All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 0/4] mwifiex: throughput enhancements for TX
@ 2015-02-25 16:50 Avinash Patil
  2015-02-25 16:50 ` [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing Avinash Patil
                   ` (3 more replies)
  0 siblings, 4 replies; 11+ messages in thread
From: Avinash Patil @ 2015-02-25 16:50 UTC (permalink / raw)
  To: linux-wireless; +Cc: akarwar, cluo, liuzy, szli, Avinash Patil

This patch series includes patches which improve TX throughputs.

Marc Yang (1):
  mwifiex: avoid queue_work while work is ongoing

Zhaoyang Liu (3):
  mwifiex: get rid of BA setup helper functions
  mwifiex: remove_bss_prio_lock
  mwifiex: preprocess packets from TX queue

 drivers/net/wireless/mwifiex/11n.c           |  18 +++-
 drivers/net/wireless/mwifiex/11n.h           |  32 -------
 drivers/net/wireless/mwifiex/11n_aggr.c      |  14 ++-
 drivers/net/wireless/mwifiex/11n_rxreorder.c |   7 +-
 drivers/net/wireless/mwifiex/cfg80211.c      |  33 +++++++
 drivers/net/wireless/mwifiex/decl.h          |   2 +
 drivers/net/wireless/mwifiex/init.c          |   5 ++
 drivers/net/wireless/mwifiex/main.c          |  71 ++++++++++++---
 drivers/net/wireless/mwifiex/main.h          |  21 +++--
 drivers/net/wireless/mwifiex/pcie.c          |   2 +-
 drivers/net/wireless/mwifiex/txrx.c          | 126 +++++++++++++++++++++++++++
 drivers/net/wireless/mwifiex/usb.c           |   4 +-
 drivers/net/wireless/mwifiex/wmm.c           |  49 ++++++-----
 drivers/net/wireless/mwifiex/wmm.h           |   2 +
 14 files changed, 305 insertions(+), 81 deletions(-)

-- 
1.8.1.4


^ permalink raw reply	[flat|nested] 11+ messages in thread
* [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing
  2015-02-25 16:50 [PATCH 0/4] mwifiex: throughput enhancements for TX Avinash Patil
@ 2015-02-25 16:50 ` Avinash Patil
  2015-03-03 12:37   ` Kalle Valo
  2015-02-25 16:50 ` [PATCH 2/4] mwifiex: get rid of BA setup helper functions Avinash Patil
                   ` (2 subsequent siblings)
  3 siblings, 1 reply; 11+ messages in thread
From: Avinash Patil @ 2015-02-25 16:50 UTC (permalink / raw)
  To: linux-wireless; +Cc: akarwar, cluo, liuzy, szli, Marc Yang, Avinash Patil

From: Marc Yang <yangyang@marvell.com>

Current code does not check whether main_work_queue or
rx_work_queue is running when preparing to do queue_work,
this code fix add check before calling queue_work, reducing
unnecessary queue_work switch.

This change instead sets more_task flag to ensure we run main_process
superloop once again.

Signed-off-by: Marc Yang <yangyang@marvell.com>
Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
Signed-off-by: Shengzhen Li <szli@marvell.com>
Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
---
 drivers/net/wireless/mwifiex/main.c | 38 +++++++++++++++++++++++++++++++------
 drivers/net/wireless/mwifiex/main.h |  1 +
 drivers/net/wireless/mwifiex/pcie.c |  2 +-
 drivers/net/wireless/mwifiex/usb.c  |  4 ++--
 4 files changed, 36 insertions(+), 9 deletions(-)

diff --git a/drivers/net/wireless/mwifiex/main.c b/drivers/net/wireless/mwifiex/main.c
index 74488ab..447eb6f 100644
--- a/drivers/net/wireless/mwifiex/main.c
+++ b/drivers/net/wireless/mwifiex/main.c
@@ -131,6 +131,34 @@ static int mwifiex_unregister(struct mwifiex_adapter *adapter)
 	return 0;
 }
 
+void mwifiex_queue_main_work(struct mwifiex_adapter *adapter)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adapter->main_proc_lock, flags);
+	if (adapter->mwifiex_processing) {
+		adapter->more_task_flag = true;
+		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
+	} else {
+		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
+		queue_work(adapter->workqueue, &adapter->main_work);
+	}
+}
+EXPORT_SYMBOL_GPL(mwifiex_queue_main_work);
+
+static void mwifiex_queue_rx_work(struct mwifiex_adapter *adapter)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
+	if (adapter->rx_processing) {
+		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
+	} else {
+		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
+		queue_work(adapter->rx_workqueue, &adapter->rx_work);
+	}
+}
+
 static int mwifiex_process_rx(struct mwifiex_adapter *adapter)
 {
 	unsigned long flags;
@@ -154,7 +182,7 @@ static int mwifiex_process_rx(struct mwifiex_adapter *adapter)
 			if (adapter->if_ops.submit_rem_rx_urbs)
 				adapter->if_ops.submit_rem_rx_urbs(adapter);
 			adapter->delay_main_work = false;
-			queue_work(adapter->workqueue, &adapter->main_work);
+			mwifiex_queue_main_work(adapter);
 		}
 		mwifiex_handle_rx_packet(adapter, skb);
 	}
@@ -214,9 +242,7 @@ process_start:
 		if (atomic_read(&adapter->rx_pending) >= HIGH_RX_PENDING &&
 		    adapter->iface_type != MWIFIEX_USB) {
 			adapter->delay_main_work = true;
-			if (!adapter->rx_processing)
-				queue_work(adapter->rx_workqueue,
-					   &adapter->rx_work);
+			mwifiex_queue_rx_work(adapter);
 			break;
 		}
 
@@ -229,7 +255,7 @@ process_start:
 		}
 
 		if (adapter->rx_work_enabled && adapter->data_received)
-			queue_work(adapter->rx_workqueue, &adapter->rx_work);
+			mwifiex_queue_rx_work(adapter);
 
 		/* Need to wake up the card ? */
 		if ((adapter->ps_state == PS_STATE_SLEEP) &&
@@ -606,7 +632,7 @@ int mwifiex_queue_tx_pkt(struct mwifiex_private *priv, struct sk_buff *skb)
 	atomic_inc(&priv->adapter->tx_pending);
 	mwifiex_wmm_add_buf_txqueue(priv, skb);
 
-	queue_work(priv->adapter->workqueue, &priv->adapter->main_work);
+	mwifiex_queue_main_work(priv->adapter);
 
 	return 0;
 }
diff --git a/drivers/net/wireless/mwifiex/main.h b/drivers/net/wireless/mwifiex/main.h
index 16be45e..801a23b 100644
--- a/drivers/net/wireless/mwifiex/main.h
+++ b/drivers/net/wireless/mwifiex/main.h
@@ -1422,6 +1422,7 @@ u8 mwifiex_adjust_data_rate(struct mwifiex_private *priv,
 
 void mwifiex_dump_drv_info(struct mwifiex_adapter *adapter);
 void *mwifiex_alloc_rx_buf(int rx_len, gfp_t flags);
+void mwifiex_queue_main_work(struct mwifiex_adapter *adapter);
 
 #ifdef CONFIG_DEBUG_FS
 void mwifiex_debugfs_init(void);
diff --git a/drivers/net/wireless/mwifiex/pcie.c b/drivers/net/wireless/mwifiex/pcie.c
index 4b463c3..c0ff33e 100644
--- a/drivers/net/wireless/mwifiex/pcie.c
+++ b/drivers/net/wireless/mwifiex/pcie.c
@@ -2101,7 +2101,7 @@ static irqreturn_t mwifiex_pcie_interrupt(int irq, void *context)
 		goto exit;
 
 	mwifiex_interrupt_status(adapter);
-	queue_work(adapter->workqueue, &adapter->main_work);
+	mwifiex_queue_main_work(adapter);
 
 exit:
 	return IRQ_HANDLED;
diff --git a/drivers/net/wireless/mwifiex/usb.c b/drivers/net/wireless/mwifiex/usb.c
index 2238730..3adf1f6 100644
--- a/drivers/net/wireless/mwifiex/usb.c
+++ b/drivers/net/wireless/mwifiex/usb.c
@@ -193,7 +193,7 @@ static void mwifiex_usb_rx_complete(struct urb *urb)
 		dev_dbg(adapter->dev, "info: recv_length=%d, status=%d\n",
 			recv_length, status);
 		if (status == -EINPROGRESS) {
-			queue_work(adapter->workqueue, &adapter->main_work);
+			mwifiex_queue_main_work(adapter);
 
