[OE-core] [PATCH V4] dbus: fix ptest failure

[richard.purdie at linuxfoundation.org]

On Thu, 2019-04-18 at 09:59 +0800, Changqing Li wrote:
> 
> On 4/18/19 6:01 AM, Richard Purdie wrote:
> > On Wed, 2019-04-17 at 16:38 +0800, changqing.li at windriver.com
> > wrote:
> > > From: Changqing Li <changqing.li at windriver.com>
> > > 
> > > 1. since one bug in run-ptest, testcase test-bus have never been
> > > actually run (althrough it's result is PASS).
> > > 
> > > After commit 0828850, test-bus can actually run but it
> > > did not install:
> > >   test-service, test-shell-service, test-segfault, and
> > >   dbus-daemon-launch-helper-test
> > > Add the configure flag:
> > >   --enable-embedded-tests
> > > to generate binary dbus-daemon-launch-helper-test, then install
> > > them so that test-bus will now pass.
> > > 
> > > 2. fix testcase test-dbus-daemon failed
> > > we enable --enable-verbose-mode in recipe dbus-test, and don't
> > > enable it in recipe dbus. This will make below test code get
> > > unexpect result of have_verbose and assert.
> > > disable --enable-verbose-mode for recipe dbus-test to fix it.
> > > 
> > >  #ifdef DBUS_ENABLE_STATS
> > >   g_assert_true (have_stats);
> > >  #else
> > >   g_assert_false (have_stats);
> > >  #endif
> > > 
> > > Signed-off-by: Changqing Li <changqing.li at windriver.com>
> > > ---
> > >  meta/recipes-core/dbus/dbus-test_1.12.12.bb | 13 ++++++++++---
> > >  meta/recipes-core/dbus/dbus/run-ptest       | 16 ++++++++++++---
> > > -
> > >  2 files changed, 22 insertions(+), 7 deletions(-)
> > 
> > I'm wondering if some of these tests were intentionally not run due
> > to
> > the length of time they take? It takes the dbus-ptest time from 26s
> > to
> > 250+s. Can you see which test that is and why its taking so long
> > please?
> 
> test-bus take most of the time, it include several sub tests. My test
> result:
> all test runed: 
> real    2m59.637s
> user    0m59.494s
> sys    0m41.952s
> skipped test-bus:
> real    0m13.125s
> user    0m12.230s
> sys    0m0.162s
> 
> Previously we don't intentionally skip test-bus,  just because we had
> a bug in run-ptest before,
> so test-bus  accidentally not run,  just return PASS. 
> After below commit fix this bug, test-bus can be runned.
> https://git.openembedded.org/openembedded-core/commit/?id=0828850fd09f738572ae8259384af07eeb81182b
> 
> -for i in `ls test/test-*`; do ./$i ./test/data
> DBUS_TEST_HOMEDIR=./test >/dev/null; output; done
> 
> "DBUS_TEST_HOMEDIR=./test" will take as an argument,  make no sub
> test is runned. 


Ross, do you remember if we intentionally skip this dbus test for
taking around 200s, taking the overall test time from 20s to 220s?

I'm strongly tempted to explicitly disable this test unless someone can
convince me it tests someting critical...

Cheers,

Richard