netdev - Re: [PATCH iproute2-next] Add tc-BPF example for a TCP pure ack recognizer

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: <d1c86b02-04e8-f02a-2fe6-ac69517e1cf7@gmail.com>
Date:   Wed, 2 May 2018 09:22:24 -0600
From:   David Ahern <dsahern@...il.com>
To:     Dave Taht <dave.taht@...il.com>
Cc:     Linux Kernel Network Developers <netdev@...r.kernel.org>
Subject: Re: [PATCH iproute2-next] Add tc-BPF example for a TCP pure ack
 recognizer

On 5/1/18 10:51 PM, Dave Taht wrote:
> On Tue, May 1, 2018 at 9:12 PM, David Ahern <dsahern@...il.com> wrote:
>> On 5/1/18 12:32 PM, Dave Taht wrote:
>>> ack_recognize can shift pure tcp acks into another flowid.
>>> ---
>>>  examples/bpf/ack_recognize.c | 98 ++++++++++++++++++++++++++++++++++++++++++++
>>>  1 file changed, 98 insertions(+)
>>>  create mode 100644 examples/bpf/ack_recognize.c
>>>
>>> diff --git a/examples/bpf/ack_recognize.c b/examples/bpf/ack_recognize.c
>>> new file mode 100644
>>> index 0000000..5da620c
>>> --- /dev/null
>>> +++ b/examples/bpf/ack_recognize.c
>>> @@ -0,0 +1,98 @@
>>> +/* SPDX-License-Identifier: GPL-2.0 */
>>> +/*
>>> + * Copyright 2017 Google Inc.
>>
>> 2017?? it's 2018. Did you write this last year and just now posting?
> 
> November, 2017? Shortly prior to you taking iproute2-next off of steven's hands.

ok, just checking

> 
> It was the first stage of a proof of concept showing (eventually) that
> a simple ack decimator/filter (like "drop every other ack", or simple
> "drop head" in the supplied example) had a ton of problems, and that
> to filter out acks correctly to any extent, it needed to peer back
> into the queue. (see the sch_cake ack-filter controversy on-going on
> this list)
> 
> While it's better than what is in wondershaper, I wouldn't recommend
> anyone use it. It was, however, useful in acquiring a gut feel for
> what several other broken ack filters might be doing in the field. I
> submitted it as a possibly useful example for showing off tc-bpf and
> to add fuel to the ack-filter fire under cake.
> 
> I can clean it up, SPF it, etc, if you want it. Otherwise, sorry for the noise.

I'm fine with the example, just prefer magic numbers to be clarified.