The gnome-desktop-testing-runner has a default timeout of 300s. With
qemuriscv64, some tests in the glib-2.0 test suite (notably
codegen.py.test and gi-compile-repository.py.test) can take a long time
and exceed this timeout limit, resulting in intermittent test failures.
To avoid this problem, double the timeout by passing '-t 600' to
gnome-desktop-testing-runner.
Although not a perfect comparison (I have to use 'taskset --cpu-list 0'
to simulate loading in order to make the tests fail consistently on my
local machine), it's worth noting that the timeout increase does result
in a slightly longer test run. Here is an example of the duration when a
failure happens under 'taskset --cpu-list 0 runqemu nographic snapshot'::
|SUMMARY: total=298; passed=296; skipped=1; failed=1; user=606.7s; system=1388.3s; maxrss=170976
|FAIL: glib/codegen.py.test (Child process killed by signal 9)
|
|ERROR: Exit status is 2
|DURATION: 1368
|END: /usr/lib/glib-2.0/ptest
|2025-09-18T19:04
|STOP: ptest-runner
|TOTAL: 1 FAIL: 1
and a pass:
|SUMMARY: total=298; passed=297; skipped=1; failed=0; user=682.2s; system=1295.0s; maxrss=170476
|+ userdel glib2-test
|DURATION: 1402
|END: /usr/lib/glib-2.0/ptest
|2025-09-18T17:23
|STOP: ptest-runner
|TOTAL: 1 FAIL: 0
[YOCTO #15891]
Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
fi
useradd glib2-test
cd /tmp
-su glib2-test -c 'G_TEST_TMPDIR=`readlink -f /tmp` gnome-desktop-testing-runner glib'
+su glib2-test -c 'G_TEST_TMPDIR=`readlink -f /tmp` gnome-desktop-testing-runner -t 600 glib'
userdel glib2-test