Re: Back to Merkle Hash Trees...

----- Original Message ----- 
From: "Olaf van der Spek" <[email protected]>
Subject: Re: [BitTorrent] Back to Merkle Hash Trees...

> Why not?
> If I have file A, calculate the merkle tree and root hash rh(A), give
> you rh(A), how can you give me file B with rh(B) = rh(A)?

Depends on the hash used. If you use SHA-0 yes, if you use MD5 yes, SHA-1 is 
the next in that series and is getting close, SHA-256/384/512 already have 
weaknesses in this area shown against them.

Actually in an effort to get this as good as possible I have to admit that 
my proposal is not optimally secure either. While it does not suffer from 
the preimage order attack of the original it does suffer from a first 
preimage attack, to correct this is fairly easy change my proposal from:

Segment #
Size
Hash
Parent by id
nodeID
number of direct children

to:
{{Segment #, size, parentId, nodeId, number of direct children}, 
hash({Segment #, size, parentId, nodeId, number of direct children}| 
children)}

This pushes the problem to a complete preimage compromise, also known as a 
complete hash break (specifically SHA-0 would actually still be viable, MD5 
would not, SHA-1 is in great shape).



I think it would be better if we move the security discussion to a thread 
specifically on the security discussion, that should make filtering easier 
for those that don't want to be bothered.
                    Joe 



 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/BitTorrent/

<*> To unsubscribe from this group, send an email to:
    [email protected]

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/