 			/* urb for data_ep is re-submitted now;
 			 * urb for cmd_ep will be re-submitted in callback
@@ -262,7 +262,7 @@ static void mwifiex_usb_tx_complete(struct urb *urb)
 					    urb->status ? -1 : 0);
 	}
 
-	queue_work(adapter->workqueue, &adapter->main_work);
+	mwifiex_queue_main_work(adapter);
 
 	return;
 }
-- 
1.8.1.4


^ permalink raw reply related	[flat|nested] 11+ messages in thread
* [PATCH 2/4] mwifiex: get rid of BA setup helper functions
  2015-02-25 16:50 [PATCH 0/4] mwifiex: throughput enhancements for TX Avinash Patil
  2015-02-25 16:50 ` [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing Avinash Patil
@ 2015-02-25 16:50 ` Avinash Patil
  2015-02-25 16:50 ` [PATCH 3/4] mwifiex: remove_bss_prio_lock Avinash Patil
  2015-02-25 16:50 ` [PATCH 4/4] mwifiex: preprocess packets from TX queue Avinash Patil
  3 siblings, 0 replies; 11+ messages in thread
From: Avinash Patil @ 2015-02-25 16:50 UTC (permalink / raw)
  To: linux-wireless; +Cc: akarwar, cluo, liuzy, szli, Avinash Patil

From: Zhaoyang Liu <liuzy@marvell.com>

This patch removes BA setup helper routines
mwifiex_is_bastream_setup and  mwifiex_is_amsdu_in_ampdu_allowed.

Current code will use two functions to check bastream setup and
amsdu in ampdu. This patch change these functions to flags, thus
avoiding redundant spin_lock check while dequeuing TX packets.

Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
Signed-off-by: Shengzhen Li <szli@marvell.com>
Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
---
 drivers/net/wireless/mwifiex/11n.c           | 18 +++++++++++++++-
 drivers/net/wireless/mwifiex/11n.h           | 32 ----------------------------
 drivers/net/wireless/mwifiex/11n_rxreorder.c |  7 +++++-
 drivers/net/wireless/mwifiex/main.h          | 13 ++++++-----
 drivers/net/wireless/mwifiex/wmm.c           | 18 +++++++++-------
 drivers/net/wireless/mwifiex/wmm.h           |  2 ++
 6 files changed, 43 insertions(+), 47 deletions(-)

diff --git a/drivers/net/wireless/mwifiex/11n.c b/drivers/net/wireless/mwifiex/11n.c
index 543148d..433bd68 100644
--- a/drivers/net/wireless/mwifiex/11n.c
+++ b/drivers/net/wireless/mwifiex/11n.c
@@ -159,6 +159,7 @@ int mwifiex_ret_11n_addba_req(struct mwifiex_private *priv,
 	int tid;
 	struct host_cmd_ds_11n_addba_rsp *add_ba_rsp = &resp->params.add_ba_rsp;
 	struct mwifiex_tx_ba_stream_tbl *tx_ba_tbl;
+	struct mwifiex_ra_list_tbl *ra_list;
 	u16 block_ack_param_set = le16_to_cpu(add_ba_rsp->block_ack_param_set);
 
 	add_ba_rsp->ssn = cpu_to_le16((le16_to_cpu(add_ba_rsp->ssn))
@@ -166,7 +167,13 @@ int mwifiex_ret_11n_addba_req(struct mwifiex_private *priv,
 
 	tid = (block_ack_param_set & IEEE80211_ADDBA_PARAM_TID_MASK)
 	       >> BLOCKACKPARAM_TID_POS;
+	ra_list = mwifiex_wmm_get_ralist_node(priv, tid, add_ba_rsp->
+		peer_mac_addr);
 	if (le16_to_cpu(add_ba_rsp->status_code) != BA_RESULT_SUCCESS) {
+		if (ra_list) {
+			ra_list->ba_status = BA_SETUP_NONE;
+			ra_list->amsdu_in_ampdu = false;
+		}
 		mwifiex_del_ba_tbl(priv, tid, add_ba_rsp->peer_mac_addr,
 				   TYPE_DELBA_SENT, true);
 		if (add_ba_rsp->add_rsp_result != BA_RESULT_TIMEOUT)
@@ -185,6 +192,10 @@ int mwifiex_ret_11n_addba_req(struct mwifiex_private *priv,
 			tx_ba_tbl->amsdu = true;
 		else
 			tx_ba_tbl->amsdu = false;
+		if (ra_list) {
+			ra_list->amsdu_in_ampdu = tx_ba_tbl->amsdu;
+			ra_list->ba_status = BA_SETUP_COMPLETE;
+		}
 	} else {
 		dev_err(priv->adapter->dev, "BA stream not created\n");
 	}
@@ -515,6 +526,7 @@ void mwifiex_create_ba_tbl(struct mwifiex_private *priv, u8 *ra, int tid,
 			   enum mwifiex_ba_status ba_status)
 {
 	struct mwifiex_tx_ba_stream_tbl *new_node;
+	struct mwifiex_ra_list_tbl *ra_list;
 	unsigned long flags;
 
 	if (!mwifiex_get_ba_tbl(priv, tid, ra)) {
@@ -522,7 +534,11 @@ void mwifiex_create_ba_tbl(struct mwifiex_private *priv, u8 *ra, int tid,
 				   GFP_ATOMIC);
 		if (!new_node)
 			return;
-
+		ra_list = mwifiex_wmm_get_ralist_node(priv, tid, ra);
+		if (ra_list) {
+			ra_list->ba_status = ba_status;
+			ra_list->amsdu_in_ampdu = false;
+		}
 		INIT_LIST_HEAD(&new_node->list);
 
 		new_node->tid = tid;
diff --git a/drivers/net/wireless/mwifiex/11n.h b/drivers/net/wireless/mwifiex/11n.h
index 8e2e394..afdd58a 100644
--- a/drivers/net/wireless/mwifiex/11n.h
+++ b/drivers/net/wireless/mwifiex/11n.h
@@ -77,22 +77,6 @@ mwifiex_is_station_ampdu_allowed(struct mwifiex_private *priv,
 	return (node->ampdu_sta[tid] != BA_STREAM_NOT_ALLOWED) ? true : false;
 }
 
-/* This function checks whether AMSDU is allowed for BA stream. */
-static inline u8
-mwifiex_is_amsdu_in_ampdu_allowed(struct mwifiex_private *priv,
-				  struct mwifiex_ra_list_tbl *ptr, int tid)
-{
-	struct mwifiex_tx_ba_stream_tbl *tx_tbl;
-
-	if (is_broadcast_ether_addr(ptr->ra))
-		return false;
-	tx_tbl = mwifiex_get_ba_tbl(priv, tid, ptr->ra);
-	if (tx_tbl)
-		return tx_tbl->amsdu;
-
-	return false;
-}
-
 /* This function checks whether AMPDU is allowed or not for a particular TID. */
 static inline u8
 mwifiex_is_ampdu_allowed(struct mwifiex_private *priv,
@@ -182,22 +166,6 @@ mwifiex_find_stream_to_delete(struct mwifiex_private *priv, int ptr_tid,
 }
 
