[PATCH v2 08/10] app/test: add some unattached tests to fast-test suite

[David Marchand]

On Fri, 5 Dec 2025 at 13:53, Bruce Richardson
<bruce.richardson at intel.com> wrote:
>
> On Fri, Dec 05, 2025 at 01:50:03PM +0100, David Marchand wrote:
> > On Fri, 5 Dec 2025 at 13:44, Bruce Richardson
> > <bruce.richardson at intel.com> wrote:
> > >
> > > On Fri, Dec 05, 2025 at 01:37:06PM +0100, David Marchand wrote:
> > > > On Thu, 4 Dec 2025 at 19:22, Bruce Richardson
> > > > <bruce.richardson at intel.com> wrote:
> > > > > diff --git a/app/test/test_red.c b/app/test/test_red.c
> > > > > index 7f38ed1469..4b17173999 100644
> > > > > --- a/app/test/test_red.c
> > > > > +++ b/app/test/test_red.c
> > > > > @@ -1872,6 +1872,6 @@ test_red_all(void)
> > > > >
> > > > >  #endif /* !RTE_EXEC_ENV_WINDOWS */
> > > > >
> > > > > -REGISTER_TEST_COMMAND(red_autotest, test_red);
> > > > > +REGISTER_FAST_TEST(red_autotest, NOHUGE_OK, ASAN_OK, test_red);
> > > > >  REGISTER_PERF_TEST(red_perf, test_red_perf);
> > > > >  REGISTER_PERF_TEST(red_all, test_red_all);
> > > >
> > > > The commit 50823f30f0c8 ("test: build using per-file dependencies")
> > > > removed some context on why some tests were not part of a testsuite.
> > > > -# Tests known to have issues or which don't belong in other tests lists.
> > > > -extra_test_names = [
> > > > -        'alarm_autotest', # ee00af60170b ("test: remove strict timing
> > > > requirements some tests")
> > > > -        'red_autotest', # https://bugs.dpdk.org/show_bug.cgi?id=826
> > > > -]
> > > >
> > > > I would prefer we don't put an unreliable test in fast-tests.
> > > >
> > > Agreed, but for many cases for developers running these on their systems,
> > > it's probably worth having them in. What would you think about having
> > > checks in the tests themselves to detect if they are running in github CI
> > > and to skip themselves in that case?
> >
> > rte_red API has been left without tests for years.
> > I would prefer we remove the library completely rather than having to
> > work more on it.
> >
> >
> > IIRC, the issue was seen at UNH, not in GHA and it was quite random.
> > If the goal is just about removing a build warning, let's put this
> > test in a "attic" testsuite.
> >
>
> Attic is it so. What about alarm autotest? I still see possibility of value
> in having it, but I get the fact that it is timing dependent. Attic also?

Stephen stabilised it: 31fcb7d7dbf0 ("test/alarm: make the test more reliable").
And it was re-added to fast tests.
app/test/test_alarm.c:REGISTER_FAST_TEST(alarm_autotest, true, true,
test_alarm);


-- 
David Marchand