[PATCH 0/3] rt2x00 update

[PATCH 0/3] rt2x00 update

[Ivo van Doorn]

Hi John,

The following 3 patches are intended on top of
the patch series "mac80211 interface (config) cleanups"
which were send to the list by Johannes.

The first one is a mac80211 bugfix for drivers that require
the sequence counting to be done in software.
The second is a minor update for CTS protection support in
rt2x00lib. There aren't any users for it at this time, but it is
small, harmless and not really worth the effor for keepin it
seperate. ;)
The last one cleans up beacon handling in rt2x00 which was
possible because of Johannes' patch series.

Ivo

[PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

Currently only beacons generated in AP mode have the software
sequence number inserted. This means IBSS and Mesh mode are broken
for all hardware that require software sequence numbers.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
---
 net/mac80211/tx.c |    5 +++++
 1 files changed, 5 insertions(+), 0 deletions(-)

diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index 359cb36..0e7e1f3 100644
--- a/net/mac80211/tx.c
+++ b/net/mac80211/tx.c
@@ -1876,6 +1876,8 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
 		hdr->frame_control = IEEE80211_FC(IEEE80211_FTYPE_MGMT,
 						  IEEE80211_STYPE_BEACON);
 
+		ieee80211_include_sequence(sdata, hdr);
+
 		num_beacons = &ifsta->num_beacons;
 	} else if (ieee80211_vif_is_mesh(&sdata->vif)) {
 		/* headroom, head length, tail length and maximum TIM length */
@@ -1896,6 +1898,9 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
 			cpu_to_le16(local->hw.conf.beacon_int);
 		mgmt->u.beacon.capab_info = 0x0; /* 0x0 for MPs */
 
+		ieee80211_include_sequence(sdata,
+					   (struct ieee80211_hdr *)skb->data);
+
 		pos = skb_put(skb, 2);
 		*pos++ = WLAN_EID_SSID;
 		*pos++ = 0x0;
-- 
1.5.6.1

[PATCH 2/3] rt2x00: Add support for CTS protection in rt2x00lib

[Ivo van Doorn]

Inform drivers about the changed CTS protection settings.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
---
 drivers/net/wireless/rt2x00/rt2x00.h       |    1 +
 drivers/net/wireless/rt2x00/rt2x00config.c |    2 ++
 drivers/net/wireless/rt2x00/rt2x00mac.c    |    2 +-
 3 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/drivers/net/wireless/rt2x00/rt2x00.h b/drivers/net/wireless/rt2x00/rt2x00.h
index 9ad3ce4..b32fedf 100644
--- a/drivers/net/wireless/rt2x00/rt2x00.h
+++ b/drivers/net/wireless/rt2x00/rt2x00.h
@@ -434,6 +434,7 @@ struct rt2x00lib_conf {
  */
 struct rt2x00lib_erp {
 	int short_preamble;
+	int cts_protection;
 
 	int ack_timeout;
 	int ack_consume_time;
diff --git a/drivers/net/wireless/rt2x00/rt2x00config.c b/drivers/net/wireless/rt2x00/rt2x00config.c
index 48608e8..f20ca71 100644
--- a/drivers/net/wireless/rt2x00/rt2x00config.c
+++ b/drivers/net/wireless/rt2x00/rt2x00config.c
@@ -84,6 +84,8 @@ void rt2x00lib_config_erp(struct rt2x00_dev *rt2x00dev,
 	memset(&erp, 0, sizeof(erp));
 
 	erp.short_preamble = bss_conf->use_short_preamble;
+	erp.cts_protection = bss_conf->use_cts_prot;
+
 	erp.ack_timeout = PLCP + get_duration(ACK_SIZE, 10);
 	erp.ack_consume_time = SIFS + PLCP + get_duration(ACK_SIZE, 10);
 
diff --git a/drivers/net/wireless/rt2x00/rt2x00mac.c b/drivers/net/wireless/rt2x00/rt2x00mac.c
index 84b51f4..ff853c4 100644
--- a/drivers/net/wireless/rt2x00/rt2x00mac.c
+++ b/drivers/net/wireless/rt2x00/rt2x00mac.c
@@ -511,7 +511,7 @@ void rt2x00mac_bss_info_changed(struct ieee80211_hw *hw,
 	 * When the erp information has changed, we should perform
 	 * additional configuration steps. For all other changes we are done.
 	 */
-	if (changes & BSS_CHANGED_ERP_PREAMBLE) {
+	if (changes & (BSS_CHANGED_ERP_PREAMBLE | BSS_CHANGED_ERP_CTS_PROT)) {
 		if (!test_bit(DRIVER_REQUIRE_SCHEDULED, &rt2x00dev->flags))
 			rt2x00lib_config_erp(rt2x00dev, intf, bss_conf);
 		else
-- 
1.5.6.1

[PATCH 3/3] rt2x00: Reorganize beacon handling

[Ivo van Doorn]

With the new beacon handling from mac80211 we can
reorganize the beacon handling in rt2x00 as well.
This patch will move the function to the TX handlers,
and move all duplicate code into rt2x00queue.c.

After this change the descriptor helper functions
from rt2x00queue.c no longer need to be exported
outside of rt2x00lib and can be declared static.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
---
 drivers/net/wireless/rt2x00/rt2400pci.c   |   89 +++++++----------
 drivers/net/wireless/rt2x00/rt2500pci.c   |   90 +++++++----------
 drivers/net/wireless/rt2x00/rt2500usb.c   |  154 +++++++++++------------------
 drivers/net/wireless/rt2x00/rt2x00.h      |   36 +-------
 drivers/net/wireless/rt2x00/rt2x00dev.c   |    9 +--
 drivers/net/wireless/rt2x00/rt2x00lib.h   |    8 ++
 drivers/net/wireless/rt2x00/rt2x00mac.c   |   31 ++----
 drivers/net/wireless/rt2x00/rt2x00queue.c |   64 +++++++++++-
 drivers/net/wireless/rt2x00/rt61pci.c     |  103 +++++++-------------
 drivers/net/wireless/rt2x00/rt73usb.c     |  115 ++++++++-------------
 10 files changed, 289 insertions(+), 410 deletions(-)

diff --git a/drivers/net/wireless/rt2x00/rt2400pci.c b/drivers/net/wireless/rt2x00/rt2400pci.c
index d0f226e..4c0538d 100644
--- a/drivers/net/wireless/rt2x00/rt2400pci.c
+++ b/drivers/net/wireless/rt2x00/rt2400pci.c
@@ -1069,6 +1069,40 @@ static void rt2400pci_write_tx_desc(struct rt2x00_dev *rt2x00dev,
 /*
  * TX data initialization
  */
+static void rt2400pci_write_beacon(struct queue_entry *entry)
+{
+	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
+	struct queue_entry_priv_pci *entry_priv = entry->priv_data;
+	struct skb_frame_desc *skbdesc = get_skb_frame_desc(entry->skb);
+	u32 word;
+	u32 reg;
+
+	/*
+	 * Disable beaconing while we are reloading the beacon data,
+	 * otherwise we might be sending out invalid data.
+	 */
+	rt2x00pci_register_read(rt2x00dev, CSR14, &reg);
+	rt2x00_set_field32(&reg, CSR14_TSF_COUNT, 0);
+	rt2x00_set_field32(&reg, CSR14_TBCN, 0);
+	rt2x00_set_field32(&reg, CSR14_BEACON_GEN, 0);
+	rt2x00pci_register_write(rt2x00dev, CSR14, reg);
+
+	/*
+	 * Replace rt2x00lib allocated descriptor with the
+	 * pointer to the _real_ hardware descriptor.
+	 * After that, map the beacon to DMA and update the
+	 * descriptor.
+	 */
+	memcpy(entry_priv->desc, skbdesc->desc, skbdesc->desc_len);
+	skbdesc->desc = entry_priv->desc;
+
+	rt2x00queue_map_txskb(rt2x00dev, entry->skb);
+
+	rt2x00_desc_read(entry_priv->desc, 1, &word);
+	rt2x00_set_field32(&word, TXD_W1_BUFFER_ADDRESS, skbdesc->skb_dma);
+	rt2x00_desc_write(entry_priv->desc, 1, word);
+}
+
 static void rt2400pci_kick_tx_queue(struct rt2x00_dev *rt2x00dev,
 				    const enum data_queue_qid queue)
 {
@@ -1515,59 +1549,6 @@ static u64 rt2400pci_get_tsf(struct ieee80211_hw *hw)
 	return tsf;
 }
 
