;
+}
+
 static int ath10k_htt_rx_get_csum_state(struct sk_buff *skb)
 {
 	struct htt_rx_desc *rxd;
@@ -877,6 +901,57 @@ static int ath10k_htt_rx_get_csum_state(struct sk_buff *skb)
 	return CHECKSUM_UNNECESSARY;
 }
 
+static int ath10k_unchain_msdu(struct sk_buff *msdu_head)
+{
+	struct sk_buff *next = msdu_head->next;
+	struct sk_buff *to_free = next;
+	int space;
+	int total_len = 0;
+
+	/* TODO:  Might could optimize this by using
+	 * skb_try_coalesce or similar method to
+	 * decrease copying, or maybe get mac80211 to
+	 * provide a way to just receive a list of
+	 * skb?
+	 */
+
+	msdu_head->next = NULL;
+
+	/* Allocate total length all at once. */
+	while (next) {
+		total_len += next->len;
+		next = next->next;
+	}
+
+	space = total_len - skb_tailroom(msdu_head);
+	if ((space > 0) &&
+	    (pskb_expand_head(msdu_head, 0, space, GFP_ATOMIC) < 0)) {
+		/* TODO:  bump some rx-oom error stat */
+		/* put it back together so we can free the
+		 * whole list at once.
+		 */
+		msdu_head->next = to_free;
+		return -1;
+	}
+
+	/* Walk list again, copying contents into
+	 * msdu_head
+	 */
+	next = to_free;
+	while (next) {
+		skb_copy_from_linear_data(next, skb_put(msdu_head, next->len),
+					  next->len);
+		next = next->next;
+	}
+
+	/* If here, we have consolidated skb.  Free the
+	 * fragments and pass the main skb on up the
+	 * stack.
+	 */
+	ath10k_htt_rx_free_msdu_chain(to_free);
+	return 0;
+}
+
 static void ath10k_htt_rx_handler(struct ath10k_htt *htt,
 				  struct htt_rx_indication *rx)
 {
@@ -888,6 +963,8 @@ static void ath10k_htt_rx_handler(struct ath10k_htt *htt,
 	u8 *fw_desc;
 	int i, j;
 
+	lockdep_assert_held(&htt->rx_ring.lock);
+
 	memset(&info, 0, sizeof(info));
 
 	fw_desc_len = __le16_to_cpu(rx->prefix.fw_rx_desc_bytes);
@@ -940,7 +1017,8 @@ static void ath10k_htt_rx_handler(struct ath10k_htt *htt,
 			status = info.status;
 
 			/* Skip mgmt frames while we handle this in WMI */
-			if (status == HTT_RX_IND_MPDU_STATUS_MGMT_CTRL) {
+			if (status == HTT_RX_IND_MPDU_STATUS_MGMT_CTRL ||
+			    ath10k_htt_rx_is_mgmt(msdu_head)) {
 				ath10k_dbg(ATH10K_DBG_HTT, "htt rx mgmt ctrl\n");
 				ath10k_htt_rx_free_msdu_chain(msdu_head);
 				continue;
@@ -964,10 +1042,8 @@ static void ath10k_htt_rx_handler(struct ath10k_htt *htt,
 				continue;
 			}
 
-			/* FIXME: we do not support chaining yet.
-			 * this needs investigation */
-			if (msdu_chaining) {
-				ath10k_warn("htt rx msdu_chaining is true\n");
+			if (msdu_chaining &&
+			    (ath10k_unchain_msdu(msdu_head) < 0)) {
 				ath10k_htt_rx_free_msdu_chain(msdu_head);
 				continue;
 			}
@@ -990,6 +1066,7 @@ static void ath10k_htt_rx_handler(struct ath10k_htt *htt,
 			info.rate.info0 = rx->ppdu.info0;
 			info.rate.info1 = __le32_to_cpu(rx->ppdu.info1);
 			info.rate.info2 = __le32_to_cpu(rx->ppdu.info2);
+			info.tsf = __le32_to_cpu(rx->ppdu.tsf);
 
 			hdr = ath10k_htt_rx_skb_get_hdr(msdu_head);
 
@@ -1023,8 +1100,11 @@ static void ath10k_htt_rx_frag_handler(struct ath10k_htt *htt,
 
 	msdu_head = NULL;
 	msdu_tail = NULL;
+
+	spin_lock_bh(&htt->rx_ring.lock);
 	msdu_chaining = ath10k_htt_rx_amsdu_pop(htt, &fw_desc, &fw_desc_len,
 						&msdu_head, &msdu_tail);
+	spin_unlock_bh(&htt->rx_ring.lock);
 
 	ath10k_dbg(ATH10K_DBG_HTT_DUMP, "htt rx frag ahead\n");
 
@@ -1116,6 +1196,45 @@ end:
 	}
 }
 
+static void ath10k_htt_rx_frm_tx_compl(struct ath10k *ar,
+				       struct sk_buff *skb)
+{
+	struct ath10k_htt *htt = &ar->htt;
+	struct htt_resp *resp = (struct htt_resp *)skb->data;
+	struct htt_tx_done tx_done = {};
+	int status = MS(resp->data_tx_completion.flags, HTT_DATA_TX_STATUS);
+	__le16 msdu_id;
+	int i;
+
+	lockdep_assert_held(&htt->tx_lock);
+
+	switch (status) {
+	case HTT_DATA_TX_STATUS_NO_ACK:
+		tx_done.no_ack = true;
+		break;
+	case HTT_DATA_TX_STATUS_OK:
+		break;
+	case HTT_DATA_TX_STATUS_DISCARD:
+	case HTT_DATA_TX_STATUS_POSTPONE:
+	case HTT_DATA_TX_STATUS_DOWNLOAD_FAIL:
+		tx_done.discard = true;
+		break;
+	default:
+		ath10k_warn("unhandled tx completion status %d\n", status);
+		tx_done.discard = true;
+		break;
+	}
+
+	ath10k_dbg(ATH10K_DBG_HTT, "htt tx completion num_msdus %d\n",
+		   resp->data_tx_completion.num_msdus);
+
+	for (i = 0; i < resp->data_tx_completion.num_msdus; i++) {
+		msdu_id = resp->data_tx_completion.msdus[i];
+		tx_done.msdu_id = __le16_to_cpu(msdu_id);
+		ath10k_txrx_tx_unref(htt, &tx_done);
+	}
+}
+
 void ath10k_htt_t2h_msg_handler(struct ath10k *ar, struct sk_buff *skb)
 {
 	struct ath10k_htt *htt = &ar->htt;
@@ -1134,10 +1253,12 @@ void ath10k_htt_t2h_msg_handler(struct ath10k *ar, struct sk_buff *skb)
 		complete(&htt->target_version_received);
 		break;
 	}
-	case HTT_T2H_MSG_TYPE_RX_IND: {
-		ath10k_htt_rx_handler(htt, &resp->rx_ind);
-		break;
-	}
+	case HTT_T2H_MSG_TYPE_RX_IND:
+		spin_lock_bh(&htt->rx_ring.lock);
+		__skb_queue_tail(&htt->rx_compl_q, skb);
+		spin_unlock_bh(&htt->rx_ring.lock);
+		tasklet_schedule(&htt->txrx_compl_task);
+		return;
 	case HTT_T2H_MSG_TYPE_PEER_MAP: {
 		struct htt_peer_map_event ev = {
 			.vdev_id = resp->peer_map.vdev_id,
@@ -1172,44 +1293,17 @@ void ath10k_htt_t2h_msg_handler(struct ath10k *ar, struct sk_buff *skb)
 			break;
 		}
 
+		spin_lock_bh(&htt->tx_lock);
 		ath10k_txrx_tx_unref(htt, &tx_done);
+		spin_unlock_bh(&htt->tx_lock);
 		break;
 	}
-	case HTT_T2H_MSG_TYPE_TX_COMPL_IND: {
-		struct htt_tx_done tx_done = {};
-		int status = MS(resp->data_tx_completion.flags,
-				HTT_DATA_TX_STATUS);
-		__le16 msdu_id;
-		int i;
-
-		switch (status) {
-		case HTT_DATA_TX_STATUS_NO_ACK:
-			tx_done.no_ack = true;
-			break;
-		case HTT_DATA_TX_STATUS_OK:
-			break;
-		case HTT_DATA_TX_STATUS_DISCARD:
-		case HTT_DATA_TX_STATUS_POSTPONE:
-		case HTT_DATA_TX_STATUS_DOWNLOAD_FAIL:
-			tx_done.discard = true;
-			break;
-		default:
-			ath10k_warn("unhandled tx completion status %d\n",
-				    status);
-			tx_done.discard = true;
-			break;
-		}
-
-		ath10k_dbg(ATH10K_DBG_HTT, "htt tx completion num_msdus %d\n",
-			   resp->data_tx_completion.num_msdus);
-
-		for (i = 0; i < resp->data_tx_completion.num_msdus; i++) {
-			msdu_id = resp->data_tx_completion.msdus[i];
-			tx_done.msdu_id = __le16_to_cpu(msdu_id);
-			ath10k_txrx_tx_unref(htt, &tx_done);
-		}
-		break;
-	}
+	case HTT_T2H_MSG_TYPE_TX_COMPL_IND:
+		spin_lock_bh(&htt->tx_lock);
+		__skb_queue_tail(&htt->tx_compl_q, skb);
+		spin_unlock_bh(&htt->tx_lock);
+		tasklet_schedule(&htt->txrx_compl_task);
+		return;
 	case HTT_T2H_MSG_TYPE_SEC_IND: {
 		struct ath10k *ar = htt->ar;
 		struct htt_security_indication *ev = &resp->security_indication;
@@ -1249,3 +1343,25 @@ void ath10k_htt_t2h_msg_handler(struct ath10k *ar, struct sk_buff *skb)
 	/* Free the indication buffer */
 	dev_kfree_skb_any(skb);
 }
+
+static void ath10k_htt_txrx_compl_task(unsigned long ptr)
+{
+	struct ath10k_htt *htt = (struct ath10k_htt *)ptr;
+	struct htt_resp *resp;
+	struct sk_buff *skb;
+
+	spin_lock_bh(&htt->tx_lock);
+	while ((skb = __skb_dequeue(&htt->tx_compl_q))) {
+		ath10k_htt_rx_frm_tx_compl(htt->ar, skb);
+		dev_kfree_skb_any(skb);
+	}
+	spin_unlock_bh(&htt->tx_lock);
+
+	spin_lock_bh(&htt->rx_ring.lock);
+	while ((skb = __skb_dequeue(&htt->rx_compl_q))) {
+		resp = (struct htt_resp *)skb->data;
+		ath10k_htt_rx_handler(htt, &resp->rx_ind);
+		dev_kfree_skb_any(skb);
+	}
+	spin_unlock_bh(&htt->rx_ring.lock);
+}
diff --git a/drivers/net/wireless/ath/ath10k/htt_tx.c b/drivers/net/wireless/ath/ath10k/htt_tx.c
index acaa046dc93b..7a3e2e40dd5c 100644
--- a/drivers/net/wireless/ath/ath10k/htt_tx.c
+++ b/drivers/net/wireless/ath/ath10k/htt_tx.c
@@ -109,6 +109,14 @@ int ath10k_htt_tx_attach(struct ath10k_htt *htt)
 		return -ENOMEM;
 	}
 
