In message <20030823044218.BDDB891@coconut.itojun.org>
Jun-ichiro itojun Hagino writes:

>> That said: I'm assuming it is OK to add m_tag_delete_nonopersistent(),
>> to call it before reflecting an ICMP packet, and that FreeBSD (at
>> least) will likely do so, too?
>
>	no.  it is not decided that we introduce m_tag_delete_nonopersistent()
>	API, so it is inappropriate for you to add them now.


Really? I see a unanimous consensus to drop [all NetBSD] tags, in the
specific case of icmp-reflection. Also in the general case, whenever
an in-kernel protocol re-uses an inbound mbuf as an outbound packet,
instead of allocating a new chain for the outbound packet.

Nobody else objects to m_tag_delete_nonpersistent(). It helps maintain
consistency, going forward, with other *BSDs, who do need it.

Unless there's a substantive objection, it seems like time to update
the comment (referencing actual users of persistent tags), then commit
the change.

Or (since there's a genuine problem with if_loop) should I commit an
m_tag_delete(m, NULL) to the icmp code now, then change it to
m_tag_delete_nonpersistent() later, as-and-when?