-static int rt2400pci_beacon_update(struct ieee80211_hw *hw, struct sk_buff *skb)
-{
-	struct rt2x00_dev *rt2x00dev = hw->priv;
-	struct ieee80211_tx_info *tx_info = IEEE80211_SKB_CB(skb);
-	struct rt2x00_intf *intf = vif_to_intf(tx_info->control.vif);
-	struct queue_entry_priv_pci *entry_priv;
-	struct skb_frame_desc *skbdesc;
-	struct txentry_desc txdesc;
-	u32 reg;
-
-	if (unlikely(!intf->beacon))
-		return -ENOBUFS;
-	entry_priv = intf->beacon->priv_data;
-
-	/*
-	 * Copy all TX descriptor information into txdesc,
-	 * after that we are free to use the skb->cb array
-	 * for our information.
-	 */
-	intf->beacon->skb = skb;
-	rt2x00queue_create_tx_descriptor(intf->beacon, &txdesc);
-
-	/*
-	 * Fill in skb descriptor
-	 */
-	skbdesc = get_skb_frame_desc(skb);
-	memset(skbdesc, 0, sizeof(*skbdesc));
-	skbdesc->desc = entry_priv->desc;
-	skbdesc->desc_len = intf->beacon->queue->desc_size;
-	skbdesc->entry = intf->beacon;
-
-	/*
-	 * Disable beaconing while we are reloading the beacon data,
-	 * otherwise we might be sending out invalid data.
-	 */
-	rt2x00pci_register_read(rt2x00dev, CSR14, &reg);
-	rt2x00_set_field32(&reg, CSR14_TSF_COUNT, 0);
-	rt2x00_set_field32(&reg, CSR14_TBCN, 0);
-	rt2x00_set_field32(&reg, CSR14_BEACON_GEN, 0);
-	rt2x00pci_register_write(rt2x00dev, CSR14, reg);
-
-	/*
-	 * Enable beacon generation.
-	 * Write entire beacon with descriptor to register,
-	 * and kick the beacon generator.
-	 */
-	rt2x00queue_map_txskb(rt2x00dev, intf->beacon->skb);
-	rt2x00queue_write_tx_descriptor(intf->beacon, &txdesc);
-	rt2x00dev->ops->lib->kick_tx_queue(rt2x00dev, QID_BEACON);
-
-	return 0;
-}
-
 static int rt2400pci_tx_last_beacon(struct ieee80211_hw *hw)
 {
 	struct rt2x00_dev *rt2x00dev = hw->priv;
@@ -1609,9 +1590,9 @@ static const struct rt2x00lib_ops rt2400pci_rt2x00_ops = {
 	.link_tuner		= rt2400pci_link_tuner,
 	.write_tx_desc		= rt2400pci_write_tx_desc,
 	.write_tx_data		= rt2x00pci_write_tx_data,
+	.write_beacon		= rt2400pci_write_beacon,
 	.kick_tx_queue		= rt2400pci_kick_tx_queue,
 	.fill_rxdone		= rt2400pci_fill_rxdone,
-	.beacon_update		= rt2400pci_beacon_update,
 	.config_filter		= rt2400pci_config_filter,
 	.config_intf		= rt2400pci_config_intf,
 	.config_erp		= rt2400pci_config_erp,
diff --git a/drivers/net/wireless/rt2x00/rt2500pci.c b/drivers/net/wireless/rt2x00/rt2500pci.c
index 7ee6b74..aa6dfb8 100644
--- a/drivers/net/wireless/rt2x00/rt2500pci.c
+++ b/drivers/net/wireless/rt2x00/rt2500pci.c
@@ -1227,6 +1227,40 @@ static void rt2500pci_write_tx_desc(struct rt2x00_dev *rt2x00dev,
 /*
  * TX data initialization
  */
+static void rt2500pci_write_beacon(struct queue_entry *entry)
+{
+	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
+	struct queue_entry_priv_pci *entry_priv = entry->priv_data;
+	struct skb_frame_desc *skbdesc = get_skb_frame_desc(entry->skb);
+	u32 word;
+	u32 reg;
+
+	/*
+	 * Disable beaconing while we are reloading the beacon data,
+	 * otherwise we might be sending out invalid data.
+	 */
+	rt2x00pci_register_read(rt2x00dev, CSR14, &reg);
+	rt2x00_set_field32(&reg, CSR14_TSF_COUNT, 0);
+	rt2x00_set_field32(&reg, CSR14_TBCN, 0);
+	rt2x00_set_field32(&reg, CSR14_BEACON_GEN, 0);
+	rt2x00pci_register_write(rt2x00dev, CSR14, reg);
+
+	/*
+	 * Replace rt2x00lib allocated descriptor with the
+	 * pointer to the _real_ hardware descriptor.
+	 * After that, map the beacon to DMA and update the
+	 * descriptor.
+	 */
+	memcpy(entry_priv->desc, skbdesc->desc, skbdesc->desc_len);
+	skbdesc->desc = entry_priv->desc;
+
+	rt2x00queue_map_txskb(rt2x00dev, entry->skb);
+
+	rt2x00_desc_read(entry_priv->desc, 1, &word);
+	rt2x00_set_field32(&word, TXD_W1_BUFFER_ADDRESS, skbdesc->skb_dma);
+	rt2x00_desc_write(entry_priv->desc, 1, word);
+}
+
 static void rt2500pci_kick_tx_queue(struct rt2x00_dev *rt2x00dev,
 				    const enum data_queue_qid queue)
 {
@@ -1808,60 +1842,6 @@ static u64 rt2500pci_get_tsf(struct ieee80211_hw *hw)
 	return tsf;
 }
 
-static int rt2500pci_beacon_update(struct ieee80211_hw *hw, struct sk_buff *skb)
-{
-	struct rt2x00_dev *rt2x00dev = hw->priv;
-	struct ieee80211_tx_info *tx_info = IEEE80211_SKB_CB(skb);
-	struct rt2x00_intf *intf = vif_to_intf(tx_info->control.vif);
-	struct queue_entry_priv_pci *entry_priv;
-	struct skb_frame_desc *skbdesc;
-	struct txentry_desc txdesc;
-	u32 reg;
-
-	if (unlikely(!intf->beacon))
-		return -ENOBUFS;
-
-	entry_priv = intf->beacon->priv_data;
-
-	/*
-	 * Copy all TX descriptor information into txdesc,
-	 * after that we are free to use the skb->cb array
-	 * for our information.
-	 */
-	intf->beacon->skb = skb;
-	rt2x00queue_create_tx_descriptor(intf->beacon, &txdesc);
-
-	/*
-	 * Fill in skb descriptor
-	 */
-	skbdesc = get_skb_frame_desc(skb);
-	memset(skbdesc, 0, sizeof(*skbdesc));
-	skbdesc->desc = entry_priv->desc;
-	skbdesc->desc_len = intf->beacon->queue->desc_size;
-	skbdesc->entry = intf->beacon;
-
-	/*
-	 * Disable beaconing while we are reloading the beacon data,
-	 * otherwise we might be sending out invalid data.
-	 */
-	rt2x00pci_register_read(rt2x00dev, CSR14, &reg);
-	rt2x00_set_field32(&reg, CSR14_TSF_COUNT, 0);
-	rt2x00_set_field32(&reg, CSR14_TBCN, 0);
-	rt2x00_set_field32(&reg, CSR14_BEACON_GEN, 0);
-	rt2x00pci_register_write(rt2x00dev, CSR14, reg);
-
-	/*
-	 * Enable beacon generation.
-	 * Write entire beacon with descriptor to register,
-	 * and kick the beacon generator.
-	 */
-	rt2x00queue_map_txskb(rt2x00dev, intf->beacon->skb);
-	rt2x00queue_write_tx_descriptor(intf->beacon, &txdesc);
-	rt2x00dev->ops->lib->kick_tx_queue(rt2x00dev, QID_BEACON);
-
-	return 0;
-}
-
 static int rt2500pci_tx_last_beacon(struct ieee80211_hw *hw)
 {
 	struct rt2x00_dev *rt2x00dev = hw->priv;
@@ -1903,9 +1883,9 @@ static const struct rt2x00lib_ops rt2500pci_rt2x00_ops = {
 	.link_tuner		= rt2500pci_link_tuner,
 	.write_tx_desc		= rt2500pci_write_tx_desc,
 	.write_tx_data		= rt2x00pci_write_tx_data,
+	.write_beacon		= rt2500pci_write_beacon,
 	.kick_tx_queue		= rt2500pci_kick_tx_queue,
 	.fill_rxdone		= rt2500pci_fill_rxdone,
-	.beacon_update		= rt2500pci_beacon_update,
 	.config_filter		= rt2500pci_config_filter,
 	.config_intf		= rt2500pci_config_intf,
 	.config_erp		= rt2500pci_config_erp,
diff --git a/drivers/net/wireless/rt2x00/rt2500usb.c b/drivers/net/wireless/rt2x00/rt2500usb.c
index c6d78e2..3558cb2 100644
--- a/drivers/net/wireless/rt2x00/rt2500usb.c
+++ b/drivers/net/wireless/rt2x00/rt2500usb.c
@@ -1107,6 +1107,65 @@ static void rt2500usb_write_tx_desc(struct rt2x00_dev *rt2x00dev,
 	rt2x00_desc_write(txd, 0, word);
 }
 
+/*
+ * TX data initialization
+ */
+static void rt2500usb_beacondone(struct urb *urb);
+
+static void rt2500usb_write_beacon(struct queue_entry *entry)
+{
+	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
+	struct usb_device *usb_dev = to_usb_device_intf(rt2x00dev->dev);
+	struct queue_entry_priv_usb_bcn *bcn_priv = entry->priv_data;
+	struct skb_frame_desc *skbdesc = get_skb_frame_desc(entry->skb);
+	int pipe = usb_sndbulkpipe(usb_dev, 1);
+	int length;
+	u16 reg;
+
+	/*
+	 * Add the descriptor in front of the skb.
+	 */
+	skb_push(entry->skb, entry->queue->desc_size);
+	memcpy(entry->skb->data, skbdesc->desc, skbdesc->desc_len);
+	skbdesc->desc = entry->skb->data;
+
+	/*
+	 * Disable beaconing while we are reloading the beacon data,
+	 * otherwise we might be sending out invalid data.
+	 */
+	rt2500usb_register_read(rt2x00dev, TXRX_CSR19, &reg);
+	rt2x00_set_field16(&reg, TXRX_CSR19_TSF_COUNT, 0);
+	rt2x00_set_field16(&reg, TXRX_CSR19_TBCN, 0);
+	rt2x00_set_field16(&reg, TXRX_CSR19_BEACON_GEN, 0);
+	rt2500usb_register_write(rt2x00dev, TXRX_CSR19, reg);
+
+	/*
+	 * USB devices cannot blindly pass the skb->len as the
+	 * length of the data to usb_fill_bulk_urb. Pass the skb
+	 * to the driver to determine what the length should be.
+	 */
+	length = rt2x00dev->ops->lib->get_tx_data_len(rt2x00dev, entry->skb);
+
+	usb_fill_bulk_urb(bcn_priv->urb, usb_dev, pipe,
+			  entry->skb->data, length, rt2500usb_beacondone,
+			  entry);
+
+	/*
+	 * Second we need to create the guardian byte.
+	 * We only need a single byte, so lets recycle
+	 * the 'flags' field we are not using for beacons.
+	 */
+	bcn_priv->guardian_data = 0;
+	usb_fill_bulk_urb(bcn_priv->guardian_urb, usb_dev, pipe,
+			  &bcn_priv->guardian_data, 1, rt2500usb_beacondone,
+			  entry);
+
+	/*
+	 * Send out the guardian byte.
+	 */
+	usb_submit_urb(bcn_priv->guardian_urb, GFP_ATOMIC);
+}
+
 static int rt2500usb_get_tx_data_len(struct rt2x00_dev *rt2x00dev,
 				     struct sk_buff *skb)
 {
@@ -1122,9 +1181,6 @@ static int rt2500usb_get_tx_data_len(struct rt2x00_dev *rt2x00dev,
 	return length;
 }
 
