OS Status Meeting Minutes

December 10, 2007

Steve Bergwerff

• Introductions to Scott Miller, a new hire in the kernel support team joining the meeting today, along with Peter Luscher and Gordon Rice.
• Important Trinity 2 dates to be aware of:
  • M3 test results review is today
  • M4 build is on Tuesday (Dec 11).
  • M5 build is scheduled for Jan 16

Doug Bailey

• Worked on a number of PRs last week
  • PR 53959 for trace event with user string when a bad string pointer is used. Provided a fix.
  • ISR causing a crash. Will give an experimental proc to dump out additional info.
  • PR 24729 was fixed for SH PVR
  • Investigating PR 40853 for a regression test failure in Trinity 2 that was originally reported on arm then showed up on sh.
  • Problem with EIDE driver for RP1. Added a comment to PR to get more info and assigned to Hugh.
  • PR 53263 for a customer backport for mapinfo. Submitted the change for code inspection.
• This week will wrap up the experimental load for customer and will get trace event fix. Will then work on the RP1 PR.

Peter Mitsis

• Last couple of weeks spent some time writing test cases for PR 19321 and 46995.
• Rest of time spent on PRs
  • PR 53519 ended up being a dup.
  • PR 45631 fix is submitted for CI
  • Looked into an apparent regression failure for Trinity 2 during file system testing but couldn’t reproduce it and discovered that the test was newly added and doesn’t apply to the older file system
  • Investigating PR 53580 for a data compare error. Not sure if it is a file system issue or memory manager.
  • PR 53181 is for a warning that only applies to older drivers. Will add a release note to indicate under what conditions the warning will be reported.
• Going through fixed PRs to determine exactly which ones would be included in an update to the file system.
• For the analysis on the apparent file system read/write performance slowdown, it was discovered that the customer had not enabled L2 cache. Once this was done the numbers were as expected.

Mosa Yeung

• Wrote some test cases for Trinity 2.
• Still fixing lint warnings against Trinity 2 and almost ready for CI for services/systems. There will be 59 files.

Brian Stecher

• Last week a good part of the time was spent working on PRs and triaging incoming PRs
• Working on a customer issue that was reported on 6.3.0 SP2. If add a couple of no-ops at specific points, then it starts working. Almost at the point of being able to prove that it is an errata and not the OS
• Working on a wired tlb fix for Trinity 2. Have a pretty good idea of how to fix it.
• Applied a workaround to an errata for the 8641D. Required a proc change and startup changes which are now on the head.
• This week, will continue working on the customer issue (suspected errata) and wired TLB, and a few other issues reported by a customer that is working with Trinity 2.

Shiv Nagarajan

• Spent some time looking at several small PRs that have now been applied to head and a few others that are waiting on CI requests
• Worked on three customer reported issues for Trinity 2.
  • One was for a process getting blocked on an mmap_peer call. Built a debug proc and gave to the customer to try.
  • Second is a select() issue with telnet. Have Sean involved with the investigation on this.
  • Third is a reported crash triggered by a qnet test. Need more info to try and reproduce it here.
• Also committed a change for PR 54054 to support mmap() NOINIT flag on Trinity 2..
• Have some other PRs for Trinity 2 that will try to complete and send diffs for CI.

Attilla Danko

• The test case for the cs_time change in Trinity 2 uncovered a problem. Fixed the issue and the test case now passes.
• Was looking at 3 PRs that were marked as high priority for Trinity 2. These are all addressed now.
• Spent some further time looking at async message code to see if some of the remaining issues can be fixed without a rewrite of the code.
• Asked Kerry to give a tech talk about Athens 6.4.0.
• This week will spend 2 days in meetings with consultants regarding additional process certification.

Colin Burgess

• Working on various PRs
  • Have a fix checked in for PR 11211.
  • Have a couple of waitfor() fixes which are almost ready. Got the CI back from Peter V with some good comments that need to be implemented.
• Customer using Trinity 2 is having an issue with trying to read process memory and getting a server fault. Pretty sure that it’s a bug in the customer code but will continue to work with them this week.
• Also the same customer reported an issue using wired TLB on their hardware. After investigation with Brian, discovered a logic error that could cause it but haven’t been able to reproduce it here yet. The customer made a change for the shared library and is no longer held up by this but we’ll continue to work on it.
• Looking into a Trinity 2 customer reported issue with devctl call on SMP.
• Working on tiger team for getting a full head branch using the gcc 4.2.1 compiler.
• Attended PR review on all that are assigned to me. Have another meeting this week to complete it.

Marian Kos

• Running Trinity 2 APM regression tests, changing and fixing test cases, creating new ones, and generating PRs,
• Finding issues with running on SMP.

Linda Deng

• Setting up and running tinderbox on Trinity 2. Getting familiar with the build and analyzing and going thru the errors.
• Created some new test cases for Trinity 2.
• Resolved some file system regression test issues for Trinity 2.

Angelina Gomes

• Mostly working on doc updates for the orange team right now.
• Did some work on terminology updates for Memory Partitioning doc. Will work with Attilla to get his FAQ added for APM.

Sunil Kittur and Mike Kisel

• Not in attendance.