Status of the autopkgtest regressions for pipewire 1.6.2-1ubuntu1.3
after the retries run on 2026-10-13/14 (thanks for those):
10 of the 14 tests listed on the excuses page now pass:
fluidsynth/amd64, gnome-remote-desktop/amd64, gnome-settings-
daemon/amd64, gnome-shell/armhf and s390x, kodi/arm64, mutter/arm64,
pipewire/s390x, speech-dispatcher/arm64 and weston/armhf. qemu/ppc64el
and qemu-hwe/ppc64el are no longer listed (britney now classes them
ALWAYSFAIL): the migration-reference runs of 2026-10-12 fail with the
same test-qemu-user.sh error (busybox cannot map libresolv.so.2), and no
run of either test on ppc64el has passed since 2026-07-29.
The remaining four:
- mutter/armhf and qemu-hwe/s390x: the retries ended after 2 s with
"testbed failure: eof from the virtualisation server" (lxc launch/delete
failed), so no test ran. They need another retry.
- mutter/amd64v3: mutter-18/wayland-stable-rounding.test timed out after
300 s again, 55/56 tests passed. This test is flaky on this architecture
independent of pipewire: the same test failed with trigger
systemd/259.9-0ubuntu0.1 on 2026-10-02 and passed on retry on
2026-10-06, and it passed with trigger libadwaita-1/1.9.1-0ubuntu0.2 on
2026-10-08 (same mutter 50.1-0ubuntu2.4). The mutter testbed does not
install libspa-0.2-bluetooth.
- gnome-shell/amd64: build-and-test failed in gnome-
shell:closeWithActiveWindows. The shell ran the scenario to completion
("Shutting down GNOME Shell", no criticals); what failed afterwards is
the test tool's cleanup of its temporary XDG home
(tempfile.TemporaryDirectory.__exit__ -> shutil.rmtree): OSError: [Errno
39] Directory not empty: '.../cache/mesa_shader_cache'. The run with
trigger mutter/50.1-0ubuntu2.4 on 2026-09-09 failed in the same scenario
with the same traceback (OSError: [Errno 39] Directory not empty) and
passed on retry on 2026-09-10. In the 2026-10-14 run the shell-tests
part, which runs the same scripts against the installed gnome-shell with
the same pipewire packages, passed, and so did the pipewire-triggered
gnome-shell runs on amd64v3, arm64, armhf, ppc64el and s390x. (The
2026-10-08 run had failed differently, with the shutdown-time
"MessageTray has been already disposed" critical in gnome-
shell:minimal/basic; the same kind of shutdown critical failed the
systemd/259.9-0ubuntu0.1 run on 2026-09-30 and passed on retry.)
The two patches in this upload only touch spa/plugins/bluez5/
(bluez5-dbus.c and defs.h, shipped in libspa-0.2-bluetooth). That
package is installed in the gnome-shell testbed and the test harness
starts pipewire and wireplumber, but on a private bus whose mocks cover
logind, UPower, NetworkManager, polkit and friends, not BlueZ, and bluez
is not installed, so no BlueZ adapter ever appears and the changed code
(adapter registration/removal) does not run in these tests.
Could the four remaining tests be retried once more?
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2160089
Title:
bluez5: WirePlumber SIGABRT (use-after-free) when BT adapter is
removed with pending RegisterApplication calls
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/pipewire/+bug/2160089/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
