On 25 February 2016 at 06:15, Ilya Maximets <i.maxim...@samsung.com> wrote: > This prevents failure of test scenarios 'make check-kernel' > and 'make check-system-userspace' if creation of veth pairs is not > supported by kernel (or module isn't loaded). > > Signed-off-by: Ilya Maximets <i.maxim...@samsung.com>
All of the tests require support for namespaces and veths, so I think the main use for this patch is to change the "fail" for all tests to "skip" if you're running the testsuite in a setup without veths. Rather than adding an extra macro which must be specified for every test, I suggest that this should be integrated into either OVS_TRAFFIC_VSWITCHD_START() or ADD_VETH(). If I'm not mistaken, this should provide the same result with a 1-liner patch. > --- > tests/system-common-macros.at | 9 +++++++++ > tests/system-traffic.at | 45 > +++++++++++++++++++++++++++++++++++++++++++ > 2 files changed, 54 insertions(+) > > diff --git a/tests/system-common-macros.at b/tests/system-common-macros.at > index fcd7792..107900a 100644 > --- a/tests/system-common-macros.at > +++ b/tests/system-common-macros.at > @@ -155,3 +155,12 @@ m4_define([NETNS_DAEMONIZE], > # Do basic check for vxlan functionality, skip the test if it's not there. > m4_define([OVS_CHECK_VXLAN], > [AT_SKIP_IF([! ip link add foo type vxlan help 2>&1 | grep dstport > >/dev/null])]) > + > +# OVS_CHECK_VETH() > +# > +# Do basic check for veth functionality, skip the test if it's not there. > +m4_define([OVS_CHECK_VETH], > + [ADD_NAMESPACES(at_vethtest) > + AT_SKIP_IF([! ip netns exec at_vethtest ip link add type veth peer]) > + ] > +) You shouldn't need to add a namespace to check whether you can create a veth. I don't think any of the tests actually adds a veth inside a namespace, either. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
