RE: [PATCH v7 08/24] rtw89: implement mac80211 ops

[Pkshih <pkshih@realtek.com>]

> -----Original Message-----
> From: Toke Høiland-Jørgensen <toke@redhat.com>
> Sent: Saturday, October 9, 2021 1:45 AM
> To: Pkshih <pkshih@realtek.com>; kvalo@codeaurora.org
> Cc: linux-wireless@vger.kernel.org
> Subject: Re: [PATCH v7 08/24] rtw89: implement mac80211 ops
> 
> Ping-Ke Shih <pkshih@realtek.com> writes:
> 
> > To yield better TX performance, wake TX queue is implemented.
> 
> I went looking for this, and was happy to find it! :)
> 
> Do you also support airtime reporting and/or AQL? And if not, any plans
> to do so?

Not yet. But, I can estimate the cost to make a plan.

Could I know the requirements of hardware? For now, this driver
only reports simple information like TX/RX rate, RSSI and etc. Is it
enough?

It seems like AQL configures weights of stations. Could I know how
to verify this function?


> 
> One small nit below:
> 
> > +static void rtw89_ops_wake_tx_queue(struct ieee80211_hw *hw,
> > +				    struct ieee80211_txq *txq)
> > +{
> > +	struct rtw89_dev *rtwdev = hw->priv;
> > +
> > +	ieee80211_schedule_txq(hw, txq);
> > +	queue_work(rtwdev->txq_wq, &rtwdev->txq_work);
> 
> Why the call to ieee80211_schedule_txq()? mac80211 calls that before
> calling wake_tx_queue() (through the schedule_and_wake_txq() helper), so
> it seems a bit superfluous here?
> 

You are right. I will remove it after v7 get merged.

--
Ping-Ke