unittest addCleanup calls are called in reverse order, so we need to reverse
the order of the calls as well, otherwise we won't correctly terminate the
subprocess children
An assertion that is raised within a callback will not be swallowed by
the C code that called the function. To ensure that errors will be
noticable, pass the result back to the surrounding scope and check it
there.
This excercises the path where we early-report a result and the
VerifyStop call must wait for the operation to complete intenernally.
Note that this test cannot fail right now due to the FpImageDevice
internal code still trying to hide the deactivation delay internally.
See libfprint!174
An application terminating because of a signal like SIGSEGV, SIGABRT and
friends, will exit with a signal number that is 128 + $SIGNAL_NUMBER, so
let's ensure that the daemon has not been terminated because of a such error
This makes even more sense with address sanitizer builds, as the daemon
would exit with abort.
The test doesn't need any assertion because we're calling DeleteEnrolledFingers
and in case it fails a net.reactivated.Fprint.Error.PermissionDenied error
would be thrown, and thus an exception would be raised at python level, making
the test to fail.
test_claim_from_other_client_is_released_when_vanished would fail on
the CI but work on a local system because we wouldn't want long enough
for the "vanished" code path to be taken into account. Add a small
timeout to make sure it works on the CI as well.
enroll_image() was always waiting for enroll-completed rather than for
what the caller expected as the result.
======================================================================
FAIL: test_enroll_invalid_storage_dir (__main__.FPrintdVirtualDeviceClaimedTest)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/hadess/Projects/jhbuild/fprintd/tests/fprintd.py", line 661, in test_enroll_invalid_storage_dir
self.enroll_image('whorl', expected_result='enroll-failed')
File "/home/hadess/Projects/jhbuild/fprintd/tests/fprintd.py", line 384, in enroll_image
self.wait_for_result('enroll-completed')
File "/home/hadess/Projects/jhbuild/fprintd/tests/fprintd.py", line 373, in wait_for_result
self.assertEqual(self._last_result, expected)
AssertionError: 'enroll-failed' != 'enroll-completed'
- enroll-failed
+ enroll-completed
The test file calls self.daemon_start() in order to start fprintd and
locate the virtual image device that's needed for the test to run.
However, since the virtual image driver is not available on all
libfprint installations, the test should be skipped if the driver is not
available.
The skipping mechanism used to work, by checking if self.device is None.
This is based on the assumption that self.device would be set to None in
cases where the driver is not available. However, even in the past
self.device is only set to None in the tearDown method and not in setUp,
so presumably in edge cases it didn't entirely work.
However, since 0fb4f3b021 which
consistently removes the self.device attribute rather than setting it to
None, the "self.device is None" check no longer works. In particular,
the following error message is shown:
test_manager_get_default_device (__main__.FPrintdManagerTests) ...
Did not find virtual device! Probably libfprint was build without
the corresponding driver!
ERROR
After this patch, the following is shown:
test_manager_get_default_device (__main__.FPrintdManagerTests) ...
Did not find virtual device! Probably libfprint was build without
the corresponding driver!
skipped 'Need virtual_image device to run the test'
We fix this bug by consistently setting self.device to None, in both the
setUp method before daemon_start gets called, and in tearDown. We also
make the same change to self.manager for consistency.
The issue was not caught on CI, as the CI configuration always installs
a libfprint version that has the virtual_image device explicitly enabled
in the preparation phase.
When we run tests in a system with real devices, we may try to initialize
the real ones, while we can just ignore them all in tests.
We do it in setUp instead of setUpClass to allow tests to change this if
they need to, but just for temporary.
When creating a new unit we used to get the system bus via Gio.bus_get_sync,
however this has a singleton implementation, and so would always return the
same connection, creating issues in tests when a new test suite is added
because the newly got connection would be already closed.
So, just manually create a new bus connection, also close the bus and
cleanup the test bus in dbus.
Instead of automatically replying with the 'whorl' image for every enroll
state signal with result 'enroll-stage-passed', only perform the number
of required enroll stages and ensure that we get the expected results.
This also will allow to manually perform enroll steps in other tests.
When coverage is enabled fprintd test won't generate any .gcda file and so
apparently no data, this happens because gcov doesn't handle properly the
process termination when SIGTERM is used, and so when in fprintd.py we
terminate the process no coverage data is reported.
To avoid this, quit the main loop cleanly on SIGTERM, so that we will exit
from the main function cleanly, making libc to perform a gcov flush when we
exit the program.
