netdev - Re: [net-next PATCH 1/1 V4] qdisc: bulk dequeue support for qdiscs with TCQ_F

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: <1411754797.16953.134.camel@edumazet-glaptop2.roam.corp.google.com>
Date:	Fri, 26 Sep 2014 11:06:37 -0700
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Dave Taht <dave.taht@...il.com>
Cc:	Tom Herbert <therbert@...gle.com>,
	Jesper Dangaard Brouer <brouer@...hat.com>,
	Linux Netdev List <netdev@...r.kernel.org>,
	"David S. Miller" <davem@...emloft.net>,
	Alexander Duyck <alexander.h.duyck@...el.com>,
	Toke Høiland-Jørgensen <toke@...e.dk>,
	Florian Westphal <fw@...len.de>,
	Jamal Hadi Salim <jhs@...atatu.com>,
	John Fastabend <john.r.fastabend@...el.com>,
	Daniel Borkmann <dborkman@...hat.com>,
	Hannes Frederic Sowa <hannes@...essinduktion.org>
Subject: Re: [net-next PATCH 1/1 V4] qdisc: bulk dequeue support for qdiscs
 with TCQ_F_ONETXQUEUE

On Wed, 2014-09-24 at 19:58 -0700, Dave Taht wrote:

> I have long hoped that the actual BQL limit in play would feed into
> TCP small queues when there are a lot of flows to make each tcp
> "small" queue gradually smaller...

Yep, but what do you do if you have 64 TX queues, each one having
different BQL limits ? This looks like a NP problem when you also have
millions of tcp flows.

Basically BQL limit is more a sign of local host being able to drain TX
completions fast or not.

Here at Google, we have a very tiny queue of at most 2 packets per flow,
and srtt being in usec units gives us a very dynamic signal, based on
end to end measures.

I am experimenting moving TSQ processing from tasklet to workqueue,
because we need faster response to {soft}irq when we so often enter
TCP stack.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html