-/*
- * TX data initialization
- */
 static void rt2500usb_kick_tx_queue(struct rt2x00_dev *rt2x00dev,
 				    const enum data_queue_qid queue)
 {
@@ -1679,96 +1735,6 @@ static int rt2500usb_probe_hw(struct rt2x00_dev *rt2x00dev)
 	return 0;
 }
 
-/*
- * IEEE80211 stack callback functions.
- */
-static int rt2500usb_beacon_update(struct ieee80211_hw *hw, struct sk_buff *skb)
-{
-	struct rt2x00_dev *rt2x00dev = hw->priv;
-	struct usb_device *usb_dev = to_usb_device_intf(rt2x00dev->dev);
-	struct ieee80211_tx_info *tx_info = IEEE80211_SKB_CB(skb);
-	struct rt2x00_intf *intf = vif_to_intf(tx_info->control.vif);
-	struct queue_entry_priv_usb_bcn *bcn_priv;
-	struct skb_frame_desc *skbdesc;
-	struct txentry_desc txdesc;
-	int pipe = usb_sndbulkpipe(usb_dev, 1);
-	int length;
-	u16 reg;
-
-	if (unlikely(!intf->beacon))
-		return -ENOBUFS;
-
-	bcn_priv = intf->beacon->priv_data;
-
-	/*
-	 * Copy all TX descriptor information into txdesc,
-	 * after that we are free to use the skb->cb array
-	 * for our information.
-	 */
-	intf->beacon->skb = skb;
-	rt2x00queue_create_tx_descriptor(intf->beacon, &txdesc);
-
-	/*
-	 * Add the descriptor in front of the skb.
-	 */
-	skb_push(skb, intf->beacon->queue->desc_size);
-	memset(skb->data, 0, intf->beacon->queue->desc_size);
-
-	/*
-	 * Fill in skb descriptor
-	 */
-	skbdesc = get_skb_frame_desc(skb);
-	memset(skbdesc, 0, sizeof(*skbdesc));
-	skbdesc->desc = skb->data;
-	skbdesc->desc_len = intf->beacon->queue->desc_size;
-	skbdesc->entry = intf->beacon;
-
-	/*
-	 * Disable beaconing while we are reloading the beacon data,
-	 * otherwise we might be sending out invalid data.
-	 */
-	rt2500usb_register_read(rt2x00dev, TXRX_CSR19, &reg);
-	rt2x00_set_field16(&reg, TXRX_CSR19_TSF_COUNT, 0);
-	rt2x00_set_field16(&reg, TXRX_CSR19_TBCN, 0);
-	rt2x00_set_field16(&reg, TXRX_CSR19_BEACON_GEN, 0);
-	rt2500usb_register_write(rt2x00dev, TXRX_CSR19, reg);
-
-	rt2x00queue_write_tx_descriptor(intf->beacon, &txdesc);
-
-	/*
-	 * USB devices cannot blindly pass the skb->len as the
-	 * length of the data to usb_fill_bulk_urb. Pass the skb
-	 * to the driver to determine what the length should be.
-	 */
-	length = rt2500usb_get_tx_data_len(rt2x00dev, skb);
-
-	usb_fill_bulk_urb(bcn_priv->urb, usb_dev, pipe,
-			  skb->data, length, rt2500usb_beacondone,
-			  intf->beacon);
-
-	/*
-	 * Second we need to create the guardian byte.
-	 * We only need a single byte, so lets recycle
-	 * the 'flags' field we are not using for beacons.
-	 */
-	bcn_priv->guardian_data = 0;
-	usb_fill_bulk_urb(bcn_priv->guardian_urb, usb_dev, pipe,
-			  &bcn_priv->guardian_data, 1, rt2500usb_beacondone,
-			  intf->beacon);
-
-	/*
-	 * Send out the guardian byte.
-	 */
-	usb_submit_urb(bcn_priv->guardian_urb, GFP_ATOMIC);
-
-	/*
-	 * Enable beacon generation.
-	 */
-	rt2500usb_kick_tx_queue(rt2x00dev, QID_BEACON);
-
-	return 0;
-}
-
 static const struct ieee80211_ops rt2500usb_mac80211_ops = {
 	.tx			= rt2x00mac_tx,
 	.start			= rt2x00mac_start,
@@ -1796,10 +1762,10 @@ static const struct rt2x00lib_ops rt2500usb_rt2x00_ops = {
 	.link_tuner		= rt2500usb_link_tuner,
 	.write_tx_desc		= rt2500usb_write_tx_desc,
 	.write_tx_data		= rt2x00usb_write_tx_data,
+	.write_beacon		= rt2500usb_write_beacon,
 	.get_tx_data_len	= rt2500usb_get_tx_data_len,
 	.kick_tx_queue		= rt2500usb_kick_tx_queue,
 	.fill_rxdone		= rt2500usb_fill_rxdone,
-	.beacon_update		= rt2500usb_beacon_update,
 	.config_filter		= rt2500usb_config_filter,
 	.config_intf		= rt2500usb_config_intf,
 	.config_erp		= rt2500usb_config_erp,
diff --git a/drivers/net/wireless/rt2x00/rt2x00.h b/drivers/net/wireless/rt2x00/rt2x00.h
index b32fedf..f0d7e08 100644
--- a/drivers/net/wireless/rt2x00/rt2x00.h
+++ b/drivers/net/wireless/rt2x00/rt2x00.h
@@ -521,6 +521,7 @@ struct rt2x00lib_ops {
 			       struct sk_buff *skb,
 			       struct txentry_desc *txdesc);
 	int (*write_tx_data) (struct queue_entry *entry);
+	void (*write_beacon) (struct queue_entry *entry);
 	int (*get_tx_data_len) (struct rt2x00_dev *rt2x00dev,
 				struct sk_buff *skb);
 	void (*kick_tx_queue) (struct rt2x00_dev *rt2x00dev,
@@ -535,8 +536,6 @@ struct rt2x00lib_ops {
 	/*
 	 * Configuration handlers.
 	 */
-	int (*beacon_update) (struct ieee80211_hw *hw, struct sk_buff *bcn);
-
 	void (*config_filter) (struct rt2x00_dev *rt2x00dev,
 			       const unsigned int filter_flags);
 	void (*config_intf) (struct rt2x00_dev *rt2x00dev,
@@ -913,39 +912,6 @@ static inline u16 get_duration_res(const unsigned int size, const u8 rate)
 void rt2x00queue_map_txskb(struct rt2x00_dev *rt2x00dev, struct sk_buff *skb);
 
 /**
- * rt2x00queue_create_tx_descriptor - Create TX descriptor from mac80211 input
- * @entry: The entry which will be used to transfer the TX frame.
- * @txdesc: rt2x00 TX descriptor which will be initialized by this function.
- *
- * This function will initialize the &struct txentry_desc based on information
- * from mac80211. This descriptor can then be used by rt2x00lib and the drivers
- * to correctly initialize the hardware descriptor.
- * Note that before calling this function the skb->cb array must be untouched
- * by rt2x00lib. Only after this function completes will it be save to
- * overwrite the skb->cb information.
- * The reason for this is that mac80211 writes its own tx information into
- * the skb->cb array, and this function will use that information to initialize
- * the &struct txentry_desc structure.
- */
-void rt2x00queue_create_tx_descriptor(struct queue_entry *entry,
-				      struct txentry_desc *txdesc);
-
-/**
- * rt2x00queue_write_tx_descriptor - Write TX descriptor to hardware
- * @entry: The entry which will be used to transfer the TX frame.
- * @txdesc: TX descriptor which will be used to write hardware descriptor
- *
- * This function will write a TX descriptor initialized by
- * &rt2x00queue_create_tx_descriptor to the hardware. After this call
- * has completed the frame is now owned by the hardware, the hardware
- * queue will have automatically be kicked unless this frame was generated
- * by rt2x00lib, in which case the frame is "special" and must be kicked
- * by the caller.
- */
-void rt2x00queue_write_tx_descriptor(struct queue_entry *entry,
-				     struct txentry_desc *txdesc);
-
-/**
  * rt2x00queue_get_queue - Convert queue index to queue pointer
  * @rt2x00dev: Pointer to &struct rt2x00_dev.
  * @queue: rt2x00 queue index (see &enum data_queue_qid).
diff --git a/drivers/net/wireless/rt2x00/rt2x00dev.c b/drivers/net/wireless/rt2x00/rt2x00dev.c
index 1b7f877..8c93eb8 100644
--- a/drivers/net/wireless/rt2x00/rt2x00dev.c
+++ b/drivers/net/wireless/rt2x00/rt2x00dev.c
@@ -434,13 +434,8 @@ static void rt2x00lib_intf_scheduled_iter(void *data, u8 *mac,
 	if (!test_bit(DEVICE_ENABLED_RADIO, &rt2x00dev->flags))
 		return;
 
-	if (delayed_flags & DELAYED_UPDATE_BEACON) {
-		struct ieee80211_if_conf conf;
-		conf.bssid = conf.ssid = NULL;
-		conf.ssid_len = 0;
-		conf.changed = IEEE80211_IFCC_BEACON;
-		rt2x00dev->ops->hw->config_interface(rt2x00dev->hw, vif, &conf);
-	}
+	if (delayed_flags & DELAYED_UPDATE_BEACON)
+		rt2x00queue_update_beacon(rt2x00dev, vif);
 
 	if (delayed_flags & DELAYED_CONFIG_ERP)
 		rt2x00lib_config_erp(rt2x00dev, intf, &conf);
diff --git a/drivers/net/wireless/rt2x00/rt2x00lib.h b/drivers/net/wireless/rt2x00/rt2x00lib.h
index eae5ce1..f2c9b0e 100644
--- a/drivers/net/wireless/rt2x00/rt2x00lib.h
+++ b/drivers/net/wireless/rt2x00/rt2x00lib.h
@@ -139,6 +139,14 @@ void rt2x00queue_free_skb(struct rt2x00_dev *rt2x00dev, struct sk_buff *skb);
 int rt2x00queue_write_tx_frame(struct data_queue *queue, struct sk_buff *skb);
 
