netdev - RE: rhashtable: Add cap on number of elements in hash table

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Tue, 19 May 2015 10:25:22 +0000
From:	David Laight <David.Laight@...LAB.COM>
To:	'Herbert Xu' <herbert@...dor.apana.org.au>,
	David Miller <davem@...emloft.net>
CC:	"johannes@...solutions.net" <johannes@...solutions.net>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"kaber@...sh.net" <kaber@...sh.net>,
	"tgraf@...g.ch" <tgraf@...g.ch>,
	"johannes.berg@...el.com" <johannes.berg@...el.com>
Subject: RE: rhashtable: Add cap on number of elements in hash table

From: Herbert Xu
> Sent: 18 May 2015 23:35
> On Mon, May 18, 2015 at 04:12:01PM -0400, David Miller wrote:
> >
> > Ok, this of course depends upon the distribution of the input data
> > and the strength/suitability of the hash function.
> >
> > I'm a little bit disappointed in what Thomas found.  I would expect
> > the distribution to be at least a little bit better.
> 
> Just to make it clear this is not a problem with uneven distribution.
> The same thing would happen with any hash function.  It's essentially
> a generalised birthday paradox.

Which means that with real data and a real hash function it is very
likely to be worse.
Did anyone look at what happens to the longest chains if you put
2 or 4 times as many items into the hash table?
Clearly there will be more short chains, but the longest might not
get proportionally longer (guessing at probability theory).

Personally I don't like the idea of rejecting inserts because chains
might get long, and certainly not because a malloc() fails in the
middle of the resize operation.

I'm sure there are some tables where slightly longer chains don't matter.

	David

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html