Re: PARASITIC COMPUTING - WHAT'S NEXT

From: 'ken'@FTU
Date: 10/23/01 

Message-ID: <3BD4B4AA.2090604@yahoo.com>
Date: Mon, 22 Oct 2001 20:07:06 -0400
From: "'ken'@FTU" <franklin_tech_bulletins@yahoo.com>
To: Kurt Seifried <bugtraq@seifried.org>
Subject: Re: PARASITIC COMPUTING - WHAT'S NEXT

The process occurs at the TCP level, ultimately. When the server
validates the TCP checksum, it also validates whether or not the data in
that packet contains a correct solution to the np-complete
satisfiability problem. If the server replies, then that means it is a
valid solution. If the machine outputting the packets to be validated
either does not receive a reply (timeout) or receives a message that the
checksum is not correct that packet does not contain valid solution.
More specifically: Instead of taking your data and computing the
checksum, one computes what the checksum to the answer should be and
then inserts the data that needs to be tested. This basically makes TCP
a data validator.

The exploit occurs because I use your TCP/IP stack and CPU to validate
checksums. Although you give me permission to use your TCP/IP stack for
your service (HTTP or FTP) you do not give me permission to do my
np-complete calculations. So basically, I use your processer without
your permission.

If this is incorrect, please feel free to correct me.

'ken'

Kurt Seifried wrote:

> This raises some interesting question. If you connect to my server, can I
> simply send back a disclaimer of some sort and then ask you to process
> stuff? I.e. at what level? "By accessing this website you agree todownload a
> java application that will run Seti @home (to name one possible example)". A
> reply ICMP ping packet containing the text followed by packets that require
> the processing? What if you initiate a connection to me but because of
> something I did, like I visit your website which results in your webserver
> looking up in-addr-arpa info on my IP?
>
> I think it ultimately doesn't matter much, like many things people will
> simply do it, and damn the consequences. Thus the onus (unfortunately) falls
> onto the end user, much like anti-virus software, personal firewall,s
> keeping software up to date, etc to protect themselves (rmemeber napster? my
> ISP sent out email to the effect of "this is a warning, turn off napster if
> you are not using it, otherwise we will cheerfully send you a large bill
> when you blow your upload limit away. A lot of people were still surprised
> and unhappy.). Even if it is a crime that won't stop people from doing it,
> and unless large amounts of proovable damage are done you won't get law
> enforcement very interested anytime soon.
>
> Kurt Seifried, kurt@seifried.org
> A15B BEE5 B391 B9AD B0EF
> AEB0 AD63 0B4E AD56 E574
> http://www.seifried.org/security/
>
>
>
>
>

Relevant Pages

  • Re: xinetd and fortran
    ... I tried calling flushafter each write by the server - no change in ... Running tcpdump yields nothing I can understand. ... it fails I get a packet I don't understand repeated every now and then ... bad checksum is used to prevent the packet from influencing the ...
    (comp.lang.fortran)
  • Re: XL driver checksum producing corrupted but checksum-correct packets
    ... you might want to try hooking up the UDP checksum drop case to ... :I may have misread: my reading was that the if_xl card marks the packet as ... the checksum check on my server on reception is succeeding). ... is because the client is no longer sending a corrupt packet. ...
    (freebsd-hackers)
  • Re: Help calculate checksum (newbie)
    ... Here's the sample output from server: ... I cannot figure out where this checksum come from, can anyone post a sample in vb.net or C#? ... Doesn't the packet description contain information on how to calculate the checksum? ...
    (microsoft.public.dotnet.languages.vb)
  • Re: Help calculate checksum (newbie)
    ... Here's the sample output from server: ... I cannot figure out where this checksum come from, can anyone post a sample in vb.net or C#? ... Doesn't the packet description contain information on how to calculate the checksum? ...
    (microsoft.public.dotnet.languages.vb)
  • Re: Same SID Issue
    ... why don't you compare the current image checksum with the initial checksum of the same original image? ... Then if you save the checksum on the server assigning to to this particular device in the devic list you will always be able to ... Assuming, of course, you have EWF running and enabled on the client image. ... >>>> SID and hostname are different things and they may cause different ...
    (microsoft.public.windowsxp.embedded)