 /**
+ * rt2x00queue_update_beacon - Send new beacon from mac80211 to hardware
+ * @rt2x00dev: Pointer to &struct rt2x00_dev.
+ * @vif: Interface for which the beacon should be updated.
+ */
+int rt2x00queue_update_beacon(struct rt2x00_dev *rt2x00dev,
+			      struct ieee80211_vif *vif);
+
+/**
  * rt2x00queue_index_inc - Index incrementation function
  * @queue: Queue (&struct data_queue) to perform the action on.
  * @index: Index type (&enum queue_index) to perform the action on.
diff --git a/drivers/net/wireless/rt2x00/rt2x00mac.c b/drivers/net/wireless/rt2x00/rt2x00mac.c
index ff853c4..214130e 100644
--- a/drivers/net/wireless/rt2x00/rt2x00mac.c
+++ b/drivers/net/wireless/rt2x00/rt2x00mac.c
@@ -348,8 +348,8 @@ int rt2x00mac_config_interface(struct ieee80211_hw *hw,
 {
 	struct rt2x00_dev *rt2x00dev = hw->priv;
 	struct rt2x00_intf *intf = vif_to_intf(vif);
-	struct sk_buff *beacon;
-	int status;
+	int update_bssid = 0;
+	int status = 0;
 
 	/*
 	 * Mac80211 might be calling this function while we are trying
@@ -361,15 +361,13 @@ int rt2x00mac_config_interface(struct ieee80211_hw *hw,
 	spin_lock(&intf->lock);
 
 	/*
-	 * If the interface does not work in master mode,
-	 * then the bssid value in the interface structure
-	 * should now be set.
-	 *
 	 * conf->bssid can be NULL if coming from the internal
 	 * beacon update routine.
 	 */
-	if (conf->bssid && vif->type != IEEE80211_IF_TYPE_AP)
+	if (conf->changed & IEEE80211_IFCC_BSSID && conf->bssid) {
+		update_bssid = 1;
 		memcpy(&intf->bssid, conf->bssid, ETH_ALEN);
+	}
 
 	spin_unlock(&intf->lock);
 
@@ -379,23 +377,14 @@ int rt2x00mac_config_interface(struct ieee80211_hw *hw,
 	 * values as arguments we make keep access to rt2x00_intf thread safe
 	 * even without the lock.
 	 */
-	rt2x00lib_config_intf(rt2x00dev, intf, vif->type, NULL, conf->bssid);
+	rt2x00lib_config_intf(rt2x00dev, intf, vif->type, NULL,
+			      update_bssid ? conf->bssid : NULL);
 
 	/*
-	 * We only need to initialize the beacon when in master/ibss mode.
+	 * Update the beacon.
 	 */
-	if ((vif->type != IEEE80211_IF_TYPE_AP &&
-	     vif->type != IEEE80211_IF_TYPE_IBSS) ||
-	    !(conf->changed & IEEE80211_IFCC_BEACON))
-		return 0;
-
-	beacon = ieee80211_beacon_get(rt2x00dev->hw, vif);
-	if (!beacon)
-		return -ENOMEM;
-
-	status = rt2x00dev->ops->lib->beacon_update(rt2x00dev->hw, beacon);
-	if (status)
-		dev_kfree_skb(beacon);
+	if (conf->changed & IEEE80211_IFCC_BEACON)
+		status = rt2x00queue_update_beacon(rt2x00dev, vif);
 
 	return status;
 }
diff --git a/drivers/net/wireless/rt2x00/rt2x00queue.c b/drivers/net/wireless/rt2x00/rt2x00queue.c
index ecf57f8..7f44203 100644
--- a/drivers/net/wireless/rt2x00/rt2x00queue.c
+++ b/drivers/net/wireless/rt2x00/rt2x00queue.c
@@ -115,8 +115,8 @@ void rt2x00queue_free_skb(struct rt2x00_dev *rt2x00dev, struct sk_buff *skb)
 	dev_kfree_skb_any(skb);
 }
 
-void rt2x00queue_create_tx_descriptor(struct queue_entry *entry,
-				      struct txentry_desc *txdesc)
+static void rt2x00queue_create_tx_descriptor(struct queue_entry *entry,
+					     struct txentry_desc *txdesc)
 {
 	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
 	struct ieee80211_tx_info *tx_info = IEEE80211_SKB_CB(entry->skb);
@@ -240,10 +240,9 @@ void rt2x00queue_create_tx_descriptor(struct queue_entry *entry,
 			txdesc->signal |= 0x08;
 	}
 }
-EXPORT_SYMBOL_GPL(rt2x00queue_create_tx_descriptor);
 
-void rt2x00queue_write_tx_descriptor(struct queue_entry *entry,
-				     struct txentry_desc *txdesc)
+static void rt2x00queue_write_tx_descriptor(struct queue_entry *entry,
+					    struct txentry_desc *txdesc)
 {
 	struct data_queue *queue = entry->queue;
 	struct rt2x00_dev *rt2x00dev = queue->rt2x00dev;
@@ -273,7 +272,6 @@ void rt2x00queue_write_tx_descriptor(struct queue_entry *entry,
 	    !test_bit(ENTRY_TXD_BURST, &txdesc->flags))
 		rt2x00dev->ops->lib->kick_tx_queue(rt2x00dev, queue->qid);
 }
-EXPORT_SYMBOL_GPL(rt2x00queue_write_tx_descriptor);
 
 int rt2x00queue_write_tx_frame(struct data_queue *queue, struct sk_buff *skb)
 {
@@ -323,6 +321,60 @@ int rt2x00queue_write_tx_frame(struct data_queue *queue, struct sk_buff *skb)
 	return 0;
 }
 
+int rt2x00queue_update_beacon(struct rt2x00_dev *rt2x00dev,
+			      struct ieee80211_vif *vif)
+{
+	struct rt2x00_intf *intf = vif_to_intf(vif);
+	struct skb_frame_desc *skbdesc;
+	struct txentry_desc txdesc;
+	__le32 desc[16];
+
+	if (unlikely(!intf->beacon))
+		return -ENOBUFS;
+
+	intf->beacon->skb = ieee80211_beacon_get(rt2x00dev->hw, vif);
+	if (!intf->beacon->skb)
+		return -ENOMEM;
+
+	/*
+	 * Copy all TX descriptor information into txdesc,
+	 * after that we are free to use the skb->cb array
+	 * for our information.
+	 */
+	rt2x00queue_create_tx_descriptor(intf->beacon, &txdesc);
+
+	/*
+	 * For the descriptor we use a local array from where the
+	 * driver can move it to the correct location required for
+	 * the hardware.
+	 */
+	memset(desc, 0, sizeof(desc));
+
+	/*
+	 * Fill in skb descriptor
+	 */
+	skbdesc = get_skb_frame_desc(intf->beacon->skb);
+	memset(skbdesc, 0, sizeof(*skbdesc));
+	skbdesc->desc = desc;
+	skbdesc->desc_len = intf->beacon->queue->desc_size;
+	skbdesc->entry = intf->beacon;
+
+	/*
+	 * Write TX descriptor into reserved room in front of the beacon.
+	 */
+	rt2x00queue_write_tx_descriptor(intf->beacon, &txdesc);
+
+	/*
+	 * Send beacon to hardware.
+	 * Also enable beacon generation, which might have been disabled
+	 * by the driver during the config_beacon() callback function.
+	 */
+	rt2x00dev->ops->lib->write_beacon(intf->beacon);
+	rt2x00dev->ops->lib->kick_tx_queue(rt2x00dev, QID_BEACON);
+
+	return 0;
+}
+
 struct data_queue *rt2x00queue_get_queue(struct rt2x00_dev *rt2x00dev,
 					 const enum data_queue_qid queue)
 {
diff --git a/drivers/net/wireless/rt2x00/rt61pci.c b/drivers/net/wireless/rt2x00/rt61pci.c
index 34d0fd0..70ef7bf 100644
--- a/drivers/net/wireless/rt2x00/rt61pci.c
+++ b/drivers/net/wireless/rt2x00/rt61pci.c
@@ -1600,6 +1600,41 @@ static void rt61pci_write_tx_desc(struct rt2x00_dev *rt2x00dev,
 /*
  * TX data initialization
  */
+static void rt61pci_write_beacon(struct queue_entry *entry)
+{
+	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
+	struct skb_frame_desc *skbdesc = get_skb_frame_desc(entry->skb);
+	unsigned int beacon_base;
+	u32 reg;
+
+	/*
+	 * Disable beaconing while we are reloading the beacon data,
+	 * otherwise we might be sending out invalid data.
+	 */
+	rt2x00pci_register_read(rt2x00dev, TXRX_CSR9, &reg);
+	rt2x00_set_field32(&reg, TXRX_CSR9_TSF_TICKING, 0);
+	rt2x00_set_field32(&reg, TXRX_CSR9_TBTT_ENABLE, 0);
+	rt2x00_set_field32(&reg, TXRX_CSR9_BEACON_GEN, 0);
+	rt2x00pci_register_write(rt2x00dev, TXRX_CSR9, reg);
+
+	/*
+	 * Write entire beacon with descriptor to register.
+	 */
+	beacon_base = HW_BEACON_OFFSET(entry->entry_idx);
+	rt2x00pci_register_multiwrite(rt2x00dev,
+				      beacon_base,
+				      skbdesc->desc, skbdesc->desc_len);
+	rt2x00pci_register_multiwrite(rt2x00dev,
+				      beacon_base + skbdesc->desc_len,
+				      entry->skb->data, entry->skb->len);
+
+	/*
+	 * Clean up beacon skb.
+	 */
+	dev_kfree_skb_any(entry->skb);
+	entry->skb = NULL;
+}
+
 static void rt61pci_kick_tx_queue(struct rt2x00_dev *rt2x00dev,
 				  const enum data_queue_qid queue)
 {
@@ -2355,72 +2390,6 @@ static u64 rt61pci_get_tsf(struct ieee80211_hw *hw)
 	return tsf;
 }
 
-static int rt61pci_beacon_update(struct ieee80211_hw *hw, struct sk_buff *skb)
-{
-	struct rt2x00_dev *rt2x00dev = hw->priv;
-	struct ieee80211_tx_info *tx_info = IEEE80211_SKB_CB(skb);
-	struct rt2x00_intf *intf = vif_to_intf(tx_info->control.vif);
-	struct queue_entry_priv_pci *entry_priv;
-	struct skb_frame_desc *skbdesc;
-	struct txentry_desc txdesc;
-	unsigned int beacon_base;
-	u32 reg;
-
-	if (unlikely(!intf->beacon))
-		return -ENOBUFS;
-
-	/*
-	 * Copy all TX descriptor information into txdesc,
-	 * after that we are free to use the skb->cb array
-	 * for our information.
-	 */
-	intf->beacon->skb = skb;
-	rt2x00queue_create_tx_descriptor(intf->beacon, &txdesc);
-
-	entry_priv = intf->beacon->priv_data;
-	memset(entry_priv->desc, 0, intf->beacon->queue->desc_size);
-
-	/*
-	 * Fill in skb descriptor
-	 */
-	skbdesc = get_skb_frame_desc(skb);
-	memset(skbdesc, 0, sizeof(*skbdesc));
-	skbdesc->desc = entry_priv->desc;
-	skbdesc->desc_len = intf->beacon->queue->desc_size;
-	skbdesc->entry = intf->beacon;
-
-	/*
-	 * Disable beaconing while we are reloading the beacon data,
-	 * otherwise we might be sending out invalid data.
-	 */
-	rt2x00pci_register_read(rt2x00dev, TXRX_CSR9, &reg);
-	rt2x00_set_field32(&reg, TXRX_CSR9_TSF_TICKING, 0);
-	rt2x00_set_field32(&reg, TXRX_CSR9_TBTT_ENABLE, 0);
-	rt2x00_set_field32(&reg, TXRX_CSR9_BEACON_GEN, 0);
-	rt2x00pci_register_write(rt2x00dev, TXRX_CSR9, reg);
-
-	/*
-	 * Write entire beacon with descriptor to register,
-	 * and kick the beacon generator.
-	 */
-	rt2x00queue_write_tx_descriptor(intf->beacon, &txdesc);
-	beacon_base = HW_BEACON_OFFSET(intf->beacon->entry_idx);
-	rt2x00pci_register_multiwrite(rt2x00dev, beacon_base,
-				      skbdesc->desc, skbdesc->desc_len);
-	rt2x00pci_register_multiwrite(rt2x00dev,
-				      beacon_base + skbdesc->desc_len,
-				      skb->data, skb->len);
-	rt61pci_kick_tx_queue(rt2x00dev, QID_BEACON);
-
-	/*
-	 * Clean up beacon skb.
-	 */
-	dev_kfree_skb_any(skb);
-	intf->beacon->skb = NULL;
-
-	return 0;
-}
-
 static const struct ieee80211_ops rt61pci_mac80211_ops = {
 	.tx			= rt2x00mac_tx,
 	.start			= rt2x00mac_start,
@@ -2455,9 +2424,9 @@ static const struct rt2x00lib_ops rt61pci_rt2x00_ops = {
 	.link_tuner		= rt61pci_link_tuner,
 	.write_tx_desc		= rt61pci_write_tx_desc,
 	.write_tx_data		= rt2x00pci_write_tx_data,
+	.write_beacon		= rt61pci_write_beacon,
 	.kick_tx_queue		= rt61pci_kick_tx_queue,
 	.fill_rxdone		= rt61pci_fill_rxdone,
-	.beacon_update		= rt61pci_beacon_update,
 	.config_filter		= rt61pci_config_filter,
 	.config_intf		= rt61pci_config_intf,
 	.config_erp		= rt61pci_config_erp,
diff --git a/drivers/net/wireless/rt2x00/rt73usb.c b/drivers/net/wireless/rt2x00/rt73usb.c
index 501c2d9..34c6ff2 100644
--- a/drivers/net/wireless/rt2x00/rt73usb.c
+++ b/drivers/net/wireless/rt2x00/rt73usb.c
@@ -1343,6 +1343,49 @@ static void rt73usb_write_tx_desc(struct rt2x00_dev *rt2x00dev,
 	rt2x00_desc_write(txd, 0, word);
 }
 
+/*
+ * TX data initialization
+ */
+static void rt73usb_write_beacon(struct queue_entry *entry)
+{
+	struct rt2x00_dev *rt2x00dev = entry->queue->rt2x00dev;
+	struct skb_frame_desc *skbdesc = get_skb_frame_desc(entry->skb);
+	unsigned int beacon_base;
+	u32 reg;
+
+	/*
+	 * Add the descriptor in front of the skb.
+	 */
+	skb_push(entry->skb, entry->queue->desc_size);
+	memcpy(entry->skb->data, skbdesc->desc, skbdesc->desc_len);
+	skbdesc->desc = entry->skb->data;
+
+	/*
+	 * Disable beaconing while we are reloading the beacon data,
+	 * otherwise we might be sending out invalid data.
+	 */
+	rt73usb_register_read(rt2x00dev, TXRX_CSR9, &reg);
+	rt2x00_set_field32(&reg, TXRX_CSR9_TSF_TICKING, 0);
+	rt2x00_set_field32(&reg, TXRX_CSR9_TBTT_ENABLE, 0);
+	rt2x00_set_field32(&reg, TXRX_CSR9_BEACON_GEN, 0);
+	rt73usb_register_write(rt2x00dev, TXRX_CSR9, reg);
+
+	/*
+	 * Write entire beacon with descriptor to register.
+	 */
+	beacon_base = HW_BEACON_OFFSET(entry->entry_idx);
+	rt2x00usb_vendor_request(rt2x00dev, USB_MULTI_WRITE,
+				 USB_VENDOR_REQUEST_OUT, beacon_base, 0,
+				 entry->skb->data, entry->skb->len,
+				 REGISTER_TIMEOUT32(entry->skb->len));
+
+	/*
+	 * Clean up the beacon skb.
+	 */
+	dev_kfree_skb(entry->skb);
+	entry->skb = NULL;
+}
+
 static int rt73usb_get_tx_data_len(struct rt2x00_dev *rt2x00dev,
 				   struct sk_buff *skb)
 {
@@ -1358,9 +1401,6 @@ static int rt73usb_get_tx_data_len(struct rt2x00_dev *rt2x00dev,
 	return length;
 }
 
-/*
- * TX data initialization
- */
 static void rt73usb_kick_tx_queue(struct rt2x00_dev *rt2x00dev,
 				  const enum data_queue_qid queue)
 {
@@ -1958,73 +1998,6 @@ static u64 rt73usb_get_tsf(struct ieee80211_hw *hw)
 #define rt73usb_get_tsf	NULL
 #endif
 
-static int rt73usb_beacon_update(struct ieee80211_hw *hw, struct sk_buff *skb)
-{
-	struct rt2x00_dev *rt2x00dev = hw->priv;
-	struct ieee80211_tx_info *tx_info = IEEE80211_SKB_CB(skb);
-	struct rt2x00_intf *intf = vif_to_intf(tx_info->control.vif);
-	struct skb_frame_desc *skbdesc;
-	struct txentry_desc txdesc;
-	unsigned int beacon_base;
-	u32 reg;
-
-	if (unlikely(!intf->beacon))
-		return -ENOBUFS;
-
-	/*
-	 * Copy all TX descriptor information into txdesc,
-	 * after that we are free to use the skb->cb array
-	 * for our information.
-	 */
-	intf->beacon->skb = skb;
-	rt2x00queue_create_tx_descriptor(intf->beacon, &txdesc);
-
-	/*
-	 * Add the descriptor in front of the skb.
-	 */
-	skb_push(skb, intf->beacon->queue->desc_size);
-	memset(skb->data, 0, intf->beacon->queue->desc_size);
-
-	/*
-	 * Fill in skb descriptor
-	 */
-	skbdesc = get_skb_frame_desc(skb);
-	memset(skbdesc, 0, sizeof(*skbdesc));
-	skbdesc->desc = skb->data;
-	skbdesc->desc_len = intf->beacon->queue->desc_size;
-	skbdesc->entry = intf->beacon;
-
-	/*
-	 * Disable beaconing while we are reloading the beacon data,
-	 * otherwise we might be sending out invalid data.
-	 */
-	rt73usb_register_read(rt2x00dev, TXRX_CSR9, &reg);
-	rt2x00_set_field32(&reg, TXRX_CSR9_TSF_TICKING, 0);
-	rt2x00_set_field32(&reg, TXRX_CSR9_TBTT_ENABLE, 0);
-	rt2x00_set_field32(&reg, TXRX_CSR9_BEACON_GEN, 0);
-	rt73usb_register_write(rt2x00dev, TXRX_CSR9, reg);
-
-	/*
-	 * Write entire beacon with descriptor to register,
-	 * and kick the beacon generator.
-	 */
-	rt2x00queue_write_tx_descriptor(intf->beacon, &txdesc);
-	beacon_base = HW_BEACON_OFFSET(intf->beacon->entry_idx);
-	rt2x00usb_vendor_request(rt2x00dev, USB_MULTI_WRITE,
-				 USB_VENDOR_REQUEST_OUT, beacon_base, 0,
-				 skb->data, skb->len,
-				 REGISTER_TIMEOUT32(skb->len));
-	rt73usb_kick_tx_queue(rt2x00dev, QID_BEACON);
-
-	/*
-	 * Clean up the beacon skb.
-	 */
-	dev_kfree_skb(skb);
-	intf->beacon->skb = NULL;
-
-	return 0;
-}
-
 static const struct ieee80211_ops rt73usb_mac80211_ops = {
 	.tx			= rt2x00mac_tx,
 	.start			= rt2x00mac_start,
@@ -2057,10 +2030,10 @@ static const struct rt2x00lib_ops rt73usb_rt2x00_ops = {
 	.link_tuner		= rt73usb_link_tuner,
 	.write_tx_desc		= rt73usb_write_tx_desc,
 	.write_tx_data		= rt2x00usb_write_tx_data,
+	.write_beacon		= rt73usb_write_beacon,
 	.get_tx_data_len	= rt73usb_get_tx_data_len,
 	.kick_tx_queue		= rt73usb_kick_tx_queue,
 	.fill_rxdone		= rt73usb_fill_rxdone,
-	.beacon_update		= rt73usb_beacon_update,
 	.config_filter		= rt73usb_config_filter,
 	.config_intf		= rt73usb_config_intf,
 	.config_erp		= rt73usb_config_erp,
-- 
1.5.6.1

