On Wed, Nov 27, 2019 at 3:16 PM Van Haaren, Harry
<harry.van.haa...@intel.com> wrote:
>
> > -----Original Message-----
> > From: Aaron Conole <acon...@redhat.com>
> > Sent: Wednesday, November 27, 2019 2:10 PM
> > To: Van Haaren, Harry <harry.van.haa...@intel.com>
> > Cc: dev@dpdk.org
> > Subject: Re: [PATCH] test/service: fix wait for service core
> >
> > Harry van Haaren <harry.van.haa...@intel.com> writes:
> >
> > > This commit fixes a sporadic failure of the service_autotest
> > > unit test, as seen in the DPDK CI. The failure occurs as the main test
> > > thread did not wait on the service-thread to return, and allowing it
> > > to read a flag before the service was able to write to it.
> > >
> > > The fix changes the wait API call to specific the service-core ID,
> > > and this waits for cores with both ROLE_RTE and ROLE_SERVICE.
> > >
> > > The rte_eal_mp_wait_lcore() call does not (and should not) wait
> > > for service cores, so must not be used to wait on service-cores.
> > >
> > > Fixes: f038a81e1c56 ("service: add unit tests")
> > >
> > > Reported-by: Aaron Conole <acon...@redhat.com>
> > > Signed-off-by: Harry van Haaren <harry.van.haa...@intel.com>
> > >
> > > ---
> >
> > It might also be good to document this behavior in the API area. It's
> > unclear that the lcore wait function which takes a core id will work,
> > but the broad wait will not.
>
> Yes agreed that docs can improve here - different patch.
>
>
> > > Given this is a fix in the unit test, and not a functional change
> > > I'm not sure its worth backporting to LTS / stable releases?
> > > I've not added stable on CC yet.
> >
> > I think it's worth it if the LTS / stable branches use the unit tests
> > (otherwise, they will observe sporadic failures).
>
> Ok, I've added sta...@dpdk.org on CC now
>
>
> > > app/test/test_service_cores.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/app/test/test_service_cores.c b/app/test/test_service_cores.c
> > > index 9fe38f5e0..a922c7ddc 100644
> > > --- a/app/test/test_service_cores.c
> > > +++ b/app/test/test_service_cores.c
> > > @@ -483,7 +483,7 @@ service_lcore_en_dis_able(void)
> > > int ret = rte_eal_remote_launch(service_remote_launch_func, NULL,
> > > slcore_id);
> > > TEST_ASSERT_EQUAL(0, ret, "Ex-service core remote launch failed.");
> > > - rte_eal_mp_wait_lcore();
> > > + rte_eal_wait_lcore(slcore_id);
> > > TEST_ASSERT_EQUAL(1, service_remote_launch_flag,
> > > "Ex-service core function call had no effect.");
> >
> > Should we also have some change like the following (just a guess):
> >
> > diff --git a/app/test/test_service_cores.c b/app/test/test_service_cores.c
> > index 9fe38f5e08..695c35ac6c 100644
> > --- a/app/test/test_service_cores.c
> > +++ b/app/test/test_service_cores.c
> > @@ -773,7 +773,7 @@ service_app_lcore_poll_impl(const int mt_safe)
> >
> > /* flag done, then wait for the spawned 2nd core to return */
> > params[0] = 1;
> > - rte_eal_mp_wait_lcore();
> > + rte_eal_wait_lcore(app_core2);
> >
> > /* core two gets launched first - and should hold the service lock */
> > TEST_ASSERT_EQUAL(0, app_core2_ret,
>
>
> I reviewed this usage of the function, and I believe it waits on application
> cores (aka, ROLE_RTE, not ROLE_SERVICE). Hence this usage is actually correct.
> Please review and double check my logic though - more eyes is good.
It seems to be the case, yes.
My overall feeling is that the services stuff is a giant hack, so
better documentation will prove me wrong :-).
As I said I am for taking this change in 19.11 now, as it only impacts
this test and it seems to solve the random failures.
Acked-by: David Marchand <david.march...@redhat.com>
--
David Marchand
