Date: Wed, 9 Oct 2013 08:06:30 -0500 From: Brooks Davis <brooks@freebsd.org> To: Thomas Mueller <mueller6727@bellsouth.net>, freebsd-current@freebsd.org Subject: Re: Unable to installworld after successful buildworld and buildkernel Message-ID: <20131009130630.GB71891@lor.one-eyed-alien.net> In-Reply-To: <20131009055213.GM56872@funkthat.com> References: <3F.9A.16944.166A3525@cdptpa-oedge02> <20131008141501.GA69028@lor.one-eyed-alien.net> <28.45.16944.64B74525@cdptpa-oedge02> <20131008214809.GC69269@lor.one-eyed-alien.net> <44.C1.16944.5E6C4525@cdptpa-oedge02> <20131009055213.GM56872@funkthat.com>
next in thread | previous in thread | raw e-mail | index | archive | help
--jRHKVT23PllUwdXP
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Tue, Oct 08, 2013 at 10:52:13PM -0700, John-Mark Gurney wrote:
> Thomas Mueller wrote this message on Wed, Oct 09, 2013 at 03:00 +0000:
> > > > But where is DB_FROM_SRC documented? I never saw it anywhere. UPD=
ATING file needs to be updated for installing FreeBSD on a partition where =
there is no OS installed.
> >=20
> > > It's documented in Makefile.inc1 where the less commonly used options
> > > are documented. It should be documented in the Handbook along with t=
he
> > > process of cross installation and image creation.
> >=20
> > > I disagree that UPDATING should document the process of installing to
> > > other locations. That would be a distraction from it's purpose of
> > > letting users update their systems. It's already far too complicated.
> >=20
> > -- Brooks
> >=20
> > UPDATING file has a section on cross-installing current into a separate=
partition.
> >=20
> > This part ought to be complete, including DB_FROM_SRC, rather than leav=
ing a user lost at sea when a new user (unbound) sneaks in.
> >=20
> > UPDATING documented when user auditdistd was added, so why not for unbo=
und?
>=20
> How does this look:
> Index: UPDATING
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> --- UPDATING (revision 256024)
> +++ UPDATING (working copy)
> @@ -1899,7 +1899,7 @@
> make buildkernel KERNCONF=3DYOUR_KERNEL_HERE
> <maybe newfs current's root partition>
> <mount current's root partition on directory ${CURRENT_ROOT}>
> - make installworld DESTDIR=3D${CURRENT_ROOT}
> + make installworld DESTDIR=3D${CURRENT_ROOT} -DDB_FROM_SRC
> make distribution DESTDIR=3D${CURRENT_ROOT} # if newfs'd
> make installkernel KERNCONF=3DYOUR_KERNEL_HERE DESTDIR=3D${CURREN=
T_ROOT}
> cp /etc/fstab ${CURRENT_ROOT}/etc/fstab # if n=
ewfs'd
>=20
>=20
> I tested distribution and it appears not to need it.
Distribution could easily need it in the future so I'd suggest adding in
there as well. If nothing else, if building on a box with a highly
non-standard /etc/passwd or /etc/group the results could be bizzare and
inconsistant if -DDB_FROM_SRC is not used in both cases. installkernel
is probably safe since people who renumber root/wheel will get what they
deserve if they don't audit all of the make infrastructure.
-- Brooks
P.S. This probably shouldn't be documented because it is dangerous in
some situations: but for administrators who keep their passwd and group
files entirely in sync with FreeBSD's you can skip the whole mergemaster
-p annoyance by specifying DESTDIR=3D/ -DDB_FROM_SRC.
--jRHKVT23PllUwdXP
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (FreeBSD)
iD8DBQFSVVTVXY6L6fI4GtQRAnYEAKC10EJEHdA8tg52t27mzzZELsMK+QCbBu8H
phcxRxcAuYm6Y3G+cPo+u78=
=xalx
-----END PGP SIGNATURE-----
--jRHKVT23PllUwdXP--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20131009130630.GB71891>