netdev - Re: [PATCH net-next 6/6] sched: Eliminate use of flow_keys in sch

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: <1425326547.5130.129.camel@edumazet-glaptop2.roam.corp.google.com>
Date:	Mon, 02 Mar 2015 12:02:27 -0800
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Tom Herbert <therbert@...gle.com>
Cc:	netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next 6/6] sched: Eliminate use of flow_keys in
 sch_sfq

On Mon, 2015-03-02 at 11:51 -0800, Tom Herbert wrote:
> On Mon, Mar 2, 2015 at 11:43 AM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> > On Mon, 2015-03-02 at 11:34 -0800, Tom Herbert wrote:
> >
> >> Spraying would be quite easy by modulating txhash from TCP and using
> >> either  UDP encapsulation like GUE or IPv6 flow labels. Only problem
> >> is that this would constantly be changing rxhash at the receiver for
> >> the connection, although flow labels might 'just work' with IPv6
> >> Toeplitz in NICs.
> >
> > When I said 'regular IPv4/tcp traffic' this did not include any added
> > encapsulation.
> >
> > Most NIC are not able to properly offload TCP (TSO) with such
> > encapsulations.
> >
> IPv6 flow labels should work through TSO, they would just be
> replicated for each segment. You would need switches that can hash
> based on flow label though, and of course IPv6 deployed in the network
> :-).


Yeah, but I explicitly talked about native IPv4 + TCP ;)

not IPv6 + TCP, or IPv6 + IPV4 + TCP or some encapsulation...

Trend is to generalize random packet spraying, even within a flow.

https://www.cs.purdue.edu/homes/kompella/publications/infocom13.pdf

This forces us to build a resilient 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