netdev - Re: [RFC][PATCH] QoS TBF and latency configuration misbehavior

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]

Date:	Tue, 31 Aug 2010 23:47:20 +0200
From:	Jarek Poplawski <jarkao2@...il.com>
To:	Dan Kruchinin <dkruchinin@....org>
Cc:	netdev@...r.kernel.org, Alexey Kuznetsov <kuznet@....inr.ac.ru>,
	Stephen Hemminger <shemminger@...l.org>
Subject: Re: [RFC][PATCH] QoS TBF and latency configuration misbehavior

On Wed, Sep 01, 2010 at 01:00:53AM +0400, Dan Kruchinin wrote:
> On Tue, Aug 31, 2010 at 11:57 PM, Jarek Poplawski <jarkao2@...il.com> wrote:
> > Dan Kruchinin wrote, On 08/31/2010 07:01 PM:
> >
> >> I'm sorry it seems my email client has broken patch formating.
> >> Here is properly formated one:
> >> diff --git a/tc/q_tbf.c b/tc/q_tbf.c
> >> index dc556fe..850e6db 100644
> >> --- a/tc/q_tbf.c
> >> +++ b/tc/q_tbf.c
> >> @@ -178,7 +178,7 @@ static int tbf_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
> >>       }
> >>
> >>       if (opt.limit == 0) {
> >> -             double lim = opt.rate.rate*(double)latency/TIME_UNITS_PER_SEC + buffer;
> >> +             double lim = opt.rate.rate*(double)latency/TIME_UNITS_PER_SEC;
> >
> > The way limit is calculated here from latency suggests some safety defaults
> > are taken wrt. the implementation, which could be omitted while setting the
> > limit directly. You try to change/fix this to adhere to the documentation,
> > but such a change would definitely break many user configs, so I doubt it's
> > the right solution here. Probably you should rather think about fixing the
> > manual.
> 
> Thank you for comments.
> The only thing that embarrasses me abut documentation fixing is that
> the latency loses all its sense.

Hmm... As a matter of fact, I'm not too picky about docs (and happy if
they don't skip some params at all), but your test with such a high
burst wrt. the rate didn't make much sense to me either. ;-)

> Documentation describes latency as something intuitively clear: "the
> maximum amount of time a packet can sit in the TBF"
> but tc implementation handles it something like: "an additional time
> packet can sit int the TBF after main waiting queue which size is
> equal to the burst size is completely full.". It doesn't seem to have
> any sense.
> Moreover, user can pass "limit" value to tc directly and if this limit
> value is less than burst size then latency will be improperly
> calculated(it'll have a negative value) which is not good as well.

Yes, many params could be misused/abused in tc without a warning, so
people have to test their configs first, and often learn this way how
it really works. So the docs are often secondary, which is not right
of course.

> By the way, the following descriptions of TBF algorithm are proposed
> the same the manual says:
> http://www.opalsoft.net/qos/DS-24.htm
> ww.docum.org/docum.org/docs/other/tbf02_kw.ps
> 
> If the documentation should be fixed I'm not sure that latency can be
> somehow logically described to have any sense at all.

I guess there could be added a warning there is such a difference.

Jarek P.
--
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