Skip to contentSkip to conversation

p54: refractor beacon_update and stop path

PATCH In review Christian Lamparter 3 participants 3 messages Mar 5, 2009 – Mar 17, 2009 Networking drivers (wireless)

lore
Share & tools
lore
#0 lore raw

PATCH 6/7 p54: refractor beacon_update and stop path

According to the LMAC API Section 3.2.2, we don't have to cancel the previous beacon manually. The firmware will automatically return it to the driver through the feedback mechanism.
And while we're at it, we can remove some left over code as well.

---

drivers/net/wireless/p54/p54common.c +11 -11
diff -Nurp a/drivers/net/wireless/p54/p54common.c b/drivers/net/wireless/p54/p54common.c
@@ -2015,18 +2015,14 @@ static int p54_beacon_update(struct ieee
 	struct sk_buff *beacon;
 	int ret;
 
-	if (priv->cached_beacon) {
-		p54_tx_cancel(dev, priv->cached_beacon);
-		/* wait for the last beacon the be freed */
-		msleep(10);
-	}
-
 	beacon = ieee80211_beacon_get(dev, vif);
 	if (!beacon)
 		return -ENOMEM;
 	ret = p54_beacon_tim(beacon);
 	if (ret)
 		return ret;
+
+	/* the firmware will automatically cancel old beacons. */ 
 	ret = p54_tx(dev, beacon);
 	if (ret)
 		return ret;
@@ -2072,17 +2068,21 @@ out:
 static void p54_stop(struct ieee80211_hw *dev)
 {
 	struct p54_common *priv = dev->priv;
-	struct sk_buff *skb;
 
 	mutex_lock(&priv->conf_mutex);
 	priv->mode = NL80211_IFTYPE_UNSPECIFIED;
+	p54_setup_mac(dev);
 	cancel_delayed_work_sync(&priv->work);
-	if (priv->cached_beacon)
-		p54_tx_cancel(dev, priv->cached_beacon);
+
+	/*
+	 * p54spi uses mac80211's workqueue to write the frames into
+	 * device's memory. Therefore, we should give the queue a chance
+	 * to complete all outstanding work, before the device is stopped.
+	 */
+	msleep(25);
 
 	priv->stop(dev);
-	while ((skb = skb_dequeue(&priv->tx_queue)))
-		kfree_skb(skb);
+	skb_queue_purge(&priv->tx_queue);
 	priv->cached_beacon = NULL;
 	priv->tsf_high32 = priv->tsf_low32 = 0;
 	mutex_unlock(&priv->conf_mutex);
#1 lore raw
Quoted 40 lines · On Thursday 05 March 2009 21:33:32 Christian Lamparter wrote:

According to the LMAC API Section 3.2.2, we don't have to cancel the previous beacon manually. The firmware will automatically return it to the driver through the feedback mechanism.
And while we're at it, we can remove some left over code as well.

---
diff -Nurp a/drivers/net/wireless/p54/p54common.c b/drivers/net/wireless/p54/p54common.c

drivers/net/wireless/p54/p54common.c +3 -7
@@ -2015,18 +2015,14 @@ static int p54_beacon_update(struct ieee
 	struct sk_buff *beacon;
 	int ret;
 
-	if (priv->cached_beacon) {
-		p54_tx_cancel(dev, priv->cached_beacon);
-		/* wait for the last beacon the be freed */
-		msleep(10);
-	}
-
 	beacon = ieee80211_beacon_get(dev, vif);
 	if (!beacon)
 		return -ENOMEM;
 	ret = p54_beacon_tim(beacon);
 	if (ret)
 		return ret;
+
+	/* the firmware will automatically cancel old beacons. */ 
 	ret = p54_tx(dev, beacon);
 	if (ret)
 		return ret;
@@ -2072,17 +2068,21 @@ out:
 static void p54_stop(struct ieee80211_hw *dev)
 {
 	struct p54_common *priv = dev->priv;
-	struct sk_buff *skb;
 
 	mutex_lock(&priv->conf_mutex);
 	priv->mode = NL80211_IFTYPE_UNSPECIFIED;
+	p54_setup_mac(dev);

drivers/net/wireless/p54/p54common.c out: · line 2082

Quoted 10 lines
drivers/net/wireless/p54/p54common.c +7 -2
 	cancel_delayed_work_sync(&priv->work);
-	if (priv->cached_beacon)
-		p54_tx_cancel(dev, priv->cached_beacon);
+
+	/*
+	 * p54spi uses mac80211's workqueue to write the frames into
+	 * device's memory. Therefore, we should give the queue a chance
+	 * to complete all outstanding work, before the device is stopped.
+	 */
+	msleep(25);

What about flush_workqueue()?

But is it even possible for the work to execute here? Does the work take the conf_mutex?

Quoted 14 lines
drivers/net/wireless/p54/p54common.c +1 -3
 	priv->stop(dev);
-	while ((skb = skb_dequeue(&priv->tx_queue)))
-		kfree_skb(skb);
+	skb_queue_purge(&priv->tx_queue);
 	priv->cached_beacon = NULL;
 	priv->tsf_high32 = priv->tsf_low32 = 0;
 	mutex_unlock(&priv->conf_mutex);
--

To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Signature · 1 line
Greetings, Michael.
#2 lore raw
Quoted 38 lines · Michael Buesch
Quoted 37 lines · On Thursday 05 March 2009 21:33:32 Christian Lamparter wrote:

According to the LMAC API Section 3.2.2, we don't have to cancel the previous beacon manually. The firmware will automatically return it to the driver through the feedback mechanism.
And while we're at it, we can remove some left over code as well.

---
diff -Nurp a/drivers/net/wireless/p54/p54common.c b/drivers/net/wireless/p54/p54common.c

drivers/net/wireless/p54/p54common.c +2 -7
@@ -2015,18 +2015,14 @@ static int p54_beacon_update(struct ieee
 	struct sk_buff *beacon;
 	int ret;
 
-	if (priv->cached_beacon) {
-		p54_tx_cancel(dev, priv->cached_beacon);
-		/* wait for the last beacon the be freed */
-		msleep(10);
-	}
-
 	beacon = ieee80211_beacon_get(dev, vif);
 	if (!beacon)
 		return -ENOMEM;
 	ret = p54_beacon_tim(beacon);
 	if (ret)
 		return ret;
+
+	/* the firmware will automatically cancel old beacons. */ 
 	ret = p54_tx(dev, beacon);
 	if (ret)
 		return ret;
@@ -2072,17 +2068,21 @@ out:
 static void p54_stop(struct ieee80211_hw *dev)
 {
 	struct p54_common *priv = dev->priv;
-	struct sk_buff *skb;
 
Quoted 15 lines

drivers/net/wireless/p54/p54common.c out: · line 2082

Quoted 13 lines
drivers/net/wireless/p54/p54common.c +8 -2
 	mutex_lock(&priv->conf_mutex);
 	priv->mode = NL80211_IFTYPE_UNSPECIFIED;
+	p54_setup_mac(dev);
 	cancel_delayed_work_sync(&priv->work);
-	if (priv->cached_beacon)
-		p54_tx_cancel(dev, priv->cached_beacon);
+
+	/*
+	 * p54spi uses mac80211's workqueue to write the frames into
+	 * device's memory. Therefore, we should give the queue a chance
+	 * to complete all outstanding work, before the device is stopped.
+	 */
+	msleep(25);

What about flush_workqueue()?

I agree, we should use an actual API here and not just "wait long enough (we hope)"...

John

Signature · 2 lines
John W. Linville		Someday the world will need a hero, and you
[email protected]			might be all we have.  Be ready.