[PATCH v2] wifi: mac80211: export ieee80211_tpt_led_trig_tx/rx for driver

[PATCH v2] wifi: mac80211: export ieee80211_tpt_led_trig_tx/rx for driver

[Yi-Chia Hsieh]

Whenever the H/W path is enabled and traffic is in the binding state,
mac80211 is not aware of the traffic. Consequently, the LED does not
blink for that reason.

The ieee80211_tpt_led_trig_tx/rx functions are exported for the driver
so that we can report the tx and rx bytes from the driver when
the H/W path is being used.

Signed-off-by: Yi-Chia Hsieh <yi-chia.hsieh@mediatek.com>
---
v2: split series and use CONFIG_MAC80211_LEDS
---
 include/net/mac80211.h | 17 +++++++++++++++++
 net/mac80211/led.c     | 18 ++++++++++++++++++
 net/mac80211/led.h     | 18 ------------------
 net/mac80211/rx.c      |  2 +-
 net/mac80211/tx.c      |  4 ++--
 5 files changed, 38 insertions(+), 21 deletions(-)

diff --git a/include/net/mac80211.h b/include/net/mac80211.h
index 3a8a2d2c58c3..2106b6b29e2d 100644
--- a/include/net/mac80211.h
+++ b/include/net/mac80211.h
@@ -4729,6 +4729,8 @@ __ieee80211_create_tpt_led_trigger(struct ieee80211_hw *hw,
 				   unsigned int flags,
 				   const struct ieee80211_tpt_blink *blink_table,
 				   unsigned int blink_table_len);
+void __ieee80211_tpt_led_trig_tx(struct ieee80211_hw *hw, int bytes);
+void __ieee80211_tpt_led_trig_rx(struct ieee80211_hw *hw, int bytes);
 #endif
 /**
  * ieee80211_get_tx_led_name - get name of TX LED
@@ -4839,6 +4841,21 @@ ieee80211_create_tpt_led_trigger(struct ieee80211_hw *hw, unsigned int flags,
 #endif
 }
 
+static inline void
+ieee80211_tpt_led_trig_tx(struct ieee80211_hw *hw, int bytes)
+{
+#ifdef CONFIG_MAC80211_LEDS
+	__ieee80211_tpt_led_trig_tx(hw, bytes);
+#endif
+}
+
+static inline void
+ieee80211_tpt_led_trig_rx(struct ieee80211_hw *hw, int bytes)
+{
+#ifdef CONFIG_MAC80211_LEDS
+	__ieee80211_tpt_led_trig_rx(hw, bytes);
+#endif
+}
 /**
  * ieee80211_unregister_hw - Unregister a hardware device
  *
diff --git a/net/mac80211/led.c b/net/mac80211/led.c
index 2dc732147e85..af03a2ef5c6a 100644
--- a/net/mac80211/led.c
+++ b/net/mac80211/led.c
@@ -319,6 +319,24 @@ __ieee80211_create_tpt_led_trigger(struct ieee80211_hw *hw,
 }
 EXPORT_SYMBOL(__ieee80211_create_tpt_led_trigger);
 
