DCI released an “Informational Bulletin” this month, outlining its general policy towards changes in NIST standards. The bulletin is online at http://www.dcimovies.com/FIPS/nov-11-2010-FIPS_NIST_IB.html.
Several important points are made:
- DCI is relying on the SMPTE NIST Revisions Study Group to address a specific fix for the dual key use problem. Manufacturers of new media block designs are told to digitally sign logs and TLS sessions with a 2nd certificate.
- DCI intends to remove specific security algorithm requirements from its specification, and will defer specifically to NIST standards instead. This action should reduce the occurrence of future conflicts between NIST and the DCI specification. But it requires unnamed others to find suitable security fixes that comply with future NIST-imposed changes.
- DCI publicly states that NIST has not provided a timetable by which fixes are required for the dual key use problem, but it does expect NIST to introduce the new FIPS 140-3 standard in the first quarter of 2011.
Usually the first question raised is what happens to current product? The answer to this was stated in DCI’s Compliance Test Plan Change Policy Statement in October:
“When a new version of the CTP is published and has taken effect based on the above, any device undergoing actual compliance testing or re-testing, as defined by an executed testing agreement, the described device being in the possession of the DCI-licensed testing facility, and at least one test has been initiated, may continue to test or re-test to completion using the version of the CTP in effect when that device’s testing or re-testing began.”
Once a product has passed the DCI CTP, it will retain its status as DCI compliant. However, if a change is made to the product that requires re-testing for FIPS compliance, then all bets are off. The changed product will have to meet the FIPS compliance standards in effect at that time.
The significance of this policy is that product design changes, as well as changes to standards, will strive to avoid modification within the secure boundary of the media block. This has already impacted proposed changes in SMPTE, for example. It is possible to distribute an odd number of audio channels. However, not all media blocks can accept an odd number of audio channels, as the AES (a different AES – Audio Engineering Society) requires an even number of channels when using its popular AES-3 digital audio transmission method. There was a request during the recent revision of ST429-2, the core SMPTE DCP standard, to require all media blocks to accept any count of audio channels, whether odd or even in number. However, it was learned that this would require a change within the secure boundary of some media blocks, To accommodate, the requirement was changed instead to force mastering houses to package an even number of audio channels in the DCP. As time goes on, the industry will find that new advances cannot be introduced without introducing problems within the FIPS security boundary. For those who are worried that digital cinema standards will change with some frequency causing early obsolescence, this should be comforting news.
Interestingly, the SMPTE NIST Revisions Study Group mentioned by DCI has only held one conference call, which took place in November. It was decided that the dual key usage problem would be solved by recording, in the security log, the “thumbprint” of the digital certificate used to decrypt the KDM. This was not one of the methods originally proposed by DCI’s security consultant, but the manufacturing sector preferred it as it produces the least impact to product design and to the management of KDMs.
This points to the “new” way that DCI intends future NIST-related issues to be solved. DCI wants others to fix these problems. In the case of dual usage of keys, it appears that the accepted fix will come from a SMPTE sub-group. But that is only the beginning, as standards will need to be created or modified.
Today, manufacturers are bound by contract to comply with DCI and NIST due to the VPF deployment agreements that exist around the world. But after 2020, there may no longer by any deployment agreements to create such loyalty. Should a more difficult fix be required in the future, there may no longer be the willingness and manpower from the manufacturing community to address it.
DCI wants to disengage from having to regularly maintain its specification. The good news is that the changes that result will cause as little disruption as possible, as the manufacturers will not have it otherwise. But by disengaging, it is very likely that DCI will one day lose control of its spec.