 /*
- * This function checks whether BA stream is set up or not.
- */
-static inline int
-mwifiex_is_ba_stream_setup(struct mwifiex_private *priv,
-			   struct mwifiex_ra_list_tbl *ptr, int tid)
-{
-	struct mwifiex_tx_ba_stream_tbl *tx_tbl;
-
-	tx_tbl = mwifiex_get_ba_tbl(priv, tid, ptr->ra);
-	if (tx_tbl && IS_BASTREAM_SETUP(tx_tbl))
-		return true;
-
-	return false;
-}
-
-/*
  * This function checks whether associated station is 11n enabled
  */
 static inline int mwifiex_is_sta_11n_enabled(struct mwifiex_private *priv,
diff --git a/drivers/net/wireless/mwifiex/11n_rxreorder.c b/drivers/net/wireless/mwifiex/11n_rxreorder.c
index a2e8817..f75f8ac 100644
--- a/drivers/net/wireless/mwifiex/11n_rxreorder.c
+++ b/drivers/net/wireless/mwifiex/11n_rxreorder.c
@@ -659,6 +659,7 @@ mwifiex_del_ba_tbl(struct mwifiex_private *priv, int tid, u8 *peer_mac,
 {
 	struct mwifiex_rx_reorder_tbl *tbl;
 	struct mwifiex_tx_ba_stream_tbl *ptx_tbl;
+	struct mwifiex_ra_list_tbl *ra_list;
 	u8 cleanup_rx_reorder_tbl;
 	unsigned long flags;
 
@@ -686,7 +687,11 @@ mwifiex_del_ba_tbl(struct mwifiex_private *priv, int tid, u8 *peer_mac,
 				"event: TID, RA not found in table\n");
 			return;
 		}
-
+		ra_list = mwifiex_wmm_get_ralist_node(priv, tid, peer_mac);
+		if (ra_list) {
+			ra_list->amsdu_in_ampdu = false;
+			ra_list->ba_status = BA_SETUP_NONE;
+		}
 		spin_lock_irqsave(&priv->tx_ba_stream_tbl_lock, flags);
 		mwifiex_11n_delete_tx_ba_stream_tbl_entry(priv, ptx_tbl);
 		spin_unlock_irqrestore(&priv->tx_ba_stream_tbl_lock, flags);
diff --git a/drivers/net/wireless/mwifiex/main.h b/drivers/net/wireless/mwifiex/main.h
index 801a23b..b47bffa 100644
--- a/drivers/net/wireless/mwifiex/main.h
+++ b/drivers/net/wireless/mwifiex/main.h
@@ -210,6 +210,12 @@ struct mwifiex_tx_aggr {
 	u8 amsdu;
 };
 
+enum mwifiex_ba_status {
+	BA_SETUP_NONE = 0,
+	BA_SETUP_INPROGRESS,
+	BA_SETUP_COMPLETE
+};
+
 struct mwifiex_ra_list_tbl {
 	struct list_head list;
 	struct sk_buff_head skb_head;
@@ -218,6 +224,8 @@ struct mwifiex_ra_list_tbl {
 	u16 max_amsdu;
 	u16 ba_pkt_count;
 	u8 ba_packet_thr;
+	enum mwifiex_ba_status ba_status;
+	u8 amsdu_in_ampdu;
 	u16 total_pkt_count;
 	bool tdls_link;
 };
@@ -601,11 +609,6 @@ struct mwifiex_private {
 	struct mwifiex_11h_intf_state state_11h;
 };
 
-enum mwifiex_ba_status {
-	BA_SETUP_NONE = 0,
-	BA_SETUP_INPROGRESS,
-	BA_SETUP_COMPLETE
-};
 
 struct mwifiex_tx_ba_stream_tbl {
 	struct list_head list;
diff --git a/drivers/net/wireless/mwifiex/wmm.c b/drivers/net/wireless/mwifiex/wmm.c
index ef717ac..ba75a45 100644
--- a/drivers/net/wireless/mwifiex/wmm.c
+++ b/drivers/net/wireless/mwifiex/wmm.c
@@ -157,6 +157,8 @@ void mwifiex_ralist_add(struct mwifiex_private *priv, const u8 *ra)
 
 		ra_list->is_11n_enabled = 0;
 		ra_list->tdls_link = false;
+		ra_list->ba_status = BA_SETUP_NONE;
+		ra_list->amsdu_in_ampdu = false;
 		if (!mwifiex_queuing_ra_based(priv)) {
 			if (mwifiex_get_tdls_link_status(priv, ra) ==
 			    TDLS_SETUP_COMPLETE) {
@@ -574,7 +576,7 @@ mwifiex_clean_txrx(struct mwifiex_private *priv)
  * This function retrieves a particular RA list node, matching with the
  * given TID and RA address.
  */
-static struct mwifiex_ra_list_tbl *
+struct mwifiex_ra_list_tbl *
 mwifiex_wmm_get_ralist_node(struct mwifiex_private *priv, u8 tid,
 			    const u8 *ra_addr)
 {
@@ -1276,14 +1278,14 @@ mwifiex_dequeue_tx_packet(struct mwifiex_adapter *adapter)
 	}
 
 	if (!ptr->is_11n_enabled ||
-	    mwifiex_is_ba_stream_setup(priv, ptr, tid) ||
-	    priv->wps.session_enable) {
+		ptr->ba_status ||
+		priv->wps.session_enable) {
 		if (ptr->is_11n_enabled &&
-		    mwifiex_is_ba_stream_setup(priv, ptr, tid) &&
-		    mwifiex_is_amsdu_in_ampdu_allowed(priv, ptr, tid) &&
-		    mwifiex_is_amsdu_allowed(priv, tid) &&
-		    mwifiex_is_11n_aggragation_possible(priv, ptr,
-							adapter->tx_buf_size))
+			ptr->ba_status &&
+			ptr->amsdu_in_ampdu &&
+			mwifiex_is_amsdu_allowed(priv, tid) &&
+			mwifiex_is_11n_aggragation_possible(priv, ptr,
+							  adapter->tx_buf_size))
 			mwifiex_11n_aggregate_pkt(priv, ptr, ptr_index, flags);
 			/* ra_list_spinlock has been freed in
 			 * mwifiex_11n_aggregate_pkt()
diff --git a/drivers/net/wireless/mwifiex/wmm.h b/drivers/net/wireless/mwifiex/wmm.h
index 569bd73..48ece0b 100644
--- a/drivers/net/wireless/mwifiex/wmm.h
+++ b/drivers/net/wireless/mwifiex/wmm.h
@@ -127,4 +127,6 @@ mwifiex_wmm_get_queue_raptr(struct mwifiex_private *priv, u8 tid,
 			    const u8 *ra_addr);
 u8 mwifiex_wmm_downgrade_tid(struct mwifiex_private *priv, u32 tid);
 
+struct mwifiex_ra_list_tbl *mwifiex_wmm_get_ralist_node(struct mwifiex_private
+					*priv, u8 tid, const u8 *ra_addr);
 #endif /* !_MWIFIEX_WMM_H_ */
-- 
1.8.1.4


^ permalink raw reply related	[flat|nested] 11+ messages in thread
* [PATCH 3/4] mwifiex: remove_bss_prio_lock
  2015-02-25 16:50 [PATCH 0/4] mwifiex: throughput enhancements for TX Avinash Patil
  2015-02-25 16:50 ` [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing Avinash Patil
  2015-02-25 16:50 ` [PATCH 2/4] mwifiex: get rid of BA setup helper functions Avinash Patil
@ 2015-02-25 16:50 ` Avinash Patil
  2015-03-03 12:39   ` Kalle Valo
  2015-02-25 16:50 ` [PATCH 4/4] mwifiex: preprocess packets from TX queue Avinash Patil
  3 siblings, 1 reply; 11+ messages in thread
From: Avinash Patil @ 2015-02-25 16:50 UTC (permalink / raw)
  To: linux-wireless; +Cc: akarwar, cluo, liuzy, szli, Amitkumar Karwar

From: Zhaoyang Liu <liuzy@marvell.com>

This patch does away with spinlock in
mwifiex_wmm_get_highest_priolist_ptr in order to improve TP.

Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
Signed-off-by: Shengzhen Li <szli@marvell.com>
Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Amitkumar Karwar <patila@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
---
 drivers/net/wireless/mwifiex/cfg80211.c | 33 +++++++++++++++++++++++++++++++++
 drivers/net/wireless/mwifiex/main.c     |  2 +-
 drivers/net/wireless/mwifiex/main.h     |  1 +
 drivers/net/wireless/mwifiex/wmm.c      | 11 ++---------
 4 files changed, 37 insertions(+), 10 deletions(-)

diff --git a/drivers/net/wireless/mwifiex/cfg80211.c b/drivers/net/wireless/mwifiex/cfg80211.c
index 5f3c1d3..fb55092 100644
--- a/drivers/net/wireless/mwifiex/cfg80211.c
+++ b/drivers/net/wireless/mwifiex/cfg80211.c
@@ -717,6 +717,9 @@ mwifiex_cfg80211_init_p2p_go(struct mwifiex_private *priv)
 
 static int mwifiex_deinit_priv_params(struct mwifiex_private *priv)
 {
+	struct mwifiex_adapter *adapter = priv->adapter;
+	unsigned long flags;
+
 	priv->mgmt_frame_mask = 0;
 	if (mwifiex_send_cmd(priv, HostCmd_CMD_MGMT_FRAME_REG,
 			     HostCmd_ACT_GEN_SET, 0,
@@ -727,6 +730,25 @@ static int mwifiex_deinit_priv_params(struct mwifiex_private *priv)
 	}
 
 	mwifiex_deauthenticate(priv, NULL);
+
+	spin_lock_irqsave(&adapter->main_proc_lock, flags);
+	adapter->main_locked = true;
+	if (adapter->mwifiex_processing) {
+		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
+		flush_workqueue(adapter->workqueue);
+	} else {
+		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
+	}
+
+	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
+	adapter->rx_locked = true;
+	if (adapter->rx_processing) {
+		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
+		flush_workqueue(adapter->rx_workqueue);
+	} else {
+		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
+	}
+
 	mwifiex_free_priv(priv);
 	priv->wdev.iftype = NL80211_IFTYPE_UNSPECIFIED;
 	priv->bss_mode = NL80211_IFTYPE_UNSPECIFIED;
@@ -740,6 +762,9 @@ mwifiex_init_new_priv_params(struct mwifiex_private *priv,
 			     struct net_device *dev,
 			     enum nl80211_iftype type)
 {
+	struct mwifiex_adapter *adapter = priv->adapter;
+	unsigned long flags;
+
 	mwifiex_init_priv(priv);
 
 	priv->bss_mode = type;
@@ -770,6 +795,14 @@ mwifiex_init_new_priv_params(struct mwifiex_private *priv,
 		return -EOPNOTSUPP;
 	}
 
+	spin_lock_irqsave(&adapter->main_proc_lock, flags);
+	adapter->main_locked = false;
+	spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
+
+	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
+	adapter->rx_locked = false;
+	spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
+
 	return 0;
 }
 
diff --git a/drivers/net/wireless/mwifiex/main.c b/drivers/net/wireless/mwifiex/main.c
index 447eb6f..46e1789 100644
--- a/drivers/net/wireless/mwifiex/main.c
+++ b/drivers/net/wireless/mwifiex/main.c
@@ -217,7 +217,7 @@ int mwifiex_main_process(struct mwifiex_adapter *adapter)
 	spin_lock_irqsave(&adapter->main_proc_lock, flags);
 
 	/* Check if already processing */
-	if (adapter->mwifiex_processing) {
+	if (adapter->mwifiex_processing || adapter->main_locked) {
 		adapter->more_task_flag = true;
 		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
 		goto exit_main_proc;
diff --git a/drivers/net/wireless/mwifiex/main.h b/drivers/net/wireless/mwifiex/main.h
index b47bffa..cc6a623 100644
--- a/drivers/net/wireless/mwifiex/main.h
+++ b/drivers/net/wireless/mwifiex/main.h
@@ -773,6 +773,7 @@ struct mwifiex_adapter {
 	bool rx_work_enabled;
 	bool rx_processing;
 	bool delay_main_work;
+	bool main_locked;
 	bool rx_locked;
 	struct mwifiex_bss_prio_tbl bss_prio_tbl[MWIFIEX_MAX_BSS_NUM];
 	/* spin lock for init/shutdown */
diff --git a/drivers/net/wireless/mwifiex/wmm.c b/drivers/net/wireless/mwifiex/wmm.c
index ba75a45..ce747f7 100644
--- a/drivers/net/wireless/mwifiex/wmm.c
+++ b/drivers/net/wireless/mwifiex/wmm.c
@@ -944,14 +944,11 @@ mwifiex_wmm_get_highest_priolist_ptr(struct mwifiex_adapter *adapter,
 	struct mwifiex_ra_list_tbl *ptr;
 	struct mwifiex_tid_tbl *tid_ptr;
 	atomic_t *hqp;
-	unsigned long flags_bss, flags_ra;
+	unsigned long flags_ra;
 	int i, j;
 
 	/* check the BSS with highest priority first */
 	for (j = adapter->priv_num - 1; j >= 0; --j) {
-		spin_lock_irqsave(&adapter->bss_prio_tbl[j].bss_prio_lock,
-				  flags_bss);
-
 		/* iterate over BSS with the equal priority */
 		list_for_each_entry(adapter->bss_prio_tbl[j].bss_prio_cur,
 				    &adapter->bss_prio_tbl[j].bss_prio_head,
@@ -987,19 +984,15 @@ mwifiex_wmm_get_highest_priolist_ptr(struct mwifiex_adapter *adapter,
 			}
 		}
 
-		spin_unlock_irqrestore(&adapter->bss_prio_tbl[j].bss_prio_lock,
-				       flags_bss);
 	}
 
 	return NULL;
 
 found:
-	/* holds bss_prio_lock / ra_list_spinlock */
+	/* holds ra_list_spinlock */
 	if (atomic_read(hqp) > i)
 		atomic_set(hqp, i);
 	spin_unlock_irqrestore(&priv_tmp->wmm.ra_list_spinlock, flags_ra);
-	spin_unlock_irqrestore(&adapter->bss_prio_tbl[j].bss_prio_lock,
-			       flags_bss);
 
 	*priv = priv_tmp;
 	*tid = tos_to_tid[i];
-- 
1.8.1.4


^ permalink raw reply related	[flat|nested] 11+ messages in thread
* [PATCH 4/4] mwifiex: preprocess packets from TX queue
  2015-02-25 16:50 [PATCH 0/4] mwifiex: throughput enhancements for TX Avinash Patil
                   ` (2 preceding siblings ...)
  2015-02-25 16:50 ` [PATCH 3/4] mwifiex: remove_bss_prio_lock Avinash Patil
@ 2015-02-25 16:50 ` Avinash Patil
  3 siblings, 0 replies; 11+ messages in thread
From: Avinash Patil @ 2015-02-25 16:50 UTC (permalink / raw)
  To: linux-wireless; +Cc: akarwar, cluo, liuzy, szli, Marc Yang, Avinash Patil

From: Zhaoyang Liu <liuzy@marvell.com>

During profiling, we discovered that driver remains idle for time
when pakcet is downloaded to FW but no TX_DONE has been received
i.e. while data_sent is true.

This patch adds enhancement to TX routine where we preprocess
packets from TX queue, make them ready for TX and add them to
separate TX queue.

Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
Signed-off-by: Marc Yang <yangyang@marvell.com>
Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
---
 drivers/net/wireless/mwifiex/11n_aggr.c |  14 +++-
 drivers/net/wireless/mwifiex/decl.h     |   2 +
 drivers/net/wireless/mwifiex/init.c     |   5 ++
 drivers/net/wireless/mwifiex/main.c     |  31 ++++++--
 drivers/net/wireless/mwifiex/main.h     |   6 ++
 drivers/net/wireless/mwifiex/txrx.c     | 126 ++++++++++++++++++++++++++++++++
 drivers/net/wireless/mwifiex/wmm.c      |  20 ++++-
 7 files changed, 189 insertions(+), 15 deletions(-)

diff --git a/drivers/net/wireless/mwifiex/11n_aggr.c b/drivers/net/wireless/mwifiex/11n_aggr.c
index 9b983b5..d8558a6 100644
--- a/drivers/net/wireless/mwifiex/11n_aggr.c
+++ b/drivers/net/wireless/mwifiex/11n_aggr.c
@@ -175,6 +175,7 @@ mwifiex_11n_aggregate_pkt(struct mwifiex_private *priv,
 	struct txpd *ptx_pd = NULL;
 	struct timeval tv;
 	int headroom = adapter->iface_type == MWIFIEX_USB ? 0 : INTF_HEADER_LEN;
+	int aggr_num = 0;
 