+	htt->tx_pool = dma_pool_create("ath10k htt tx pool", htt->ar->dev,
+				       sizeof(struct ath10k_htt_txbuf), 4, 0);
+	if (!htt->tx_pool) {
+		kfree(htt->used_msdu_ids);
+		kfree(htt->pending_tx);
+		return -ENOMEM;
+	}
+
 	return 0;
 }
 
@@ -117,9 +125,7 @@ static void ath10k_htt_tx_cleanup_pending(struct ath10k_htt *htt)
 	struct htt_tx_done tx_done = {0};
 	int msdu_id;
 
-	/* No locks needed. Called after communication with the device has
-	 * been stopped. */
-
+	spin_lock_bh(&htt->tx_lock);
 	for (msdu_id = 0; msdu_id < htt->max_num_pending_tx; msdu_id++) {
 		if (!test_bit(msdu_id, htt->used_msdu_ids))
 			continue;
@@ -132,6 +138,7 @@ static void ath10k_htt_tx_cleanup_pending(struct ath10k_htt *htt)
 
 		ath10k_txrx_tx_unref(htt, &tx_done);
 	}
+	spin_unlock_bh(&htt->tx_lock);
 }
 
 void ath10k_htt_tx_detach(struct ath10k_htt *htt)
@@ -139,6 +146,7 @@ void ath10k_htt_tx_detach(struct ath10k_htt *htt)
 	ath10k_htt_tx_cleanup_pending(htt);
 	kfree(htt->pending_tx);
 	kfree(htt->used_msdu_ids);
+	dma_pool_destroy(htt->tx_pool);
 	return;
 }
 
@@ -334,7 +342,9 @@ int ath10k_htt_mgmt_tx(struct ath10k_htt *htt, struct sk_buff *msdu)
 		goto err_free_msdu_id;
 	}
 
-	res = ath10k_skb_map(dev, msdu);
+	skb_cb->paddr = dma_map_single(dev, msdu->data, msdu->len,
+				       DMA_TO_DEVICE);
+	res = dma_mapping_error(dev, skb_cb->paddr);
 	if (res)
 		goto err_free_txdesc;
 
@@ -348,8 +358,7 @@ int ath10k_htt_mgmt_tx(struct ath10k_htt *htt, struct sk_buff *msdu)
 	memcpy(cmd->mgmt_tx.hdr, msdu->data,
 	       min_t(int, msdu->len, HTT_MGMT_FRM_HDR_DOWNLOAD_LEN));
 
-	skb_cb->htt.frag_len = 0;
-	skb_cb->htt.pad_len = 0;
+	skb_cb->htt.txbuf = NULL;
 
 	res = ath10k_htc_send(&htt->ar->htc, htt->eid, txdesc);
 	if (res)
@@ -358,7 +367,7 @@ int ath10k_htt_mgmt_tx(struct ath10k_htt *htt, struct sk_buff *msdu)
 	return 0;
 
 err_unmap_msdu:
-	ath10k_skb_unmap(dev, msdu);
+	dma_unmap_single(dev, skb_cb->paddr, msdu->len, DMA_TO_DEVICE);
 err_free_txdesc:
 	dev_kfree_skb_any(txdesc);
 err_free_msdu_id:
@@ -375,19 +384,19 @@ err:
 int ath10k_htt_tx(struct ath10k_htt *htt, struct sk_buff *msdu)
 {
 	struct device *dev = htt->ar->dev;
-	struct htt_cmd *cmd;
-	struct htt_data_tx_desc_frag *tx_frags;
 	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)msdu->data;
 	struct ath10k_skb_cb *skb_cb = ATH10K_SKB_CB(msdu);
-	struct sk_buff *txdesc = NULL;
-	bool use_frags;
-	u8 vdev_id = ATH10K_SKB_CB(msdu)->vdev_id;
-	u8 tid;
-	int prefetch_len, desc_len;
-	int msdu_id = -1;
+	struct ath10k_hif_sg_item sg_items[2];
+	struct htt_data_tx_desc_frag *frags;
+	u8 vdev_id = skb_cb->vdev_id;
+	u8 tid = skb_cb->htt.tid;
+	int prefetch_len;
 	int res;
-	u8 flags0;
-	u16 flags1;
+	u8 flags0 = 0;
+	u16 msdu_id, flags1 = 0;
+	dma_addr_t paddr;
+	u32 frags_paddr;
+	bool use_frags;
 
 	res = ath10k_htt_tx_inc_pending(htt);
 	if (res)
@@ -406,114 +415,120 @@ int ath10k_htt_tx(struct ath10k_htt *htt, struct sk_buff *msdu)
 	prefetch_len = min(htt->prefetch_len, msdu->len);
 	prefetch_len = roundup(prefetch_len, 4);
 
-	desc_len = sizeof(cmd->hdr) + sizeof(cmd->data_tx) + prefetch_len;
-
-	txdesc = ath10k_htc_alloc_skb(desc_len);
-	if (!txdesc) {
-		res = -ENOMEM;
-		goto err_free_msdu_id;
-	}
-
 	/* Since HTT 3.0 there is no separate mgmt tx command. However in case
 	 * of mgmt tx using TX_FRM there is not tx fragment list. Instead of tx
 	 * fragment list host driver specifies directly frame pointer. */
 	use_frags = htt->target_version_major < 3 ||
 		    !ieee80211_is_mgmt(hdr->frame_control);
 
-	if (!IS_ALIGNED((unsigned long)txdesc->data, 4)) {
-		ath10k_warn("htt alignment check failed. dropping packet.\n");
-		res = -EIO;
-		goto err_free_txdesc;
-	}
+	skb_cb->htt.txbuf = dma_pool_alloc(htt->tx_pool, GFP_ATOMIC,
+					   &paddr);
+	if (!skb_cb->htt.txbuf)
+		goto err_free_msdu_id;
+	skb_cb->htt.txbuf_paddr = paddr;
 
-	if (use_frags) {
-		skb_cb->htt.frag_len = sizeof(*tx_frags) * 2;
-		skb_cb->htt.pad_len = (unsigned long)msdu->data -
-				      round_down((unsigned long)msdu->data, 4);
+	skb_cb->paddr = dma_map_single(dev, msdu->data, msdu->len,
+				       DMA_TO_DEVICE);
+	res = dma_mapping_error(dev, skb_cb->paddr);
+	if (res)
+		goto err_free_txbuf;
 
-		skb_push(msdu, skb_cb->htt.frag_len + skb_cb->htt.pad_len);
-	} else {
-		skb_cb->htt.frag_len = 0;
-		skb_cb->htt.pad_len = 0;
-	}
+	if (likely(use_frags)) {
+		frags = skb_cb->htt.txbuf->frags;
 
-	res = ath10k_skb_map(dev, msdu);
-	if (res)
-		goto err_pull_txfrag;
-
-	if (use_frags) {
-		dma_sync_single_for_cpu(dev, skb_cb->paddr, msdu->len,
-					DMA_TO_DEVICE);
-
-		/* tx fragment list must be terminated with zero-entry */
-		tx_frags = (struct htt_data_tx_desc_frag *)msdu->data;
-		tx_frags[0].paddr = __cpu_to_le32(skb_cb->paddr +
-						  skb_cb->htt.frag_len +
-						  skb_cb->htt.pad_len);
-		tx_frags[0].len   = __cpu_to_le32(msdu->len -
-						  skb_cb->htt.frag_len -
-						  skb_cb->htt.pad_len);
-		tx_frags[1].paddr = __cpu_to_le32(0);
-		tx_frags[1].len   = __cpu_to_le32(0);
-
-		dma_sync_single_for_device(dev, skb_cb->paddr, msdu->len,
-					   DMA_TO_DEVICE);
-	}
+		frags[0].paddr = __cpu_to_le32(skb_cb->paddr);
+		frags[0].len = __cpu_to_le32(msdu->len);
+		frags[1].paddr = 0;
+		frags[1].len = 0;
 
-	ath10k_dbg(ATH10K_DBG_HTT, "tx-msdu 0x%llx\n",
-		   (unsigned long long) ATH10K_SKB_CB(msdu)->paddr);
-	ath10k_dbg_dump(ATH10K_DBG_HTT_DUMP, NULL, "tx-msdu: ",
-			msdu->data, msdu->len);
+		flags0 |= SM(ATH10K_HW_TXRX_NATIVE_WIFI,
+			     HTT_DATA_TX_DESC_FLAGS0_PKT_TYPE);
 
-	skb_put(txdesc, desc_len);
-	cmd = (struct htt_cmd *)txdesc->data;
+		frags_paddr = skb_cb->htt.txbuf_paddr;
+	} else {
+		flags0 |= SM(ATH10K_HW_TXRX_MGMT,
+			     HTT_DATA_TX_DESC_FLAGS0_PKT_TYPE);
 
-	tid = ATH10K_SKB_CB(msdu)->htt.tid;
+		frags_paddr = skb_cb->paddr;
+	}
 
-	ath10k_dbg(ATH10K_DBG_HTT, "htt data tx using tid %hhu\n", tid);
+	/* Normally all commands go through HTC which manages tx credits for
+	 * each endpoint and notifies when tx is completed.
+	 *
+	 * HTT endpoint is creditless so there's no need to care about HTC
+	 * flags. In that case it is trivial to fill the HTC header here.
+	 *
+	 * MSDU transmission is considered completed upon HTT event. This
+	 * implies no relevant resources can be freed until after the event is
+	 * received. That's why HTC tx completion handler itself is ignored by
+	 * setting NULL to transfer_context for all sg items.
+	 *
+	 * There is simply no point in pushing HTT TX_FRM through HTC tx path
+	 * as it's a waste of resources. By bypassing HTC it is possible to
+	 * avoid extra memory allocations, compress data structures and thus
+	 * improve performance. */
+
+	skb_cb->htt.txbuf->htc_hdr.eid = htt->eid;
+	skb_cb->htt.txbuf->htc_hdr.len = __cpu_to_le16(
+			sizeof(skb_cb->htt.txbuf->cmd_hdr) +
+			sizeof(skb_cb->htt.txbuf->cmd_tx) +
+			prefetch_len);
+	skb_cb->htt.txbuf->htc_hdr.flags = 0;
 
-	flags0  = 0;
 	if (!ieee80211_has_protected(hdr->frame_control))
 		flags0 |= HTT_DATA_TX_DESC_FLAGS0_NO_ENCRYPT;
-	flags0 |= HTT_DATA_TX_DESC_FLAGS0_MAC_HDR_PRESENT;
 
-	if (use_frags)
-		flags0 |= SM(ATH10K_HW_TXRX_NATIVE_WIFI,
-			     HTT_DATA_TX_DESC_FLAGS0_PKT_TYPE);
-	else
-		flags0 |= SM(ATH10K_HW_TXRX_MGMT,
-			     HTT_DATA_TX_DESC_FLAGS0_PKT_TYPE);
+	flags0 |= HTT_DATA_TX_DESC_FLAGS0_MAC_HDR_PRESENT;
 
-	flags1  = 0;
 	flags1 |= SM((u16)vdev_id, HTT_DATA_TX_DESC_FLAGS1_VDEV_ID);
 	flags1 |= SM((u16)tid, HTT_DATA_TX_DESC_FLAGS1_EXT_TID);
 	flags1 |= HTT_DATA_TX_DESC_FLAGS1_CKSUM_L3_OFFLOAD;
 	flags1 |= HTT_DATA_TX_DESC_FLAGS1_CKSUM_L4_OFFLOAD;
 
