Subject: Re: Obsolete files?
To: Martin S. Weber <Ephaeton@gmx.net>
From: Quentin Garnier <cube@cubidou.net>
List: current-users
Date: 07/23/2007 17:53:23
--J5MfuwkIyy7RmF4Q
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Mon, Jul 23, 2007 at 11:14:04AM +0200, Martin S. Weber wrote:
> Moin,
>=20
> On Mon, Jul 23, 2007 at 04:05:22AM +0200, Quentin Garnier wrote:
> > On Mon, Jul 23, 2007 at 02:15:16AM +0200, Johnny Billquist wrote:
> > > Why is postinstall listing old share libraries? I don't think it's a=
=20
> > > good idea to remove old shared libraries.
> > > Can anyone tell me why it's a good idea?
> > > Just upgraded one machine from 4.99.19 to 4.99.24, and among others,=
=20
> > > libc.so.12.150 came up on my list...
> >=20
> > It won't remove lib<lib>.major, but removing lib<lib>.major.minor is
> > fine.
> >=20
>=20
> I'm an idiot, I know, but when I installed 4.0 beta2 on a machine
> previously running 4.99.17, I killed my libc.12.149 with=20
> postinstall obsolete and machine was borked, then... I don't
> remember what "other" libc was still there; but even the tools
> in /rescue complained (probably by calling something else - I
> couldn't use /rescue/shutdown for example [wall failed?]).

That's a different kind of issue.  Newer (static) binaries will fail
because the kernel doesn't provide the interfaces they want.  The same
goes for newer dynamic libraries:  the libc.so.12 they find, which has
been linked to the 4.0_BETA2 one by the "upgrade" process, doesn't have
the symbols it needs.

> ...which is surprising because they should come from one of
> the .tgz's
> ...whose contents should not depend on an "obsolete" lib ??

The thing is, the name lib<lib>.so.<M>.<m> is not registered anywhere.
Dynamic binaries only depend on the major number, so what libc.so.12
points to is what matters.  But then again, if the kernel doesn't have
the back-end for the newer symbols, binaries will still fail.

Downgrading is not an easy process, you have to be very careful.

--=20
Quentin Garnier - cube@cubidou.net - cube@NetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.

--J5MfuwkIyy7RmF4Q
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (NetBSD)

iQEVAwUBRqTO89goQloHrPnoAQJobAf9Ed40LhSMgimKPm0fgn8QlaPN8wBX2IMy
t8mYanbmYAvCMxJABay5l3lVCTairCmaGnwD9R4V94blo6n3JnBtwE2dIDXViYqb
LXCYQ3RKLaaZbweOCyWs0ow+ab2rasYoYESHQWXET5ZA8iZ6PFXvwQwVNRfnj+Pi
yFe/7hTQwXJZe2S6YK5htpbx7u92zzPusG3NQQvBhjlCGyA8VZp9tdiiRF09DfSU
8FplpiCqj+56n8C752KWYDN4l3E3V4pqMAXkAl73syO4THWJttK8jEcYFGasethY
89cr5Mwy6Pe1minU9bhNu0jNCcECMLjsq85Mvxxqvy8R9XChpd8JBQ==
=Rk8w
-----END PGP SIGNATURE-----

--J5MfuwkIyy7RmF4Q--