Re: kern/55164 (net/ndp/t_ndp:ndp_rtm test case fails again)

[kre%NetBSD.org@localhost]

Synopsis: net/ndp/t_ndp:ndp_rtm test case fails again

Responsible-Changed-From-To: christos->kre
Responsible-Changed-By: kre%NetBSD.org@localhost
Responsible-Changed-When: Sun, 12 Apr 2020 06:20:00 +0000
Responsible-Changed-Why:
I have been looking into this one...

One thing for certain is that none of the commits mentioned in
the PR is relevant - those were the ones that fixed the
"setsockopt on non-socket" issue, but in the process broke
twice as many tests as was fixed.

This test (net/t_ndp/ndp_rtm) was one of those that broke.

Those were subsequently fixed by these commits:

commit 2020.04.03.16.20.51 martin src/usr.bin/netstat/Attic/prog_ops.h 1.4
commit 2020.04.03.16.20.51 martin src/usr.bin/netstat/netstat_hostops.c 1.3
commit 2020.04.03.16.20.51 martin src/usr.bin/netstat/netstat_rumpops.c 1.3
commit 2020.04.03.16.20.51 martin src/usr.bin/netstat/prog_ops.h 1.4
commit 2020.04.03.16.20.52 martin src/sbin/route/prog_ops.h 1.5
commit 2020.04.03.16.20.52 martin src/sbin/route/route_hostops.c 1.3
commit 2020.04.03.16.20.52 martin src/sbin/route/route_rumpops.c 1.3

which unified netstat & route more than they had before, and in the
process seems to have lost printing the AUTHOR info (flag and addr)
when present.

I am still testing to confirm that the info is actually in the RTM_MISS
packet when it is sent to the route command, and is just not being
printed (rather than being missing completely in the msg from the kernel).

I have confirmed that (at least directly) rump is not involved in this
issue, when I do the equivalent commands on real hardware (actually
configuring the ULA, and pinging a non-existant ULA (which would be
on the same link as the source - just as the test does - and with
"route monitor" running) I don't see the author info in that case either.

Expect more soon...