-	cmd->hdr.msg_type        = HTT_H2T_MSG_TYPE_TX_FRM;
-	cmd->data_tx.flags0      = flags0;
-	cmd->data_tx.flags1      = __cpu_to_le16(flags1);
-	cmd->data_tx.len         = __cpu_to_le16(msdu->len -
-						 skb_cb->htt.frag_len -
-						 skb_cb->htt.pad_len);
-	cmd->data_tx.id          = __cpu_to_le16(msdu_id);
-	cmd->data_tx.frags_paddr = __cpu_to_le32(skb_cb->paddr);
-	cmd->data_tx.peerid      = __cpu_to_le32(HTT_INVALID_PEERID);
-
-	memcpy(cmd->data_tx.prefetch, hdr, prefetch_len);
+	skb_cb->htt.txbuf->cmd_hdr.msg_type = HTT_H2T_MSG_TYPE_TX_FRM;
+	skb_cb->htt.txbuf->cmd_tx.flags0 = flags0;
+	skb_cb->htt.txbuf->cmd_tx.flags1 = __cpu_to_le16(flags1);
+	skb_cb->htt.txbuf->cmd_tx.len = __cpu_to_le16(msdu->len);
+	skb_cb->htt.txbuf->cmd_tx.id = __cpu_to_le16(msdu_id);
+	skb_cb->htt.txbuf->cmd_tx.frags_paddr = __cpu_to_le32(frags_paddr);
+	skb_cb->htt.txbuf->cmd_tx.peerid = __cpu_to_le32(HTT_INVALID_PEERID);
+
+	ath10k_dbg(ATH10K_DBG_HTT,
+		   "htt tx flags0 %hhu flags1 %hu len %d id %hu frags_paddr %08x, msdu_paddr %08x vdev %hhu tid %hhu\n",
+		   flags0, flags1, msdu->len, msdu_id, frags_paddr,
+		   (u32)skb_cb->paddr, vdev_id, tid);
+	ath10k_dbg_dump(ATH10K_DBG_HTT_DUMP, NULL, "htt tx msdu: ",
+			msdu->data, msdu->len);
 
-	res = ath10k_htc_send(&htt->ar->htc, htt->eid, txdesc);
+	sg_items[0].transfer_id = 0;
+	sg_items[0].transfer_context = NULL;
+	sg_items[0].vaddr = &skb_cb->htt.txbuf->htc_hdr;
+	sg_items[0].paddr = skb_cb->htt.txbuf_paddr +
+			    sizeof(skb_cb->htt.txbuf->frags);
+	sg_items[0].len = sizeof(skb_cb->htt.txbuf->htc_hdr) +
+			  sizeof(skb_cb->htt.txbuf->cmd_hdr) +
+			  sizeof(skb_cb->htt.txbuf->cmd_tx);
+
+	sg_items[1].transfer_id = 0;
+	sg_items[1].transfer_context = NULL;
+	sg_items[1].vaddr = msdu->data;
+	sg_items[1].paddr = skb_cb->paddr;
+	sg_items[1].len = prefetch_len;
+
+	res = ath10k_hif_tx_sg(htt->ar,
+			       htt->ar->htc.endpoint[htt->eid].ul_pipe_id,
+			       sg_items, ARRAY_SIZE(sg_items));
 	if (res)
 		goto err_unmap_msdu;
 
 	return 0;
 
 err_unmap_msdu:
-	ath10k_skb_unmap(dev, msdu);
-err_pull_txfrag:
-	skb_pull(msdu, skb_cb->htt.frag_len + skb_cb->htt.pad_len);
-err_free_txdesc:
-	dev_kfree_skb_any(txdesc);
+	dma_unmap_single(dev, skb_cb->paddr, msdu->len, DMA_TO_DEVICE);
+err_free_txbuf:
+	dma_pool_free(htt->tx_pool,
+		      skb_cb->htt.txbuf,
+		      skb_cb->htt.txbuf_paddr);
 err_free_msdu_id:
 	spin_lock_bh(&htt->tx_lock);
 	htt->pending_tx[msdu_id] = NULL;
diff --git a/drivers/net/wireless/ath/ath10k/mac.c b/drivers/net/wireless/ath/ath10k/mac.c
index e17f5d732b5a..511a2f81e7af 100644
--- a/drivers/net/wireless/ath/ath10k/mac.c
+++ b/drivers/net/wireless/ath/ath10k/mac.c
@@ -323,13 +323,15 @@ static int ath10k_peer_create(struct ath10k *ar, u32 vdev_id, const u8 *addr)
 
 	ret = ath10k_wmi_peer_create(ar, vdev_id, addr);
 	if (ret) {
-		ath10k_warn("Failed to create wmi peer: %i\n", ret);
+		ath10k_warn("Failed to create wmi peer %pM on vdev %i: %i\n",
+			    addr, vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_wait_for_peer_created(ar, vdev_id, addr);
 	if (ret) {
-		ath10k_warn("Failed to wait for created wmi peer: %i\n", ret);
+		ath10k_warn("Failed to wait for created wmi peer %pM on vdev %i: %i\n",
+			    addr, vdev_id, ret);
 		return ret;
 	}
 	spin_lock_bh(&ar->data_lock);
@@ -349,7 +351,8 @@ static int ath10k_mac_set_kickout(struct ath10k_vif *arvif)
 	ret = ath10k_wmi_pdev_set_param(ar, param,
 					ATH10K_KICKOUT_THRESHOLD);
 	if (ret) {
-		ath10k_warn("Failed to set kickout threshold: %d\n", ret);
+		ath10k_warn("Failed to set kickout threshold on vdev %i: %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
@@ -357,8 +360,8 @@ static int ath10k_mac_set_kickout(struct ath10k_vif *arvif)
 	ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, param,
 					ATH10K_KEEPALIVE_MIN_IDLE);
 	if (ret) {
-		ath10k_warn("Failed to set keepalive minimum idle time : %d\n",
-			    ret);
+		ath10k_warn("Failed to set keepalive minimum idle time on vdev %i : %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
@@ -366,8 +369,8 @@ static int ath10k_mac_set_kickout(struct ath10k_vif *arvif)
 	ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, param,
 					ATH10K_KEEPALIVE_MAX_IDLE);
 	if (ret) {
-		ath10k_warn("Failed to set keepalive maximum idle time: %d\n",
-			    ret);
+		ath10k_warn("Failed to set keepalive maximum idle time on vdev %i: %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
@@ -375,8 +378,8 @@ static int ath10k_mac_set_kickout(struct ath10k_vif *arvif)
 	ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, param,
 					ATH10K_KEEPALIVE_MAX_UNRESPONSIVE);
 	if (ret) {
-		ath10k_warn("Failed to set keepalive maximum unresponsive time: %d\n",
-			    ret);
+		ath10k_warn("Failed to set keepalive maximum unresponsive time on vdev %i: %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
@@ -529,13 +532,15 @@ static int ath10k_vdev_start(struct ath10k_vif *arvif)
 
 	ret = ath10k_wmi_vdev_start(ar, &arg);
 	if (ret) {
-		ath10k_warn("WMI vdev start failed: ret %d\n", ret);
+		ath10k_warn("WMI vdev %i start failed: ret %d\n",
+			    arg.vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_vdev_setup_sync(ar);
 	if (ret) {
-		ath10k_warn("vdev setup failed %d\n", ret);
+		ath10k_warn("vdev %i setup failed %d\n",
+			    arg.vdev_id, ret);
 		return ret;
 	}
 
@@ -553,13 +558,15 @@ static int ath10k_vdev_stop(struct ath10k_vif *arvif)
 
 	ret = ath10k_wmi_vdev_stop(ar, arvif->vdev_id);
 	if (ret) {
-		ath10k_warn("WMI vdev stop failed: ret %d\n", ret);
+		ath10k_warn("WMI vdev %i stop failed: ret %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_vdev_setup_sync(ar);
 	if (ret) {
-		ath10k_warn("vdev setup failed %d\n", ret);
+		ath10k_warn("vdev %i setup sync failed %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
@@ -597,19 +604,22 @@ static int ath10k_monitor_start(struct ath10k *ar, int vdev_id)
 
 	ret = ath10k_wmi_vdev_start(ar, &arg);
 	if (ret) {
-		ath10k_warn("Monitor vdev start failed: ret %d\n", ret);
+		ath10k_warn("Monitor vdev %i start failed: ret %d\n",
+			    vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_vdev_setup_sync(ar);
 	if (ret) {
-		ath10k_warn("Monitor vdev setup failed %d\n", ret);
+		ath10k_warn("Monitor vdev %i setup failed %d\n",
+			    vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_wmi_vdev_up(ar, vdev_id, 0, ar->mac_addr);
 	if (ret) {
-		ath10k_warn("Monitor vdev up failed: %d\n", ret);
+		ath10k_warn("Monitor vdev %i up failed: %d\n",
+			    vdev_id, ret);
 		goto vdev_stop;
 	}
 
@@ -621,7 +631,8 @@ static int ath10k_monitor_start(struct ath10k *ar, int vdev_id)
 vdev_stop:
 	ret = ath10k_wmi_vdev_stop(ar, ar->monitor_vdev_id);
 	if (ret)
-		ath10k_warn("Monitor vdev stop failed: %d\n", ret);
+		ath10k_warn("Monitor vdev %i stop failed: %d\n",
+			    ar->monitor_vdev_id, ret);
 
 	return ret;
 }
@@ -644,15 +655,18 @@ static int ath10k_monitor_stop(struct ath10k *ar)
 
 	ret = ath10k_wmi_vdev_down(ar, ar->monitor_vdev_id);
 	if (ret)
-		ath10k_warn("Monitor vdev down failed: %d\n", ret);
+		ath10k_warn("Monitor vdev %i down failed: %d\n",
+			    ar->monitor_vdev_id, ret);
 
 	ret = ath10k_wmi_vdev_stop(ar, ar->monitor_vdev_id);
 	if (ret)
-		ath10k_warn("Monitor vdev stop failed: %d\n", ret);
+		ath10k_warn("Monitor vdev %i stop failed: %d\n",
+			    ar->monitor_vdev_id, ret);
 
 	ret = ath10k_vdev_setup_sync(ar);
 	if (ret)
-		ath10k_warn("Monitor_down sync failed: %d\n", ret);
+		ath10k_warn("Monitor_down sync failed, vdev %i: %d\n",
+			    ar->monitor_vdev_id, ret);
 
 	ar->monitor_enabled = false;
 	return ret;
@@ -682,7 +696,8 @@ static int ath10k_monitor_create(struct ath10k *ar)
 				     WMI_VDEV_TYPE_MONITOR,
 				     0, ar->mac_addr);
 	if (ret) {
-		ath10k_warn("WMI vdev monitor create failed: ret %d\n", ret);
+		ath10k_warn("WMI vdev %i monitor create failed: ret %d\n",
+			    ar->monitor_vdev_id, ret);
 		goto vdev_fail;
 	}
 
@@ -711,7 +726,8 @@ static int ath10k_monitor_destroy(struct ath10k *ar)
 
 	ret = ath10k_wmi_vdev_delete(ar, ar->monitor_vdev_id);
 	if (ret) {
-		ath10k_warn("WMI vdev monitor delete failed: %d\n", ret);
+		ath10k_warn("WMI vdev %i monitor delete failed: %d\n",
+			    ar->monitor_vdev_id, ret);
 		return ret;
 	}
 
