netdev - Re: [Make-wifi-fast] [RFCv2 1/3] mac80211: implement fq

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: <CA+BoTQ=PWCK6cq83sthqRRt0xYWBOORbb=0+QHM6XVaWCWVhDw@mail.gmail.com>
Date:	Tue, 22 Mar 2016 07:51:06 +0100
From:	Michal Kazior <michal.kazior@...to.com>
To:	David Lang <david@...g.hm>
Cc:	linux-wireless <linux-wireless@...r.kernel.org>,
	Emmanuel Grumbach <emmanuel.grumbach@...el.com>,
	Network Development <netdev@...r.kernel.org>,
	"ath10k@...ts.infradead.org" <ath10k@...ts.infradead.org>,
	codel@...ts.bufferbloat.net, make-wifi-fast@...ts.bufferbloat.net,
	Johannes Berg <johannes@...solutions.net>,
	Tim Shepard <shep@...m.mit.edu>
Subject: Re: [Make-wifi-fast] [RFCv2 1/3] mac80211: implement fq_codel for
 software queuing

On 22 March 2016 at 02:35, David Lang <david@...g.hm> wrote:
> On Wed, 16 Mar 2016, Michal Kazior wrote:
>
>> Since 11n aggregation become important to get the
>> best out of txops. However aggregation inherently
>> requires buffering and queuing. Once variable
>> medium conditions to different associated stations
>> is considered it became apparent that bufferbloat
>> can't be simply fought with qdiscs for wireless
>> drivers.
>
> If the network is quiet enough, don't do any buffering, but in almost all
> situations you are going to need to buffer starting no later than the second
> packet you try to send.
>
> Don't try to make queueing occur, just deal with the queues that form
> naturally because you can't transmit data any faster (and work to keep them
> under some semblence of control)
[...]

This is what already happens. Queues typically start to build up when
hardware tx fifos/queues become busy so by the time they become
available you might have a bunch of frames you can aggregate.

The patch is more about getting rid of qdiscs because it's inherently
hard to teach them how 802.11 aggregation works (per-station per-tid)
and the ever-changing nature of per-station tx conditions.

I'll update the commit log to better reflect what is being done.


Michał