In D28645#648051, @glebius wrote:

1. Keep the historical behavior.

LTGM

Would it not better to not setting the callback on shutdown?

Straightforward.

This will result in STP over Vlan (or PVST ala Cisco) if applied to vlan interfaces.
LTGM

I like, that the ugly "reference" paramenter is gone. It was never used.

Is this a problem of the kernel or the laziness of the arp code?

In D28516#644387, @afedorov wrote:

Did you mean D28559?

• Document behavior in the man page.

• Fix man page

• Add documentation in man page

In D28645#641097, @rgrimes wrote:

Infact it is probalby safer to assume a /24 than a /32.

Is the destroy callout guaranteed to run outside (or on the border) of epoch? Can this callout check the refcount and chase to abort itself?

Inverted logic changes are hard to review.

• Reword some comments.
• Fix loopCount setting.

• Fix move logic after rebase to modifications to the parent review.

• Expose error handling.

• New code uses current best practise for constants.

• Sanity checks on input

• rebase after changes

• Rebase after some changes

• Fix use after lost reference.

Compiles fine (backported to stable/13), but causes kernel panic

• Factor out a send and statistic routine

Overcome by D28559
All write access to this variable will be moved to calls under the WRITER lock.

Rebase

In D28545#639301, @rgrimes wrote:

In D28545#639178, @donner wrote:

In D28545#639176, @rgrimes wrote:

In D28545#638947, @donner wrote:

• Get rid of "hent" instead of "host".

This types of change just cause lots of noise in this review that makes it harder to see the real change.

That's why this change is a secondary modification.
Please choose the relevant part in the revision tab "History"

I do not see what your saying, and what is committed is all of what appears here as a single commit, no seperation, and for that mater, not even a mention that you renamed hent -> host in the commit.

In D28545#639176, @rgrimes wrote:

In D28545#638947, @donner wrote:

• Get rid of "hent" instead of "host".

This types of change just cause lots of noise in this review that makes it harder to see the real change.

OTOH it's bad, because it prevents the code from running under const constraints.

• Revert "Switch staleness to counter framework."
• Comment why it's safe to do so.

In D28546#639111, @kp wrote:

I don't know if counter(9) makes sense here.

• Get rid of "hent" instead of "host".