@@ -839,7 +855,9 @@ static void ath10k_control_beaconing(struct ath10k_vif *arvif,
 
 		spin_lock_bh(&arvif->ar->data_lock);
 		if (arvif->beacon) {
-			ath10k_skb_unmap(arvif->ar->dev, arvif->beacon);
+			dma_unmap_single(arvif->ar->dev,
+					 ATH10K_SKB_CB(arvif->beacon)->paddr,
+					 arvif->beacon->len, DMA_TO_DEVICE);
 			dev_kfree_skb_any(arvif->beacon);
 
 			arvif->beacon = NULL;
@@ -862,8 +880,8 @@ static void ath10k_control_beaconing(struct ath10k_vif *arvif,
 	ret = ath10k_wmi_vdev_up(arvif->ar, arvif->vdev_id, arvif->aid,
 				 arvif->bssid);
 	if (ret) {
-		ath10k_warn("Failed to bring up VDEV: %d\n",
-			    arvif->vdev_id);
+		ath10k_warn("Failed to bring up vdev %d: %i\n",
+			    arvif->vdev_id, ret);
 		ath10k_vdev_stop(arvif);
 		return;
 	}
@@ -943,8 +961,8 @@ static int ath10k_mac_vif_setup_ps(struct ath10k_vif *arvif)
 		ret = ath10k_wmi_set_sta_ps_param(ar, arvif->vdev_id, param,
 						  conf->dynamic_ps_timeout);
 		if (ret) {
-			ath10k_warn("Failed to set inactivity time for VDEV: %d\n",
-				    arvif->vdev_id);
+			ath10k_warn("Failed to set inactivity time for vdev %d: %i\n",
+				    arvif->vdev_id, ret);
 			return ret;
 		}
 	} else {
@@ -1196,8 +1214,8 @@ static int ath10k_peer_assoc_qos_ap(struct ath10k *ar,
 						 WMI_AP_PS_PEER_PARAM_UAPSD,
 						 uapsd);
 		if (ret) {
-			ath10k_warn("failed to set ap ps peer param uapsd: %d\n",
-				    ret);
+			ath10k_warn("failed to set ap ps peer param uapsd for vdev %i: %d\n",
+				    arvif->vdev_id, ret);
 			return ret;
 		}
 
@@ -1206,8 +1224,8 @@ static int ath10k_peer_assoc_qos_ap(struct ath10k *ar,
 						 WMI_AP_PS_PEER_PARAM_MAX_SP,
 						 max_sp);
 		if (ret) {
-			ath10k_warn("failed to set ap ps peer param max sp: %d\n",
-				    ret);
+			ath10k_warn("failed to set ap ps peer param max sp for vdev %i: %d\n",
+				    arvif->vdev_id, ret);
 			return ret;
 		}
 
@@ -1218,8 +1236,8 @@ static int ath10k_peer_assoc_qos_ap(struct ath10k *ar,
 		ret = ath10k_wmi_set_ap_ps_param(ar, arvif->vdev_id, sta->addr,
 					WMI_AP_PS_PEER_PARAM_AGEOUT_TIME, 10);
 		if (ret) {
-			ath10k_warn("failed to set ap ps peer param ageout time: %d\n",
-				    ret);
+			ath10k_warn("failed to set ap ps peer param ageout time for vdev %i: %d\n",
+				    arvif->vdev_id, ret);
 			return ret;
 		}
 	}
@@ -1411,8 +1429,8 @@ static void ath10k_bss_assoc(struct ieee80211_hw *hw,
 
 	ap_sta = ieee80211_find_sta(vif, bss_conf->bssid);
 	if (!ap_sta) {
-		ath10k_warn("Failed to find station entry for %pM\n",
-			    bss_conf->bssid);
+		ath10k_warn("Failed to find station entry for %pM, vdev %i\n",
+			    bss_conf->bssid, arvif->vdev_id);
 		rcu_read_unlock();
 		return;
 	}
@@ -1424,8 +1442,8 @@ static void ath10k_bss_assoc(struct ieee80211_hw *hw,
 	ret = ath10k_peer_assoc_prepare(ar, arvif, ap_sta,
 					bss_conf, &peer_arg);
 	if (ret) {
-		ath10k_warn("Peer assoc prepare failed for %pM\n: %d",
-			    bss_conf->bssid, ret);
+		ath10k_warn("Peer assoc prepare failed for %pM vdev %i\n: %d",
+			    bss_conf->bssid, arvif->vdev_id, ret);
 		rcu_read_unlock();
 		return;
 	}
@@ -1434,14 +1452,15 @@ static void ath10k_bss_assoc(struct ieee80211_hw *hw,
 
 	ret = ath10k_wmi_peer_assoc(ar, &peer_arg);
 	if (ret) {
-		ath10k_warn("Peer assoc failed for %pM\n: %d",
-			    bss_conf->bssid, ret);
+		ath10k_warn("Peer assoc failed for %pM vdev %i\n: %d",
+			    bss_conf->bssid, arvif->vdev_id, ret);
 		return;
 	}
 
 	ret = ath10k_setup_peer_smps(ar, arvif, bss_conf->bssid, &ht_cap);
 	if (ret) {
-		ath10k_warn("failed to setup peer SMPS: %d\n", ret);
+		ath10k_warn("failed to setup peer SMPS for vdev %i: %d\n",
+			    arvif->vdev_id, ret);
 		return;
 	}
 
@@ -1514,34 +1533,35 @@ static int ath10k_station_assoc(struct ath10k *ar, struct ath10k_vif *arvif,
 
 	ret = ath10k_peer_assoc_prepare(ar, arvif, sta, NULL, &peer_arg);
 	if (ret) {
-		ath10k_warn("WMI peer assoc prepare failed for %pM\n",
-			    sta->addr);
+		ath10k_warn("WMI peer assoc prepare failed for %pM vdev %i: %i\n",
+			    sta->addr, arvif->vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_wmi_peer_assoc(ar, &peer_arg);
 	if (ret) {
-		ath10k_warn("Peer assoc failed for STA %pM\n: %d",
-			    sta->addr, ret);
+		ath10k_warn("Peer assoc failed for STA %pM vdev %i: %d\n",
+			    sta->addr, arvif->vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_setup_peer_smps(ar, arvif, sta->addr, &sta->ht_cap);
 	if (ret) {
-		ath10k_warn("failed to setup peer SMPS: %d\n", ret);
+		ath10k_warn("failed to setup peer SMPS for vdev: %d\n", ret);
 		return ret;
 	}
 
 	ret = ath10k_install_peer_wep_keys(arvif, sta->addr);
 	if (ret) {
-		ath10k_warn("could not install peer wep keys (%d)\n", ret);
+		ath10k_warn("could not install peer wep keys for vdev %i: %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
 	ret = ath10k_peer_assoc_qos_ap(ar, arvif, sta);
 	if (ret) {
-		ath10k_warn("could not set qos params for STA %pM, %d\n",
-			    sta->addr, ret);
+		ath10k_warn("could not set qos params for STA %pM for vdev %i: %d\n",
+			    sta->addr, arvif->vdev_id, ret);
 		return ret;
 	}
 
@@ -1557,7 +1577,8 @@ static int ath10k_station_disassoc(struct ath10k *ar, struct ath10k_vif *arvif,
 
 	ret = ath10k_clear_peer_keys(arvif, sta->addr);
 	if (ret) {
-		ath10k_warn("could not clear all peer wep keys (%d)\n", ret);
+		ath10k_warn("could not clear all peer wep keys for vdev %i: %d\n",
+			    arvif->vdev_id, ret);
 		return ret;
 	}
 
@@ -2524,7 +2545,8 @@ static int ath10k_add_interface(struct ieee80211_hw *hw,
 	ret = ath10k_wmi_vdev_create(ar, arvif->vdev_id, arvif->vdev_type,
 				     arvif->vdev_subtype, vif->addr);
 	if (ret) {
-		ath10k_warn("WMI vdev create failed: ret %d\n", ret);
+		ath10k_warn("WMI vdev %i create failed: ret %d\n",
+			    arvif->vdev_id, ret);
 		goto err;
 	}
 
@@ -2535,7 +2557,8 @@ static int ath10k_add_interface(struct ieee80211_hw *hw,
 	ret = ath10k_wmi_vdev_set_param(ar, 0, vdev_param,
 					arvif->def_wep_key_idx);
 	if (ret) {
-		ath10k_warn("Failed to set default keyid: %d\n", ret);
+		ath10k_warn("Failed to set vdev %i default keyid: %d\n",
+			    arvif->vdev_id, ret);
 		goto err_vdev_delete;
 	}
 
@@ -2544,21 +2567,23 @@ static int ath10k_add_interface(struct ieee80211_hw *hw,
 					ATH10K_HW_TXRX_NATIVE_WIFI);
 	/* 10.X firmware does not support this VDEV parameter. Do not warn */
 	if (ret && ret != -EOPNOTSUPP) {
-		ath10k_warn("Failed to set TX encap: %d\n", ret);
+		ath10k_warn("Failed to set vdev %i TX encap: %d\n",
+			    arvif->vdev_id, ret);
 		goto err_vdev_delete;
 	}
 
 	if (arvif->vdev_type == WMI_VDEV_TYPE_AP) {
 		ret = ath10k_peer_create(ar, arvif->vdev_id, vif->addr);
 		if (ret) {
-			ath10k_warn("Failed to create peer for AP: %d\n", ret);
+			ath10k_warn("Failed to create vdev %i peer for AP: %d\n",
+				    arvif->vdev_id, ret);
 			goto err_vdev_delete;
 		}
 
 		ret = ath10k_mac_set_kickout(arvif);
 		if (ret) {
-			ath10k_warn("Failed to set kickout parameters: %d\n",
-				    ret);
+			ath10k_warn("Failed to set vdev %i kickout parameters: %d\n",
+				    arvif->vdev_id, ret);
 			goto err_peer_delete;
 		}
 	}
@@ -2569,7 +2594,8 @@ static int ath10k_add_interface(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_set_sta_ps_param(ar, arvif->vdev_id,
 						  param, value);
 		if (ret) {
-			ath10k_warn("Failed to set RX wake policy: %d\n", ret);
+			ath10k_warn("Failed to set vdev %i RX wake policy: %d\n",
+				    arvif->vdev_id, ret);
 			goto err_peer_delete;
 		}
 
@@ -2578,7 +2604,8 @@ static int ath10k_add_interface(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_set_sta_ps_param(ar, arvif->vdev_id,
 						  param, value);
 		if (ret) {
-			ath10k_warn("Failed to set TX wake thresh: %d\n", ret);
+			ath10k_warn("Failed to set vdev %i TX wake thresh: %d\n",
+				    arvif->vdev_id, ret);
 			goto err_peer_delete;
 		}
 
