On Tue, Jan 20, 2004 at 10:09:53AM +0100, Ruben de Groot wrote: > On Mon, Jan 19, 2004 at 09:49:19PM +0100, Poul-Henning Kamp typed: > > [...] > > > +# If we do not have a writable /var, create a memory > > +# filesystem for /var. We don't have /usr yet so > ^^^^^^^^^^^^^^^^^^^ > > +# use mkdir instead of touch to test. We want mount > > +# to record its mounts so we have to make sure /var/db > > +# exists before doing the mount -a. > > +# > > + > > +if (/bin/mkdir /var/.diskless 2> /dev/null); then > > + rmdir /var/.diskless > > +else > > + mkdir -p /var > > + mount_md ${varsize:=32m} /var > > +fi > > + > > +# Populate /var if it looks empty > > +if [ -d /var/run -a -d /var/db -a -d /var/empty ] ; then > > + true > > +else > > + /usr/sbin/mtree -deU -f /etc/mtree/BSD.var.dist -p /var > /dev/null 2>&1 > ^^^^^^^^^^^^^^^ > Where did we get /usr ? In rc.d/diskless, we run a mount -a before we get here which would be inappropriate in this case. In most cases we do have /usr since this comes after mountcritlocal which mounts all local file systems. Only the case of a remote /usr seperate from / doesn't work here. I think there's some argument that we should move mtree and possiably newsyslog to /sbin to support the population of /var. -- Brooks -- Any statement of the form "X is the one, true Y" is FALSE. PGP fingerprint 655D 519C 26A7 82E7 2529 9BF0 5D8E 8BE9 F238 1AD4
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:39 UTC