Thanks for fixing!
I've been running the kernel that has exactly the same fix and confirmed the problem I had was fixed.

It seems that the condition is inversed by this change.
Probably the key need to be deleted if it is set?

I will retract this since it does not completely fix the problem and I have not come up with a good fix so far.

Replace epoch_wait_preempt() with NET_EPOCH_WAIT()

@kp Thanks for the comment.
I don't know the detail, but the "NOTES" in ping.c states:

• Add lltable_delete_conditional() instead of lltable_flags_free().
• Use lltable_delete_conditional() for purging in6_multi for solicited node multicast addresses.
• Move out the sockaddr_dl setup code from nd6_ns_input() to a new function.

• Add the lltable.llt_flags member and LLT_ADDEDPROXY flag in it
• Add lltable_flags_free() that frees LLEs with specified la_flags
• Use lltable_flags_free() for freeing in6_multis in if_detach path
• Add the test script

Add the spare member to indicate explicit alignment.

In D35307#800525, @melifaro wrote:

Could you also add a test, verifying the actual behaviour? It should be relatively straight-forwarded, based on tests/sys/netinet6/forward6.sh (2 VNETs) && tests/sys/netinet6/ndp.sh (general NDP checks),

Fix spell check warnings

• changed the "mape" nvlist structure
• updated the test for the review comments

• moved to the new nvlist-based implementation
• removed timeout in the test
• changed the manual page that map-e-portset is for MAP-E CE (avoid confusion with MAP-E BR)

In D29468#665332, @kp wrote:

Thinking about this some more, I think that what we want to do here is to introduce a pfctl_pool, for use in libpfctl / pfctl. Similar to how we already have struct pfctl_rule and struct pfctl_anchor.

@kp thanks for notifying the changes.
I'd like to hear your idea for the following points to rebase the code to the nvlist-based implementation.

• rebased the patch from 12-STABLE to -current
• added the test script
• fixed 2 issues