@@ -2587,7 +2614,8 @@ static int ath10k_add_interface(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_set_sta_ps_param(ar, arvif->vdev_id,
 						  param, value);
 		if (ret) {
-			ath10k_warn("Failed to set PSPOLL count: %d\n", ret);
+			ath10k_warn("Failed to set vdev %i PSPOLL count: %d\n",
+				    arvif->vdev_id, ret);
 			goto err_peer_delete;
 		}
 	}
@@ -2651,17 +2679,19 @@ static void ath10k_remove_interface(struct ieee80211_hw *hw,
 	if (arvif->vdev_type == WMI_VDEV_TYPE_AP) {
 		ret = ath10k_peer_delete(arvif->ar, arvif->vdev_id, vif->addr);
 		if (ret)
-			ath10k_warn("Failed to remove peer for AP: %d\n", ret);
+			ath10k_warn("Failed to remove peer for AP vdev %i: %d\n",
+				    arvif->vdev_id, ret);
 
 		kfree(arvif->u.ap.noa_data);
 	}
 
-	ath10k_dbg(ATH10K_DBG_MAC, "mac vdev delete %d (remove interface)\n",
+	ath10k_dbg(ATH10K_DBG_MAC, "mac vdev %i delete (remove interface)\n",
 		   arvif->vdev_id);
 
 	ret = ath10k_wmi_vdev_delete(ar, arvif->vdev_id);
 	if (ret)
-		ath10k_warn("WMI vdev delete failed: %d\n", ret);
+		ath10k_warn("WMI vdev %i delete failed: %d\n",
+			    arvif->vdev_id, ret);
 
 	if (arvif->vdev_type == WMI_VDEV_TYPE_MONITOR)
 		ar->monitor_present = false;
@@ -2750,8 +2780,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 			   arvif->vdev_id, arvif->beacon_interval);
 
 		if (ret)
-			ath10k_warn("Failed to set beacon interval for VDEV: %d\n",
-				    arvif->vdev_id);
+			ath10k_warn("Failed to set beacon interval for vdev %d: %i\n",
+				    arvif->vdev_id, ret);
 	}
 
 	if (changed & BSS_CHANGED_BEACON) {
@@ -2763,8 +2793,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_pdev_set_param(ar, pdev_param,
 						WMI_BEACON_STAGGERED_MODE);
 		if (ret)
-			ath10k_warn("Failed to set beacon mode for VDEV: %d\n",
-				    arvif->vdev_id);
+			ath10k_warn("Failed to set beacon mode for vdev %d: %i\n",
+				    arvif->vdev_id, ret);
 	}
 
 	if (changed & BSS_CHANGED_BEACON_INFO) {
@@ -2778,8 +2808,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, vdev_param,
 						arvif->dtim_period);
 		if (ret)
-			ath10k_warn("Failed to set dtim period for VDEV: %d\n",
-				    arvif->vdev_id);
+			ath10k_warn("Failed to set dtim period for vdev %d: %i\n",
+				    arvif->vdev_id, ret);
 	}
 
 	if (changed & BSS_CHANGED_SSID &&
@@ -2799,7 +2829,7 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 			ret = ath10k_peer_create(ar, arvif->vdev_id,
 						 info->bssid);
 			if (ret)
-				ath10k_warn("Failed to add peer %pM for vdev %d when changin bssid: %i\n",
+				ath10k_warn("Failed to add peer %pM for vdev %d when changing bssid: %i\n",
 					    info->bssid, arvif->vdev_id, ret);
 
 			if (vif->type == NL80211_IFTYPE_STATION) {
@@ -2815,8 +2845,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 
 				ret = ath10k_vdev_start(arvif);
 				if (ret) {
-					ath10k_warn("failed to start vdev: %d\n",
-						    ret);
+					ath10k_warn("failed to start vdev %i: %d\n",
+						    arvif->vdev_id, ret);
 					goto exit;
 				}
 
@@ -2851,8 +2881,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, vdev_param,
 						cts_prot);
 		if (ret)
-			ath10k_warn("Failed to set CTS prot for VDEV: %d\n",
-				    arvif->vdev_id);
+			ath10k_warn("Failed to set CTS prot for vdev %d: %d\n",
+				    arvif->vdev_id, ret);
 	}
 
 	if (changed & BSS_CHANGED_ERP_SLOT) {
@@ -2870,8 +2900,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, vdev_param,
 						slottime);
 		if (ret)
-			ath10k_warn("Failed to set erp slot for VDEV: %d\n",
-				    arvif->vdev_id);
+			ath10k_warn("Failed to set erp slot for vdev %d: %i\n",
+				    arvif->vdev_id, ret);
 	}
 
 	if (changed & BSS_CHANGED_ERP_PREAMBLE) {
@@ -2889,8 +2919,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 		ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, vdev_param,
 						preamble);
 		if (ret)
-			ath10k_warn("Failed to set preamble for VDEV: %d\n",
-				    arvif->vdev_id);
+			ath10k_warn("Failed to set preamble for vdev %d: %i\n",
+				    arvif->vdev_id, ret);
 	}
 
 	if (changed & BSS_CHANGED_ASSOC) {
@@ -3021,8 +3051,8 @@ static void ath10k_set_key_h_def_keyidx(struct ath10k *ar,
 	ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, vdev_param,
 					key->keyidx);
 	if (ret)
-		ath10k_warn("failed to set group key as default key: %d\n",
-			    ret);
+		ath10k_warn("failed to set vdev %i group key as default key: %d\n",
+			    arvif->vdev_id, ret);
 }
 
 static int ath10k_set_key(struct ieee80211_hw *hw, enum set_key_cmd cmd,
@@ -3082,7 +3112,8 @@ static int ath10k_set_key(struct ieee80211_hw *hw, enum set_key_cmd cmd,
 
 	ret = ath10k_install_key(arvif, key, cmd, peer_addr);
 	if (ret) {
-		ath10k_warn("ath10k_install_key failed (%d)\n", ret);
+		ath10k_warn("key installation failed for vdev %i peer %pM: %d\n",
+			    arvif->vdev_id, peer_addr, ret);
 		goto exit;
 	}
 
@@ -3179,6 +3210,13 @@ static int ath10k_sta_state(struct ieee80211_hw *hw,
 	int max_num_peers;
 	int ret = 0;
 
+	if (old_state == IEEE80211_STA_NOTEXIST &&
+	    new_state == IEEE80211_STA_NONE) {
+		memset(arsta, 0, sizeof(*arsta));
+		arsta->arvif = arvif;
+		INIT_WORK(&arsta->update_wk, ath10k_sta_rc_update_wk);
+	}
+
 	/* cancel must be done outside the mutex to avoid deadlock */
 	if ((old_state == IEEE80211_STA_NONE &&
 	     new_state == IEEE80211_STA_NOTEXIST))
@@ -3208,10 +3246,6 @@ static int ath10k_sta_state(struct ieee80211_hw *hw,
 			   "mac vdev %d peer create %pM (new sta) num_peers %d\n",
 			   arvif->vdev_id, sta->addr, ar->num_peers);
 
-		memset(arsta, 0, sizeof(*arsta));
-		arsta->arvif = arvif;
-		INIT_WORK(&arsta->update_wk, ath10k_sta_rc_update_wk);
-
 		ret = ath10k_peer_create(ar, arvif->vdev_id, sta->addr);
 		if (ret)
 			ath10k_warn("Failed to add peer %pM for vdev %d when adding a new sta: %i\n",
@@ -3226,8 +3260,8 @@ static int ath10k_sta_state(struct ieee80211_hw *hw,
 			   arvif->vdev_id, sta->addr);
 		ret = ath10k_peer_delete(ar, arvif->vdev_id, sta->addr);
 		if (ret)
-			ath10k_warn("Failed to delete peer: %pM for VDEV: %d\n",
-				    sta->addr, arvif->vdev_id);
+			ath10k_warn("Failed to delete peer %pM for vdev %d: %i\n",
+				    sta->addr, arvif->vdev_id, ret);
 
 		if (vif->type == NL80211_IFTYPE_STATION)
 			ath10k_bss_disassoc(hw, vif);
@@ -3243,8 +3277,8 @@ static int ath10k_sta_state(struct ieee80211_hw *hw,
 
 		ret = ath10k_station_assoc(ar, arvif, sta);
 		if (ret)
-			ath10k_warn("Failed to associate station: %pM\n",
-				    sta->addr);
+			ath10k_warn("Failed to associate station %pM for vdev %i: %i\n",
+				    sta->addr, arvif->vdev_id, ret);
 	} else if (old_state == IEEE80211_STA_ASSOC &&
 		   new_state == IEEE80211_STA_AUTH &&
 		   (vif->type == NL80211_IFTYPE_AP ||
@@ -3257,8 +3291,8 @@ static int ath10k_sta_state(struct ieee80211_hw *hw,
 
 		ret = ath10k_station_disassoc(ar, arvif, sta);
 		if (ret)
-			ath10k_warn("Failed to disassociate station: %pM\n",
-				    sta->addr);
+			ath10k_warn("Failed to disassociate station: %pM vdev %i ret %i\n",
+				    sta->addr, arvif->vdev_id, ret);
 	}
 exit:
 	mutex_unlock(&ar->conf_mutex);
@@ -3539,7 +3573,8 @@ static void ath10k_flush(struct ieee80211_hw *hw, u32 queues, bool drop)
 		}), ATH10K_FLUSH_TIMEOUT_HZ);
 
 	if (ret <= 0 || skip)
-		ath10k_warn("tx not flushed\n");
+		ath10k_warn("tx not flushed (skip %i ar-state %i): %i\n",
+			    skip, ar->state, ret);
 
 skip:
 	mutex_unlock(&ar->conf_mutex);
@@ -3905,7 +3940,8 @@ static bool ath10k_get_fixed_rate_nss(const struct cfg80211_bitrate_mask *mask,
 
 static int ath10k_set_fixed_rate_param(struct ath10k_vif *arvif,
 				       u8 fixed_rate,
-				       u8 fixed_nss)
+				       u8 fixed_nss,
+				       u8 force_sgi)
 {
 	struct ath10k *ar = arvif->ar;
 	u32 vdev_param;
@@ -3914,12 +3950,16 @@ static int ath10k_set_fixed_rate_param(struct ath10k_vif *arvif,
 	mutex_lock(&ar->conf_mutex);
 
 	if (arvif->fixed_rate == fixed_rate &&
-	    arvif->fixed_nss == fixed_nss)
+	    arvif->fixed_nss == fixed_nss &&
+	    arvif->force_sgi == force_sgi)
 		goto exit;
 
 	if (fixed_rate == WMI_FIXED_RATE_NONE)
 		ath10k_dbg(ATH10K_DBG_MAC, "mac disable fixed bitrate mask\n");
 
+	if (force_sgi)
+		ath10k_dbg(ATH10K_DBG_MAC, "mac force sgi\n");
+
 	vdev_param = ar->wmi.vdev_param->fixed_rate;
 	ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id,
 					vdev_param, fixed_rate);
@@ -3945,6 +3985,19 @@ static int ath10k_set_fixed_rate_param(struct ath10k_vif *arvif,
 
 	arvif->fixed_nss = fixed_nss;
 