+void __ieee80211_tpt_led_trig_tx(struct ieee80211_hw *hw, int bytes)
+{
+	struct ieee80211_local *local = hw_to_local(hw);
+
+	if (atomic_read(&local->tpt_led_active))
+		local->tpt_led_trigger->tx_bytes += bytes;
+}
+EXPORT_SYMBOL(__ieee80211_tpt_led_trig_tx);
+
+void __ieee80211_tpt_led_trig_rx(struct ieee80211_hw *hw, int bytes)
+{
+	struct ieee80211_local *local = hw_to_local(hw);
+
+	if (atomic_read(&local->tpt_led_active))
+		local->tpt_led_trigger->rx_bytes += bytes;
+}
+EXPORT_SYMBOL(__ieee80211_tpt_led_trig_rx);
+
 static void ieee80211_start_tpt_led_trig(struct ieee80211_local *local)
 {
 	struct tpt_led_trigger *tpt_trig = local->tpt_led_trigger;
diff --git a/net/mac80211/led.h b/net/mac80211/led.h
index d25f13346b82..98db4356d0de 100644
--- a/net/mac80211/led.h
+++ b/net/mac80211/led.h
@@ -66,21 +66,3 @@ static inline void ieee80211_mod_tpt_led_trig(struct ieee80211_local *local,
 {
 }
 #endif
-
-static inline void
-ieee80211_tpt_led_trig_tx(struct ieee80211_local *local, int bytes)
-{
-#ifdef CONFIG_MAC80211_LEDS
-	if (atomic_read(&local->tpt_led_active))
-		local->tpt_led_trigger->tx_bytes += bytes;
-#endif
-}
-
-static inline void
-ieee80211_tpt_led_trig_rx(struct ieee80211_local *local, int bytes)
-{
-#ifdef CONFIG_MAC80211_LEDS
-	if (atomic_read(&local->tpt_led_active))
-		local->tpt_led_trigger->rx_bytes += bytes;
-#endif
-}
diff --git a/net/mac80211/rx.c b/net/mac80211/rx.c
index 4f707d2a160f..5747d7dac4d7 100644
--- a/net/mac80211/rx.c
+++ b/net/mac80211/rx.c
@@ -5375,7 +5375,7 @@ void ieee80211_rx_list(struct ieee80211_hw *hw, struct ieee80211_sta *pubsta,
 	if (skb) {
 		if ((status->flag & RX_FLAG_8023) ||
 			ieee80211_is_data_present(hdr->frame_control))
-			ieee80211_tpt_led_trig_rx(local, skb->len);
+			ieee80211_tpt_led_trig_rx(&local->hw, skb->len);
 
 		if (status->flag & RX_FLAG_8023)
 			__ieee80211_rx_handle_8023(hw, pubsta, skb, list);
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index 7fe7280e8437..234de8d3b8bb 100644
--- a/net/mac80211/tx.c
+++ b/net/mac80211/tx.c
@@ -4320,7 +4320,7 @@ void __ieee80211_subif_start_xmit(struct sk_buff *skb,
 	len = 0;
  out:
 	if (len)
-		ieee80211_tpt_led_trig_tx(local, len);
+		ieee80211_tpt_led_trig_tx(&local->hw, len);
 	rcu_read_unlock();
 }
 
@@ -4646,7 +4646,7 @@ static void ieee80211_8023_xmit(struct ieee80211_sub_if_data *sdata,
 	sta->deflink.tx_stats.packets[queue] += skbs;
 	sta->deflink.tx_stats.bytes[queue] += len;
 
-	ieee80211_tpt_led_trig_tx(local, len);
+	ieee80211_tpt_led_trig_tx(&local->hw, len);
 
 	ieee80211_tx_8023(sdata, skb, sta, false);
 
-- 
2.39.0

Re: [PATCH v2] wifi: mac80211: export ieee80211_tpt_led_trig_tx/rx for driver

[Johannes Berg]

> +++ b/net/mac80211/led.c
> @@ -319,6 +319,24 @@ __ieee80211_create_tpt_led_trigger(struct ieee80211_hw *hw,
>  }
>  EXPORT_SYMBOL(__ieee80211_create_tpt_led_trigger);
>  
> +void __ieee80211_tpt_led_trig_tx(struct ieee80211_hw *hw, int bytes)
> +{
> +	struct ieee80211_local *local = hw_to_local(hw);
> +
> +	if (atomic_read(&local->tpt_led_active))
> +		local->tpt_led_trigger->tx_bytes += bytes;
> +}
> +EXPORT_SYMBOL(__ieee80211_tpt_led_trig_tx);
> +
> +void __ieee80211_tpt_led_trig_rx(struct ieee80211_hw *hw, int bytes)
> +{
> +	struct ieee80211_local *local = hw_to_local(hw);
> +
> +	if (atomic_read(&local->tpt_led_active))
> +		local->tpt_led_trigger->rx_bytes += bytes;
> +}
> +EXPORT_SYMBOL(__ieee80211_tpt_led_trig_rx);
> 

It feels a bit wasteful to export not one but even two functions for
this ...

The trigger only really cares about the sum of the two, so maybe we
shouldn't really care too much, even in mac80211? I think the only
reason we currently separate RX and TX is that they can run
concurrently, but the truth is that it already can race anyway, at least
if you have multiple interfaces ...

So maybe we should just switch to a single counter, and accept that we
may sometimes lose an update?

Also this seems wasteful for mac80211 internals, so maybe instead add
triple-underscore versions that are still inline and called from the
double-underscore versions as well as the existing mac80211 callers.

What do you think?

johannes