ath5k: fix missing output in monitor mode after ifconfig up

PATCH Superseded Joerg Albert 3 participants 8 messages Aug 1, 2009 – Aug 4, 2009 Atheros ATH5K wireless driver

Share & tools

Joerg Albert

author

2009-08-01 22:33 UTC #0 lore raw

PATCH 1/1 ath5k: fix missing output in monitor mode after ifconfig up

Let ath5k_chan_set() call ath5k_reset() if ath5k_config() was told that the channel changed. This fixes the bug that we don't get any packets in monitor mode after:

ifconfig wlan0 down
iwconfig wlan0 mode monitor channel 1
ifconfig wlan0 up

but they arrive after

iwconfig wlan0 channel 2

Signed-off-by Joerg Albert <[email protected]>

---
 drivers/net/wireless/ath/ath5k/base.c |   17 +++++++++--------
 1 files changed, 9 insertions(+), 8 deletions(-)

drivers/net/wireless/ath/ath5k/base.c +9 -8

index 3a1c156..09ca89c 100644
@@ -289,7 +289,8 @@ static unsigned int ath5k_copy_channels(struct ath5k_hw *ah,
                                unsigned int max);
 static int     ath5k_setup_bands(struct ieee80211_hw *hw);
 static int     ath5k_chan_set(struct ath5k_softc *sc,
-                               struct ieee80211_channel *chan);
+                              struct ieee80211_channel *chan,
+                              int chan_changed);
 static void    ath5k_setcurmode(struct ath5k_softc *sc,
                                unsigned int mode);
 static void    ath5k_mode_setup(struct ath5k_softc *sc);

@@ -1079,14 +1080,13 @@ ath5k_setup_bands(struct ieee80211_hw *hw)
  * Called with sc->lock.
  */
 static int
-ath5k_chan_set(struct ath5k_softc *sc, struct ieee80211_channel *chan)
+ath5k_chan_set(struct ath5k_softc *sc, struct ieee80211_channel *chan,
+              int chan_changed)
 {
-       ATH5K_DBG(sc, ATH5K_DEBUG_RESET, "(%u MHz) -> (%u MHz)\n",
-               sc->curchan->center_freq, chan->center_freq);
-
-       if (chan->center_freq != sc->curchan->center_freq ||
-               chan->hw_value != sc->curchan->hw_value) {
+       ATH5K_DBG(sc, ATH5K_DEBUG_RESET, "(%u MHz) -> (%u MHz) chan_changed %x\n",
+                 sc->curchan->center_freq, chan->center_freq, chan_changed);
 
+       if (chan_changed) {
                /*
                 * To switch channels clear any pending DMA operations;
                 * wait long enough for the RX fifo to drain, reset the

@@ -2811,7 +2811,8 @@ ath5k_config(struct ieee80211_hw *hw, u32 changed)
 
        mutex_lock(&sc->lock);
 
-       ret = ath5k_chan_set(sc, conf->channel);
+       ret = ath5k_chan_set(sc, conf->channel,
+                            changed & IEEE80211_CONF_CHANGE_CHANNEL);
        if (ret < 0)
                goto unlock;

Signature · 1 line

1.6.0.4

Bob Copeland

2009-08-02 03:21 UTC #1 lore raw

Quoted 3 lines · Joerg Albert

Let ath5k_chan_set() call ath5k_reset() if ath5k_config() was told that the channel changed. This fixes the bug that we don't get any packets in monitor mode after:

Cool, thanks for the patch. Just a couple of comments:

Quoted 1 line

+ int chan_changed);

Should use bool for that.

drivers/net/wireless/ath/ath5k/base.c ath5k_config(struct ieee80211_hw *hw, u32 changed) · line 2815

Quoted 3 lines

drivers/net/wireless/ath/ath5k/base.c +2 -1

-       ret = ath5k_chan_set(sc, conf->channel);
+       ret = ath5k_chan_set(sc, conf->channel,
+                            changed & IEEE80211_CONF_CHANGE_CHANNEL);

Since this is the only place ath5k_chan_set is called, how about changing ath5k_chan_set to unconditionally do a reset, and put the test outside?

i.e.:

if (changed & IEEE80211_CONF_CHANGE_CHANNEL)
ath5k_chan_set(sc, conf->channel);

Thanks!

Signature · 1 line

Bob Copeland %% www.bobcopeland.com

Joerg Albert

author

2009-08-02 07:36 UTC #2 lore raw

Quoted 15 lines · Bob Copeland

Quoted 3 lines · Joerg Albert
Let ath5k_chan_set() call ath5k_reset() if ath5k_config() was told that the channel changed. This fixes the bug that we don't get any packets in monitor mode after:
Cool, thanks for the patch. Just a couple of comments:
Quoted 1 line
+ int chan_changed);
Should use bool for that.
drivers/net/wireless/ath/ath5k/base.c ath5k_config(struct ieee80211_hw *hw, u32 changed) · line 2815
Quoted 3 lines
drivers/net/wireless/ath/ath5k/base.c +2 -1
-       ret = ath5k_chan_set(sc, conf->channel);
+       ret = ath5k_chan_set(sc, conf->channel,
+                            changed & IEEE80211_CONF_CHANGE_CHANNEL);

Quoted 7 lines

Since this is the only place ath5k_chan_set is called, how about changing ath5k_chan_set to unconditionally do a reset, and put the test outside?
i.e.:
if (changed & IEEE80211_CONF_CHANGE_CHANNEL)
ath5k_chan_set(sc, conf->channel);

Good idea. I'll redo the patch.

BTW, it assumes that IEEE80211_CONF_CHANGE_CHANNEL is set even if only chan->hw_value has changed, e.g. by switching from A to TurboA on the same channel. AFAIK TurboX isn't currently supported by ath5k.

Regards,
Jörg.

Joerg Albert

author

2009-08-02 07:51 UTC #3 lore raw

In reply to Bob Copeland

Let ath5k_chan_set() always call ath5k_reset().
This fixes the bug that we don't
get any packets in monitor mode after:

ifconfig wlan0 down
iwconfig wlan0 mode monitor channel 1
ifconfig wlan0 up

but they arrive after

iwconfig wlan0 channel 2

Signed-off-by Joerg Albert <[email protected]>

---
 drivers/net/wireless/ath/ath5k/base.c |   35 ++++++++++++++------------------
 1 files changed, 15 insertions(+), 20 deletions(-)

drivers/net/wireless/ath/ath5k/base.c +15 ~~-20~~

index 3a1c156..5d50285 100644
@@ -1071,10 +1071,9 @@ ath5k_setup_bands(struct ieee80211_hw *hw)
 }
 
 /*
- * Set/change channels.  If the channel is really being changed,
- * it's done by reseting the chip.  To accomplish this we must
- * first cleanup any pending DMA, then restart stuff after a la
- * ath5k_init.
+ * Set/change channels. We always reset the chip.
+ * To accomplish this we must first cleanup any pending DMA,
+ * then restart stuff after a la  ath5k_init.
  *
  * Called with sc->lock.
  */

@@ -1084,19 +1083,13 @@ ath5k_chan_set(struct ath5k_softc *sc, struct ieee80211_channel *chan)
        ATH5K_DBG(sc, ATH5K_DEBUG_RESET, "(%u MHz) -> (%u MHz)\n",
                sc->curchan->center_freq, chan->center_freq);
 
-       if (chan->center_freq != sc->curchan->center_freq ||
-               chan->hw_value != sc->curchan->hw_value) {
-
-               /*
-                * To switch channels clear any pending DMA operations;
-                * wait long enough for the RX fifo to drain, reset the
-                * hardware at the new frequency, and then re-enable
-                * the relevant bits of the h/w.
-                */
-               return ath5k_reset(sc, chan);
-       }
-
-       return 0;
+       /*
+        * To switch channels clear any pending DMA operations;
+        * wait long enough for the RX fifo to drain, reset the
+        * hardware at the new frequency, and then re-enable
+        * the relevant bits of the h/w.
+        */
+       return ath5k_reset(sc, chan);
 }
 
 static void

@@ -2811,9 +2804,11 @@ ath5k_config(struct ieee80211_hw *hw, u32 changed)
 
        mutex_lock(&sc->lock);
 
-       ret = ath5k_chan_set(sc, conf->channel);
-       if (ret < 0)
-               goto unlock;
+       if (changed & IEEE80211_CONF_CHANGE_CHANNEL) {
+               ret = ath5k_chan_set(sc, conf->channel);
+               if (ret < 0)
+                       goto unlock;
+       }
 
        if ((changed & IEEE80211_CONF_CHANGE_POWER) &&
        (sc->power_level != conf->power_level)) {

Signature · 1 line

1.6.0.4

Bob Copeland

Acked-by

2009-08-04 15:13 UTC #4 lore raw

Quoted 3 lines · Joerg Albert

Let ath5k_chan_set() always call ath5k_reset().
This fixes the bug that we don't
get any packets in monitor mode after:

Thanks! Looks good.

Acked-by Bob Copeland <[email protected]>

Signature · 1 line

Bob Copeland %% www.bobcopeland.com

John W. Linville

2009-08-04 15:46 UTC #5 lore raw

In reply to Joerg Albert

Quoted 8 lines · Joerg Albert

Let ath5k_chan_set() always call ath5k_reset().
This fixes the bug that we don't
get any packets in monitor mode after:
ifconfig wlan0 down
iwconfig wlan0 mode monitor channel 1
ifconfig wlan0 up

Quoted 5 lines

but they arrive after
iwconfig wlan0 channel 2
Signed-off-by Joerg Albert <[email protected]>

patching file drivers/net/wireless/ath/ath5k/base.c
Hunk #1 FAILED at 1071.
Hunk #2 FAILED at 1083.
Hunk #3 FAILED at 2804.
3 out of 3 hunks FAILED -- saving rejects to file drivers/net/wireless/ath/ath5k/base.c.rej

Signature · 3 lines

John W. Linville		Someday the world will need a hero, and you
[email protected]			might be all we have.  Be ready.
			¡Viva Honduras Libre!

Joerg Albert

author

2009-08-04 23:46 UTC #6 lore raw

Quoted 13 lines · John W. Linville

Quoted 12 lines · Joerg Albert
Let ath5k_chan_set() always call ath5k_reset().
This fixes the bug that we don't
get any packets in monitor mode after:
ifconfig wlan0 down
iwconfig wlan0 mode monitor channel 1
ifconfig wlan0 up
but they arrive after
iwconfig wlan0 channel 2

Quoted 7 lines

Quoted 1 line
Signed-off-by Joerg Albert <[email protected]>
patching file drivers/net/wireless/ath/ath5k/base.c
Hunk #1 FAILED at 1071.
Hunk #2 FAILED at 1083.
Hunk #3 FAILED at 2804.
3 out of 3 hunks FAILED -- saving rejects to file drivers/net/wireless/ath/ath5k/base.c.rej

Sorry, seems like Thunderbird has converted the whitespaces. I'll repost the patch.

Joerg Albert

author

2009-08-04 23:52 UTC #7 lore raw

PATCH v3 1/1 ath5k: fix missing output in monitor mode after ifconfig up

In reply to John W. Linville

Let ath5k_chan_set() always call ath5k_reset().
This fixes the bug that we don't
get any packets in monitor mode after:

ifconfig wlan0 down
iwconfig wlan0 mode monitor channel 1
ifconfig wlan0 up

but they arrive after

iwconfig wlan0 channel 2

Signed-off-by Joerg Albert <[email protected]>

---

Same version as [PATCH v2] but with correct whitespaces (hopefully).

 drivers/net/wireless/ath/ath5k/base.c |   35 ++++++++++++++------------------
 1 files changed, 15 insertions(+), 20 deletions(-)

drivers/net/wireless/ath/ath5k/base.c +15 ~~-20~~

index 3a1c156..5d50285 100644
@@ -1071,10 +1071,9 @@ ath5k_setup_bands(struct ieee80211_hw *hw)
 }
 
 /*
- * Set/change channels.  If the channel is really being changed,
- * it's done by reseting the chip.  To accomplish this we must
- * first cleanup any pending DMA, then restart stuff after a la
- * ath5k_init.
+ * Set/change channels. We always reset the chip.
+ * To accomplish this we must first cleanup any pending DMA,
+ * then restart stuff after a la  ath5k_init.
  *
  * Called with sc->lock.
  */

@@ -1084,19 +1083,13 @@ ath5k_chan_set(struct ath5k_softc *sc, struct ieee80211_channel *chan)
 	ATH5K_DBG(sc, ATH5K_DEBUG_RESET, "(%u MHz) -> (%u MHz)\n",
 		sc->curchan->center_freq, chan->center_freq);
 
-	if (chan->center_freq != sc->curchan->center_freq ||
-		chan->hw_value != sc->curchan->hw_value) {
-
-		/*
-		 * To switch channels clear any pending DMA operations;
-		 * wait long enough for the RX fifo to drain, reset the
-		 * hardware at the new frequency, and then re-enable
-		 * the relevant bits of the h/w.
-		 */
-		return ath5k_reset(sc, chan);
-	}
-
-	return 0;
+	/*
+	 * To switch channels clear any pending DMA operations;
+	 * wait long enough for the RX fifo to drain, reset the
+	 * hardware at the new frequency, and then re-enable
+	 * the relevant bits of the h/w.
+	 */
+	return ath5k_reset(sc, chan);
 }
 
 static void

@@ -2811,9 +2804,11 @@ ath5k_config(struct ieee80211_hw *hw, u32 changed)
 
 	mutex_lock(&sc->lock);
 
-	ret = ath5k_chan_set(sc, conf->channel);
-	if (ret < 0)
-		goto unlock;
+	if (changed & IEEE80211_CONF_CHANGE_CHANNEL) {
+		ret = ath5k_chan_set(sc, conf->channel);
+		if (ret < 0)
+			goto unlock;
+	}
 
 	if ((changed & IEEE80211_CONF_CHANGE_POWER) &&
 	(sc->power_level != conf->power_level)) {

Signature · 1 line

1.6.0.4