Interesting email Andrew (as always!). Threat modelling is part of the security 
review process so we have formal threat models for parts of the system, but I 
don’t think we have a holistic threat model documented to the level to which 
you are describing. Certainly capturing security decisions and philosophy is 
important. I’m currently working on threat model as part of a roadmap for 
security future features so I can try to incorporate some of the ideas below 
into that. 


On 10 Sep 2014, at 5:50 pm, Andrew Sutherland <[email protected]> 
wrote:

> :pauljt's question about certified app debugging made me wonder if we have 
> any formal-ish descriptions of the threat/attack models for Firefox OS.  I 
> particularly am interested because it seems like discussions about 
> complicated issues like this frequently end up going in circles because 
> people are trying to cobble mental models together from email responses.  
> (ex: I think the "Add-on File Registration PRD" discussion at 
> https://groups.google.com/d/msg/mozilla.dev.planning/zLwLNyWm7es/n85wjv5DJ1QJ 
> would have gone much better with a representation of the attack-defense tree 
> and indication of which choices were discarded because the attacker already 
> has a valid counter-attack.)
> 
> And in some cases I think visually expressing the key assumptions that 
> underlie a lot of our protections can be helpful.  For example, in the recent 
> dev-gaia thread on "App badges via notifications", it seemed to be proposed 
> to let apps dynamically change their icon 
> (https://groups.google.com/d/msg/mozilla.dev.gaia/sHFTy6m8va8/g7-V19YusgEJ). 
> But it's my understanding that much of our defense against rogue apps is that 
> they are unable to change their name or icon.  Being able to easily see at a 
> glance that defeating a morphing impersonation attack is or is not critical 
> to us would be handy. (And a hyperlink to a wiki from the graph node that 
> discusses the trade-offs would be super handy!)
> 
> I've just tried to bring myself up to speed on this a little bit, and it 
> seems like the http://satoss.uni.lu/projects/atrees/index.php project's 
> research and papers on Attack Trees are particularly interesting[1].
> 
> Most important, this library of trees has graphical examples in .pdf form: 
> http://satoss.uni.lu/projects/atrees/library.php
> I suggest starting with a look at 
> http://satoss.uni.lu/projects/atrees/trees/confidentiality.pdf since it has a 
> defense node as a root and I like it.
> 
> There's a recent survey paper on attack/defense modeling that seems to cover 
> the very large field: http://satoss.uni.lu/members/barbara/papers/survey.pdf
> 
> The paper on Attack-Defense trees: 
> http://satoss.uni.lu/members/barbara/papers/ADT12.pdf
> Slides for those who experience academic paper-blindness or just like bullet 
> points: http://satoss.uni.lu/members/barbara/papers/slides.pdf
> 
> There's also a GPL3 tool implemented in Java at 
> http://satoss.uni.lu/members/piotr/adtool/, although unless one cares about 
> the math aspects it seems like one could probably accomplish the diagram 
> goals without involving Java
> 
> Thanks!
> Andrew
> 
> 1: Note that the specific funded effort seems to have ceased and for more 
> recent papers you'll need to see what the authors like 
> http://satoss.uni.lu/members/barbara/publications.php are up to.
> _______________________________________________
> dev-b2g mailing list
> [email protected]
> https://lists.mozilla.org/listinfo/dev-b2g

Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail

_______________________________________________
dev-b2g mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-b2g

Reply via email to