[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#59474: marked as done (dpkg downgrades without warning even though this breaks another package)



Your message dated Tue, 16 Sep 2003 14:17:10 -0400
with message-id <E19zKNy-0002NS-00@auric.debian.org>
and subject line Bug#61156: fixed in dpkg 1.10.11
has caused the attached Bug report to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what I am
talking about this indicates a serious mail system misconfiguration
somewhere.  Please contact me immediately.)

Debian bug tracking system administrator
(administrator, Debian Bugs database)

--------------------------------------
Received: (at submit) by bugs.debian.org; 2 Mar 2000 22:21:41 +0000
Received: (qmail 26215 invoked from network); 2 Mar 2000 22:21:40 -0000
Received: from mail.uni-essen.de (132.252.184.8)
  by master.debian.org with SMTP; 2 Mar 2000 22:21:40 -0000
Received: from Kappa (ascend3-109.extern.uni-essen.de [132.252.241.109])
	by mail.uni-essen.de (8.9.3/8.9.3/4) with SMTP id XAA44984
	for <submit@bugs.debian.org>; Thu, 2 Mar 2000 23:21:38 +0100
From: Sebastian Leske <sleske@informatik.uni-essen.de>
Reply-To: Sebastian.Leske@bigfoot.de
Organization: Universitaet-GH Essen
To: submit@bugs.debian.org
Subject: dpkg downgrades without warning even though this breaks another package
Date: Thu, 2 Mar 2000 21:49:24 +0000
X-Mailer: KMail [version 1.0.28]
Content-Type: text/plain
MIME-Version: 1.0
Message-Id: <00030222182700.00494@Kappa>
Content-Transfer-Encoding: quoted-printable

Package: dpkg
Version: 1.4.0.34=20
Severity: critical

Hi,

I just ran into a bug that almost ruined my day by making my system unboo=
table.

Short summary:

I was able to downgrade from libc6_2.1.2-10 to libc6_2.0.7.19981211-6 wit=
hout
a warning from dpkg, even though I had mount_2.10f-2 installed, which pre=
-depends
on libc6 (>=3D 2.1.2). This (not surprisingly, as I later realized) broke=
 my
whole system, since / could no longer be remounted rw on boot (mount segf=
aulted).

I file this as 'Severity: critical' since it can make a system unusable.


Longer story:

Until a short while ago I had a fully slink system without any new deb's =
from
potato. I then decided to upgrade libc6 and ldso to the potato versions t=
o
be able to install potato deb's (which all depend on libc6 2.1.2).

This worked without problems. Shortly afterwards I installed mount_2.10f-=
2.deb=20
as my first potato package.

Today I tried to downgrade back to the original slink libc6 and ldso, bec=
ause=20
some other program didn't work with libc6_2.1.2-10.deb.

That triggered a catastrophe: When I rebooted after the upgrade, while th=
e
system did boot, I was unable to log in anymore. Luckily I managed to rep=
air=20
things after booting from the slink CD.

Looking back, I realized my mistake: I had downgraded libc6 (from 2.1.2-1=
0 to
the original slink 2.0.7.19981211-6), even though my new mount_2.10f-2.de=
b
has a  'Pre-Depends: libc6 (>=3D 2.1.2) '. This caused mount to stop work=
ing, which
meant root filesystem could not be remounted rw on boot, which broke=20
(almost) everything.

Now, while I realize it's somewhat foolish to downgrad an essential libra=
ry like
libc6, I cannot understand why dpkg let me downgrade libc6 without a warn=
ing!

Transcript:
*******************
  [At this point I just successfully installed libc6_2.1.2-10 and the cor=
responding
   ldso.]
Kappa:/home/sleske# dpkg -i /usr/local/Packages/mount_2.10f-2.deb
(Reading database ... 77609 files and directories currently installed.)
Preparing to replace mount 2.9g-6 (using .../Packages/mount_2.10f-2.deb) =
=2E..
Unpacking replacement mount ...
Setting up mount (2.10f-2) ...
  [Okay, the new mount is now installed, which *NEEDS* libc6_2.1.2-10]
Kappa:/home/sleske# mount /cdrom
  [Now downgrade libc6]
Kappa:/home/sleske# dpkg -i /cdrom/dists/slink/main/binary-i386/base/libc=
6_2.0.7.19981211-6.deb
dpkg - warning: downgrading libc6 from 2.1.2-10 to 2.0.7.19981211-6.
(Reading database ... 77609 files and directories currently installed.)
Preparing to replace libc6 2.1.2-10 (using .../libc6_2.0.7.19981211-6.deb=
) ...
Unpacking replacement libc6 ...
dpkg: warning - unable to delete old file `/usr/share/doc/libc6': