PAWS security vulnerability

Tim Traver tt-list at simplenet.com
Thu May 19 23:24:00 PDT 2005


Ted,

I don't know your experience lately with people on this or any other 
list, but that last personal attack was WAY out of line. I am not a 
Troll, nor have I ever been one. I use freeBSD extensively on hundreds 
of servers, but I am not a FreeBSD source contributor.

Yes, I was shown this "vulnerability" by our network security person, 
read it over, and thought that it might be a legitimate exploit. I even 
picked up on the fact that Microsoft had already patched it in the 
service pack 2, which may mean that it was under wraps for a while, and 
was suspicious. So, after doing a little research on the net myself and 
not finding much, I decided to post something to the list to see if 
anyone had heard anything about it, and if the FreeBSD commiters were 
working on a possible patch.

Maybe I wrote my post wrong, but it didn't deserve you biting my fucking 
head off.

Now, you'll probably start in on "well, if you run that many servers, 
then why don't you know what you're doing?". I do know what I'm doing. I 
would very well be able to apply your patch,and compile a new system. 
Problem is, I'm afraid I don't quite understand the vulnerability enough 
to properly test what it is supposed to fix...

I would first need a way to break it, and then after applying your 
patch, verify that I couldn't break it any longer. If I knew how to 
break it, then I would be a better programmer than you, which I am not, 
and have never claimed to be. From the description of the issue, it 
sounds like a single cleverly made TCP packet with a bogus timestamp on 
it could take down ALL of the TCP commections to that machine.

To quote the article :
"A large value is set by the attacker as the packet timestamp. When the 
target computer processes this packet, the internal timer is updated to 
the large attacker supplied value. This causes all other valid packets 
that are received subsequent to an attack to be dropped as they are 
deemed to be too old, or invalid."

That sounds like it is pretty serious to me. One packet takes down ALL 
TCP services to the machine. You make it sound like its no big deal...Is 
it valid ? I don't know. I never claimed to know. I wasn't crying wolf 
here, just asking...

So, my statement of  "I'm not sure I have the ability to test out your 
patch." should really have been, "I don't have the knowledge enough of 
the vulnerability to test whether or not your patch works."

And I would hardly consider "If it works, I would submit it to the 
security list" as some sort of command that I was supposed to follow. 
After reading that email, I thought that you were going to submit it to 
the security list. After all, its your fucking patch.

I am slowly working my way into the community, and would love to help 
with these kind of things. But, like many other busy sys admins, I don't 
have a whole lot of spare time to work on things like this. Yes, if it 
was a serious problem enough to where I had to have a patch right away, 
I might have to devote some work time and give it a try for the team. 
I'm not sure that I know how serious it is, as I've already stated that 
I don't fully understand the supposed "vulnerability".

I hardly made any kind of desparate demands for someone to quickly make 
me a patch. You might want to go re-read those posts...

I can understand why you may have suspected troll because of the vague 
questions, but man, you flew off the handle awefully quick. Maybe you 
just need a vacation.

You bashed OpenBSD for their knee jerk reactions, and I think you just 
made a big one...

Tim.




Ted Mittelstaedt wrote:

