phy: use initial gain for adjustment instead of the dynamic one

Signed-off-by: Felix Fietkau <[email protected]>
This commit is contained in:
Felix Fietkau
2015-09-26 02:59:09 +02:00
parent 9ec1f17795
commit 9e972d5be6
+1 -2
View File
@@ -445,7 +445,7 @@ mt76_phy_update_channel_gain(struct mt76_dev *dev)
int rssi0 = (s8) MT76_GET(MT_BBP_AGC20_RSSI0, val);
int rssi1 = (s8) MT76_GET(MT_BBP_AGC20_RSSI1, val);
bool low_gain;
u8 gain[2], gain_delta;
u8 *gain = dev->cal.agc_gain_init, gain_delta;
dev->cal.avg_rssi[0] = (dev->cal.avg_rssi[0] * 15) / 16 + (rssi0 << 8);
dev->cal.avg_rssi[1] = (dev->cal.avg_rssi[1] * 15) / 16 + (rssi1 << 8);
@@ -462,7 +462,6 @@ mt76_phy_update_channel_gain(struct mt76_dev *dev)
else
val = 0x1836 << 16;
mt76_get_agc_gain(dev, gain);
val |= 0xf8;
if (dev->chandef.width == NL80211_CHAN_WIDTH_80)