On Tue, Jan 30, 2018 at 03:39:56PM +0300, Dan Carpenter wrote:
> Hello Felix Fietkau,
>
> The patch 23405236460b: "mt76: fix transmission of encrypted
> management frames" from Jan 18, 2018, leads to the following static
> checker warning:
>
> drivers/net/wireless/mediatek/mt76/mt76x2_tx.c:41 mt76x2_tx()
> warn: always true condition '(wcid->hw_key_idx != -1) => (0-255 != (-1))'
>
> drivers/net/wireless/mediatek/mt76/mt76x2_tx.c
> 26 void mt76x2_tx(struct ieee80211_hw *hw, struct ieee80211_tx_control *control,
> 27 struct sk_buff *skb)
> 28 {
> 29 struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
> 30 struct mt76x2_dev *dev = hw->priv;
> 31 struct ieee80211_vif *vif = info->control.vif;
> 32 struct mt76_wcid *wcid = &dev->global_wcid;
> 33
> 34 if (control->sta) {
> 35 struct mt76x2_sta *msta;
> 36
> 37 msta = (struct mt76x2_sta *) control->sta->drv_priv;
> 38 wcid = &msta->wcid;
> 39 }
> 40
> 41 if (vif || (!info->control.hw_key && wcid->hw_key_idx != -1)) {
> ^^^^^^^^^^^^^^^^^^^^^^
> We set ->hw_key_idx to -1 but it's a u8 so it gets truncated to 0xFF.
> This should probably be a define anyway.
>
> 42 struct mt76x2_vif *mvif;
> 43
> 44 mvif = (struct mt76x2_vif *) vif->drv_priv;
^^^^^^^^^^^^^
Oh, there is another static checker warning because of the "vif" check:
drivers/net/wireless/mediatek/mt76/mt76x2_tx.c:44 mt76x2_tx()
error: we previously assumed 'vif' could be null (see line 41)
Assume "vif" is NULL, and info->control.hw_key is zero and ->hw_key_idx
is not -1.
regards,
dan carpenter