>Hi Tim,
>
>  If you don't have the ability to test out the patch then LEARN!
>
>  As the advisory said "no known exploits have been released"  I also
>noticed that the only 2 vendors listed as implementing a fix were
>Cisco and Microsoft. And Microsoft was NOT on the problem list for
>ANY of their patched OSs.  I would therefore assume that the release
>of this so-called vulnerability was carefully timed to take place
>AFTER Microsoft had got it's ass covered, to make them look good,
>and everyone else look bad.  I continue therefore to assume that this
>is a political security hole, not an actual security hole.
>
>  OpenBSD is well known for knee-jerk reactions to real and supposed
>security holes, so it's not surprising they released a patch right away
>- of course, little good that did them since this advisory trashed them
>anyway.  But knee jerk reactions don't always take all variables into
>account.
>
>  I rewrite their patch because it was simple and easy to apply to the
>FreeBSD source - but I did not write the networking code in FreeBSD and
>have no idea if it is correct, or if OpenBSD even wrote the fix properly,
>or if in fact this is a real vulnerability that anyone needs to be
>concerned about.  In theory, any flat-key lock can be picked in less
>than a minute (I've seen it done that fast, and done it myself somewhat
>more slowly) but that does not stop millions of them from being sold
>at Home Depot every year.  If people went to a different type of lock
>that was much harder to pick then the burglar might not break in
>by picking the lock - but instead by kicking in the door which has
>the side effect of destroying the door and frame, and there's a couple
>thousand bucks lost right there fixing that - and if all the burgler
>does is steal a $200 TV set, then your better off with the pickable lock.
>The point is that any change in the networking code
>may have side effects that are worse than the problem.
>
>  I posted the patch in order to head off a big long dumbass trashing
>discussion, because I suspected you were trolling - but I was willing
>to give you the benefit of the doubt.  If you were really
>concerned - such as if you worked for some company that had some
>stick-up-their-ass security officer that was bigger than his britches,
>and you had to have a fix RIGHT NOW - then this would have allowed you
>to apply the patch to shut up the bigger-than-britches security officer
>so you could continue about your business.  In the meantime then the
>networking and security group could have had discussion about the
>PROPER way to handle this.  Probably that's this patch, but maybe not.
>
>  Now I find what?  Well, it surely looks to me like I just spoiled
>your troll, so your going to pretend it was no big deal, make a lame-ass
>excuse about how you really didn't need the patch anyway and can't
>apply it because your incompetent, and fade into the woodwork.  I told
>you to post the patch and info to the appropriate FreeBSD security lists,
>and you aren't the least bit interested in doing what I told you.  Why -
>because you were only interested in this silly hypothetical PAWS exploit
>as long as nobody could say "FreeBSD has a fix, shut up and apply it",
>so you can go urinate on the parade here.  Now I just handed you a
>urinal, and your going to run away and pee on someone else.
>
>  I don't want to see a fucking thing more from you unless it's:
>
>"Guys, I DID WHAT I WAS TOLD TO DO and went to the FreeBSD security and
>networking
>mailing lists and posted what I was given and this is what they said"
>
>  If you aren't willing to lift a finger to do that, your a fucking
>troll.  Don't waste anyone else's time here.  Next time you ask for code,
>you better check out the going hourly rate for custom programming.
>
>Ted
>
>  
>
>>-----Original Message-----
>>From: owner-freebsd-questions at freebsd.org
>>[mailto:owner-freebsd-questions at freebsd.org]On Behalf Of Tim Traver
>>Sent: Thursday, May 19, 2005 1:27 PM
>>To: Ted Mittelstaedt
>>Cc: bsd
>>Subject: Re: PAWS security vulnerability
>>Importance: Low
>>
>>
>>Ted,
>>
>>thanks for taking a look at this. I'm not sure I have the ability to
>>test out your patch. Maybe someone else on this fine list can ?
>>
>>But this sounds like a pretty severe DOS issue that seems to be
>>relatively simple to implement.
>>
>>Do you know if the 5.x branch is affected by this as well ?
>>
>>Tim.
>>
>>
>>Ted Mittelstaedt wrote:
>>
>>    
>>
>>>Hi Tim,
>>>
>>> Here is a slight mod of the OpenBSD patch for OpenBSD 3.6
>>>      
>>>
>>that has been
>>    
>>
>>>rewritten for FreeBSD 4.11.  YMMV  If it works I would submit
>>>      
>>>
>>it to the
>>    
>>
>>>FreeBSD
>>>security list.  The only change I made is OpenBSD defines "tiflags"
>>>FreeBSD defines
>>>"thflags" I assume they are the same thing.  The file is in
>>>/usr/src/sys/netinet
>>>
>>>Turning off the timestamps would be a good way to make your network go
>>>slow.
>>>
>>>*** tcp_input.c.original        Thu May 19 11:52:30 2005
>>>--- tcp_input.c Thu May 19 12:00:14 2005
>>>***************
>>>*** 976,984 ****
>>>--- 976,992 ----
>>>                * record the timestamp.
>>>                * NOTE that the test is modified according
>>>      
>>>
>>to the latest
>>    
>>
>>>                * proposal of the tcplw at cray.com list (Braden
>>>1993/04/26).
>>>+                * NOTE2 additional check added as a result of PAWS
>>>vulnerability
>>>+                * documented in Cisco security notice
>>>cisco-sn-20050518-tcpts
>>>+                * from OpenBSD patch for OpenBSD 3.6 015_tcp.patch
>>>                */
>>>               if ((to.to_flags & TOF_TS) != 0 &&
>>>                   SEQ_LEQ(th->th_seq, tp->last_ack_sent)) {
>>>+                       if (SEQ_LEQ(tp->last_ack_sent,
>>>      
>>>
>>th->th_seq + tlen
>>    
>>
>>>+
>>>+                               ((thflags & (TH_SYN|TH_FIN)) != 0)))
>>>+                                 tp->ts_recent = to.to_tsval;
>>>+                       else
>>>+                               tp->ts_recent = 0;
>>>                       tp->ts_recent_age = ticks;
>>>                       tp->ts_recent = to.to_tsval;
>>>               }
>>>
>>>Ted
>>>
>>>
>>>
>>>      
>>>
>>>>-----Original Message-----
>>>>From: owner-freebsd-questions at freebsd.org
>>>>[mailto:owner-freebsd-questions at freebsd.org]On Behalf Of Tim Traver
>>>>Sent: Thursday, May 19, 2005 10:09 AM
>>>>To: bsd
>>>>Subject: PAWS security vulnerability
>>>>
>>>>
>>>>Hi all,
>>>>
>>>>ok, this article was just published about a PAWS TCP DOS
>>>>vulnerability,
>>>>and lists freeBSD 4.x as affected.
>>>>
>>>>http://www.securityfocus.com/bid/13676/info/
>>>>
>>>>Does anyone know how to turn the TCP timestamps off on FreeBSD 4.x ?
>>>>
>>>>and is 5.4 affected too ?
>>>>
>>>>Tim.
>>>>
>>>>_______________________________________________
>>>>freebsd-questions at freebsd.org mailing list
>>>>http://lists.freebsd.org/mailman/listinfo/freebsd-questions
>>>>To unsubscribe, send any mail to
>>>>"freebsd-questions-unsubscribe at freebsd.org"
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>
>>>      
>>>
>>_______________________________________________
>>freebsd-questions at freebsd.org mailing list
>>http://lists.freebsd.org/mailman/listinfo/freebsd-questions
>>To unsubscribe, send any mail to
>>"freebsd-questions-unsubscribe at freebsd.org"
>>
>>    
>>
>
>  
>



More information about the freebsd-questions mailing list