 	skb_src = skb_peek(&pra_list->skb_head);
 	if (!skb_src) {
@@ -200,7 +201,8 @@ mwifiex_11n_aggregate_pkt(struct mwifiex_private *priv,
 
 	if (tx_info_src->flags & MWIFIEX_BUF_FLAG_TDLS_PKT)
 		tx_info_aggr->flags |= MWIFIEX_BUF_FLAG_TDLS_PKT;
-	skb_aggr->priority = skb_src->priority;
+		tx_info_aggr->flags |= MWIFIEX_BUF_FLAG_AGGR_PKT;
+		skb_aggr->priority = skb_src->priority;
 
 	do_gettimeofday(&tv);
 	skb_aggr->tstamp = timeval_to_ktime(tv);
@@ -211,11 +213,9 @@ mwifiex_11n_aggregate_pkt(struct mwifiex_private *priv,
 			break;
 
 		skb_src = skb_dequeue(&pra_list->skb_head);
-
 		pra_list->total_pkt_count--;
-
 		atomic_dec(&priv->wmm.tx_pkts_queued);
-
+		aggr_num++;
 		spin_unlock_irqrestore(&priv->wmm.ra_list_spinlock,
 				       ra_list_flags);
 		mwifiex_11n_form_amsdu_pkt(skb_aggr, skb_src, &pad);
@@ -251,6 +251,12 @@ mwifiex_11n_aggregate_pkt(struct mwifiex_private *priv,
 		ptx_pd = (struct txpd *)skb_aggr->data;
 
 	skb_push(skb_aggr, headroom);
+	tx_info_aggr->aggr_num = aggr_num;
+	if (adapter->data_sent || adapter->tx_lock_flag) {
+		atomic_add(aggr_num, &adapter->tx_queued);
+		skb_queue_tail(&adapter->tx_data_q, skb_aggr);
+		return 0;
+	}
 
 	if (adapter->iface_type == MWIFIEX_USB) {
 		adapter->data_sent = true;
diff --git a/drivers/net/wireless/mwifiex/decl.h b/drivers/net/wireless/mwifiex/decl.h
index cf2fa11..f530207 100644
--- a/drivers/net/wireless/mwifiex/decl.h
+++ b/drivers/net/wireless/mwifiex/decl.h
@@ -83,6 +83,7 @@
 #define MWIFIEX_BUF_FLAG_TDLS_PKT	   BIT(2)
 #define MWIFIEX_BUF_FLAG_EAPOL_TX_STATUS   BIT(3)
 #define MWIFIEX_BUF_FLAG_ACTION_TX_STATUS  BIT(4)
+#define MWIFIEX_BUF_FLAG_AGGR_PKT	   BIT(5)
 
 #define MWIFIEX_BRIDGED_PKTS_THR_HIGH      1024
 #define MWIFIEX_BRIDGED_PKTS_THR_LOW        128
@@ -179,6 +180,7 @@ struct mwifiex_txinfo {
 	u8 flags;
 	u8 bss_num;
 	u8 bss_type;
+	u8 aggr_num;
 	u32 pkt_len;
 	u8 ack_frame_id;
 	u64 cookie;
diff --git a/drivers/net/wireless/mwifiex/init.c b/drivers/net/wireless/mwifiex/init.c
index 2e1df02..2f278a6 100644
--- a/drivers/net/wireless/mwifiex/init.c
+++ b/drivers/net/wireless/mwifiex/init.c
@@ -481,6 +481,7 @@ int mwifiex_init_lock_list(struct mwifiex_adapter *adapter)
 	spin_lock_init(&adapter->rx_proc_lock);
 
 	skb_queue_head_init(&adapter->rx_data_q);
+	skb_queue_head_init(&adapter->tx_data_q);
 
 	for (i = 0; i < adapter->priv_num; ++i) {
 		INIT_LIST_HEAD(&adapter->bss_prio_tbl[i].bss_prio_head);
@@ -688,6 +689,10 @@ mwifiex_shutdown_drv(struct mwifiex_adapter *adapter)
 		}
 	}
 
+	atomic_set(&adapter->tx_queued, 0);
+	while ((skb = skb_dequeue(&adapter->tx_data_q)))
+		mwifiex_write_data_complete(adapter, skb, 0, 0);
+
 	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
 
 	while ((skb = skb_dequeue(&adapter->rx_data_q))) {
diff --git a/drivers/net/wireless/mwifiex/main.c b/drivers/net/wireless/mwifiex/main.c
index 46e1789..dcc32c1 100644
--- a/drivers/net/wireless/mwifiex/main.c
+++ b/drivers/net/wireless/mwifiex/main.c
@@ -259,10 +259,11 @@ process_start:
 
 		/* Need to wake up the card ? */
 		if ((adapter->ps_state == PS_STATE_SLEEP) &&
-		    (adapter->pm_wakeup_card_req &&
-		     !adapter->pm_wakeup_fw_try) &&
-		    (is_command_pending(adapter) ||
-		     !mwifiex_wmm_lists_empty(adapter))) {
+			(adapter->pm_wakeup_card_req &&
+			!adapter->pm_wakeup_fw_try) &&
+			(is_command_pending(adapter) ||
+			!skb_queue_empty(&adapter->tx_data_q) ||
+			!mwifiex_wmm_lists_empty(adapter))) {
 			adapter->pm_wakeup_fw_try = true;
 			mod_timer(&adapter->wakeup_timer, jiffies + (HZ*3));
 			adapter->if_ops.wakeup(adapter);
@@ -286,7 +287,8 @@ process_start:
 
 			if ((!adapter->scan_chan_gap_enabled &&
 			     adapter->scan_processing) || adapter->data_sent ||
-			    mwifiex_wmm_lists_empty(adapter)) {
+			     (skb_queue_empty(&adapter->tx_data_q) &&
+			      mwifiex_wmm_lists_empty(adapter))) {
 				if (adapter->cmd_sent || adapter->curr_cmd ||
 				    (!is_command_pending(adapter)))
 					break;
@@ -335,6 +337,19 @@ process_start:
 				break;
 			}
 		}
+		if ((adapter->scan_chan_gap_enabled ||
+		     !adapter->scan_processing) &&
+		    !skb_queue_empty(&adapter->tx_data_q) &&
+		    !adapter->data_sent) {
+			mwifiex_process_tx_queue(adapter);
+			if (adapter->hs_activated) {
+				adapter->is_hs_configured = false;
+				mwifiex_hs_activated_event
+					(mwifiex_get_priv
+					 (adapter, MWIFIEX_BSS_ROLE_ANY),
+					 false);
+			}
+		}
 
 		if ((adapter->scan_chan_gap_enabled ||
 		     !adapter->scan_processing) &&
@@ -350,8 +365,10 @@ process_start:
 		}
 
 		if (adapter->delay_null_pkt && !adapter->cmd_sent &&
-		    !adapter->curr_cmd && !is_command_pending(adapter) &&
-		    mwifiex_wmm_lists_empty(adapter)) {
+			!adapter->curr_cmd && !is_command_pending(adapter) &&
+			(mwifiex_wmm_lists_empty(adapter) &&
+			skb_queue_empty(&adapter->tx_data_q))
+		) {
 			if (!mwifiex_send_null_packet
 			    (mwifiex_get_priv(adapter, MWIFIEX_BSS_ROLE_STA),
 			     MWIFIEX_TxPD_POWER_MGMT_NULL_PACKET |
diff --git a/drivers/net/wireless/mwifiex/main.h b/drivers/net/wireless/mwifiex/main.h
index cc6a623..3da99e8 100644
--- a/drivers/net/wireless/mwifiex/main.h
+++ b/drivers/net/wireless/mwifiex/main.h
@@ -58,6 +58,8 @@ enum {
 
 #define MWIFIEX_MAX_AP				64
 
+#define MWIFIEX_MAX_PKTS_TXQ			16
+
 #define MWIFIEX_DEFAULT_WATCHDOG_TIMEOUT	(5 * HZ)
 
