On Mon, Oct 18, 2021 at 02:38:13PM +0200, Florian Westphal wrote:
> Rework the reproducer for the vrf+conntrack regression reported
> by Eugene into a selftest and also add a test for ip masquerading
> that Lahav fixed recently.
>
> With net or net-next tree, the first test fails and the latter
> two pass.
>
> With 09e856d54bda5f28 ("vrf: Reset skb conntrack connection on VRF rcv")
> reverted first test passes but the last two fail.
>
> A proper fix needs more work, for time being a revert seems to be
> the best choice, snat/masquerade did not work before the fix.
Applied, thanks.
> Link: https://lore.kernel.org/netdev/378ca299-4474-7e9a-3d36-2350c8c98995@xxxxxxxxx/T/#m95358a31810df7392f541f99d187227bc75c9963
> Reported-by: Eugene Crosser <crosser@xxxxxxxxxxx>
> Cc: Lahav Schlesinger <lschlesinger@xxxxxxxxxxxxx>
> Signed-off-by: Florian Westphal <fw@xxxxxxxxx>
> ---
> Lahav, it would be nice if you could check if this
> is a correct test of the scenario that you fixed with
> 09e856d54bda5f28. Thanks!
No news from Lahav.
