> -----Original Message-----
> From: Dmitry Antipov <dmantipov@xxxxxxxxx>
> Sent: Friday, September 22, 2023 9:54 PM
> To: Ping-Ke Shih <pkshih@xxxxxxxxxxx>
> Cc: linux-wireless@xxxxxxxxxxxxxxx
> Subject: On spatial multiplexing power save in rtlwifi (rtl8192de)
>
> Shouldn't 'rtl92de_update_hal_rate_table()' avoid hardcoded 'mimo_ps'
> but rather extract it from 'struct rtl_sta_info' like it's done in
> 'rtl92de_update_hal_rate_mask()'?
>
The initial commit 3affdf42373e ("rtlwifi: rtl8192de: Merge hardware routines")
is as it is, and I'm not clear the reason why hardcode 'mino_ps'. IIRC,
this chip has two PHY, so it could be configured as two interfaces, and each
one works as 1T1R mode. Maybe, this is the reason...
Crrently, rtlpriv->dm.useramask is always true for rtl8192de, so
rtl92de_update_hal_rate_table() should not be called:
if (rtlpriv->dm.useramask)
rtl92de_update_hal_rate_mask(hw, sta, rssi_level, update_bw);
else
rtl92de_update_hal_rate_table(hw, sta);
Since I don't have a hardware to do a real test, no idea to your question.
Ping-Ke
