Re: [Ext2-devel] Re: Shrinking ext3 directories

[Daniel Phillips <phillips@arcor.de>]

On Thursday 04 July 2002 16:15, jlnance@intrex.net wrote:
> On Thu, Jul 04, 2002 at 06:48:45AM +0200, Daniel Phillips wrote:
> > > behaviour under certain application workloads.  With the half-md4, at
> > > least we can expect decent worst-case behaviour unless we're under
> > > active attack (ie. only maliscious apps get hurt).
> > 
> > OK, anti-hash-attack is on the list of things to do, and it's fairly
> > clear how to go about it:
> 
> Is it really worth the trouble and complexity to do anti-hash-attack?
> What is the worst that could happen if someone managed to create a bunch
> of files that hashed to the same value?

Just a slowdown, but in some cases it could be a quadratic slowdown
that could conceivably be turned into a denial of service.  As risks
go, it's a minor one, but there's a straightforward solution with
insignificant cost in either efficiency or code size, so why not do
it.  The overhead is just a data move from the superblock per name 
hash.

-- 
Daniel