OS Status Meeting Minutes

November 19, 2007

Steve Bergwerff

• Important Trinity 2 dates to be aware of:
  • M2 test results review was on November 5. Some new issues raised but overall results were pretty good.
  • M3 build was on November 14 and regression tests were run over this past weekend.
  • M3 was built with gcc 2.95.3 compiler but tools group provided us with an unofficial build using gcc 4.2.1 which we also ran regression tests on this past weekend. Results were pretty good. Did see some problems on the mips be board.

Doug Bailey

• Working on a number of PRs
  • PR 52884 kernel crash PR for Trintiy 2. Fixed.
  • PR 26662 to fix some long standing for loop code. Didn’t impact anything. Fixed it in Trinity 2.
  • PR 52851 regression test failure on Trinity 2. Turned out to be a test case problem.
  • PR 26861 debug breaks causing crash on SH for Trinity 2. Investigating.
  • PR 52644 test failure on C++ , looks like a compiler bug, have a code review posted for review
  • PR 53084 customer crash PR that just came in, will look at it this week
  • PR 53196 libc dependency issue in Trinity 2, looking at it now.
• Andy asking for help with some benchmark tests done for a custom engineering project

Brian Stecher

• Worked on a number of PRs that have been marked as critical
• Worked on PR 53319, mips message pass problem on Trinity 2. Needs to be fixed for a customer that is planning to do performance testing using M2. Seem to have resolved it. Didn’t see it with earlier tests so working on improving the tests to catch this.
• Will need to fix up a script this week that is used to provide source drops to a customer. Script needs to change because the networking source has moved to a different repository

Mosa Yeung

• Working on benchmarking of interrupt latency on Trinity 2. A lot of back and forth with Robert Craig to get the tests the way he wants.
• Correcting lint warnings/errors for Trinity 2. Still a lot to do but getting through them.

Mike Kisel

• Working on fixes to Memory Partitioning for Trinity 2. Created a branch to check things into that Marian will regress. Any contentious changes waiting for CI will go on the branch. Everything else will go on the head. The branch will be merged back before M4, when everything has been CI'd.
• Reviewed the MP user documentation received from Angelina
• This week will continue working on Memory Partitioning and resolving issues.

Shiv Nagarajan

• On vacation

• fixed changes to the expect script for Memory Partitioning testing and checked it into svn. All tests are now in sn.
• Modified a number of tests and the environment based on feedback from Mike
• ran a lot of regressions on Memory Partitioning, 3 or 4 times this past week, and analyzed the results
• plan to use Mike’s branch for continued testing
• reviewing the Memory Partitioning user documentation to provide comments to Angelina

Linda Deng

• Did some variable page size benchmarking for Trinity 2. Tests showed a big improvement on ARM. Will rerun the tests to see why the others don’t show a significant gain.
• Did a POSIX test run last week against M3. Found 3 issues. One was fixed by xtang. The other two appear to be test case issues which Sunil is following up on.
• This week will write a test case for PR 50889.
• Will be doing file system testing on M3

Angelina Gomes

• Last week was spent working on Memory Partitioning documentation. Sent a new version for review.
• Update the Adaptive Partitioning FAQ based on editorial comments.

Attilla Danko

• Off today.

Colin Burgess

• Fixing a number of PRs. Worked with Brian on PR 53319 for mips message passing in Trinity 2. Found the problem on Firday.
• Will be working on the prep for a special customer drop of Trinity 2 that’s needed for this week.
• Working with Chris to set up the 7200 in the lab. Almost have it working.

Peter Mitsis

• Looking at file system content for Momentics 6.4.0 on Friday. Going thru file system PRs to correctly target any that are fixed for 6.4.0.
• Working to produce disk access benchmarks for Trinity 2. Trying to duplicate environment used at a customer site. Hope to run it on compact flash this afternoon with a static image and setup for hard disk testing. If can’t get same hardware then will send to the customer with instructions on how to run on the static image using a profiler.

Gordon Rice

• Working on a qnet issue reported by a customer as critical. Two incidents were reported with qnet flapping but on different hardware with different releases. The customer had to kill qnet and restart it so only have core dump for qnet and a bit of logging info. Working with Andrew Boyd on it last week. The customer had turned off a lot of the logging so not a lot of info is available. Going through thru different scenarios that might have lead to this problem.
• Will be off Thursday and Friday for US Thanksgiving.

Chris (via email)

• Presented OS test results at the Trinity2 M2 test report meeting
• Tested and posted results for PSP 823 and PSP 825
• Setup and ran regression on PSP 850
• Posted readme on Foundry 27 for Trinity 2 M2
• Built the M3 regression tests
• Helped Peter Mitsis with booting bigmac with 6.21B fpr disk access benchmark comparisons
• Added and tested kernel benchmarks to regression run
• Work with Sada building procntos with 421 compiler, to regress this weekend
• Helping Shahram get started with Tahi testing
• Installing 7200 into the coreOS lab
• Kicked off weekend regression testing on M3
• This week will go through Trinity2 M3 regression results.
• Will be on vacation later this week.