Re: netbsd-7 kernel and older vnconfig(8)

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Mon, 9 Nov 2015 10:08:38 -0500
    From:        christos%zoulas.com@localhost (Christos Zoulas)
    Message-ID:  <20151109150838.1594C17FDAB%rebar.astron.com@localhost>

  | We'll need to keep track of the maximum index currently in use,

In the kernel we (you NetBSD developers) already do, not that it really matters.

  | because if you destroy vnds in the middle, you get ENXIO too.

The current kernel code for VNDIOCGET (used on an opened vnd or course, which
means it exists) never returns ENXIO for anything.   That was the problem
the /dev reading was designed to overcome.   All that is needed is to put
the ENXIO back in the limiting case, and problem solved.

In general, if a kernel change (any kernel change) breaks some existing
userland program, it is *always* the wrong solution to "fix" the program.
ind a way to fix the kernel instead.

What would happen with a new vndconfig(8) running on an old kernel (netbsd-6
or something) I don't know, but nor do I care, that kind of compat isn't
required.

Someone with an environment where the fix can be tested should just try it
and see what happens....

kre