News for hackers. Stuff that matters.
Re:Unhackable Kernel? by nabla9 on Friday September 18, 2015 @07:09AM (#10238481)
Current cream of the crop:

IBM System z is EAL5 rated (Semiformally Designed and Tested)

Integrity-178B is EAL6 rated (Semiformally Verified Design and Tested). It's used in aviation systems: Airbus A380, F-16, F-22, F-35 and B-2.

Currently the only EAL7+ rated (Formally Verified Design and Tested) devices are data diodes. Maybe seL4 can be first OS to get this rating?

Re:Unhackable Kernel? by nickpsecurity on Friday September 18, 2015 @02:00PM (#10240895)
The SCOMP STOP OS, GEMSOS, ASOS, LOCK, and Boeing SNS platforms were all produced to Orange Book A1 class that's largely EAL7-equivalent. STOP, GEMSOS, and SNS were all certified at A1. They're all still available for purchase. For reasons illustrated here [1], they've either dropped to EAL5+ or not been re-certified under Common Criteria. Boeing SNS kept its EAL6-7 development assurance, though, while GEMSOS is available [2] in original form via Aesec w/ various prototype applications. The new kids on the block, under MILS and SKPP banner, are taking chances with certifications that Old Guard knows are a financial trap given U.S. govt's promises aren't reliable. Private certifications against these criteria and with open-ended, pen tests are better route for cost-benefit tradeoff.

[1] http://lukemuehlhauser.com/wp-content/uploads/Bell-Looking-B...

[2] http://aesec.com/

[3] https://web.archive.org/web/20150819095124/http://www.cis.up...

(Added No 3 to represent capability-security systems given KeyKOS was fielded and had a B3/EAL6 assurance argument. Neat architecture.)

Note: The Bell paper showed both how commercial systems couldn't be trusted on a network and potential NSA IAD subversion back then. Aesec's Evaluation Report is interesting reading both for how to do high assurance architecture plus for risks like hardware they didn't see coming back then. It needs to be put on trusted hardware or ported to strong, hardware TCB. CHERI, maybe, since GEMSOS uses segmented protections.

Re:Unhackable Kernel? by hackuser on Friday September 18, 2015 @11:21AM (#10239686)
I went to lookup EAL in Wikipedia and discovered something interesting: The exact same information as above on the EAL page (with a little on the Integrity OS page).

Either the info above is from Wikipedia, or perhaps the commenter above also wrote those sections of the Wikipedia pages, or the commenter read something else that was based on those Wikipedia pages. The web is an interesting place.

Re:Unhackable Kernel? by nickpsecurity on Friday September 18, 2015 @01:47PM (#10240782)
Use Cygnacom's breakdown as it has a nice summary of what each EAL means:

https://web.archive.org/web/20130718103347/http://cygnacom.c...