 #define MWIFIEX_TIMER_10S			10000
@@ -819,6 +821,8 @@ struct mwifiex_adapter {
 	/* spin lock for RX processing routine */
 	spinlock_t rx_proc_lock;
 	u32 scan_processing;
+	struct sk_buff_head tx_data_q;
+	atomic_t tx_queued;
 	u16 region_code;
 	struct mwifiex_802_11d_domain_reg domain_reg;
 	u16 scan_probes;
@@ -904,6 +908,8 @@ struct mwifiex_adapter {
 	bool auto_tdls;
 };
 
+void mwifiex_process_tx_queue(struct mwifiex_adapter *adapter);
+
 int mwifiex_init_lock_list(struct mwifiex_adapter *adapter);
 
 void mwifiex_set_trans_start(struct net_device *dev);
diff --git a/drivers/net/wireless/mwifiex/txrx.c b/drivers/net/wireless/mwifiex/txrx.c
index ea4549f..e55a7fa 100644
--- a/drivers/net/wireless/mwifiex/txrx.c
+++ b/drivers/net/wireless/mwifiex/txrx.c
@@ -92,6 +92,12 @@ int mwifiex_process_tx(struct mwifiex_private *priv, struct sk_buff *skb,
 	else
 		head_ptr = mwifiex_process_sta_txpd(priv, skb);
 
+	if ((adapter->data_sent || adapter->tx_lock_flag) && head_ptr) {
+		skb_queue_tail(&adapter->tx_data_q, skb);
+		atomic_inc(&adapter->tx_queued);
+		return 0;
+	}
+
 	if (head_ptr) {
 		if (GET_BSS_ROLE(priv) == MWIFIEX_BSS_ROLE_STA)
 			local_tx_pd = (struct txpd *)(head_ptr + hroom);
@@ -142,6 +148,124 @@ int mwifiex_process_tx(struct mwifiex_private *priv, struct sk_buff *skb,
 	return ret;
 }
 
+static int
+mwifiex_host_to_card(struct mwifiex_adapter *adapter, struct sk_buff *skb,
+		     struct mwifiex_tx_param *tx_param)
+{
+	struct txpd *local_tx_pd = NULL;
+	u8 *head_ptr = skb->data;
+	int ret = 0;
+	struct mwifiex_private *priv;
+	struct mwifiex_txinfo *tx_info;
+
+	tx_info = MWIFIEX_SKB_TXCB(skb);
+
+	priv = mwifiex_get_priv_by_id(adapter, tx_info->bss_num,
+				      tx_info->bss_type);
+	if (!priv) {
+		dev_err(adapter->dev, "data: priv not found. Drop TX packet\n");
+		adapter->dbg.num_tx_host_to_card_failure++;
+		mwifiex_write_data_complete(adapter, skb, 0, 0);
+		return ret;
+	}
+	if (GET_BSS_ROLE(priv) == MWIFIEX_BSS_ROLE_STA) {
+		if (adapter->iface_type == MWIFIEX_USB)
+			local_tx_pd = (struct txpd *)head_ptr;
+		else
+			local_tx_pd = (struct txpd *) (head_ptr +
+				INTF_HEADER_LEN);
+	}
+
+	if (adapter->iface_type == MWIFIEX_USB) {
+		adapter->data_sent = true;
+		ret = adapter->if_ops.host_to_card(adapter,
+						   MWIFIEX_USB_EP_DATA,
+						   skb, NULL);
+	} else {
+		ret = adapter->if_ops.host_to_card(adapter,
+						   MWIFIEX_TYPE_DATA,
+						   skb, tx_param);
+	}
+	switch (ret) {
+	case -ENOSR:
+		dev_err(adapter->dev, "data: -ENOSR is returned\n");
+		break;
+	case -EBUSY:
+		if ((GET_BSS_ROLE(priv) == MWIFIEX_BSS_ROLE_STA) &&
+		    (adapter->pps_uapsd_mode && adapter->tx_lock_flag)) {
+				priv->adapter->tx_lock_flag = false;
+			if (local_tx_pd)
+				local_tx_pd->flags = 0;
+		}
+		skb_queue_head(&adapter->tx_data_q, skb);
+
+		if (tx_info->flags & MWIFIEX_BUF_FLAG_AGGR_PKT)
+			atomic_add(tx_info->aggr_num, &adapter->tx_queued);
+		else
+			atomic_inc(&adapter->tx_queued);
+
+		dev_dbg(adapter->dev, "data: -EBUSY is returned\n");
+		break;
+	case -1:
+		if (adapter->iface_type != MWIFIEX_PCIE)
+			adapter->data_sent = false;
+		dev_err(adapter->dev, "mwifiex_write_data_async failed: 0x%X\n",
+			ret);
+		adapter->dbg.num_tx_host_to_card_failure++;
+		mwifiex_write_data_complete(adapter, skb, 0, ret);
+		break;
+	case -EINPROGRESS:
+		if (adapter->iface_type != MWIFIEX_PCIE)
+			adapter->data_sent = false;
+		break;
+	case 0:
+		mwifiex_write_data_complete(adapter, skb, 0, ret);
+		break;
+	default:
+		break;
+	}
+	return ret;
+}
+
+static int
+mwifiex_dequeue_tx_queue(struct mwifiex_adapter *adapter)
+{
+	struct sk_buff *skb, *skb_next;
+	struct mwifiex_txinfo *tx_info;
+	struct mwifiex_tx_param tx_param;
+
+	skb = skb_dequeue(&adapter->tx_data_q);
+
+	if (!skb)
+		return -1;
+
+	tx_info = MWIFIEX_SKB_TXCB(skb);
+	if (tx_info->flags & MWIFIEX_BUF_FLAG_AGGR_PKT)
+		atomic_sub(tx_info->aggr_num, &adapter->tx_queued);
+	else
+		atomic_dec(&adapter->tx_queued);
+
+	if (!skb_queue_empty(&adapter->tx_data_q))
+		skb_next = skb_peek(&adapter->tx_data_q);
+	else
+		skb_next = NULL;
+
+	tx_param.next_pkt_len = ((skb_next) ? skb_next->len : 0);
+
+	return mwifiex_host_to_card(adapter, skb, &tx_param);
+}
+
+void
+mwifiex_process_tx_queue(struct mwifiex_adapter *adapter)
+{
+	do {
+		if (adapter->data_sent || adapter->tx_lock_flag)
+			break;
+		if (mwifiex_dequeue_tx_queue(adapter))
+			break;
+	} while (!skb_queue_empty(&adapter->tx_data_q));
+}
+
 /*
  * Packet send completion callback handler.
  *
@@ -181,6 +305,8 @@ int mwifiex_write_data_complete(struct mwifiex_adapter *adapter,
 
 	if (tx_info->flags & MWIFIEX_BUF_FLAG_BRIDGED_PKT)
 		atomic_dec_return(&adapter->pending_bridged_pkts);
+	if (tx_info->flags & MWIFIEX_BUF_FLAG_AGGR_PKT)
+		goto done;
 
 	if (aggr)
 		/* For skb_aggr, do not wake up tx queue */
diff --git a/drivers/net/wireless/mwifiex/wmm.c b/drivers/net/wireless/mwifiex/wmm.c
index ce747f7..c53c769 100644
--- a/drivers/net/wireless/mwifiex/wmm.c
+++ b/drivers/net/wireless/mwifiex/wmm.c
@@ -1174,6 +1174,14 @@ mwifiex_send_processed_packet(struct mwifiex_private *priv,
 
 	skb = skb_dequeue(&ptr->skb_head);
 
+	if (adapter->data_sent || adapter->tx_lock_flag) {
+		spin_unlock_irqrestore(&priv->wmm.ra_list_spinlock,
+				       ra_list_flags);
+		skb_queue_tail(&adapter->tx_data_q, skb);
+		atomic_inc(&adapter->tx_queued);
+		return;
+	}
+
 	if (!skb_queue_empty(&ptr->skb_head))
 		skb_next = skb_peek(&ptr->skb_head);
 	else
@@ -1324,11 +1332,15 @@ void
 mwifiex_wmm_process_tx(struct mwifiex_adapter *adapter)
 {
 	do {
-		/* Check if busy */
-		if (adapter->data_sent || adapter->tx_lock_flag)
-			break;
-
 		if (mwifiex_dequeue_tx_packet(adapter))
 			break;
+		if (adapter->iface_type != MWIFIEX_SDIO) {
+			if (adapter->data_sent || adapter->tx_lock_flag)
+				break;
+		} else {
+			if (atomic_read(&adapter->tx_queued) >=
+			    MWIFIEX_MAX_PKTS_TXQ)
+				break;
+		}
 	} while (!mwifiex_wmm_lists_empty(adapter));
 }
-- 
1.8.1.4


