Cardon Kernel Rebuild Project

Retrieve the Deos Cardon (version 6.12.0) kernel and produce library baseline from the certification archive and demonstrate that they can be rebuilt.

Description

The Raytheon Designated Engineering Representative (DER) in conjunction with the Witchita FAA Aircraft Certification Office (ACO) has requested we demonstrate our ability to rebuild the Cardon kernel baseline, achieving the same integrity key (within reason) as achieved when we finished certification activities in July of 2002.

The fundamental objectives that they wish to witness us achieving:

• Retrieving the data (source and build instructions) from Configuration Management (the certification archive in our case).
• Ensuring we have the right equipment in place to do the build.
• Ensuring we have the recipe (released build procedure) for the build.
• Follow the recipe and perform the build resulting in a fine executable that is either identical or different in a predicted fashion.

Note that the use of scripts is great as long as we can find them and they are controlled.

The certification authorities will be in Phoenix the first week of May, 2006 to witness the rebuild. If it appears we will not be ready by that time, we can reschedule for the first week in June, 2006.

This project is in support of the Primus_Epic_Program.

News

• 2006-05-02: Rebuild and witness by Raytheon DERs occurred. No Findings. An Observation was made that the Deos team create a PCR to update the Cardon Kernel SAS to note that an audit of the team's ability to rebuild the kernel from the certification archive was performed and the original build procedures updated in order for the audit to succeed (PCR:3739).
• 2006-04-30: A rebuild has been performed and witnessed by a Deos SQE in accordance with the Deos SQA procedures. The results have been captured within the Cardon kernel's certification archive. We are ready to demonstrate the rebuild to the authorities at their convenience.
• 2006-04-26: From mailto:steve.barnes@honeywell.com - "Thanks, Matt. That pretty well answers my questions. It looks like you are trying to be ready the first week in May, when Joe and Jeff will be here, and, if you’re not ready, Joe has indicated to you that the first week in June is OK. That’s in line with what he told me. Can you keep me... ...informed as to your plans for next week? I think Joe and Jeff are going to be at Bell Road Monday afternoon and Tuesday, so your window will probably be Wednesday or Thursday morning, if you’re going to be ready."
• 2006-04-26: User:Stephen.smith@honeywell.com reports that he can rebuild the kernel, but to do so required some modifications to the build procedure, and will not result in an executable object code file with the same CRC. The reason why can be reasonably explained, thus said explanation will need to be captured when the official rebuild is performed. The current plan is to dry-run a rebuild using the SQA witness procedure, capturing all evidence and process deviations on that form. We'll then repeat the scenario with the certification authority present.
• 2006-04-26: Successfully recreated the produce library with an identical crc. The kernel gets a different crc due to approx. 5 bytes changing. Those 5 btyes aren't in any of the sections that are loaded as shown by a diassembly and diff of the two binaries.
• 2006-04-19: Telecon held with Jeff Knickerbocker (mailto:jeff@sunrise-cert.com) and Joe Cross (mailto:joe_cross@rac.ray.com) of Raytheon. Under guidance of the Witchita ACO, looking over previous Deos build artifacts will not be acceptable. We must demonstrate our ability to rebuild.

Tasks

• Budget: $8,160 allocated to Deos team.

• Start Date: April 19, 2006.

• Current Estimated Completion Date: May 3, 2006.

• Note: Unless noted otherwise, all estimates are in terms of effort hours.

Task	Assignee	Risk	Original Estimate	Current Estimate	Elapsed	Remaining
Retrieve and recreate build environment	User:Stephen.smith@honeywell.com	Elements needed to rebuild and or archived build procedures may be too old or incorrect to perform rebuild. The crc may not match (additional analysis of resulting binary would be needed to justify differences). Also, the build procedures are a snapshot of the procedures that were in the original VSS database They assume a tree that is not in the certification archive - the Witnesses may not like that. Disassembly matches, CRC's don't. Will have to do a binary diff and do some splaining.	152	32	32	0
Perform rebuild for credit	User:Stephen.smith@honeywell.com, User:Matt.diethelm@honeywell.com, Jeff Knickerbocker (mailto:jeff@sunrise-cert.com) and Joe Cross (mailto:joe_cross@rac.ray.com) of Raytheon.	None Identified	8	2	2	0

Timesite Administrivia

Project, Product, Etc.	Project Code	Task String	Funding Contact	Notes
Cardon_Kernel_Rebuild_Project	111980	000R16DEOS	Steve Barnes	Work needed to recreate and demonstrate rebuild.

---