Received-SPF: pass (google.com: best guess record for domain of linux-wireless-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Toke =?utf-8?Q?H=C3=B8iland-J=C3=B8rgensen?= <toke@redhat.com>
To:     Yibo Zhao <yiboz@codeaurora.org>
Cc:     ath10k@lists.infradead.org, linux-wireless@vger.kernel.org,
        linux-wireless-owner@vger.kernel.org
Subject: Re: [PATCH 2/4] mac80211: defer txqs removal from rbtree
In-Reply-To: <1b3eab1f2481e0102b284f133605c6c4@codeaurora.org>
References: <1568639388-27291-1-git-send-email-yiboz@codeaurora.org> <1568639388-27291-2-git-send-email-yiboz@codeaurora.org> <87pnjyiq7o.fsf@toke.dk> <c3ee7ece0986f1c50513cd5fdd2ee03f@codeaurora.org> <87sgothmpy.fsf@toke.dk> <8cdece5c030fd95817fb099021c38613@codeaurora.org> <87tv98fu6l.fsf@toke.dk> <1b4ab006d9b5c88035845aaac193ef48@codeaurora.org> <8736gre3bm.fsf@toke.dk> <198124204167325252fcfcd65e3f2733@codeaurora.org> <87ftkp7uuz.fsf@toke.dk> <4574cce4079f8dab2b2bf223431a6eae@codeaurora.org> <877e617qg2.fsf@toke.dk> <910d9bb5f9016b29fb2aaeb0b89bac38@codeaurora.org> <874l157nrt.fsf@toke.dk> <2935b00bf3e29ad8b2738fe98dc24a76@codeaurora.org> <87lfuf5ly2.fsf@toke.dk> <1b3eab1f2481e0102b284f133605c6c4@codeaurora.org>
Date:   Tue, 24 Sep 2019 09:26:28 +0200
Message-ID: <87r24640kb.fsf@toke.dk>
MIME-Version: 1.0
Content-Type: text/plain; charset=WINDOWS-1252
Content-Transfer-Encoding: quoted-printable
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

>> Hmm, yeah, I guess we could end up with a loop like that as well.
>> Keeping the schedule_round would be a way to fix it, but I'm not sure=20
>> we
>> should just skip that station; maybe we should just end the round
>> instead?
> I am not sure. I believe, in some cases, the rest of the nodes which=20
> could be most of the nodes in the tree will not have the chance to be=20
> scheduled in this round.

My guess would be that it doesn't really matter, because in most cases
each schedule round will only actually end up queueing packets from one
or two stations; as the driver will pull multiple packets from that one
station which will often fill up the firmware queues (especially once we
start throttling that with the AQL stuff).

So I guess we can just skip TXQs that we've already seen this scheduling
round, and let the v_t compare determine transmit eligibility :)

-Toke