Bug#948834: glib2.0: FTBFS: SIGABRT in gsocketclient-slow test

Simon McVittie smcv at debian.org
Sun Feb 9 13:15:07 GMT 2020

Control: tags -1 + moreinfo unreproducible
Control: retitle -1 glib2.0: FTBFS: SIGABRT in gsocketclient-slow test

On Mon, 13 Jan 2020 at 21:53:37 +0100, Mattia Rizzolo wrote:
> your package failed to rebuild in a standard sid chroot.

This has been working fine on the official buildds and for the
maintainers. Please could you show us the output from the specific test
that is failing for you?

> 178/261 glib:gio / gsocketclient-slow           FAIL     0.02 s (killed by signal 6 SIGABRT)

This is just part of Meson's summary of the successful and failed tests,
which is quite a long way away from the test's actual output in the log.
The test's output is what we would need to be able to have any chance of
debugging a failure.

I would guess that the failure is something related to one of the
regressions tracked at <https://gitlab.gnome.org/GNOME/glib/issues/1995>.
As such, it's possible that it will be fixed by
<https://gitlab.gnome.org/GNOME/glib/merge_requests/1339>, which we need
anyway for #948554 (CVE-2020-6750).

Here is a successful run of gsocketclient-slow on a buildd, to give you an
idea of what to look for:

(search for "Start of socket-client tests")

# Start of socket-client tests
# Start of happy-eyeballs tests
# GLib-GIO-DEBUG: _g_io_module_get_default: Found default implementation dummy (GDummyProxyResolver) for ?gio-proxy-resolver?
ok 1 /socket-client/happy-eyeballs/slow
# Start of cancellation tests
ok 2 /socket-client/happy-eyeballs/cancellation/instant
# GLib-GIO-DEBUG: IPv4 DNS error: Operation was cancelled
ok 3 /socket-client/happy-eyeballs/cancellation/delayed
# End of cancellation tests
# End of happy-eyeballs tests
# End of socket-client tests

I think the most we can do here without more information is to suppress
the FTBFS by marking this test as "flaky", which means we won't run it at
build-time, and its autopkgtest failures are tracked but ignored. However,
the failure probably does reflect a real bug, so I'd only downgrade this
rather than closing it.


More information about the pkg-gnome-maintainers mailing list