/
2020-10-07 Meeting notes

2020-10-07 Meeting notes

Oct 07, 2020

Date

Oct 7, 2020

Attendees

  • @Anmar Oueja

Minutes

  • Operations

    • We need weekly a private meeting to address confidential matters.

    • Will return the meeting to one hour. First half if private and the other is public

    • Will make the public one first

    • Will have two invitations. Status will be done in the public meeting

    • Will abandon Collaborate and stick with Google docs

      • dd: New or existing doc?

  • LAVA Lab - Dave

    • 1.5% Infra errors

    • LAVA 2020.09 upgrade

    • Maintenance window 1-3 hours

  • Kernel Validation - Daniel

    • Juno PCI problem with juno-07

      • rare and difficult to reproduce. We will continue using the Juno-07 for testing

    • NXP LAVA Lab being upgraded last week

      • It was down

    • LTP to be upgraded (20200930) soon. Sanity test will be updated.

      • Quickhit is replaced by smoke test in LTP

      • Viresh was second contributor to this release. Congrats!!

    • KUnit to be added soon

      • That's great news!!!

    • Validated upcoming changes on kselftests

    • Reported ftrace selftest NULL pointer on i386 on next-20201001

    • Reported ftrace oops on Qemu/arm on next-20201001

    • Tested KVM VMX patch for x86, now on mainline

    • Tested pidfd_wait selftest patch

    • RC Log

      • 2020-10-05

        • 4.19.150, 5.4.70, 5.8.14

          • Reported no regressions in <24h

  • Android Kernel Testing - Sumit

    • Tom has sent out the reports for 4.19.148, 4.9.238, 4.4.238. 5.4 and 4.14 results are in progress; reports will be sent out shortly.

    • Many test runs don’t seem to be counted towards the final test results in qa-reports? The command line tools report different numbers?

    • Test status

    •  

      • Seen a couple of VtsKernelProcFile test failures on all boards in 4.4 and 4.19; will investigate and report.

      • Full cts/vts results:

      •  

        • Db845c: passed 1,358,342, failed 3,694, total: 1,362,036

        • Hikey960: passed 42,900, failed 2,419, total: 45,319

      • Hikey:

      •  

        • Android-4.9-o results running Android 8 using HiKey failed all 1545 VTS tests. Looking into it.

      • Hikey 960:

      •  

        • Mostly stable other than the ones mentioned above.

      • DB845c

      •  

        • Mainline results seem consistent with previous runs.

        • ctsDeqpTests update

        •  

          • Amit pulled together a ‘closer to master’ mesa update to validate his failures, and with that, the CtsDeqp failures came down from ~2000 to 24!

          • Move to an updated mesa isn’t painless, unfortunately. So we will chart our course of action on possible mesa update to 20.3 when it releases.