BugTraq
PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 16 2002 07:55PM
NGSSoftware Insight Security Research (nisr nextgenss com) (2 replies)
Re: PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 26 2002 10:07PM
Andreas Tscharner (starfire dplanet ch)
RE: PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 16 2002 08:39PM
Stefan Esser (s esser e-matters de) (2 replies)
Re: PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 17 2002 04:56AM
Valdis Kletnieks vt edu (1 replies)
Re: PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 17 2002 06:37AM
Stefan Esser (s esser e-matters de) (2 replies)
Re: PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 18 2002 02:16PM
Andreas Borchert (bugtraq andreas-borchert de)
Re: PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 17 2002 05:44PM
der Mouse (mouse Rodents Montreal QC CA)
Re: PFinger 0.7.8 format string vulnerability (#NISR16122002B) Dec 16 2002 09:49PM
der Mouse (mouse Rodents Montreal QC CA)
>> Due to the way requests are logged the only way to exploit this
>> vulnerability is through setting the DNS name of the fingering host
>> to the attacker supplied format string.

> I really wonder how you want to exploit this... Last time I checked
> all tested resolvers (Linux/BSD/Solaris) did not allow % within
> domain names and so your format string vulnerability is not
> exploitable at all...

If your resolver does not allow "funny characters" in domains, it is
broken. If nothing else, that sort of crippling makes it approximately
impossible to investigate abuse that involves using such domain names.
(At least one spammer outfit is known to use domain names containing
control characters and I think at least one other unusual character,
prseumably in an attempt to make it harder to investigate their spam.)

0x00 octets in domain labels won't work well with APIs that use C
strings, but the resolver shouldn't misbehave when encountering them
internally - and breaking on any of the other 255 octets is a Very Bad
Idea. (At least in the resolver. Other software and protocols may of
course impose their own restrictions, of varying degrees of sanity.
But the resolver infrastructure has to support all uses of the DNS,
including "unusual" uses.)

/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML mouse (at) rodents.montreal.qc (dot) ca [email concealed]
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

[ reply ]


 

Privacy Statement
Copyright 2010, SecurityFocus