Skip to contentSkip to conversation

staging: rtl8188eu: rtw_mlme.c: fix sparse warning

PATCH In review Serguey Parkhomovsky 2 participants 2 messages Dec 29, 2014 Staging

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 staging: rtl8188eu: rtw_mlme.c: fix sparse warning

Fixes the following sparse warning for rtw_mlme.c:

drivers/staging/rtl8188eu/core/rtw_mlme.c:810:9: warning: context imbalance in 'rtw_free_assoc_resources' - different lock contexts for basic block

---
 drivers/staging/rtl8188eu/core/rtw_mlme.c | 37 ++++++++++++++++++++-----------
 1 file changed, 24 insertions(+), 13 deletions(-)
drivers/staging/rtl8188eu/core/rtw_mlme.c +24 -13
index d4632da..a47717a 100644
@@ -755,12 +755,30 @@ static void free_scanqueue(struct	mlme_priv *pmlmepriv)
 	spin_unlock_bh(&scan_queue->lock);
 }
 
+static void rtw_free_current_network(struct adapter *adapter)
+{
+	struct wlan_network *pwlan = NULL;
+	struct mlme_priv *pmlmepriv = &adapter->mlmepriv;
+	struct wlan_network *tgt_network = &pmlmepriv->cur_network;
+
+	pwlan = rtw_find_network(&pmlmepriv->scanned_queue,
+				 tgt_network->network.MacAddress);
+	if (pwlan)
+		pwlan->fixed = false;
+	else
+		RT_TRACE(_module_rtl871x_mlme_c_, _drv_err_,
+			("rtw_free_assoc_resources:pwlan==NULL\n\n"));
+
+	if (check_fwstate(pmlmepriv, WIFI_ADHOC_MASTER_STATE) &&
+			 (adapter->stapriv.asoc_sta_count == 1))
+		rtw_free_network_nolock(pmlmepriv, pwlan);
+}
+
 /*
 *rtw_free_assoc_resources: the caller has to lock pmlmepriv->lock
 */
 void rtw_free_assoc_resources(struct adapter *adapter, int lock_scanned_queue)
 {
-	struct wlan_network *pwlan = NULL;
 	struct	mlme_priv *pmlmepriv = &adapter->mlmepriv;
 	struct	sta_priv *pstapriv = &adapter->stapriv;
 	struct wlan_network *tgt_network = &pmlmepriv->cur_network;
@@ -793,20 +811,13 @@ void rtw_free_assoc_resources(struct adapter *adapter, int lock_scanned_queue)
 		rtw_init_bcmc_stainfo(adapter);
 	}
 
-	if (lock_scanned_queue)
+	if (lock_scanned_queue) {
 		spin_lock_bh(&(pmlmepriv->scanned_queue.lock));
-
-	pwlan = rtw_find_network(&pmlmepriv->scanned_queue, tgt_network->network.MacAddress);
-	if (pwlan)
-		pwlan->fixed = false;
-	else
-		RT_TRACE(_module_rtl871x_mlme_c_, _drv_err_, ("rtw_free_assoc_resources:pwlan==NULL\n\n"));
-
-	if ((check_fwstate(pmlmepriv, WIFI_ADHOC_MASTER_STATE) && (adapter->stapriv.asoc_sta_count == 1)))
-		rtw_free_network_nolock(pmlmepriv, pwlan);
-
-	if (lock_scanned_queue)
+		rtw_free_current_network(adapter);
 		spin_unlock_bh(&pmlmepriv->scanned_queue.lock);
+	} else
+		rtw_free_current_network(adapter);
+
 	pmlmepriv->key_mask = 0;
 }
 
Signature · 1 line
1.9.3
#1 lore raw
Quoted 8 lines · Serguey Parkhomovsky

Fixes the following sparse warning for rtw_mlme.c:

drivers/staging/rtl8188eu/core/rtw_mlme.c:810:9: warning: context imbalance in 'rtw_free_assoc_resources' - different lock contexts for basic block

---
  drivers/staging/rtl8188eu/core/rtw_mlme.c | 37 ++++++++++++++++++++-----------
  1 file changed, 24 insertions(+), 13 deletions(-)

Are these many changes needed just because Sparse is not smart enough to understand that the following does not really have a context imbalance:

	if (lock_scanned_queue)
		spin_lock_bh(&(pmlmepriv->scanned_queue.lock));
	.....
	.....

	if (lock_scanned_queue)

spin_unlock_bh(&pmlmepriv->scanned_queue.lock);

Nothing in the middle touches lock_scanned_queue, thus if the spin lock and unlock operations will be paired. That is all that is important. The fact that some tool objects to this construct means that the tool is broken, not the code.

In my mind, Sparse warnings/errors/checks are in place to alert you to a possible problem. If they point to white space issues, then you fix them. If they point to lines that are too long, then you see if you can shorten them without harming the readability. In any case, judgement should be used before blindly trying to silence the warning.

For those reasons, NACK.

Larry