+	vdev_param = ar->wmi.vdev_param->sgi;
+	ret = ath10k_wmi_vdev_set_param(ar, arvif->vdev_id, vdev_param,
+					force_sgi);
+
+	if (ret) {
+		ath10k_warn("Could not set sgi param %d: %d\n",
+			    force_sgi, ret);
+		ret = -EINVAL;
+		goto exit;
+	}
+
+	arvif->force_sgi = force_sgi;
+
 exit:
 	mutex_unlock(&ar->conf_mutex);
 	return ret;
@@ -3959,6 +4012,11 @@ static int ath10k_set_bitrate_mask(struct ieee80211_hw *hw,
 	enum ieee80211_band band = ar->hw->conf.chandef.chan->band;
 	u8 fixed_rate = WMI_FIXED_RATE_NONE;
 	u8 fixed_nss = ar->num_rf_chains;
+	u8 force_sgi;
+
+	force_sgi = mask->control[band].gi;
+	if (force_sgi == NL80211_TXRATE_FORCE_LGI)
+		return -EINVAL;
 
 	if (!ath10k_default_bitrate_mask(ar, band, mask)) {
 		if (!ath10k_get_fixed_rate_nss(mask, band,
@@ -3967,7 +4025,13 @@ static int ath10k_set_bitrate_mask(struct ieee80211_hw *hw,
 			return -EINVAL;
 	}
 
-	return ath10k_set_fixed_rate_param(arvif, fixed_rate, fixed_nss);
+	if (fixed_rate == WMI_FIXED_RATE_NONE && force_sgi) {
+		ath10k_warn("Could not force SGI usage for default rate settings\n");
+		return -EINVAL;
+	}
+
+	return ath10k_set_fixed_rate_param(arvif, fixed_rate,
+					   fixed_nss, force_sgi);
 }
 
 static void ath10k_channel_switch_beacon(struct ieee80211_hw *hw,
@@ -4060,6 +4124,16 @@ static void ath10k_sta_rc_update(struct ieee80211_hw *hw,
 	ieee80211_queue_work(hw, &arsta->update_wk);
 }
 
+static u64 ath10k_get_tsf(struct ieee80211_hw *hw, struct ieee80211_vif *vif)
+{
+	/*
+	 * FIXME: Return 0 for time being. Need to figure out whether FW
+	 * has the API to fetch 64-bit local TSF
+	 */
+
+	return 0;
+}
+
 static const struct ieee80211_ops ath10k_ops = {
 	.tx				= ath10k_tx,
 	.start				= ath10k_start,
@@ -4085,6 +4159,7 @@ static const struct ieee80211_ops ath10k_ops = {
 	.set_bitrate_mask		= ath10k_set_bitrate_mask,
 	.channel_switch_beacon		= ath10k_channel_switch_beacon,
 	.sta_rc_update			= ath10k_sta_rc_update,
+	.get_tsf			= ath10k_get_tsf,
 #ifdef CONFIG_PM
 	.suspend			= ath10k_suspend,
 	.resume				= ath10k_resume,
@@ -4361,7 +4436,7 @@ struct ath10k_vif *ath10k_get_arvif(struct ath10k *ar, u32 vdev_id)
 						   ath10k_get_arvif_iter,
 						   &arvif_iter);
 	if (!arvif_iter.arvif) {
-		ath10k_warn("No VIF found for VDEV: %d\n", vdev_id);
+		ath10k_warn("No VIF found for vdev %d\n", vdev_id);
 		return NULL;
 	}
 
@@ -4442,7 +4517,8 @@ int ath10k_mac_register(struct ath10k *ar)
 			IEEE80211_HW_HAS_RATE_CONTROL |
 			IEEE80211_HW_SUPPORTS_STATIC_SMPS |
 			IEEE80211_HW_WANT_MONITOR_VIF |
-			IEEE80211_HW_AP_LINK_PS;
+			IEEE80211_HW_AP_LINK_PS |
+			IEEE80211_HW_SPECTRUM_MGMT;
 
 	/* MSDU can have HTT TX fragment pushed in front. The additional 4
 	 * bytes is used for padding/alignment if necessary. */
@@ -4500,7 +4576,7 @@ int ath10k_mac_register(struct ath10k *ar)
 	ret = ath_regd_init(&ar->ath_common.regulatory, ar->hw->wiphy,
 			    ath10k_reg_notifier);
 	if (ret) {
-		ath10k_err("Regulatory initialization failed\n");
+		ath10k_err("Regulatory initialization failed: %i\n", ret);
 		goto err_free;
 	}
 
diff --git a/drivers/net/wireless/ath/ath10k/pci.c b/drivers/net/wireless/ath/ath10k/pci.c
index 34f09106f423..9d242d801d9d 100644
--- a/drivers/net/wireless/ath/ath10k/pci.c
+++ b/drivers/net/wireless/ath/ath10k/pci.c
@@ -58,12 +58,10 @@ static DEFINE_PCI_DEVICE_TABLE(ath10k_pci_id_table) = {
 static int ath10k_pci_diag_read_access(struct ath10k *ar, u32 address,
 				       u32 *data);
 
-static void ath10k_pci_process_ce(struct ath10k *ar);
 static int ath10k_pci_post_rx(struct ath10k *ar);
 static int ath10k_pci_post_rx_pipe(struct ath10k_pci_pipe *pipe_info,
 					     int num);
 static void ath10k_pci_rx_pipe_cleanup(struct ath10k_pci_pipe *pipe_info);
-static void ath10k_pci_stop_ce(struct ath10k *ar);
 static int ath10k_pci_cold_reset(struct ath10k *ar);
 static int ath10k_pci_warm_reset(struct ath10k *ar);
 static int ath10k_pci_wait_for_target_init(struct ath10k *ar);
@@ -74,7 +72,6 @@ static void ath10k_pci_free_irq(struct ath10k *ar);
 static int ath10k_pci_bmi_wait(struct ath10k_ce_pipe *tx_pipe,
 			       struct ath10k_ce_pipe *rx_pipe,
 			       struct bmi_xfer *xfer);
-static void ath10k_pci_cleanup_ce(struct ath10k *ar);
 
 static const struct ce_attr host_ce_config_wlan[] = {
 	/* CE0: host->target HTC control and raw streams */
@@ -679,34 +676,12 @@ void ath10k_do_pci_sleep(struct ath10k *ar)
 	}
 }
 
-/*
- * FIXME: Handle OOM properly.
- */
-static inline
-struct ath10k_pci_compl *get_free_compl(struct ath10k_pci_pipe *pipe_info)
-{
-	struct ath10k_pci_compl *compl = NULL;
-
-	spin_lock_bh(&pipe_info->pipe_lock);
-	if (list_empty(&pipe_info->compl_free)) {
-		ath10k_warn("Completion buffers are full\n");
-		goto exit;
-	}
-	compl = list_first_entry(&pipe_info->compl_free,
-				 struct ath10k_pci_compl, list);
-	list_del(&compl->list);
-exit:
-	spin_unlock_bh(&pipe_info->pipe_lock);
-	return compl;
-}
-
 /* Called by lower (CE) layer when a send to Target completes. */
 static void ath10k_pci_ce_send_done(struct ath10k_ce_pipe *ce_state)
 {
 	struct ath10k *ar = ce_state->ar;
 	struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);
-	struct ath10k_pci_pipe *pipe_info =  &ar_pci->pipe_info[ce_state->id];
-	struct ath10k_pci_compl *compl;
+	struct ath10k_hif_cb *cb = &ar_pci->msg_callbacks_current;
 	void *transfer_context;
 	u32 ce_data;
 	unsigned int nbytes;
@@ -715,27 +690,12 @@ static void ath10k_pci_ce_send_done(struct ath10k_ce_pipe *ce_state)
 	while (ath10k_ce_completed_send_next(ce_state, &transfer_context,
 					     &ce_data, &nbytes,
 					     &transfer_id) == 0) {
-		compl = get_free_compl(pipe_info);
-		if (!compl)
-			break;
-
-		compl->state = ATH10K_PCI_COMPL_SEND;
-		compl->ce_state = ce_state;
-		compl->pipe_info = pipe_info;
-		compl->skb = transfer_context;
-		compl->nbytes = nbytes;
-		compl->transfer_id = transfer_id;
-		compl->flags = 0;
+		/* no need to call tx completion for NULL pointers */
+		if (transfer_context == NULL)
+			continue;
 
-		/*
-		 * Add the completion to the processing queue.
-		 */
-		spin_lock_bh(&ar_pci->compl_lock);
-		list_add_tail(&compl->list, &ar_pci->compl_process);
-		spin_unlock_bh(&ar_pci->compl_lock);
+		cb->tx_completion(ar, transfer_context, transfer_id);
 	}
-
-	ath10k_pci_process_ce(ar);
 }
 
 /* Called by lower (CE) layer when data is received from the Target. */
@@ -744,77 +704,100 @@ static void ath10k_pci_ce_recv_data(struct ath10k_ce_pipe *ce_state)
 	struct ath10k *ar = ce_state->ar;
 	struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);
 	struct ath10k_pci_pipe *pipe_info =  &ar_pci->pipe_info[ce_state->id];
-	struct ath10k_pci_compl *compl;
+	struct ath10k_hif_cb *cb = &ar_pci->msg_callbacks_current;
 	struct sk_buff *skb;
 	void *transfer_context;
 	u32 ce_data;
-	unsigned int nbytes;
+	unsigned int nbytes, max_nbytes;
 	unsigned int transfer_id;
 	unsigned int flags;
+	int err;
 
 	while (ath10k_ce_completed_recv_next(ce_state, &transfer_context,
 					     &ce_data, &nbytes, &transfer_id,
 					     &flags) == 0) {
-		compl = get_free_compl(pipe_info);
-		if (!compl)
-			break;
-
-		compl->state = ATH10K_PCI_COMPL_RECV;
-		compl->ce_state = ce_state;
-		compl->pipe_info = pipe_info;
-		compl->skb = transfer_context;
-		compl->nbytes = nbytes;
-		compl->transfer_id = transfer_id;
-		compl->flags = flags;
+		err = ath10k_pci_post_rx_pipe(pipe_info, 1);
+		if (unlikely(err)) {
+			/* FIXME: retry */
+			ath10k_warn("failed to replenish CE rx ring %d: %d\n",
+				    pipe_info->pipe_num, err);
+		}
 
 		skb = transfer_context;
+		max_nbytes = skb->len + skb_tailroom(skb);
 		dma_unmap_single(ar->dev, ATH10K_SKB_CB(skb)->paddr,
-				 skb->len + skb_tailroom(skb),
-				 DMA_FROM_DEVICE);
-		/*
-		 * Add the completion to the processing queue.
-		 */
-		spin_lock_bh(&ar_pci->compl_lock);
-		list_add_tail(&compl->list, &ar_pci->compl_process);
-		spin_unlock_bh(&ar_pci->compl_lock);
-	}
+				 max_nbytes, DMA_FROM_DEVICE);
 
-	ath10k_pci_process_ce(ar);
+		if (unlikely(max_nbytes < nbytes)) {
+			ath10k_warn("rxed more than expected (nbytes %d, max %d)",
+				    nbytes, max_nbytes);
+			dev_kfree_skb_any(skb);
+			continue;
+		}
+
+		skb_put(skb, nbytes);
+		cb->rx_completion(ar, skb, pipe_info->pipe_num);
+	}
 }
 
