Subject: Re: [PATCH 1/6] mt76: use mac80211 txq scheduling
To:     =?UTF-8?Q?Toke_H=c3=b8iland-J=c3=b8rgensen?= <toke@redhat.com>,
        linux-wireless@vger.kernel.org
References: <20190316204242.73560-1-nbd@nbd.name> <87muluv4hd.fsf@toke.dk>
 <7b9a0a24-14b0-9b96-0010-ce9f96308ad0@nbd.name> <875zshviqg.fsf@toke.dk>
 <2cc1f1d4-f07e-faa0-ede7-95dd2917a64a@nbd.name> <87r2b5tb6a.fsf@toke.dk>
 <111434f9-f04c-2fe2-c2b4-de757efc884c@nbd.name> <87sgvjsud7.fsf@toke.dk>
From:   Felix Fietkau <nbd@nbd.name>
Openpgp: preference=signencrypt
Autocrypt: addr=nbd@nbd.name; prefer-encrypt=mutual; keydata=
 mQGiBEah5CcRBADIY7pu4LIv3jBlyQ/2u87iIZGe6f0f8pyB4UjzfJNXhJb8JylYYRzIOSxh
 ExKsdLCnJqsG1PY1mqTtoG8sONpwsHr2oJ4itjcGHfn5NJSUGTbtbbxLro13tHkGFCoCr4Z5
 Pv+XRgiANSpYlIigiMbOkide6wbggQK32tC20QxUIwCg4k6dtV/4kwEeiOUfErq00TVqIiEE
 AKcUi4taOuh/PQWx/Ujjl/P1LfJXqLKRPa8PwD4j2yjoc9l+7LptSxJThL9KSu6gtXQjcoR2
 vCK0OeYJhgO4kYMI78h1TSaxmtImEAnjFPYJYVsxrhay92jisYc7z5R/76AaELfF6RCjjGeP
 wdalulG+erWju710Bif7E1yjYVWeA/9Wd1lsOmx6uwwYgNqoFtcAunDaMKi9xVQW18FsUusM
 TdRvTZLBpoUAy+MajAL+R73TwLq3LnKpIcCwftyQXK5pEDKq57OhxJVv1Q8XkA9Dn1SBOjNB
 l25vJDFAT9ntp9THeDD2fv15yk4EKpWhu4H00/YX8KkhFsrtUs69+vZQwbQcRmVsaXggRmll
 dGthdSA8bmJkQG5iZC5uYW1lPohgBBMRAgAgBQJGoeQnAhsjBgsJCAcDAgQVAggDBBYCAwEC
 HgECF4AACgkQ130UHQKnbvXsvgCgjsAIIOsY7xZ8VcSm7NABpi91yTMAniMMmH7FRenEAYMa
 VrwYTIThkTlQuQINBEah5FQQCACMIep/hTzgPZ9HbCTKm9xN4bZX0JjrqjFem1Nxf3MBM5vN
 CYGBn8F4sGIzPmLhl4xFeq3k5irVg/YvxSDbQN6NJv8o+tP6zsMeWX2JjtV0P4aDIN1pK2/w
 VxcicArw0VYdv2ZCarccFBgH2a6GjswqlCqVM3gNIMI8ikzenKcso8YErGGiKYeMEZLwHaxE
 Y7mTPuOTrWL8uWWRL5mVjhZEVvDez6em/OYvzBwbkhImrryF29e3Po2cfY2n7EKjjr3/141K
 DHBBdgXlPNfDwROnA5ugjjEBjwkwBQqPpDA7AYPvpHh5vLbZnVGu5CwG7NAsrb2isRmjYoqk
 wu++3117AAMFB/9S0Sj7qFFQcD4laADVsabTpNNpaV4wAgVTRHKV/kC9luItzwDnUcsZUPdQ
 f3MueRJ3jIHU0UmRBG3uQftqbZJj3ikhnfvyLmkCNe+/hXhPu9sGvXyi2D4vszICvc1KL4RD
 aLSrOsROx22eZ26KqcW4ny7+va2FnvjsZgI8h4sDmaLzKczVRIiLITiMpLFEU/VoSv0m1F4B
 FtRgoiyjFzigWG0MsTdAN6FJzGh4mWWGIlE7o5JraNhnTd+yTUIPtw3ym6l8P+gbvfoZida0
 TspgwBWLnXQvP5EDvlZnNaKa/3oBes6z0QdaSOwZCRA3QSLHBwtgUsrT6RxRSweLrcabiEkE
 GBECAAkFAkah5FQCGwwACgkQ130UHQKnbvW2GgCfTKx80VvCR/PvsUlrvdOLsIgeRGAAn1ee
 RjMaxwtSdaCKMw3j33ZbsWS4
Message-ID: <d3bc4ad2-d4b7-d95d-dd49-105297653713@nbd.name>
Date:   Mon, 18 Mar 2019 23:37:09 +0100
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.14; rv:60.0)
 Gecko/20100101 Thunderbird/60.5.3
MIME-Version: 1.0
In-Reply-To: <87sgvjsud7.fsf@toke.dk>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

On 2019-03-18 23:14, Toke Høiland-Jørgensen wrote:
> Felix Fietkau <nbd@nbd.name> writes:
> 
>> On 2019-03-17 22:59, Toke Høiland-Jørgensen wrote:
>>> Felix Fietkau <nbd@nbd.name> writes:
>>>> I've looked at ath9k (the only user at the moment), and it seems to call
>>>> the function in that way already: at PS wake or tx status time if it has
>>>> frames in its internal retry queue.
>>>> While it does not match the current documented behavior for that
>>>> function, it nicely fits ath9k's currently unfulfilled expectations ;)
>>> 
>>> Heh, fair point :)
>> I noticed another issue: after the migration to the mac80211 txq
>> scheduling code, ath9k does not handle stations going to powersave
>> properly anymore. mac80211 keeps returning txqs for stations that have
>> gone to sleep and ath9k will send out frames for them.
> 
> Ah, right. Never did have a good grip on the powersave code...
> 
>> We could deal with this in the driver and simply not return queues for
>> stations in PS mode, or mac80211 could actively cancel them once a
>> station enters PS mode. What do you prefer?
> 
> I think the cleanest would be if mac80211 handled it and just
> un-scheduled stations when they go to sleep.
I agree. I'll send a patch tomorrow.
> BTW, I was just thinking the other day about why the retry queue is kept
> around when a station goes to sleep? Isn't the station usually sleeping
> pretty long (>100 ms), where it might make more sense to drop things
> rather than try again once i comes back?
It doesn't always sleep long. It might just be background scanning.
There's no way for the AP to know in advance.
In any case, the A-MPDU receiver reorder window still needs to be
maintained, so dropping frames means we'd have to send a BAR frame to
notify the receiver, which costs airtime as well.

- Felix