debcrafters-packages team mailing list archive

[Ubuntu SRU Bot <2119291@xxxxxxxxxxxxxxxxxx>]

All autopkgtests for the newly accepted gtk4 (4.14.5+ds-0ubuntu0.4) for noble have finished running.
The following regressions have been reported in tests triggered by the package:

game-data-packager/unknown (ppc64el)
gtk4/4.14.5+ds-0ubuntu0.4 (ppc64el)
libreoffice/4:24.2.7-0ubuntu0.24.04.4 (amd64, s390x)
libreoffice/unknown (ppc64el)
mutter/46.2-1ubuntu0.24.04.10 (ppc64el, s390x)


Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-
migration/noble/update_excuses.html#gtk4

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

-- 
You received this bug notification because you are a member of
Debcrafters packages, which is subscribed to gtk4 in Ubuntu.
https://bugs.launchpad.net/bugs/2119291

Title:
  mesa 25.0.7 SRU breaks gtk4 on s390x

Status in gtk4 package in Ubuntu:
  Fix Released
Status in mesa package in Ubuntu:
  New
Status in gtk4 source package in Noble:
  Fix Committed
Status in mesa source package in Noble:
  New
Status in gtk4 source package in Plucky:
  New
Status in mesa source package in Plucky:
  New
Status in gtk4 source package in Questing:
  Fix Released
Status in mesa source package in Questing:
  New

Bug description:
  [ Impact ]

  There are a number of gtk4 tests (build-time, and again at autopkgtest
  time) that are failing on s390x *only* when run in the combination of
  an updated mesa package (25.0.7-0ubuntu0.24.04.1).

  s390x is not a desktop platform, there are no desktop images for it,
  and gtk4 is not seeded in that architecture. While it would be best to
  fully understand what is making these tests fail, it's best to skip
  them for now in that architecture only, so that users in other
  architectures can benefit from the mesa and gtk4 updates.

  [ Test Plan ]
  The failure manifests itself in the build-time tests, which are again repeated at autopkgtest time. Therefore, the test plan consists of:
  - no build failure in any platform, and specifically s390x
  - no autopkgtest failure in any platform, and specifically s390x

  Specifically, we also don't want any change in the number of tests run
  in non-s390x architectures.

  amd64 and arm64:
  x11:
  Ok:                 655
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            13
  Timeout:            0
  wayland:
  Ok:                 657
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            11
  Timeout:            0

  armhf:
  x11:
  Ok:                 578
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            2
  Timeout:            0
  wayland:
  Ok:                 569
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            11
  Timeout:            0

  i386:
  x11:
  Ok:                 576
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            2
  Timeout:            0
  wayland:
  Ok:                 567
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            11
  Timeout:            0

  ppc64el:
  x11:
  Ok:                 655
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            13
  Timeout:            0
  wayland:
  Ok:                 657
  Expected Fail:      0
  Fail:               0
  Unexpected Pass:    0
  Skipped:            11
  Timeout:            0

  riscv64:
  Build time tests are not run:
  dh: command-omitted: The call to "debian/rules override_dh_auto_test" was omitted due to "DEB_BUILD_OPTIONS=nocheck"

  s390x should have no tests with a timeout result in proposed.

  [ Where problems could occur ]

  This SRU is skipping certain gtk4 tests in s390x only, not reducing
  the test coverage in other architectures.

  Since the real cause of the test failures was not determined yet (and
  they only fail in combination with the new mesa update, as mentioned
  in the [impact] section), it's possible that by skipping these tests
  we are overlooking something that could be important down the line.

  [ Other Info ]

  There is already a long list of tests that are being skipped on s390x,
  and we are adding more tests to that list.

  The list of failed tests was obtained like this:

  grep TIMEOUT build-s390x.log | awk '{print $5}'|sort -u | sed -r
  "s/(.*)/'\1',/"

  I used the meson.build patch because these tests are run again as
  autopkgtests, and in a slightly different way, so anything in d/rules
  would not work.

  Also note that the failure mode of these tests is different than that
  of other tests already being skipped via d/rules. Here the test just
  hangs, not producing output, whereas in the tests "skipped" via
  d/rules, the comparison check between images (which is the test) is
  relaxed enough so that it passes. That trick doesn't work here.

  [ Original Description ]

  First seen in the autopkgtest:

  https://autopkgtest.ubuntu.com/packages/g/gtk4/noble/s390x

  mesa 25.0.7-0ubuntu0.24.04.1 regresses the autopkgtest of gtk4
  4.14.5+ds-0ubuntu0.1 on s390x.

  42515s build-x11            PASS (superficial)
  42515s build-wayland        FAIL timed out
  42515s python3-gi-x11       PASS (superficial)
  42515s python3-gi-wayland   FAIL timed out
  42515s installed-tests-x11  PASS
  42515s installed-tests-wayland FAIL non-zero exit status 2
  42515s installed-tests-flaky-x11 FLAKY non-zero exit status 2
  42515s installed-tests-flaky-wayland FLAKY non-zero exit status 2

  The build-wayland and python3-gi-wayland autopkgtests do not terminate
  and eventually hit the time out.

  We have tried running the tests on a bigger instance (added to
  'big_packages') and have tried triggering the tests with 'all-
  proposed=1'.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/gtk4/+bug/2119291/+subscriptions