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1382 bytes --]

On Wed, 2008-07-09 at 15:11 +0200, Ivo van Doorn wrote:
> Currently only beacons generated in AP mode have the software
> sequence number inserted. This means IBSS and Mesh mode are broken
> for all hardware that require software sequence numbers.
> 
> Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>

Acked-by: Johannes Berg <johannes@sipsolutions.net>


> ---
>  net/mac80211/tx.c |    5 +++++
>  1 files changed, 5 insertions(+), 0 deletions(-)
> 
> diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
> index 359cb36..0e7e1f3 100644
> --- a/net/mac80211/tx.c
> +++ b/net/mac80211/tx.c
> @@ -1876,6 +1876,8 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
>  		hdr->frame_control = IEEE80211_FC(IEEE80211_FTYPE_MGMT,
>  						  IEEE80211_STYPE_BEACON);
>  
> +		ieee80211_include_sequence(sdata, hdr);
> +
>  		num_beacons = &ifsta->num_beacons;
>  	} else if (ieee80211_vif_is_mesh(&sdata->vif)) {
>  		/* headroom, head length, tail length and maximum TIM length */
> @@ -1896,6 +1898,9 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
>  			cpu_to_le16(local->hw.conf.beacon_int);
>  		mgmt->u.beacon.capab_info = 0x0; /* 0x0 for MPs */
>  
> +		ieee80211_include_sequence(sdata,
> +					   (struct ieee80211_hdr *)skb->data);
> +
>  		pos = skb_put(skb, 2);
>  		*pos++ = WLAN_EID_SSID;
>  		*pos++ = 0x0;

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Michael Buesch]

On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> Currently only beacons generated in AP mode have the software
> sequence number inserted. This means IBSS and Mesh mode are broken
> for all hardware that require software sequence numbers.

Does software seq numbering even work at all?
What about packets that get sent between the driver requested the
beacon and the driver does actually queue it?

> Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
> ---
>  net/mac80211/tx.c |    5 +++++
>  1 files changed, 5 insertions(+), 0 deletions(-)
> 
> diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
> index 359cb36..0e7e1f3 100644
> --- a/net/mac80211/tx.c
> +++ b/net/mac80211/tx.c
> @@ -1876,6 +1876,8 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
>  		hdr->frame_control = IEEE80211_FC(IEEE80211_FTYPE_MGMT,
>  						  IEEE80211_STYPE_BEACON);
>  
> +		ieee80211_include_sequence(sdata, hdr);
> +
>  		num_beacons = &ifsta->num_beacons;
>  	} else if (ieee80211_vif_is_mesh(&sdata->vif)) {
>  		/* headroom, head length, tail length and maximum TIM length */
> @@ -1896,6 +1898,9 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
>  			cpu_to_le16(local->hw.conf.beacon_int);
>  		mgmt->u.beacon.capab_info = 0x0; /* 0x0 for MPs */
>  
> +		ieee80211_include_sequence(sdata,
> +					   (struct ieee80211_hdr *)skb->data);
> +
>  		pos = skb_put(skb, 2);
>  		*pos++ = WLAN_EID_SSID;
>  		*pos++ = 0x0;



-- 
Greetings Michael.

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 654 bytes --]

On Wed, 2008-07-09 at 15:38 +0200, Michael Buesch wrote:
> On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > Currently only beacons generated in AP mode have the software
> > sequence number inserted. This means IBSS and Mesh mode are broken
> > for all hardware that require software sequence numbers.
> 
> Does software seq numbering even work at all?
> What about packets that get sent between the driver requested the
> beacon and the driver does actually queue it?

Hmm. I guess we'd have to lock the master device while requesting a
beacon... but also lock around the driver code queuing the beacon or
something?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Michael Buesch wrote:
> On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > Currently only beacons generated in AP mode have the software
> > sequence number inserted. This means IBSS and Mesh mode are broken
> > for all hardware that require software sequence numbers.
> 
> Does software seq numbering even work at all?
> What about packets that get sent between the driver requested the
> beacon and the driver does actually queue it?

For rt2x00 the beacon is requested and queued within interrupt context
or during resume when it doesn't allow mac80211 to send any frames yet.

The problem would be when the config_interface() is called, if during that
call mac80211 also sends out TX frames, then you get problems in the
sequence counting.

Ivo

> > Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
> > ---
> >  net/mac80211/tx.c |    5 +++++
> >  1 files changed, 5 insertions(+), 0 deletions(-)
> > 
> > diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
> > index 359cb36..0e7e1f3 100644
> > --- a/net/mac80211/tx.c
> > +++ b/net/mac80211/tx.c
> > @@ -1876,6 +1876,8 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
> >  		hdr->frame_control = IEEE80211_FC(IEEE80211_FTYPE_MGMT,
> >  						  IEEE80211_STYPE_BEACON);
> >  
> > +		ieee80211_include_sequence(sdata, hdr);
> > +
> >  		num_beacons = &ifsta->num_beacons;
> >  	} else if (ieee80211_vif_is_mesh(&sdata->vif)) {
> >  		/* headroom, head length, tail length and maximum TIM length */
> > @@ -1896,6 +1898,9 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
> >  			cpu_to_le16(local->hw.conf.beacon_int);
> >  		mgmt->u.beacon.capab_info = 0x0; /* 0x0 for MPs */
> >  
> > +		ieee80211_include_sequence(sdata,
> > +					   (struct ieee80211_hdr *)skb->data);
> > +
> >  		pos = skb_put(skb, 2);
> >  		*pos++ = WLAN_EID_SSID;
> >  		*pos++ = 0x0;
> 
> 
> 

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1089 bytes --]

On Wed, 2008-07-09 at 16:04 +0200, Ivo van Doorn wrote:
> On Wednesday 09 July 2008, Michael Buesch wrote:
> > On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > > Currently only beacons generated in AP mode have the software
> > > sequence number inserted. This means IBSS and Mesh mode are broken
> > > for all hardware that require software sequence numbers.
> > 
> > Does software seq numbering even work at all?
> > What about packets that get sent between the driver requested the
> > beacon and the driver does actually queue it?
> 
> For rt2x00 the beacon is requested and queued within interrupt context
> or during resume when it doesn't allow mac80211 to send any frames yet.

Ok, in interrupt context definitely works.

> The problem would be when the config_interface() is called, if during that
> call mac80211 also sends out TX frames, then you get problems in the
> sequence counting.

Well, if you care about the beacon in config_interface() you're
implementing hwseq since otherwise you'd only request it when you need
it, no?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Johannes Berg wrote:
> On Wed, 2008-07-09 at 16:04 +0200, Ivo van Doorn wrote:
> > On Wednesday 09 July 2008, Michael Buesch wrote:
> > > On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > > > Currently only beacons generated in AP mode have the software
> > > > sequence number inserted. This means IBSS and Mesh mode are broken
> > > > for all hardware that require software sequence numbers.
> > > 
> > > Does software seq numbering even work at all?
> > > What about packets that get sent between the driver requested the
> > > beacon and the driver does actually queue it?
> > 
> > For rt2x00 the beacon is requested and queued within interrupt context
> > or during resume when it doesn't allow mac80211 to send any frames yet.
> 
> Ok, in interrupt context definitely works.
> 
> > The problem would be when the config_interface() is called, if during that
> > call mac80211 also sends out TX frames, then you get problems in the
> > sequence counting.
> 
> Well, if you care about the beacon in config_interface() you're
> implementing hwseq since otherwise you'd only request it when you need
> it, no?

No not really, because from what moment can the driver start requesting beacons?
I would think when mac80211 calls config_interface with the IEEE80211_IFCC_BEACON
flag. ;)

At that moment the first beacon is loaded, and the device will trigger the interrupt
to request a new beacon every x milliseconds.

Ivo

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 663 bytes --]

On Wed, 2008-07-09 at 16:57 +0200, Ivo van Doorn wrote:

> > Well, if you care about the beacon in config_interface() you're
> > implementing hwseq since otherwise you'd only request it when you need
> > it, no?
> 
> No not really, because from what moment can the driver start requesting beacons?
> I would think when mac80211 calls config_interface with the IEEE80211_IFCC_BEACON
> flag. ;)
> 
> At that moment the first beacon is loaded, and the device will trigger the interrupt
> to request a new beacon every x milliseconds.

Ok, I'd have thought you just set the "please send beacons now" flag and
request one on the next round.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Michael Buesch]

On Wednesday 09 July 2008 16:04:50 Ivo van Doorn wrote:
> On Wednesday 09 July 2008, Michael Buesch wrote:
> > On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > > Currently only beacons generated in AP mode have the software
> > > sequence number inserted. This means IBSS and Mesh mode are broken
> > > for all hardware that require software sequence numbers.
> > 
> > Does software seq numbering even work at all?
> > What about packets that get sent between the driver requested the
> > beacon and the driver does actually queue it?
> 
> For rt2x00 the beacon is requested and queued within interrupt context

Well, another CPU could be in progress of walking down the mac80211 TX code
and aquire a sequence number in the meantime before you requested the beacon.
However that frame will be blocked by your driver locks, so the two seq
numbers of the beacon and the other frame will be swapped, as the driver
will queue the beacon first.

-- 
Greetings Michael.

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1506 bytes --]

On Wed, 2008-07-09 at 16:00 +0200, Michael Buesch wrote:
> On Wednesday 09 July 2008 16:04:50 Ivo van Doorn wrote:
> > On Wednesday 09 July 2008, Michael Buesch wrote:
> > > On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > > > Currently only beacons generated in AP mode have the software
> > > > sequence number inserted. This means IBSS and Mesh mode are broken
> > > > for all hardware that require software sequence numbers.
> > > 
> > > Does software seq numbering even work at all?
> > > What about packets that get sent between the driver requested the
> > > beacon and the driver does actually queue it?
> > 
> > For rt2x00 the beacon is requested and queued within interrupt context
> 
> Well, another CPU could be in progress of walking down the mac80211 TX code
> and aquire a sequence number in the meantime before you requested the beacon.
> However that frame will be blocked by your driver locks, so the two seq
> numbers of the beacon and the other frame will be swapped, as the driver
> will queue the beacon first.

