netdev - Re: [PATCH 5/5] mac80211: set NETIF_F_LLTX when using intermediate tx queues

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <875zrdt4qk.fsf@toke.dk>
Date:   Wed, 17 Apr 2019 09:28:51 +0100
From:   Toke Høiland-Jørgensen <toke@...hat.com>
To:     Herbert Xu <herbert@...dor.apana.org.au>
Cc:     Johannes Berg <johannes@...solutions.net>,
        Arend Van Spriel <arend.vanspriel@...adcom.com>,
        Felix Fietkau <nbd@....name>, linux-wireless@...r.kernel.org,
        Eric Dumazet <eric.dumazet@...il.com>, netdev@...r.kernel.org
Subject: Re: [PATCH 5/5] mac80211: set NETIF_F_LLTX when using intermediate tx queues

Herbert Xu <herbert@...dor.apana.org.au> writes:

> On Tue, Apr 16, 2019 at 11:02:50AM +0100, Toke Høiland-Jørgensen wrote:
>> 
>> As explained at great length here:
>> https://www.usenix.org/conference/atc17/technical-sessions/presentation/hoilan-jorgesen
>> (you already know that of course, Johannes)
>
> I can understand that wireless needs its own queueing scheme, but it
> still seems wrong to place that under net/mac80211 as opposed to
> having it as a first-class citizen under net/sched.

This is because we need to resolve the MAC-layer destination station (or
rather, TID) and tie the queueing to that, because of aggregation. We
also use the queueing structure for scheduling stations to achieve
airtime fairness. Both of these would be decidedly non-trivial to pull
up to the qdisc layer. Rather, having them in mac80211 means drivers
don't need to do their own ad-hoc queueing (which was the case before,
leading extra bufferbloat).

Most of the actual queueing structure code lives in
include/net/fq_impl.h, though, so it's not actually that
mac80211-specific. I've been thinking about porting the relevant qdiscs
to use the same code, but I'm not sure that it's worth the trouble.

-Toke