Hi again everyone, 
   I am wondering if the following behavior is.. 

* a bug
* a feature
* a misfeature
...

  :)


first...
violin# mount_union -b /usr/local/java/bin /usr/local/bin
mount_union: Operation not supported
violin# mkdir /tmp/blah
violin# mount_union -b /usr/local/java/bin /tmp/blah
violin# cd /tmp/blah


and more importantly:

cello# mount_union /usr/share/etc /etc
mount_union: Operation not supported
cello# mount_union -b /usr/share/etc /etc
cello# df /etc
Filesystem             1K-blocks     Used    Avail Capacity  Mounted on
<below>:/usr/share/etc    289094   276557    11329    96%    /etc
cello# df /usr
Filesystem          1K-blocks     Used    Avail Capacity  Mounted on
maestro:/export/usr   1035734   264935   719012    27%    /usr


In the first case, I am trying to union mount two directories from the
same NFS imported filesystem together. I can definitely live without this.
(Although it would be kinda nice..)

In the second case, I am trying to union mount an NFS imported directory
above a local directory, to simplify maintenance (by a lot!) I'd like
to have just enough /etc to boot the systems and then union mount a
more explicit /etc.. If that is at all possible.

In both cases, the NFS mounts are read-only (should be no problem with
weird write updates there...?)

NetBSD-current as of 3/17.

=======================

BTW, taking options XNS out of my kernel seems to have fixed those
network crashes, but I am still seeing this...

(89)violin:/usr/local/bin>tcpdump -vv -e host not maestro and host not violin and host not harp and host not flute and host not viola and host not cello
tcpdump: listening on ep1
10:56:39.380709 0:70:a0:4d:0:0 2:0:0:0:45:0 4011 116:
                         3580 83d7 4e02 83d7 4eff 0201 0201 005c
                         33dd 0101 0000 315f b667 0000 0000 7669
                         6f6c 696e 0000 0000 0000 0000 0000 0000
                         0000 0000 0000

Weird stuff..! The Ethernet addresses don't match anything on the network
(which consists of six NetBSD systems without any network options except
for the DARPA Internet things.. INET and GATEWAY (GATEWAY only on the NetBSD
machine that actually *is* the gateway.. it is also running routed). 
(I was told this kind of packet might have something to do with IPX..?
Frankly I am wondering if this isn't the 3C595 bug that people have been
telling me about in email rearing its ugly head?) It's not that important--
doesn't seem to be breaking anything. I'm just curious :)

  Mika