Cardon Kernel Retest Project

Execute Cardon certification baseline requirements based tests on a Primus Epic Pentium-M module, and perform relevant analysis.

Description

The Primus Epic team is changing some of their Intel Pentium II based processing modules to Pentium-M. When the Cardon kernel was originally certified in 2002, it was only tested on Pentium II based modules. We need to retrieve the Cardon tests from the vault and re-execute them on a Pentium-M module. This will help ensure the Cardon Deos Kernel is compatible with the Epic Pentium-M modules.

In addition to re-executing the tests, the Deos kernel team will need to analyze Pentium-M errata and data sheets against Cardon kernel source code and ensure the Cardon kernel is compatible with the Pentium-M.

News

• 2006-10-26: User:Matt.diethelm@honeywell.com - SAS updates complete. Project complete.
• 2006-10-10: User:Matt.diethelm@honeywell.com - Still having time-on-task issues. Resetting completion date to Oct 31.
• 2006-10-01: User:Matt.diethelm@honeywell.com - Analysis and review work complete. All that remains is to update the SAS and republish it. Setting completion date to Oct 12.
• 2006-09-18: User:Matt.diethelm@honeywell.com - Still having trouble shaking loose from other tasks enough to work this. Resetting completion date to Sep. 30.
• 2006-09-07: User:Matt.diethelm@honeywell.com - Compatibility analysis and errata analysis complete. All that's left is re-review of changed tests, and update the SAS.
• 2006-09-05: User:Matt.diethelm@honeywell.com - Pentium-M compatibility analysis captured in certification archive. The copy needs to undergo peer review. Once past that, we can consider it done.
• 2006-08-31: User:Matt.diethelm@honeywell.com - Dreams and Talk on completing by Aug. 31. Only resource assigned is Matt at this time and other perceived higher priority pop-ups keep happening. Resetting delivery to September 15.
• 2006-08-07: User:Matt.diethelm@honeywell.com - We've heard back from Intel. Pentium Errata analysis is now complete.
• 2006-07-31: User:Matt.diethelm@honeywell.com - Same story. All Deos folks working other issues. Reset delivery to Aug 31. Also, still waiting a response from Intel on an errata question.
• 2006-07-17: User:Matt.diethelm@honeywell.com We've been swamped with other higher priority pop-ups. Also, our Pentium Errata analysis is still waiting on a response from Intel. No progress, and I now estimate it will be end of July before we are done assuming we get our Intel errata question answered and I can shake loose some time to finish the paperwork (SAS update, test reviews).
• 2006-06-24: User:Matt.diethelm@honeywell.com Still suffering resource shortages. Mostly need time from me to finish Test reviews and SAS updates. Still need some effort from Pentium-M team to look over errata, but all Deos related work complete. Just Deos related paperwork remains to complete this task.
• 2006-06-05: User:Matt.diethelm@honeywell.com - 43 hours of effort remaining. Time on task will likely not be 100%, thus estimated completion date for this activity is now June 30, 2006.
• 2006-05-31: mailto:Robert.Moyle@honeywell.com: "I don't see a problem with end of June."
• 2006-05-31: User:Matt.diethelm@honeywell.com We had hoped to complete by June 1; however a recent fire has sprung up on the Airbus A-380 program, and the Deos team has been asked to swing in and help. This has left us a few bodies short in our effort to finish up the Cardon kernel retest effort. We still have some re-reviews, analysis and Software Accomplishment Summary work to do, roughly 66 man hours of effort left. If we were to slip our completion date to mid or end-of June, 2006, would that be a problem?
• 2006-05-25: User:Matt.diethelm@honeywell.com: Run for score is complete. Hardware provided by the Epic team has been returned to Jon Gilreath. All that's left is the paperwork to file. We are still trying to have this complete by May 31.
• 2006-05-22: User:Matt.diethelm@honeywell.com: Consistent test runs have been accomplished. The set of test run behaviors now match what was seen on the original Pentium-II run. Test modifications were limited to the PAL and Deos platform registries used by the test procedures. A subset of these will need to be re-reviewed for credit, but the work will be less than originally estimated. A formal, for-credit run is scheduled for this Wednesday. Paperwork will be generated and archived, hopefully by Friday. Equipment will be returned to Epic on Friday. We're hoping to wrap up this Project by Friday, May 26.
• 2006-05-19: User:Matt.diethelm@honeywell.com: Thanks to Jon Gilreath, we've identified the problems we were having getting a consistent test run. We needed a fan to keep the P-M cool enough to behave. We are now back on track.
• 2006-05-12: User:Matt.diethelm@honeywell.com: Phone card receivied from Epic team. We will likely need it for the week as we retry the tests and hopefully do an official run later in the week. Each run requires about 8 hours, so it takes a few days to iterate on the test runs.
• 2006-05-12: User:Matt.diethelm@honeywell.com: We've tweaked the Pentium-M platform configuration such that we're getting an acceptable number of tests to pass. In other words, those tests that are failing can be reasonably explained and analyzed. However, one thing that we've not been able to replicate is a complete end-to-end test run. Because we cannot do this, we're uncomfortable claiming success. We suspect the NIC module is tripping up our test execution and results communication path here. Looking back through our previous Pentium II and Pentium III run-for-score records, we've noticed that those runs were done using a "phone card" LAN network connection rather than the NIC. We'd like to try and replicate that via a "phone card" with Pentium-M and attempt another test run. If we can do this, we can comfortably claim that we've replicated the test execution environment used in the original Pentium II and III run-for-score and witnessed equivalent test behavior and results. If we cannot do this, we will be left with some doubt that we'll have to reconcile some way.
• 2006-05-08: User:Matt.diethelm@honeywell.com: Code compatibility analysis complete enough for review. Also, additional progress has been made on getting original Cardon kernel tests running at Kutta.
• 2006-05-04: User:Matt.diethelm@honeywell.com: Code compatibility analysis has begun.
• 2006-04-30: User:Matt.diethelm@honeywell.com: Progress has been slow this last week. We're stuck on getting kernel tests that require a special "test" PAL. We believe we're past it, but Kutta will need to give it a try on their bench to be sure. I'm sliding the completion date for test re-execution to May 31.
• 2006-04-24 - From User:Matt.diethelm@honeywell.com Kutta has not been able to dedicate 100% of their time on this effort (only expended 20 hours last week). Coupled with some unforseen technical hiccups on getting all tests to pass, I'm worried our "Test Cases and Procedures" completion date is in jeopardy. I'm inclined to reset it to May 31, but am going to wait until the end of this week to see what the progress is.
• 2006-04-12 - From Kutta: "I was able to complete the first run of the tests last night. This morning when I came in to work all the tests had run... I had 80 tests that passed and 119 tests that failed. From looking at the report all the tests after test 126 failed which I think is suspicious. I checked the PM proc card and I was able to ftp to it and it was in normal mode. I’ll investigate further and let you know what I find."
• 2006-04-07: Analysis against Pentium-M data sheets and errata sheets has not yet started due to resource availability conflicts. Start date is likely pushed out until May 1. This implies analysis work will not complete until mid-May at best, pushing project end date into late May.
• 2006-04-06: Kutta consulting is in posession of a working P-M module and fixture. The current struggle is to get the Cardon tests rebuilt with Pentium-M specific platform settings. This was hoped for by the end of this week, but it's Thursday and we're still stuck trying to get the tests to build. Stephen Smith has been pulled off his current task and sent to Kutta to assist.
• 2006-03-30: Project has now officially started with delivery of Epic P-M module to Kutta. Thanks to help from some key Epic technicians and engineers, the module is up and pinging at Kutta and ready to serve as a test host.

