Re: [PATCH] wifi: rtl8xxxu: fixing transmisison failure for rtl8192eu

[Bitterblue Smith]

On 20/12/2022 07:44, Ping-Ke Shih wrote:
> 
> 
>> -----Original Message-----
>> From: Jun ASAKA <JunASAKA@xxxxxxxxxxxxx>
>> Sent: Saturday, December 17, 2022 11:07 AM
>> To: Jes.Sorensen@xxxxxxxxx
>> Cc: kvalo@xxxxxxxxxx; davem@xxxxxxxxxxxxx; edumazet@xxxxxxxxxx; kuba@xxxxxxxxxx; pabeni@xxxxxxxxxx;
>> linux-wireless@xxxxxxxxxxxxxxx; netdev@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Jun ASAKA
>> <JunASAKA@xxxxxxxxxxxxx>
>> Subject: [PATCH] wifi: rtl8xxxu: fixing transmisison failure for rtl8192eu
>>
>> Fixing transmission failure which results in
>> "authentication with ... timed out". This can be
>> fixed by disable the REG_TXPAUSE.
>>
>> Signed-off-by: Jun ASAKA <JunASAKA@xxxxxxxxxxxxx>
>> ---
>>  drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c | 5 +++++
>>  1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
>> b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
>> index a7d76693c02d..9d0ed6760cb6 100644
>> --- a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
>> +++ b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
>> @@ -1744,6 +1744,11 @@ static void rtl8192e_enable_rf(struct rtl8xxxu_priv *priv)
>>  	val8 = rtl8xxxu_read8(priv, REG_PAD_CTRL1);
>>  	val8 &= ~BIT(0);
>>  	rtl8xxxu_write8(priv, REG_PAD_CTRL1, val8);
>> +
>> +	/*
>> +	 * Fix transmission failure of rtl8192e.
>> +	 */
>> +	rtl8xxxu_write8(priv, REG_TXPAUSE, 0x00);
> 
> I trace when rtl8xxxu set REG_TXPAUSE=0xff that will stop TX.
> The occasions include RF calibration, LPS mode (called by power off), and
> going to stop. So, I think RF calibration does TX pause but not restore
> settings after calibration, and causes TX stuck. As the flow I traced,
> this patch looks reasonable. But, I wonder why other people don't meet
> this problem.
> 
Other people have this problem too:
https://bugzilla.kernel.org/show_bug.cgi?id=196769
https://bugzilla.kernel.org/show_bug.cgi?id=216746

The RF calibration does restore REG_TXPAUSE at the end. What happens is
when you plug in the device, something (mac80211? wpa_supplicant?) calls
rtl8xxxu_start(), then rtl8xxxu_stop(), then rtl8xxxu_start() again.
rtl8xxxu_stop() sets REG_TXPAUSE to 0xff and nothing sets it back to 0.