[lvc-project] [PATCH rtw-next v2 7/7] wifi: rtw89: process TX wait skbs for USB via C2H handler

[Ping-Ke Shih]

Fedor Pchelkin <pchelkin at ispras.ru> wrote:
> On Tue, 07. Oct 08:07, Ping-Ke Shih wrote:
> > Fedor Pchelkin <pchelkin at ispras.ru> wrote:
> > > @@ -1173,7 +1173,8 @@ int rtw89_core_tx_kick_off_and_wait(struct rtw89_dev *rtwdev, struct sk_buff *sk
> > >
> > >         if (time_left == 0) {
> > >                 ret = -ETIMEDOUT;
> > > -               list_add_tail(&wait->list, &rtwdev->tx_waits);
> > > +               if (!rtwdev->hci.tx_rpt_enable)
> > > +                       list_add_tail(&wait->list, &rtwdev->tx_waits);
> >
> > Oh. You avoid using rtwdev->tx_waits for USB. But I'd like to have the same
> > behavior as PCIE.
> 
> I may be confused but doesn't it conflict with the comment [1] you've
> posted to the previous version?  I've treated that as we should use
> rtwdev->tx_rpt_queue for both TX wait and IEEE80211_TX_CTL_REQ_TX_STATUS
> frames...

Yes. I got this thought after reviewing whole v2.

> 
> I'm all for following the PCIe-style as possible, too, but then initial
> comment [1] becomes irrelevant, right?
> 
> [1]: https://lore.kernel.org/linux-wireless/c2c40bed311c4f05948cf2541c64ea30@realtek.com/

At the [1], I wanted to iterate skb's in rtwdev->tx_rpt_queue, and then call
rtw89_core_tx_wait_complete() if 'wait' existing in RTW89_TX_SKB_CB().

I think the idea is similar, but I might not have clear picture at v1 review.
If you feel they are conflict, just check and discuss with new comments.