-/* Send the first nbytes bytes of the buffer */
-static int ath10k_pci_hif_send_head(struct ath10k *ar, u8 pipe_id,
-				    unsigned int transfer_id,
-				    unsigned int bytes, struct sk_buff *nbuf)
+static int ath10k_pci_hif_tx_sg(struct ath10k *ar, u8 pipe_id,
+				struct ath10k_hif_sg_item *items, int n_items)
 {
-	struct ath10k_skb_cb *skb_cb = ATH10K_SKB_CB(nbuf);
 	struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);
-	struct ath10k_pci_pipe *pipe_info = &(ar_pci->pipe_info[pipe_id]);
-	struct ath10k_ce_pipe *ce_hdl = pipe_info->ce_hdl;
-	unsigned int len;
-	u32 flags = 0;
-	int ret;
+	struct ath10k_pci_pipe *pci_pipe = &ar_pci->pipe_info[pipe_id];
+	struct ath10k_ce_pipe *ce_pipe = pci_pipe->ce_hdl;
+	struct ath10k_ce_ring *src_ring = ce_pipe->src_ring;
+	unsigned int nentries_mask = src_ring->nentries_mask;
+	unsigned int sw_index = src_ring->sw_index;
+	unsigned int write_index = src_ring->write_index;
+	int err, i;
 
-	len = min(bytes, nbuf->len);
-	bytes -= len;
+	spin_lock_bh(&ar_pci->ce_lock);
 
-	if (len & 3)
-		ath10k_warn("skb not aligned to 4-byte boundary (%d)\n", len);
+	if (unlikely(CE_RING_DELTA(nentries_mask,
+				   write_index, sw_index - 1) < n_items)) {
+		err = -ENOBUFS;
+		goto unlock;
+	}
 
-	ath10k_dbg(ATH10K_DBG_PCI,
-		   "pci send data vaddr %p paddr 0x%llx len %d as %d bytes\n",
-		   nbuf->data, (unsigned long long) skb_cb->paddr,
-		   nbuf->len, len);
-	ath10k_dbg_dump(ATH10K_DBG_PCI_DUMP, NULL,
-			"ath10k tx: data: ",
-			nbuf->data, nbuf->len);
-
-	ret = ath10k_ce_send(ce_hdl, nbuf, skb_cb->paddr, len, transfer_id,
-			     flags);
-	if (ret)
-		ath10k_warn("failed to send sk_buff to CE: %p\n", nbuf);
+	for (i = 0; i < n_items - 1; i++) {
+		ath10k_dbg(ATH10K_DBG_PCI,
+			   "pci tx item %d paddr 0x%08x len %d n_items %d\n",
+			   i, items[i].paddr, items[i].len, n_items);
+		ath10k_dbg_dump(ATH10K_DBG_PCI_DUMP, NULL, "item data: ",
+				items[i].vaddr, items[i].len);
 
-	return ret;
+		err = ath10k_ce_send_nolock(ce_pipe,
+					    items[i].transfer_context,
+					    items[i].paddr,
+					    items[i].len,
+					    items[i].transfer_id,
+					    CE_SEND_FLAG_GATHER);
+		if (err)
+			goto unlock;
+	}
+
+	/* `i` is equal to `n_items -1` after for() */
+
+	ath10k_dbg(ATH10K_DBG_PCI,
+		   "pci tx item %d paddr 0x%08x len %d n_items %d\n",
+		   i, items[i].paddr, items[i].len, n_items);
+	ath10k_dbg_dump(ATH10K_DBG_PCI_DUMP, NULL, "item data: ",
+			items[i].vaddr, items[i].len);
+
+	err = ath10k_ce_send_nolock(ce_pipe,
+				    items[i].transfer_context,
+				    items[i].paddr,
+				    items[i].len,
+				    items[i].transfer_id,
+				    0);
+	if (err)
+		goto unlock;
+
+	err = 0;
+unlock:
+	spin_unlock_bh(&ar_pci->ce_lock);
+	return err;
 }
 
 static u16 ath10k_pci_hif_get_free_queue_number(struct ath10k *ar, u8 pipe)
@@ -903,52 +886,6 @@ static void ath10k_pci_hif_set_callbacks(struct ath10k *ar,
 	       sizeof(ar_pci->msg_callbacks_current));
 }
 
-static int ath10k_pci_alloc_compl(struct ath10k *ar)
-{
-	struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);
-	const struct ce_attr *attr;
-	struct ath10k_pci_pipe *pipe_info;
-	struct ath10k_pci_compl *compl;
-	int i, pipe_num, completions;
-
-	spin_lock_init(&ar_pci->compl_lock);
-	INIT_LIST_HEAD(&ar_pci->compl_process);
-
-	for (pipe_num = 0; pipe_num < CE_COUNT; pipe_num++) {
-		pipe_info = &ar_pci->pipe_info[pipe_num];
-
-		spin_lock_init(&pipe_info->pipe_lock);
-		INIT_LIST_HEAD(&pipe_info->compl_free);
-
-		/* Handle Diagnostic CE specially */
-		if (pipe_info->ce_hdl == ar_pci->ce_diag)
-			continue;
-
-		attr = &host_ce_config_wlan[pipe_num];
-		completions = 0;
-
-		if (attr->src_nentries)
-			completions += attr->src_nentries;
-
-		if (attr->dest_nentries)
-			completions += attr->dest_nentries;
-
-		for (i = 0; i < completions; i++) {
-			compl = kmalloc(sizeof(*compl), GFP_KERNEL);
-			if (!compl) {
-				ath10k_warn("No memory for completion state\n");
-				ath10k_pci_cleanup_ce(ar);
-				return -ENOMEM;
-			}
-
-			compl->state = ATH10K_PCI_COMPL_FREE;
-			list_add_tail(&compl->list, &pipe_info->compl_free);
-		}
-	}
-
-	return 0;
-}
-
 static int ath10k_pci_setup_ce_irq(struct ath10k *ar)
 {
 	struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);
@@ -993,147 +930,6 @@ static void ath10k_pci_kill_tasklet(struct ath10k *ar)
 		tasklet_kill(&ar_pci->pipe_info[i].intr);
 }
 
-static void ath10k_pci_stop_ce(struct ath10k *ar)
-{
-	struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);
-	struct ath10k_pci_compl *compl;
-	struct sk_buff *skb;
-
-	/* Mark pending completions as aborted, so that upper layers free up
-	 * their associated resources */
-	spin_lock_bh(&ar_pci->compl_lock);
-	list_for_each_entry(compl, &ar_pci->compl_process, list) {
-		skb = compl->skb;
-		ATH10K_SKB_CB(skb)->is_aborted = true;
-	}
-	spin_unlock_bh(&ar_pci->compl_lock);
-}
-
-static void ath10k_pci_cleanup_ce(struct ath10k *ar)
-{
-	struct ath10k_pci *ar_pci = ath10k_pci_priv(ar);
-	struct ath10k_pci_compl *compl, *tmp;
-	struct ath10k_pci_pipe *pipe_info;
-	struct sk_buff *netbuf;
-	int pipe_num;
-
-	/* Free pending completions. */
-	spin_lock_bh(&ar_pci->compl_lock);
-	if (!list_empty(&ar_pci->compl_process))
-		ath10k_warn("pending completions still present! possible memory leaks.\n");
-
-	list_for_each_entry_safe(compl, tmp, &ar_pci->compl_process, list) {
-		list_del(&compl->list);
-		netbuf = compl->skb;
-		dev_kfree_skb_any(netbuf);
-		kfree(compl);
-	}
-	spin_unlock_bh(&ar_pci->compl_lock);
-
-	/* Free unused completions for each pipe. */
-	for (pipe_num = 0; pipe_num < CE_COUNT; pipe_num++) {
-		pipe_info = &ar_pci->pipe_info[pipe_num];
-
-		spin_lock_bh(&pipe_info->pipe_lock);
-		list_for_each_entry_safe(compl, tmp,
-					 &pipe_info->compl_free, list) {
-			list_del(&compl->list);
-			kfree(compl);
-		}
-		spin_unlock_bh(&pipe_info->pipe_lock);
-	}
-}
-
-static void ath10k_pci_process_ce(struct ath10k *ar)
-{
-	struct ath10k_pci *ar_pci = ar->hif.priv;
-	struct ath10k_hif_cb *cb = &ar_pci->msg_callbacks_current;
-	struct ath10k_pci_compl *compl;
-	struct sk_buff *skb;
-	unsigned int nbytes;
-	int ret, send_done = 0;
-
-	/* Upper layers aren't ready to handle tx/rx completions in parallel so
-	 * we must serialize all completion processing. */
-
-	spin_lock_bh(&ar_pci->compl_lock);
-	if (ar_pci->compl_processing) {
-		spin_unlock_bh(&ar_pci->compl_lock);
-		return;
-	}
-	ar_pci->compl_processing = true;
-	spin_unlock_bh(&ar_pci->compl_lock);
-
-	for (;;) {
-		spin_lock_bh(&ar_pci->compl_lock);
-		if (list_empty(&ar_pci->compl_process)) {
-			spin_unlock_bh(&ar_pci->compl_lock);
-			break;
-		}
-		compl = list_first_entry(&ar_pci->compl_process,
-					 struct ath10k_pci_compl, list);
-		list_del(&compl->list);
-		spin_unlock_bh(&ar_pci->compl_lock);
-
-		switch (compl->state) {
-		case ATH10K_PCI_COMPL_SEND:
-			cb->tx_completion(ar,
-					  compl->skb,
-					  compl->transfer_id);
-			send_done = 1;
-			break;
-		case ATH10K_PCI_COMPL_RECV:
-			ret = ath10k_pci_post_rx_pipe(compl->pipe_info, 1);
-			if (ret) {
-				ath10k_warn("failed to post RX buffer for pipe %d: %d\n",
-					    compl->pipe_info->pipe_num, ret);
-				break;
-			}
-
-			skb = compl->skb;
-			nbytes = compl->nbytes;
-
-			ath10k_dbg(ATH10K_DBG_PCI,
-				   "ath10k_pci_ce_recv_data netbuf=%p  nbytes=%d\n",
-				   skb, nbytes);
-			ath10k_dbg_dump(ATH10K_DBG_PCI_DUMP, NULL,
-					"ath10k rx: ", skb->data, nbytes);
-
-			if (skb->len + skb_tailroom(skb) >= nbytes) {
-				skb_trim(skb, 0);
-				skb_put(skb, nbytes);
-				cb->rx_completion(ar, skb,
-						  compl->pipe_info->pipe_num);
-			} else {
-				ath10k_warn("rxed more than expected (nbytes %d, max %d)",
-					    nbytes,
-					    skb->len + skb_tailroom(skb));
-			}
-			break;
-		case ATH10K_PCI_COMPL_FREE:
-			ath10k_warn("free completion cannot be processed\n");
-			break;
-		default:
-			ath10k_warn("invalid completion state (%d)\n",
-				    compl->state);
-			break;
-		}
-
-		compl->state = ATH10K_PCI_COMPL_FREE;
-
-		/*
-		 * Add completion back to the pipe's free list.
-		 */
-		spin_lock_bh(&compl->pipe_info->pipe_lock);
-		list_add_tail(&compl->list, &compl->pipe_info->compl_free);
-		spin_unlock_bh(&compl->pipe_info->pipe_lock);
-	}
-
-	spin_lock_bh(&ar_pci->compl_lock);
-	ar_pci->compl_processing = false;
-	spin_unlock_bh(&ar_pci->compl_lock);
-}
-
 /* TODO - temporary mapping while we have too few CE's */
 static int ath10k_pci_hif_map_service_to_pipe(struct ath10k *ar,
 					      u16 service_id, u8 *ul_pipe,
@@ -1305,17 +1101,11 @@ static int ath10k_pci_hif_start(struct ath10k *ar)
 	ath10k_pci_free_early_irq(ar);
 	ath10k_pci_kill_tasklet(ar);
 