Indeed, I was wrong.

I wonder if we should remove the hwseq support completely. It's much
easier for the driver to do this, especially since we pass a vif pointer
with driver-private data to all relevant functions and the driver could
keep the current sequence number in there. Or, just like hwseq would,
keep a global sequence number?

How does that affect multi-bss support btw? Do we have to use sw seqno
for that?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Johannes Berg wrote:
> On Wed, 2008-07-09 at 16:00 +0200, Michael Buesch wrote:
> > On Wednesday 09 July 2008 16:04:50 Ivo van Doorn wrote:
> > > On Wednesday 09 July 2008, Michael Buesch wrote:
> > > > On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > > > > Currently only beacons generated in AP mode have the software
> > > > > sequence number inserted. This means IBSS and Mesh mode are broken
> > > > > for all hardware that require software sequence numbers.
> > > > 
> > > > Does software seq numbering even work at all?
> > > > What about packets that get sent between the driver requested the
> > > > beacon and the driver does actually queue it?
> > > 
> > > For rt2x00 the beacon is requested and queued within interrupt context
> > 
> > Well, another CPU could be in progress of walking down the mac80211 TX code
> > and aquire a sequence number in the meantime before you requested the beacon.
> > However that frame will be blocked by your driver locks, so the two seq
> > numbers of the beacon and the other frame will be swapped, as the driver
> > will queue the beacon first.

If rt2x00 would use a global lock to block all TX and beacons, then yes.
But rt2x00 uses per-queue locking. When a beacon is being updated rt2x00
will still allow regular frames to be queued.

But we can't halt the TX queues for the beacon update either, because
ieee80211_stop_queues() should only be called from within the TX path.

> Indeed, I was wrong.
> 
> I wonder if we should remove the hwseq support completely. It's much
> easier for the driver to do this, especially since we pass a vif pointer
> with driver-private data to all relevant functions and the driver could
> keep the current sequence number in there. Or, just like hwseq would,
> keep a global sequence number?

I think that would be moving the same problem to the driver side,
and I am not really a big fan of the idea to put all queue handling under
1 big lock instead of the per-queue locking which is currently the case.

> How does that affect multi-bss support btw? Do we have to use sw seqno
> for that?

As far as rt2x00 is concerned, all hardware that supports multi-bss also
support HW sequence counters. rt2400pci and rt2500pci are the only ones
requiring SW sequence counters, and they can't do multi-bss.

Ivo

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 624 bytes --]


> If rt2x00 would use a global lock to block all TX and beacons, then yes.
> But rt2x00 uses per-queue locking. When a beacon is being updated rt2x00
> will still allow regular frames to be queued.

That just widens the window. And if you have multiple queues then you
can't do sw sequence numbers anyway because the hardware might reorder
the frames.

> As far as rt2x00 is concerned, all hardware that supports multi-bss also
> support HW sequence counters. rt2400pci and rt2500pci are the only ones
> requiring SW sequence counters, and they can't do multi-bss.

Do they keep per-BSS counters?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Johannes Berg wrote:
> 
> > If rt2x00 would use a global lock to block all TX and beacons, then yes.
> > But rt2x00 uses per-queue locking. When a beacon is being updated rt2x00
> > will still allow regular frames to be queued.
> 
> That just widens the window. And if you have multiple queues then you
> can't do sw sequence numbers anyway because the hardware might reorder
> the frames.

That doesn't seem to be a problem in the legacy drivers, so I guess the hardware
does *something* to prevent problems.

> > As far as rt2x00 is concerned, all hardware that supports multi-bss also
> > support HW sequence counters. rt2400pci and rt2500pci are the only ones
> > requiring SW sequence counters, and they can't do multi-bss.
> 
> Do they keep per-BSS counters?

Not sure what you mean, but rt2400pci and rt2500pci don't support multi-bss,
so they wouldn't have per-BSS counters.
rt61pci and rt73usb do support multi-bss and keep the sequence counting
completely in firmware/hardware and I am not sure how they can be read
by the driver. So I don't know how those are stored either.


Ivo

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Michael Buesch]

On Wednesday 09 July 2008 17:08:04 Ivo van Doorn wrote:
> On Wednesday 09 July 2008, Johannes Berg wrote:
> > 
> > > If rt2x00 would use a global lock to block all TX and beacons, then yes.
> > > But rt2x00 uses per-queue locking. When a beacon is being updated rt2x00
> > > will still allow regular frames to be queued.
> > 
> > That just widens the window. And if you have multiple queues then you
> > can't do sw sequence numbers anyway because the hardware might reorder
> > the frames.
> 
> That doesn't seem to be a problem in the legacy drivers, so I guess the hardware
> does *something* to prevent problems.

Well, legacy drivers don't use mac80211. Sequence counting can trivially be fixed
by doing it _right_ before queueing the packet to the hardware in the driver,
with the TX queue lock held, that must already be there.

-- 
Greetings Michael.

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Michael Buesch wrote:
> On Wednesday 09 July 2008 17:08:04 Ivo van Doorn wrote:
> > On Wednesday 09 July 2008, Johannes Berg wrote:
> > > 
> > > > If rt2x00 would use a global lock to block all TX and beacons, then yes.
> > > > But rt2x00 uses per-queue locking. When a beacon is being updated rt2x00
> > > > will still allow regular frames to be queued.
> > > 
> > > That just widens the window. And if you have multiple queues then you
> > > can't do sw sequence numbers anyway because the hardware might reorder
> > > the frames.
> > 
> > That doesn't seem to be a problem in the legacy drivers, so I guess the hardware
> > does *something* to prevent problems.
> 
> Well, legacy drivers don't use mac80211. Sequence counting can trivially be fixed
> by doing it _right_ before queueing the packet to the hardware in the driver,
> with the TX queue lock held, that must already be there.

True, I have to double check it, but the legacy drivers use per-queue locking as well.
And it would still be possible for the driver to send out frames with a higher
sequence number before one with a lower number.

But I'll put out a request on the rt2400-devel list to see if anybody is interested
in looking into this issue for rt2400pci and rt2500pci. In if there is nobody, I'll
disable adhoc and master mode for those drivers to prevent problems.

Ivo

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 747 bytes --]

On Wed, 2008-07-09 at 17:36 +0200, Ivo van Doorn wrote:

> True, I have to double check it, but the legacy drivers use per-queue locking as well.
> And it would still be possible for the driver to send out frames with a higher
> sequence number before one with a lower number.

The legacy drivers hopefully also implement proper sequence numbering
per TID. I'll send another mail on the topic.

> But I'll put out a request on the rt2400-devel list to see if anybody is interested
> in looking into this issue for rt2400pci and rt2500pci. In if there is nobody, I'll
> disable adhoc and master mode for those drivers to prevent problems.

Heh, that works. Just disable it and tell folks who want it to work to
fix it :)

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Johannes Berg wrote:
> On Wed, 2008-07-09 at 17:36 +0200, Ivo van Doorn wrote:
> 
> > True, I have to double check it, but the legacy drivers use per-queue locking as well.
> > And it would still be possible for the driver to send out frames with a higher
> > sequence number before one with a lower number.
> 
> The legacy drivers hopefully also implement proper sequence numbering
> per TID. I'll send another mail on the topic.

Heh, I seriously doubt that. As far as I know it is working using a single
counter in the main interface structure. Just the same way it currently
works in mac80211 (which isn't surprising, since I based the mac80211 patch
on the legacy driver implementation).

> > But I'll put out a request on the rt2400-devel list to see if anybody is interested
> > in looking into this issue for rt2400pci and rt2500pci. In if there is nobody, I'll
> > disable adhoc and master mode for those drivers to prevent problems.
> 
> Heh, that works. Just disable it and tell folks who want it to work to
> fix it :)

Well I am putting out an in advance notice first. ;)

Ivo

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1283 bytes --]

On Wed, 2008-07-09 at 17:48 +0200, Ivo van Doorn wrote:
> On Wednesday 09 July 2008, Johannes Berg wrote:
> > On Wed, 2008-07-09 at 17:36 +0200, Ivo van Doorn wrote:
> > 
> > > True, I have to double check it, but the legacy drivers use per-queue locking as well.
> > > And it would still be possible for the driver to send out frames with a higher
> > > sequence number before one with a lower number.
> > 
> > The legacy drivers hopefully also implement proper sequence numbering
> > per TID. I'll send another mail on the topic.
> 
> Heh, I seriously doubt that. As far as I know it is working using a single
> counter in the main interface structure. Just the same way it currently
> works in mac80211 (which isn't surprising, since I based the mac80211 patch
> on the legacy driver implementation).

Ok so the thing is this: 802.11-2007 7.1.3.4.1 requires that we keep
sequence counters per TID (and one for "not QoS"), which we only assign
rather late in the wme code. Because even b43 hardware only keeps track
of a single sequence counter, it requires not using that counter for QoS
frames. I suspect no driver except the overly smart iwlwifi gets this
right.

Do you have flags to turn off hwseq support for the other ralink
devices?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 936 bytes --]

On Wed, 2008-07-09 at 17:49 +0200, Johannes Berg wrote:

> Ok so the thing is this: 802.11-2007 7.1.3.4.1 requires that we keep
> sequence counters per TID (and one for "not QoS"), which we only assign
> rather late in the wme code. Because even b43 hardware only keeps track
> of a single sequence counter, it requires not using that counter for QoS
> frames. I suspect no driver except the overly smart iwlwifi gets this
> right.

Even if we fix this, we still haven't solved the problem of the beacon
sequence number. I'm currently favouring putting the burden on the
driver since only rt2400pci and rt2500pci seem to require it and it's
fairly simple to keep a race-free per-VIF counter for non-QoS frames in
the driver (you can just do it under the queue lock), but doing it
race-free in mac80211 will need a bunch of extra locking.

