RE: [fw-wiz] CERT vulnerability note VU# 539363

From: Stephen Gill (
Date: 10/16/02

From: "Stephen Gill" <>
To: "'R. DuFresne'" <>
Date: Wed Oct 16 12:03:01 2002

Sadly the information provided from NS is about 1% of what should have
been stated. As of May, certain Netscreen versions were vulnerable
though fortunately bug fixes and other features were put in place to
improve mitigation methods. No mention is made here on what features
were implemented or what code/platforms were affected. This is

Customers should be urged to upgrade _from_ vulnerable versions.

-- steve

-----Original Message-----
From: R. DuFresne []
Sent: Wednesday, October 16, 2002 10:34 AM
To: Stephen Gill
Cc: 'Mikael Olsson';
Subject: RE: [fw-wiz] CERT vulnerability note VU# 539363

I've looed a bit more at the CERT site. And find a few things

First note, there are so many 'unknowns' in the listing, which,
is far from complete <even fw-1 seems to be missing, and they skip over
some of the linux *bsd distributions>. Then I looked specifically at
vendor notes for a number of vendors there, Netscreen is listed as
vulnerable, but, their notes state specifically:

Vendor Statement

   NetScreen has studied the issues raised in this vulnerability alert.
   With proper configuration of their firewalls, customers can virtually
   eliminate the impact of any of the proposed DoS attacks.
   customers are strongly advised to utilize NetScreen's SYN Flood
   protection and UDP rate limiter. On some platforms, default timeout
   parameters might need to be changed in order to have a more
   response to these attacks. Please refer to NetScreen's support site
   for more information.


Of course on another note of interest, most the vendors seem to be
responding only to the tcp/upd 'syn' flooding issue raised, and are
ignoring Stephen's crc bypass flood. I only noted IPFilter directly
addressing this, though I admit I did not check each vendor response
<though, then again I maywell have since so many are listed as

Cray is listed here, seperate from SGI, who, according to my last
information owned cray, so this seems strange <perhaps cray's been
outside SGI in the past few years and my information is old (?)>:

<cray quote>

Vendor Statement

   Cray, Inc. is not vulnerable as we provide no software that performs
   this type of function.


<SGI quote>

Vendor Statement

   No statement is currently available from the vendor regarding this


I see cray always listed in these certs as not having an issue cause
don't provide anything in the category to 'handle' these things. Course
their TCP/ip stack is still sitting there, as much as SGI's stack has to
be, at least the SGI side has decided to not prvide vendor input at this

Sadly, it seems one can't fully trust the vendors statemnts in some
and sadder yet, folks would have to actually read the statemtns to see
they reflect what CERT claims about the vendors status <if we acept
netscreens claims on the single side of the 'two' attack methods


Ron DuFresne

On Wed, 16 Oct 2002, Stephen Gill wrote:

> Sadly and amazingly the TCP/SYN flood isn't handled well. Or at least
> wasn't when the paper was written. Hopefully things have improved
> somewhat in some vendors.
> -- steve
> -----Original Message-----
> From: R. DuFresne []
> Sent: Wednesday, October 16, 2002 8:52 AM
> To: Stephen Gill
> Cc: 'Mikael Olsson';
> Subject: RE: [fw-wiz] CERT vulnerability note VU# 539363
> Of course the attacks mentioned in this CERT advisory are not really
> traffic limit overloads, but, resource exhaustion techniques. The
> tcp/syn
> flood method of exhhaustion should be well handled by most firewalls
> these
> days. But, the newer CRC related method is something even more
> interesting. And seems to support the claims of Marcus and Mikeal and
> Paul and others about the real depth and breath of the packet logic in
> filtering and stateful as well as proxied gateways. From how I read
> CERT, it seems you can have speed and performance, or you can have a
> full
> examination of the packets and all their settings, but, perhaps not
> at the sametime, so vendors shoot for the former.
> Thanks,
> Ron DuFresne
> On Wed, 16 Oct 2002, Stephen Gill wrote:
> > In my opinion if a stateful firewall claims it can filter at rate X
> > (64byte packets, etc...), it should be able to filter at that rate
> under
> > all conditions. Clearly a 100MB firewall that can be overloaded
> > 1MB of traffic is not good. I'd argue that if a 100MB firewall can
> > overloaded with 34MB of traffic, it's also not a good thing. But
> > again, even 100MB of filtering won't save you in a 100MB DoS which
> > not all that uncommon.
> >
> > I'd like to learn some of the other methods being used for
> > amongst vendors.
> >
> > -- steve
> >
> > -----Original Message-----
> > From: Mikael Olsson []
> > Sent: Wednesday, October 16, 2002 7:44 AM
> > To: Stephen Gill
> > Cc:
> > Subject: Re: [fw-wiz] CERT vulnerability note VU# 539363
> >
> >
> > Stephen Gill wrote:
> > >
> > > Thought I'd pass this along.
> > >
> > >
> >
> > Although this is something that people need to keep in mind when
> > picking / designing a firewall, I'd argue that anything north of
> > a stateless packet filter is going to be vulnerable to these sort
> > of attacks.
> >
> > If you keep state, you will be vulnerable to state table overflows.
> > Period. The only real question is: how much work does the attacker
> > need to put in before it becomes painful for the networks that the
> > firewall is protecting? Is being able to resist a 1 Mbps stream
> > (~4500 pps) "Not vulnerable"? Is being able resist a 34 Mbps stream
> > (~150 kpps) "Not vulnerable"? Or should every single firewall
> > vendor report in and say "Vulnerable", and describe what the limit
> >
> >
> > And, yes, ALG-only firewalls can also be overloaded. It's just a
> > different type of 'state'.
> >
> >

        admin & senior security consultant:
"Cutting the space budget really restores my faith in humanity.  It
eliminates dreams, goals, and ideals and lets us get straight to the
business of hate, debauchery, and self-annihilation."
                -- Johnny Hart
testing, only testing, and damn good at it too!

Relevant Pages