| Message ID | 1404145105-15954-1-git-send-email-andrea.merello@gmail.com (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
diff --git a/drivers/net/wireless/rtl818x/rtl8180/dev.c b/drivers/net/wireless/rtl818x/rtl8180/dev.c index e2dcede..73baf26 100644 --- a/drivers/net/wireless/rtl818x/rtl8180/dev.c +++ b/drivers/net/wireless/rtl818x/rtl8180/dev.c @@ -879,6 +879,8 @@ static int rtl8180_init_hw(struct ieee80211_hw *dev) reg = rtl818x_ioread8(priv, &priv->map->CONFIG3); rtl818x_iowrite8(priv, &priv->map->CONFIG3, reg | (1 << 2)); rtl818x_iowrite8(priv, &priv->map->EEPROM_CMD, RTL818X_EEPROM_CMD_NORMAL); + /* fix eccessive IFS after CTS-to-self */ + rtl818x_iowrite8(priv, REG_ADDR1(0x1ff), 0x35); } if (priv->chip_family == RTL818X_CHIP_FAMILY_RTL8187SE) {
Measuring time between _end_ of CTS-to-self and _end_ of datapacket (with a prism54 board and mac80211 hacked to let the MAC timestamp stay untouched in the radiotap header) resulted in about 300uS, while the datapacket itself should be by far shorter (less than 100uS) and IFS should be SIFS (10uS). This measure was confirmed whith a scope: about 250uS IFS has been seen between the two packets. This situation causes the CTS-to-self protection mechanism to work incorrectly due to the NAV expiring during, or even before beginning, the packet transmission, and it also causes the performances to be anyway reduced due to time waste. This problem has been seen at every packet TXed with CTS-to-self enabled on rtl8185 board. rtl8187se seems not affected (and rtl8180, being a 802.11b card, does not have CTS-to-self mechaninsm). This patch fixes this by adding a magic register write, making the board wait for correct SIFS after CTS-to-self packet. Signed-off-by: Andrea Merello <andrea.merello@gmail.com> --- drivers/net/wireless/rtl818x/rtl8180/dev.c | 2 ++ 1 file changed, 2 insertions(+)