On Mon, 21 Nov 2011 10:03:21 PST, Dan Kaminsky said: > 15.3M lines of code != 15.3M lines of code in use on any one system != > 15.3M lines of code that can ever involve a security boundary. Yes, but the vast majority of it is on use on *some* system (heck, there's still code in there to support the 3 or so NCR Voyager systems still in existence). And the biggest hassle with security boundaries is that often the place the failure actually occurs is nowhere near where the boundary should have been enforced. So just because there are only (for example) 500K lines of code involved with the security boundary doesn't mean you can simply ignore the other 14.8M lines of code, as you may have to do some hunting to find the 500K you're interested in (in particular, a lot of ioctl parameter checks are pushed down into drivers because the high-level VFS code has no *clue* what the parameters mean or how to validate them). It's kind of saying "We're doing an easter egg hunt, and since we only care about the 250 1-foot square areas that actually contain eggs, we're going to gloss over the fact that the areas are hidded all over 5 acres of dense woods and underbrush".
Attachment:
pgphNd1x7qvm9.pgp
Description: PGP signature
_______________________________________________ Full-Disclosure - We believe in it. Charter: http://lists.grok.org.uk/full-disclosure-charter.html Hosted and sponsored by Secunia - http://secunia.com/