bolt ftbfs in eoan (test failures)

[Matthias Klose]

seen in the recent eoan test rebuild, all architectures

Ok: 39
Expected Fail: 0
Fail: 2
Unexpected Pass: 0
Skipped: 0
Timeout: 0
dh_auto_test: cd obj-x86_64-linux-gnu && LC_ALL=C.UTF-8 MESON_TESTTHREADS=4 ninja test returned exit code 1
make[1]: *** [debian/rules:21: override_dh_auto_test] Error 255
make[1]: Leaving directory '/<<PKGBUILDDIR>>'
make: *** [debian/rules:7: build] Error 2

[Christian Kellner (gicmo)]

Are the detailed logs somewhere to be found (i.e. which tests fail)?

[Matthias Klose (doko)]

https://people.canonical.com/~doko/ftbfs-report/test-rebuild-20190906-eoan.html

[Christian Kellner (gicmo)]

The problem here kinda is the buildsystem, which uses an old kernel but a very new glibc: "4.4.0-161-generic #189-Ubuntu" and "2.30-0ubuntu1". bolt uses the "copy_file_range", which was introduced in 4.5. glibc had an emulation layer for it but that was removed with 2.30[1], and now the combination of a rather old kernel (4.4) and a very new glibc (2.30) will make "copy_file_range" return ENOSYS and thus will make bolt's "bolt_copy_bytes" function fail.

Not entirely sure how to handle it. Given that bolt's functionality itself is kinda depended on 4.13[2] it seems counter-productive to add fallback code that will *only* get executed in the test environment but never run in production (because eoan ships with linux 5.x) and the actual code that will run will indeed never get tested. Might as well skip the relevant tests.

[1] https://sourceware.org/ml/libc-alpha/2019-08/msg00029.html
[2] Thunderbolt security levels were only introduced in 4.13.

[Christian Kellner (gicmo)]

I have create a MR upstream to skip the affected tests (https://gitlab.freedesktop.org/bolt/bolt/merge_requests/187). NB: It does not skip the test if the build was configured with tests-speed=slow (which is not the default).

[Launchpad Janitor (janitor)]

This bug was fixed in the package bolt - 0.8-3~ubuntu3

---------------
bolt (0.8-3~ubuntu3) eoan; urgency=medium

  * Really apply the change this time...

 -- Sebastien Bacher <email address hidden> Sat, 21 Sep 2019 12:31:41 +0200