On Wed, Nov 26, 2014 at 01:18:53PM -0500, Stefan Berger wrote:
On 11/26/2014 11:53 AM, Eric Blake wrote:
> On 11/26/2014 09:26 AM, Stefan Berger wrote:
>> Some of the nwfilter tests are now failing since --concurrent shows
>> up in the ebtables command. To avoid this, implement a function
>> preventing the probing for lock support in the eb/iptables tools
>> and use it in the tests.
> Now that I've read Martin and Prerna's exchange, I'm wondering if we
> should instead make this override force the locking flags ON, and adjust
> the expected test output to expect the -w/--concurrent.
>
Not sure... this patch is definitely 'simpler' and that one call we add
to virsocket.h would only to be called by test cases. What we would test
if we considered the --concurrent and -w flag is basically not much,
just that some probing happened.
I must say I agree with Stefan (or at least for now). I started
working on the second variant (forcing the flag "on" for tests), and
not only I've found out that the tests themselves (not talking about
the outputs) are not adapted to that, but even though it's still not
fully fixed, I already have a diff of 1051(+), 1014(-).
I can finish it or at least share somewhere if someone else wants to
continue.
Martin