On Thu, Dec 18, 2014 at 09:50:34AM +0000, Maxim Uvarov wrote: > On 12/18/2014 12:42 PM, Stuart Haslam wrote: > > On Thu, Dec 18, 2014 at 03:22:29AM +0000, Bill Fischofer wrote: > >> > >> On Wed, Dec 17, 2014 at 6:30 PM, Stuart Haslam > >> <[email protected]<mailto:[email protected]>> wrote: > >> On Wed, Dec 17, 2014 at 10:41:52PM +0000, Maxim Uvarov wrote: > >>> Signed-off-by: Maxim Uvarov > >>> <[email protected]<mailto:[email protected]>> > >>> --- > >>> test/validation/odp_pktio.c | 105 > >>> ++++++++++++++++++++++++++++++++++++++++---- > >>> 1 file changed, 96 insertions(+), 9 deletions(-) > >>> > >>> diff --git a/test/validation/odp_pktio.c b/test/validation/odp_pktio.c > >>> index 0ba9938..02c4a7b 100644 > >>> --- a/test/validation/odp_pktio.c > >>> +++ b/test/validation/odp_pktio.c > >>> @@ -398,6 +398,90 @@ static void test_odp_pktio_sched_multi(void) > >>> pktio_test_txrx(ODP_QUEUE_TYPE_SCHED, 4); > >>> } > >>> > >>> +static void pktio_test_mtu(void) > >>> +{ > >>> + int i; > >>> + int ret; > >>> + int def; > >>> + odp_pktio_t pktio = create_pktio(iface_name[0]); > >>> + > >>> + printf("testing mtu for %s\n", iface_name[0]); > >> This printf isn't needed, cunit tells you which test is being run, same > >> with the others below. > >> > >>> + > >>> + def = odp_pktio_mtu(pktio); > >>> + CU_ASSERT(def > 0); > >>> + > >>> + for (i = 9000; i > 100; i /= 2) { > >> I don't think support for jumbo frames should be mandatory. > >> > >> > >> I can't think of any 1Gb or higher NIC/SoC that doesn't support jumbo > >> frames. This is pretty standard for higher data rate I/O these days, so > >> no harm in verifying that here. > >> > > I agree this is pretty standard on a "real" interface, but I was thinking > > more about on a development platform. The interface on my TC2 doesn't > > support >1500, nor does my beaglebone black, does that mean they can't > > support ODP? Obviously they aren't really target platforms but they > > would pass all other unit tests. > > > > This test isn't really intended to check which MTU values are supported, > > just that the API allows it to be configured. Without some other API to > > discover the capabilities of the interface (which we do need) I think > > it's better to stay within the safe zone <=1500, or change the behaviour > > above 1500 to not fail an assertion. > > > I can use <= 1500 for CU_ASSERT and print other values just in test output > if they are supported or not. Does it make any sense? > > Maxim. >
Yes, makes sense. -- Stuart. _______________________________________________ lng-odp mailing list [email protected] http://lists.linaro.org/mailman/listinfo/lng-odp