^ permalink raw reply related	[flat|nested] 11+ messages in thread
* Re: [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing
  2015-02-25 16:50 ` [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing Avinash Patil
@ 2015-03-03 12:37   ` Kalle Valo
  2015-03-05  6:38     ` Avinash Patil
  0 siblings, 1 reply; 11+ messages in thread
From: Kalle Valo @ 2015-03-03 12:37 UTC (permalink / raw)
  To: Avinash Patil; +Cc: linux-wireless, akarwar, cluo, liuzy, szli, Marc Yang

Avinash Patil <patila@marvell.com> writes:

> From: Marc Yang <yangyang@marvell.com>
>
> Current code does not check whether main_work_queue or
> rx_work_queue is running when preparing to do queue_work,
> this code fix add check before calling queue_work, reducing
> unnecessary queue_work switch.
>
> This change instead sets more_task flag to ensure we run main_process
> superloop once again.
>
> Signed-off-by: Marc Yang <yangyang@marvell.com>
> Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
> Signed-off-by: Shengzhen Li <szli@marvell.com>
> Signed-off-by: Cathy Luo <cluo@marvell.com>
> Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
> Signed-off-by: Avinash Patil <patila@marvell.com>

Really, it took six persons to write this patch? Or do you just dump the
same names to each patch?

> +void mwifiex_queue_main_work(struct mwifiex_adapter *adapter)
> +{
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&adapter->main_proc_lock, flags);
> +	if (adapter->mwifiex_processing) {
> +		adapter->more_task_flag = true;
> +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
> +	} else {
> +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
> +		queue_work(adapter->workqueue, &adapter->main_work);
> +	}
> +}
> +EXPORT_SYMBOL_GPL(mwifiex_queue_main_work);
> +
> +static void mwifiex_queue_rx_work(struct mwifiex_adapter *adapter)
> +{
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
> +	if (adapter->rx_processing) {
> +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
> +	} else {
> +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
> +		queue_work(adapter->rx_workqueue, &adapter->rx_work);
> +	}
> +}

I can apply this patch, but to me this looks like a horrible hack.

-- 
Kalle Valo

^ permalink raw reply	[flat|nested] 11+ messages in thread
* Re: [PATCH 3/4] mwifiex: remove_bss_prio_lock
  2015-02-25 16:50 ` [PATCH 3/4] mwifiex: remove_bss_prio_lock Avinash Patil
@ 2015-03-03 12:39   ` Kalle Valo
  0 siblings, 0 replies; 11+ messages in thread
From: Kalle Valo @ 2015-03-03 12:39 UTC (permalink / raw)
  To: Avinash Patil; +Cc: linux-wireless, akarwar, cluo, liuzy, szli

Avinash Patil <patila@marvell.com> writes:

> From: Zhaoyang Liu <liuzy@marvell.com>
>
> This patch does away with spinlock in
> mwifiex_wmm_get_highest_priolist_ptr in order to improve TP.
>
> Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
> Signed-off-by: Shengzhen Li <szli@marvell.com>
> Signed-off-by: Cathy Luo <cluo@marvell.com>
> Signed-off-by: Amitkumar Karwar <patila@marvell.com>
> Signed-off-by: Avinash Patil <patila@marvell.com>

[...]

> @@ -727,6 +730,25 @@ static int mwifiex_deinit_priv_params(struct mwifiex_private *priv)
>  	}
>  
>  	mwifiex_deauthenticate(priv, NULL);
> +
> +	spin_lock_irqsave(&adapter->main_proc_lock, flags);
> +	adapter->main_locked = true;
> +	if (adapter->mwifiex_processing) {
> +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
> +		flush_workqueue(adapter->workqueue);
> +	} else {
> +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
> +	}
> +
> +	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
> +	adapter->rx_locked = true;
> +	if (adapter->rx_processing) {
> +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
> +		flush_workqueue(adapter->rx_workqueue);
> +	} else {
> +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
> +	}

And here the horrible hack just grows to other functions.

-- 
Kalle Valo

^ permalink raw reply	[flat|nested] 11+ messages in thread
* RE: [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing
  2015-03-03 12:37   ` Kalle Valo
@ 2015-03-05  6:38     ` Avinash Patil
  2015-03-05 11:33       ` Kalle Valo
  0 siblings, 1 reply; 11+ messages in thread
From: Avinash Patil @ 2015-03-05  6:38 UTC (permalink / raw)
  To: Kalle Valo
  Cc: linux-wireless, Amitkumar Karwar, Cathy Luo, Zhaoyang Liu, Shengzhen Li

Hi Kalle,

> -----Original Message-----
> From: Kalle Valo [mailto:kvalo@codeaurora.org]
> Sent: Tuesday, March 03, 2015 6:08 PM
> To: Avinash Patil
> Cc: linux-wireless@vger.kernel.org; Amitkumar Karwar; Cathy Luo; Zhaoyang Liu;
> Shengzhen Li; Marc Yang
> Subject: Re: [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing
> 
> Avinash Patil <patila@marvell.com> writes:
> 
> > From: Marc Yang <yangyang@marvell.com>
> >
> > Current code does not check whether main_work_queue or rx_work_queue
> > is running when preparing to do queue_work, this code fix add check
> > before calling queue_work, reducing unnecessary queue_work switch.
> >
> > This change instead sets more_task flag to ensure we run main_process
> > superloop once again.
> >
> > Signed-off-by: Marc Yang <yangyang@marvell.com>
> > Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
> > Signed-off-by: Shengzhen Li <szli@marvell.com>
> > Signed-off-by: Cathy Luo <cluo@marvell.com>
> > Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
> > Signed-off-by: Avinash Patil <patila@marvell.com>
> 
> Really, it took six persons to write this patch? Or do you just dump the same
> names to each patch?

This should have been "Reviewed-by". Apologies.
 

> > +void mwifiex_queue_main_work(struct mwifiex_adapter *adapter) {
> > +	unsigned long flags;
> > +
> > +	spin_lock_irqsave(&adapter->main_proc_lock, flags);
> > +	if (adapter->mwifiex_processing) {
> > +		adapter->more_task_flag = true;
> > +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
> > +	} else {
> > +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
> > +		queue_work(adapter->workqueue, &adapter->main_work);
> > +	}
> > +}
> > +EXPORT_SYMBOL_GPL(mwifiex_queue_main_work);
> > +
> > +static void mwifiex_queue_rx_work(struct mwifiex_adapter *adapter) {
> > +	unsigned long flags;
> > +
> > +	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
> > +	if (adapter->rx_processing) {
> > +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
> > +	} else {
> > +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
> > +		queue_work(adapter->rx_workqueue, &adapter->rx_work);
> > +	}
> > +}
> 
> I can apply this patch, but to me this looks like a horrible hack.