I'll put off the fix until after davem's TX MQ rework though.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Johannes Berg wrote:
> On Wed, 2008-07-09 at 17:49 +0200, Johannes Berg wrote:
>=20
> > Ok so the thing is this: 802.11-2007 7.1.3.4.1 requires that we kee=
p
> > sequence counters per TID (and one for "not QoS"), which we only as=
sign
> > rather late in the wme code. Because even b43 hardware only keeps t=
rack
> > of a single sequence counter, it requires not using that counter fo=
r QoS
> > frames. =EF=BB=BFI suspect no driver except the overly smart iwlwif=
i gets this
> > right.
>=20
> Even if we fix this, we still haven't solved the problem of the beaco=
n
> sequence number. I'm currently favouring putting the burden on the
> driver since only =EF=BB=BFrt2400pci and rt2500pci seem to require it=
 and it's
> fairly simple to keep a race-free per-VIF counter for non-QoS frames =
in
> the driver (you can just do it under the queue lock), but doing it
> race-free in mac80211 will need a bunch of extra locking.
>=20
> I'll put off the fix until after davem's TX MQ rework though.

Ok, that would allow some time for rt2x00 to come up with the
implementation for the rt2400pci and rt2500pci drivers.

And this is extra motivation for a volunteer to look into this,
because when mac80211 removes the seq counting, it is not
just adhoc and master mode that are broken. ;)

Ivo
--
To unsubscribe from this list: send the line "unsubscribe linux-wireles=
s" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 612 bytes --]

On Wed, 2008-07-09 at 18:12 +0200, Ivo van Doorn wrote:

> > I'll put off the fix until after davem's TX MQ rework though.

Actually, I don't have to.

> Ok, that would allow some time for rt2x00 to come up with the
> implementation for the rt2400pci and rt2500pci drivers.
> 
> And this is extra motivation for a volunteer to look into this,
> because when mac80211 removes the seq counting, it is not
> just adhoc and master mode that are broken. ;)

Here's a patch to try it out:

http://johannes.sipsolutions.net/patches/kernel/all/2008-07-09-17%3a38/022-mac80211-fix-hwseq.patch

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Johannes Berg wrote:
> On Wed, 2008-07-09 at 18:12 +0200, Ivo van Doorn wrote:
> 
> > > I'll put off the fix until after davem's TX MQ rework though.
> 
> Actually, I don't have to.
> 
> > Ok, that would allow some time for rt2x00 to come up with the
> > implementation for the rt2400pci and rt2500pci drivers.
> > 
> > And this is extra motivation for a volunteer to look into this,
> > because when mac80211 removes the seq counting, it is not
> > just adhoc and master mode that are broken. ;)
> 
> Here's a patch to try it out:
> 
> http://johannes.sipsolutions.net/patches/kernel/all/2008-07-09-17%3a38/022-mac80211-fix-hwseq.patch

Personally I have my doubts about the meaningfullness of the flag
	IEEE80211_TX_CTL_QOS_FRAME
since drivers will probably use ieee80211_is_data_qos() instead.
Perhaps it is better to have the flag named something like:
	IEEE80211_TX_CTS_SW_SEQUENCE
or
	IEEE80211_TX_CTS_HAS_SEQUENCE

That way it is a bit clearer what the intention of the flag is.

Ivo

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 685 bytes --]

On Wed, 2008-07-09 at 20:12 +0200, Ivo van Doorn wrote:

> > http://johannes.sipsolutions.net/patches/kernel/all/2008-07-09-17%3a38/022-mac80211-fix-hwseq.patch
> 
> Personally I have my doubts about the meaningfullness of the flag
> 	IEEE80211_TX_CTL_QOS_FRAME
> since drivers will probably use ieee80211_is_data_qos() instead.
> Perhaps it is better to have the flag named something like:
> 	IEEE80211_TX_CTS_SW_SEQUENCE
> or
> 	IEEE80211_TX_CTS_HAS_SEQUENCE
> 
> That way it is a bit clearer what the intention of the flag is.

Yeah, that's probably true, though I'd like people to be aware of the
TID sequence stuff... Guess we need to write more docs.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Ivo van Doorn]

On Wednesday 09 July 2008, Johannes Berg wrote:
> On Wed, 2008-07-09 at 17:48 +0200, Ivo van Doorn wrote:
> > On Wednesday 09 July 2008, Johannes Berg wrote:
> > > On Wed, 2008-07-09 at 17:36 +0200, Ivo van Doorn wrote:
> > >=20
> > > > True, I have to double check it, but the legacy drivers use per=
-queue locking as well.
> > > > And it would still be possible for the driver to send out frame=
s with a higher
> > > > sequence number before one with a lower number.
> > >=20
> > > The legacy drivers hopefully also implement proper sequence numbe=
ring
> > > per TID. I'll send another mail on the topic.
> >=20
> > Heh, I seriously doubt that. As far as I know it is working using a=
 single
> > counter in the main interface structure. Just the same way it curre=
ntly
> > works in mac80211 (which isn't surprising, since I based the mac802=
11 patch
> > on the legacy driver implementation).
>=20
> Ok so the thing is this: 802.11-2007 7.1.3.4.1 requires that we keep
> sequence counters per TID (and one for "not QoS"), which we only assi=
gn
> rather late in the wme code. Because even b43 hardware only keeps tra=
ck
> of a single sequence counter, it requires not using that counter for =
QoS
> frames. =EF=BB=BFI suspect no driver except the overly smart iwlwifi =
gets this
> right.
>=20
> Do you have flags to turn off hwseq support for the other ralink
> devices?

rt2500usb: Not per-frame, it can only be disabled globally
rt61pci: Both per-frame as well as globally
rt73usb: Both per-frame as well as globally
rt2800pci: Not per-frame, not sure about globally
rt2800usb: Not per-frame, not sure about globally

Note that I don't know if these devices handle the TID and the not QOS =
sequence
counting in the hardware. I'll have to consult the specsheets for that.

Ivo
--
To unsubscribe from this list: send the line "unsubscribe linux-wireles=
s" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 718 bytes --]

On Wed, 2008-07-09 at 18:08 +0200, Ivo van Doorn wrote:

> rt2500usb: Not per-frame, it can only be disabled globally
> rt61pci: Both per-frame as well as globally
> rt73usb: Both per-frame as well as globally
> rt2800pci: Not per-frame, not sure about globally
> rt2800usb: Not per-frame, not sure about globally

Interesting. The reason that broadcom firmware allows to disable
per-frame is actually to implement the per-TID counters in software.

> Note that I don't know if these devices handle the TID and the not QOS sequence
> counting in the hardware. I'll have to consult the specsheets for that.

Not much we can do about it either way unless you want to always disable
globally.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1256 bytes --]

On Wed, 2008-07-09 at 17:08 +0200, Ivo van Doorn wrote:

> > That just widens the window. And if you have multiple queues then you
> > can't do sw sequence numbers anyway because the hardware might reorder
> > the frames.
> 
> That doesn't seem to be a problem in the legacy drivers, so I guess the hardware
> does *something* to prevent problems.

Actually, looking at 7.1.3.4.1 (Sequence Number field) this all seems to
be wrong anyway, we should be keeping a sequence number counter
_per_TID_ and one for non-QoS data/mgmt frames.

> > > As far as rt2x00 is concerned, all hardware that supports multi-bss also
> > > support HW sequence counters. rt2400pci and rt2500pci are the only ones
> > > requiring SW sequence counters, and they can't do multi-bss.
> > 
> > Do they keep per-BSS counters?
> 
> Not sure what you mean, but rt2400pci and rt2500pci don't support multi-bss,
> so they wouldn't have per-BSS counters.
> rt61pci and rt73usb do support multi-bss and keep the sequence counting
> completely in firmware/hardware and I am not sure how they can be read
> by the driver. So I don't know how those are stored either.

Ok. Since the hw is capable of mBSS I'd guess it has sequence numbers
for each BSS.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[Michael Buesch]

On Wednesday 09 July 2008 16:37:01 Ivo van Doorn wrote:
> On Wednesday 09 July 2008, Johannes Berg wrote:
> > On Wed, 2008-07-09 at 16:00 +0200, Michael Buesch wrote:
> > > On Wednesday 09 July 2008 16:04:50 Ivo van Doorn wrote:
> > > > On Wednesday 09 July 2008, Michael Buesch wrote:
> > > > > On Wednesday 09 July 2008 15:11:45 Ivo van Doorn wrote:
> > > > > > Currently only beacons generated in AP mode have the software
> > > > > > sequence number inserted. This means IBSS and Mesh mode are broken
> > > > > > for all hardware that require software sequence numbers.
> > > > > 
> > > > > Does software seq numbering even work at all?
> > > > > What about packets that get sent between the driver requested the
> > > > > beacon and the driver does actually queue it?
> > > > 
> > > > For rt2x00 the beacon is requested and queued within interrupt context
> > > 
> > > Well, another CPU could be in progress of walking down the mac80211 TX code
> > > and aquire a sequence number in the meantime before you requested the beacon.
> > > However that frame will be blocked by your driver locks, so the two seq
> > > numbers of the beacon and the other frame will be swapped, as the driver
> > > will queue the beacon first.
> 
> If rt2x00 would use a global lock to block all TX and beacons, then yes.
> But rt2x00 uses per-queue locking. When a beacon is being updated rt2x00
> will still allow regular frames to be queued.

It still leaves a race condition window. That's actually worse than
consistently breaking ;)

> But we can't halt the TX queues for the beacon update either, because
> ieee80211_stop_queues() should only be called from within the TX path.

I think the problem can only be fixed by doing sequence numbering in the driver.

-- 
Greetings Michael.

Re: [PATCH 1/3] mac80211: Include sequence number in IBSS and Mesh beacons

[John W. Linville]

On Wed, Jul 09, 2008 at 03:11:45PM +0200, Ivo van Doorn wrote:
> Currently only beacons generated in AP mode have the software
> sequence number inserted. This means IBSS and Mesh mode are broken
> for all hardware that require software sequence numbers.
> 
> Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>

As you are aware, Johannes posted a patch that removed
ieee80211_include_sequence.  I'm dropping this (i.e. Ivo's) patch.

John
-- 
John W. Linville
linville@tuxdriver.com

[PATCH 0/3] rt2x00 update

[Ivo van Doorn]

Hi John,

Here are 3 patches for rt2x00 intended for the 2.6.27 kernel.
It finally fixes the ifdown & ifup breakage in rt61pci, as well
as a BUG_ON() which was triggered during such an ifdown & ifup
cycle.

Ivo