On Thu, Dec 08, 2016 at 06:50:32PM -0800, Daniele Di Proietto wrote:
> The order of the options in the packet generated by ovs-controller
> depends on the hash function.  I believe that murmur hash (our default)
> produces different outputs depending on the endianness of the system.
> 
> This commit fixes the test by reordering the options in the packet
> before checking them.
> 
> This was reported before as a failure of the test on x86 with sse42
> enabled, I'm fixing it now because it affects other targets build by
> distros by default.
> 
> Reported-at: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=840770
> Signed-off-by: Daniele Di Proietto <[email protected]>

I agree that this would solve the problem.

I think that it might be nicer to instead sort the options in
build_dhcpv4_action() and build_dhcpv6_action() in ovn-northd.c.  Then
the generated logical actions won't differ from one architecture to
another, so that if we one day push a test to a higher level we won't
have an underlying source of differences.  There's even a helper
smap_sort() that could do most of the work.

Are you willing to make that change?

Thanks,

Ben.
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to