We are here trying to avoid requeueing another work when work item is being executed. We set more_task flag to true if work item is being executed and mwifiex_main_process would execute superloop once again if more_task it set. work_pending() cannot be of much help here since we want to avoid queing only when work item is being executed. Could you please let us know if you think of any better way to handle this?
> --
> Kalle Valo

-Avinash

^ permalink raw reply	[flat|nested] 11+ messages in thread
* Re: [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing
  2015-03-05  6:38     ` Avinash Patil
@ 2015-03-05 11:33       ` Kalle Valo
  2015-03-05 11:39         ` Avinash Patil
  0 siblings, 1 reply; 11+ messages in thread
From: Kalle Valo @ 2015-03-05 11:33 UTC (permalink / raw)
  To: Avinash Patil
  Cc: linux-wireless, Amitkumar Karwar, Cathy Luo, Zhaoyang Liu, Shengzhen Li

Avinash Patil <patila@marvell.com> writes:

>> > From: Marc Yang <yangyang@marvell.com>
>> >
>> > Current code does not check whether main_work_queue or rx_work_queue
>> > is running when preparing to do queue_work, this code fix add check
>> > before calling queue_work, reducing unnecessary queue_work switch.
>> >
>> > This change instead sets more_task flag to ensure we run main_process
>> > superloop once again.
>> >
>> > Signed-off-by: Marc Yang <yangyang@marvell.com>
>> > Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
>> > Signed-off-by: Shengzhen Li <szli@marvell.com>
>> > Signed-off-by: Cathy Luo <cluo@marvell.com>
>> > Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
>> > Signed-off-by: Avinash Patil <patila@marvell.com>
>> 
>> Really, it took six persons to write this patch? Or do you just dump the same
>> names to each patch?
>
> This should have been "Reviewed-by". Apologies.

Ok, Reviewed-by would make more sense here. Please use that in the
future (no need to change anything for these patches).

>> > +void mwifiex_queue_main_work(struct mwifiex_adapter *adapter) {
>> > +	unsigned long flags;
>> > +
>> > +	spin_lock_irqsave(&adapter->main_proc_lock, flags);
>> > +	if (adapter->mwifiex_processing) {
>> > +		adapter->more_task_flag = true;
>> > +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
>> > +	} else {
>> > +		spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
>> > +		queue_work(adapter->workqueue, &adapter->main_work);
>> > +	}
>> > +}
>> > +EXPORT_SYMBOL_GPL(mwifiex_queue_main_work);
>> > +
>> > +static void mwifiex_queue_rx_work(struct mwifiex_adapter *adapter) {
>> > +	unsigned long flags;
>> > +
>> > +	spin_lock_irqsave(&adapter->rx_proc_lock, flags);
>> > +	if (adapter->rx_processing) {
>> > +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
>> > +	} else {
>> > +		spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
>> > +		queue_work(adapter->rx_workqueue, &adapter->rx_work);
>> > +	}
>> > +}
>> 
>> I can apply this patch, but to me this looks like a horrible hack.
>
> We are here trying to avoid requeueing another work when work item is
> being executed. We set more_task flag to true if work item is being
> executed and mwifiex_main_process would execute superloop once again
> if more_task it set. work_pending() cannot be of much help here since
> we want to avoid queing only when work item is being executed. Could
> you please let us know if you think of any better way to handle this?

It would be best to modify workqueue code to support this instead of
doing this in the driver. But this was just an observation I made while
reviewing your patches, I'll apply these anyway.

-- 
Kalle Valo

^ permalink raw reply	[flat|nested] 11+ messages in thread
* RE: [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing
  2015-03-05 11:33       ` Kalle Valo
@ 2015-03-05 11:39         ` Avinash Patil
  2015-03-05 11:54           ` Kalle Valo
  0 siblings, 1 reply; 11+ messages in thread
From: Avinash Patil @ 2015-03-05 11:39 UTC (permalink / raw)
  To: Kalle Valo
  Cc: linux-wireless, Amitkumar Karwar, Cathy Luo, Zhaoyang Liu, Shengzhen Li

Hi Kalle,

Please hold on.
I am  planning to divide patch3 into 2 patches - I will send updated v2.

Thanks,
Avinash

________________________________________
From: Kalle Valo [kvalo@codeaurora.org]
Sent: Thursday, March 05, 2015 5:03 PM
To: Avinash Patil
Cc: linux-wireless@vger.kernel.org; Amitkumar Karwar; Cathy Luo; Zhaoyang Liu; Shengzhen Li
Subject: Re: [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing

Avinash Patil <patila@marvell.com> writes:

>> > From: Marc Yang <yangyang@marvell.com>
>> >
>> > Current code does not check whether main_work_queue or rx_work_queue
>> > is running when preparing to do queue_work, this code fix add check
>> > before calling queue_work, reducing unnecessary queue_work switch.
>> >
>> > This change instead sets more_task flag to ensure we run main_process
>> > superloop once again.
>> >
>> > Signed-off-by: Marc Yang <yangyang@marvell.com>
>> > Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
>> > Signed-off-by: Shengzhen Li <szli@marvell.com>
>> > Signed-off-by: Cathy Luo <cluo@marvell.com>
>> > Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
>> > Signed-off-by: Avinash Patil <patila@marvell.com>
>>
>> Really, it took six persons to write this patch? Or do you just dump the same
>> names to each patch?
>
> This should have been "Reviewed-by". Apologies.

Ok, Reviewed-by would make more sense here. Please use that in the
future (no need to change anything for these patches).

>> > +void mwifiex_queue_main_work(struct mwifiex_adapter *adapter) {
>> > +  unsigned long flags;
>> > +
>> > +  spin_lock_irqsave(&adapter->main_proc_lock, flags);
>> > +  if (adapter->mwifiex_processing) {
>> > +          adapter->more_task_flag = true;
>> > +          spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
>> > +  } else {
>> > +          spin_unlock_irqrestore(&adapter->main_proc_lock, flags);
>> > +          queue_work(adapter->workqueue, &adapter->main_work);
>> > +  }
>> > +}
>> > +EXPORT_SYMBOL_GPL(mwifiex_queue_main_work);
>> > +
>> > +static void mwifiex_queue_rx_work(struct mwifiex_adapter *adapter) {
>> > +  unsigned long flags;
>> > +
>> > +  spin_lock_irqsave(&adapter->rx_proc_lock, flags);
>> > +  if (adapter->rx_processing) {
>> > +          spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
>> > +  } else {
>> > +          spin_unlock_irqrestore(&adapter->rx_proc_lock, flags);
>> > +          queue_work(adapter->rx_workqueue, &adapter->rx_work);
>> > +  }
>> > +}
>>
>> I can apply this patch, but to me this looks like a horrible hack.
>
> We are here trying to avoid requeueing another work when work item is
> being executed. We set more_task flag to true if work item is being
> executed and mwifiex_main_process would execute superloop once again
> if more_task it set. work_pending() cannot be of much help here since
> we want to avoid queing only when work item is being executed. Could
> you please let us know if you think of any better way to handle this?

It would be best to modify workqueue code to support this instead of
doing this in the driver. But this was just an observation I made while
reviewing your patches, I'll apply these anyway.

--
Kalle Valo

^ permalink raw reply	[flat|nested] 11+ messages in thread
* Re: [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing
  2015-03-05 11:39         ` Avinash Patil
@ 2015-03-05 11:54           ` Kalle Valo
  0 siblings, 0 replies; 11+ messages in thread
From: Kalle Valo @ 2015-03-05 11:54 UTC (permalink / raw)
  To: Avinash Patil
  Cc: linux-wireless, Amitkumar Karwar, Cathy Luo, Zhaoyang Liu, Shengzhen Li

Avinash Patil <patila@marvell.com> writes:

> Please hold on.
> I am  planning to divide patch3 into 2 patches - I will send updated v2.

Ok, I'll wait for v2.

-- 
Kalle Valo

^ permalink raw reply	[flat|nested] 11+ messages in thread
end of thread, other threads:[~2015-03-05 11:54 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-02-25 16:50 [PATCH 0/4] mwifiex: throughput enhancements for TX Avinash Patil
2015-02-25 16:50 ` [PATCH 1/4] mwifiex: avoid queue_work while work is ongoing Avinash Patil
2015-03-03 12:37   ` Kalle Valo
2015-03-05  6:38     ` Avinash Patil
2015-03-05 11:33       ` Kalle Valo
2015-03-05 11:39         ` Avinash Patil
2015-03-05 11:54           ` Kalle Valo
2015-02-25 16:50 ` [PATCH 2/4] mwifiex: get rid of BA setup helper functions Avinash Patil
2015-02-25 16:50 ` [PATCH 3/4] mwifiex: remove_bss_prio_lock Avinash Patil
2015-03-03 12:39   ` Kalle Valo
2015-02-25 16:50 ` [PATCH 4/4] mwifiex: preprocess packets from TX queue Avinash Patil

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.