Tasks

Initial Budget: $29,000 allocated to Kutta, $43,200 allocated to the Deos team.

YTD: $19,939 (Deos Team Only)

Current ETC: $0

Current EAC: $19,939 (Does not include Kutta charges)

Milestones	Due Date	Estimated Delivery	Delivered	Percentage Complete
Test Cases and Procedures	30-JUN-2006	31-OCT-2006	26-OCT-2006	100
Cardon Kernel Compatibility Analysis	30-JUN-2006	15-SEP-2006	07-SEP-2006	100
Kutta Technical Support	31-MAY-2006	31-MAY-2006	25-May-2006	100

Test Cases and Procedures

Task	Assignee	Risk	Original Estimate	Current Estimate	Elapsed	Remaining
Attempt sample test procedure run on P-M. This should help determine how many tests will need to be altered to work on P-M.	Bruce Elliott of Kutta Consulting	Recreating the test environment and test build infrastructure	80	76	76	0
Modify test procedures or test settings for P-M platform	Bruce Elliott of Kutta Consulting	More tests or test settings will need to be modified than expected.	80	70	70	0
Re-review tests that were altered (if needed).	user:matt.diethelm@honeywell.com	None identified.	40	10	10	0
Perform run-for-score	Kutta Consulting, with assist from user:matt.diethelm@honeywell.com	None identified	8	8	8	0
Capture test evidence in Deos certification archive and update Cardon Kernel Software Accomplishment Summary	user:matt.diethelm@honeywell.com	None identified	16	22	22	0

Kutta Technical Support

This is a level of effort task. Various Deos team members may be asked to assist Kutta with any technical roadblocks. The hours spent doing that should be accumulated here in order for us to see how much we're spending.

Original Estimate	Elapsed
100	67

Cardon Kernel Compatibility Analysis

These tasks will be performed by the Deos team.

Task	Assignee	Risk	Original Estimate	Current Estimate	Elapsed	Remaining
Analyze Cardon Kernel Code against Pentium-M Errata	user:aaron.larson@honeywell.com	Low risk - there could be an incompatibility that renders Cardon use on Pentium-M undesirable	40	40	40	0
Analyze Cardon Kernel Code against Pentium-M Data Sheet	user:aaron.larson@honeywell.com and or user:ryan.roffelsen@honeywell.com	low risk - We detect an incompatibility that renders Cardon use on Pentium-M undesirable	40	20	20	0

Timesite Administrivia

See Timesite_Codes

---