Foundations Team Updates - Thursday 22 September 2022

Last week’s status can be found here: Foundations Team Updates - Thursday 15 September 2022

1 Like



  • Lots of interviews and take home test reviews
  • Sponsored systemd for @enr0n
  • json-glib:
  • failed to understand adsys timeout on riscv, so filed bug report:
  • snapd is broken on riscv kinetic: actually all Go applications are and all applications that use “pointer tagging” method. IMO, we need to disable sv57, I provided a patch for that to the kernel team which I tested successfully, still waiting for feedback.
  • livecd-rootfs PR for Icicle and LicheeRV
  • investigating issue in Ubuntu Core when loading licheerv dkms driver
  • trying to reproduce issue encountered a few months ago which led to disable CONFIG_VMAP_STACK from the kernel (problem with EFI page table not synchronized)



  • Bug triage on September 15 2022


  • OpenJDK provides a JIT compiler for RISC-V. But it is not enabled in Kinetic.
  • Fixed building with Hotspot enabled in schroot (takes 28 h on Unmatched board). Still waiting on Launchpad to succeed (60 h expected).
  • Next step: create Debian bug.

Product roadmap sprint

  • Updated slides


  • systemd-boot for RISC-V has been released

LicheeRV gadget snap

  • Fixed a build issue with core20.


  • I filed my MOTU application.

The Ubuntu QA Team’s status can be found at Ubuntu QA Team Weekly Status Update

1 Like
  • Finished the cargo and rustc upgrades to 0.62 and 1.61 respectively in preparation for the upcoming Firefox
  • Looked at patches for the glibc upload
  • Interviews & Take-Home tests
  • Investigated various breakages after having upgraded to Kinetic (email still broken :/)
  • Bug triage
  • Subiquity
    • PR: 1428 (LP: #1989353) - fix issue where the kernel configured by autoinstall would not be used. There is a feature where the kernel meta-package can be chosen by setting the name to /run/kernel-meta-package, but autoinstall should take precedence over that value (and wasn’t).
    • Review PRs
    • Working on documentation on usage of Autoinstall with live-server to produce a Desktop install
  • +1 - See full report for more details
    • more things on Audacity LP: #1983862 trying to get the build sorted out
    • kiwi vs kexec-tools (LP: #1988966) - adjust kiwi binary packages architectures to fix dependency on non-existing kexec-tools risc-v package. Later we should be able to reverse this but we need some risc-v patches stabilized first.
    • dolfin vs fenics and mshr - mshr wanted a no-change rebuild to match the recent dolfin builds, which fixed tests for both fenics and mshr.
1 Like
  • Writing SRU bug for nullboot
  • Paperwork
  • Thinking about APT download sandboxing
  • New scary APT bug in jammy, regression from the new phased updater (LP: #1990525)?
  • APT merge request triage
  • Fixed regression in recent grub2{,-unsigned} uploads in kinetic
  • Some APT bionic SRU verification
  • Paperwork
  • Secret work :smiley:
1 Like


  • Revisited PR 1421 - add support for cloud-init in without going through an autoinstall.
  • Opened & merged PR 1426 to address multiple issues in the snaplist screen:
    • LP: #1989198 - snap info pre-fetching is broken
    • LP: #1989193 - race condition causing the list of snaps to be (partially) empty
    • LP: #1989179 - “Try again” mechanism comes with no guarantee that snaps will be installed
  • Revisited and merged PR 1418 to fix a crash when a LVM logical volume is edited - LP: #1988407
  • Opened & merged PR 1429- cdrom unmount issue caused by recent change for FDE
  • Opened PR 1430 to update list of restricted usernames (e.g., kvm). Using such a username would prevent the user from being created because a system group with the same name exists at the time of the user creation.
  • Submitted .debdiff to user-setup to add sgx to the list of restricted usernames. A system group named sgx is created by udev.postinst (src:systemd). Also forwarded the patch to Debian. LP: #1990488. Sponsoring needed :slight_smile:
  • Currently investigating LP: #1989977 where Subiquity crashes if an extended partition is created with no size specified.
1 Like
  • Wrote post about Hyperpixel configuration (pending)
  • Worked on support for a couple of eInk displays (PaPiRus and Inky)
  • Working on RPi.GPIO->lgpio shim
  • Finally verified jammy SRU of raspi-config (LP: #1972982)
  • QEMU patches merged upstream for flash-kernel
  • Pre-installed image discussions
  • Investigated FDE possibilities on rpi desktop images
  • Investigated reports of pi-bluetooth not working on Core; more work required
  • Did a little docs work on ubuntu-image
  • Firefox meeting
1 Like

short week, out on sick leave last week.


  • Addressed discussion on ubuntu-devel about digikam being ftbfs due to qtav removal (it wasn’t).
  • Filed LP: #1990187 on systemd about libnss-resolve being pulled into the base system again, with some knock-on consequences.
  • assisted Ubuntu Unity team with getting the right packages seeded so gnome-shell isn’t pulled in where it’s unwanted/unused.
  • found that livecd-rootfs has had a bug for the past 4 years where it calls germinate with main and restricted regardless of the components specified for the flavor. Uploaded a fix for this
  • also uploaded a hack to lightdm so that when building ubuntu-unity images, apt traverses dependencies in the desired order and doesn’t pull gnome-shell into the image


  • ffmpeg

    • processed a number of removal requests, including audacity
    • removed performous-composer, already removed from Debian testing
    • patched openjfx to not use the ffmpeg bits. This is not great since it breaks API but probably better than ripping all the reverse-dependencies out of the release.
    • tracked down a removal request for libdlna, as ubuntu-archive had not been subscribed to it (otherwise I was going to file one myself). Removed.
    • that’s the last of the reverse-dependencies, so all the ffmpeg libraries can finally be removed once openjfx migrates!
  • libsvtav1enc0

    • tracked down a regression in libyuv due to changes to enable more SIMD exploitation on amd64 and arm64, which cause the amd64 build to fail. This was blocking libavif, so reverted the upstream patch for now. Eventually figured out that arm64 build was failing for a different reason, and that it was not a regression (the -proposed version has build-time test suite enabled and the release version does not), so disabled the testsuite on arm64 to
      finally get this in.


1 Like



I spent most of my time on building/testing audacity 3.1.3 and fighting against multiple issues.