At Fri, 19 Jun 2009 20:15:32 -0400, Matthew Mondor
<mm_lists%pulsar-zone.net@localhost> wrote:
Subject: Re: FreeBSD devfs support on NetBSD 5.0
>
> I wrote down a few ideas yesterday night about implications of an
> implementation. I unfortunately doubt that I can actively work on
> this, but if these notes can serve as inspiration or as the start of a
> technical discussion on the subject, here they are:
>
>
> <URL:http://cvs.pulsar-zone.net/cgi-bin/cvsweb.cgi/mmondor/netbsd/devfs_ideas.txt?rev=1.1;content-type=text%2Fplain>
>
Holy Complexity Batman!
Personally I think local permissions and ownership changes, symbolic
links, and the like need to be supported in a completely transparent
way, but that's almost trivial to do with a simplified type of unified
filesystem overlay mount, one which stores only meta-data overrides for
underlying files (or almost like an extension to the umapfs, one which
stores the new meta-data). Such filesystem layering is already well
supported and relatively well understood. With this kind of mechanism
there are no configuration files whatsoever to mess with (especially not
any requiring in-kernel parsers or whatever) to store such information,
or any other non-standard or non-intuitive aspect to it whatsoever. You
could still grab all this meta-data from the union FS by simply
collecting it with standard filesystem management tools, such as "pax"
(so long as they have some way to restrict their examination to only the
top layer). I.e. a union overlay filesystem which is basically
meta-data only is all you need -- that is your "configuration file".
The code is almost all already written.
Other than that devfs should simply be a live representation as a
filesystem of the autoconf tree. Device files appear if they are
configured, and disappear if they are not, dynamically if the bus and
device drivers support such changes.
--
Greg A. Woods
Planix, Inc.
<woods%planix.com@localhost> +1 416 218-0099 http://www.planix.com/
Attachment:
pgpN9_q2pMXqB.pgp
Description: PGP signature