-	ret = ath10k_pci_alloc_compl(ar);
-	if (ret) {
-		ath10k_warn("failed to allocate CE completions: %d\n", ret);
-		goto err_early_irq;
-	}
-
 	ret = ath10k_pci_request_irq(ar);
 	if (ret) {
 		ath10k_warn("failed to post RX buffers for all pipes: %d\n",
 			    ret);
-		goto err_free_compl;
+		goto err_early_irq;
 	}
 
 	ret = ath10k_pci_setup_ce_irq(ar);
@@ -1339,10 +1129,6 @@ err_stop:
 	ath10k_ce_disable_interrupts(ar);
 	ath10k_pci_free_irq(ar);
 	ath10k_pci_kill_tasklet(ar);
-	ath10k_pci_stop_ce(ar);
-	ath10k_pci_process_ce(ar);
-err_free_compl:
-	ath10k_pci_cleanup_ce(ar);
 err_early_irq:
 	/* Though there should be no interrupts (device was reset)
 	 * power_down() expects the early IRQ to be installed as per the
@@ -1413,18 +1199,10 @@ static void ath10k_pci_tx_pipe_cleanup(struct ath10k_pci_pipe *pipe_info)
 
 	while (ath10k_ce_cancel_send_next(ce_hdl, (void **)&netbuf,
 					  &ce_data, &nbytes, &id) == 0) {
-		/*
-		 * Indicate the completion to higer layer to free
-		 * the buffer
-		 */
-
-		if (!netbuf) {
-			ath10k_warn("invalid sk_buff on CE %d - NULL pointer. firmware crashed?\n",
-				    ce_hdl->id);
+		/* no need to call tx completion for NULL pointers */
+		if (!netbuf)
 			continue;
-		}
 
-		ATH10K_SKB_CB(netbuf)->is_aborted = true;
 		ar_pci->msg_callbacks_current.tx_completion(ar,
 							    netbuf,
 							    id);
@@ -1482,7 +1260,6 @@ static void ath10k_pci_hif_stop(struct ath10k *ar)
 
 	ath10k_pci_free_irq(ar);
 	ath10k_pci_kill_tasklet(ar);
-	ath10k_pci_stop_ce(ar);
 
 	ret = ath10k_pci_request_early_irq(ar);
 	if (ret)
@@ -1492,8 +1269,6 @@ static void ath10k_pci_hif_stop(struct ath10k *ar)
 	 * not DMA nor interrupt. We process the leftovers and then free
 	 * everything else up. */
 
-	ath10k_pci_process_ce(ar);
-	ath10k_pci_cleanup_ce(ar);
 	ath10k_pci_buffer_cleanup(ar);
 
 	/* Make the sure the device won't access any structures on the host by
@@ -2269,7 +2044,7 @@ static int ath10k_pci_hif_resume(struct ath10k *ar)
 #endif
 
 static const struct ath10k_hif_ops ath10k_pci_hif_ops = {
-	.send_head		= ath10k_pci_hif_send_head,
+	.tx_sg			= ath10k_pci_hif_tx_sg,
 	.exchange_bmi_msg	= ath10k_pci_hif_exchange_bmi_msg,
 	.start			= ath10k_pci_hif_start,
 	.stop			= ath10k_pci_hif_stop,
diff --git a/drivers/net/wireless/ath/ath10k/pci.h b/drivers/net/wireless/ath/ath10k/pci.h
index a4f32038c440..b43fdb4f7319 100644
--- a/drivers/net/wireless/ath/ath10k/pci.h
+++ b/drivers/net/wireless/ath/ath10k/pci.h
@@ -43,23 +43,6 @@ struct bmi_xfer {
 	u32 resp_len;
 };
 
-enum ath10k_pci_compl_state {
-	ATH10K_PCI_COMPL_FREE = 0,
-	ATH10K_PCI_COMPL_SEND,
-	ATH10K_PCI_COMPL_RECV,
-};
-
-struct ath10k_pci_compl {
-	struct list_head list;
-	enum ath10k_pci_compl_state state;
-	struct ath10k_ce_pipe *ce_state;
-	struct ath10k_pci_pipe *pipe_info;
-	struct sk_buff *skb;
-	unsigned int nbytes;
-	unsigned int transfer_id;
-	unsigned int flags;
-};
-
 /*
  * PCI-specific Target state
  *
@@ -175,9 +158,6 @@ struct ath10k_pci_pipe {
 	/* protects compl_free and num_send_allowed */
 	spinlock_t pipe_lock;
 
-	/* List of free CE completion slots */
-	struct list_head compl_free;
-
 	struct ath10k_pci *ar_pci;
 	struct tasklet_struct intr;
 };
@@ -205,14 +185,6 @@ struct ath10k_pci {
 	atomic_t keep_awake_count;
 	bool verified_awake;
 
-	/* List of CE completions to be processed */
-	struct list_head compl_process;
-
-	/* protects compl_processing and compl_process */
-	spinlock_t compl_lock;
-
-	bool compl_processing;
-
 	struct ath10k_pci_pipe pipe_info[CE_COUNT_MAX];
 
 	struct ath10k_hif_cb msg_callbacks_current;
diff --git a/drivers/net/wireless/ath/ath10k/txrx.c b/drivers/net/wireless/ath/ath10k/txrx.c
index ec6f82521b0e..0541dd939ce9 100644
--- a/drivers/net/wireless/ath/ath10k/txrx.c
+++ b/drivers/net/wireless/ath/ath10k/txrx.c
@@ -51,7 +51,8 @@ void ath10k_txrx_tx_unref(struct ath10k_htt *htt,
 	struct ieee80211_tx_info *info;
 	struct ath10k_skb_cb *skb_cb;
 	struct sk_buff *msdu;
-	int ret;
+
+	lockdep_assert_held(&htt->tx_lock);
 
 	ath10k_dbg(ATH10K_DBG_HTT, "htt tx completion msdu_id %u discard %d no_ack %d\n",
 		   tx_done->msdu_id, !!tx_done->discard, !!tx_done->no_ack);
@@ -65,12 +66,12 @@ void ath10k_txrx_tx_unref(struct ath10k_htt *htt,
 	msdu = htt->pending_tx[tx_done->msdu_id];
 	skb_cb = ATH10K_SKB_CB(msdu);
 
-	ret = ath10k_skb_unmap(dev, msdu);
-	if (ret)
-		ath10k_warn("data skb unmap failed (%d)\n", ret);
+	dma_unmap_single(dev, skb_cb->paddr, msdu->len, DMA_TO_DEVICE);
 
-	if (skb_cb->htt.frag_len)
-		skb_pull(msdu, skb_cb->htt.frag_len + skb_cb->htt.pad_len);
+	if (skb_cb->htt.txbuf)
+		dma_pool_free(htt->tx_pool,
+			      skb_cb->htt.txbuf,
+			      skb_cb->htt.txbuf_paddr);
 
 	ath10k_report_offchan_tx(htt->ar, msdu);
 
@@ -92,13 +93,11 @@ void ath10k_txrx_tx_unref(struct ath10k_htt *htt,
 	/* we do not own the msdu anymore */
 
 exit:
-	spin_lock_bh(&htt->tx_lock);
 	htt->pending_tx[tx_done->msdu_id] = NULL;
 	ath10k_htt_tx_free_msdu_id(htt, tx_done->msdu_id);
 	__ath10k_htt_tx_dec_pending(htt);
 	if (htt->num_pending_tx == 0)
 		wake_up(&htt->empty_tx_wq);
-	spin_unlock_bh(&htt->tx_lock);
 }
 
 static const u8 rx_legacy_rate_idx[] = {
@@ -258,6 +257,12 @@ void ath10k_process_rx(struct ath10k *ar, struct htt_rx_info *info)
 	status->band = ch->band;
 	status->freq = ch->center_freq;
 
+	if (info->rate.info0 & HTT_RX_INDICATION_INFO0_END_VALID) {
+		/* TSF available only in 32-bit */
+		status->mactime = info->tsf & 0xffffffff;
+		status->flag |= RX_FLAG_MACTIME_END;
+	}
+
 	ath10k_dbg(ATH10K_DBG_DATA,
 		   "rx skb %p len %u %s%s%s%s%s %srate_idx %u vht_nss %u freq %u band %u flag 0x%x fcs-err %i\n",
 		   info->skb,
@@ -378,7 +383,8 @@ void ath10k_peer_unmap_event(struct ath10k_htt *htt,
 	spin_lock_bh(&ar->data_lock);
 	peer = ath10k_peer_find_by_id(ar, ev->peer_id);
 	if (!peer) {
-		ath10k_warn("unknown peer id %d\n", ev->peer_id);
+		ath10k_warn("peer-unmap-event: unknown peer id %d\n",
+			    ev->peer_id);
 		goto exit;
 	}
 
diff --git a/drivers/net/wireless/ath/ath10k/wmi.c b/drivers/net/wireless/ath/ath10k/wmi.c
index 91e501b5499e..cb1f7b5bcf4c 100644
--- a/drivers/net/wireless/ath/ath10k/wmi.c
+++ b/drivers/net/wireless/ath/ath10k/wmi.c
@@ -1360,7 +1360,7 @@ static void ath10k_wmi_event_host_swba(struct ath10k *ar, struct sk_buff *skb)
 	struct wmi_bcn_info *bcn_info;
 	struct ath10k_vif *arvif;
 	struct sk_buff *bcn;
-	int vdev_id = 0;
+	int ret, vdev_id = 0;
 
 	ath10k_dbg(ATH10K_DBG_MGMT, "WMI_HOST_SWBA_EVENTID\n");
 
@@ -1435,16 +1435,27 @@ static void ath10k_wmi_event_host_swba(struct ath10k *ar, struct sk_buff *skb)
 				ath10k_warn("SWBA overrun on vdev %d\n",
 					    arvif->vdev_id);
 
-			ath10k_skb_unmap(ar->dev, arvif->beacon);
+			dma_unmap_single(arvif->ar->dev,
+					 ATH10K_SKB_CB(arvif->beacon)->paddr,
+					 arvif->beacon->len, DMA_TO_DEVICE);
 			dev_kfree_skb_any(arvif->beacon);
 		}
 
-		ath10k_skb_map(ar->dev, bcn);
+		ATH10K_SKB_CB(bcn)->paddr = dma_map_single(arvif->ar->dev,
+							   bcn->data, bcn->len,
+							   DMA_TO_DEVICE);
+		ret = dma_mapping_error(arvif->ar->dev,
+					ATH10K_SKB_CB(bcn)->paddr);
+		if (ret) {
+			ath10k_warn("failed to map beacon: %d\n", ret);
+			goto skip;
+		}
 
 		arvif->beacon = bcn;
 		arvif->beacon_sent = false;
 
 		ath10k_wmi_tx_beacon_nowait(arvif);
+skip:
 		spin_unlock_bh(&ar->data_lock);
 	}
 }
@@ -3382,7 +3393,6 @@ int ath10k_wmi_scan_chan_list(struct ath10k *ar,
 		ci->max_power         = ch->max_power;
 		ci->reg_power         = ch->max_reg_power;
 		ci->antenna_max       = ch->max_antenna_gain;
-		ci->antenna_max       = 0;
 
 		/* mode & flags share storage */
 		ci->mode              = ch->mode;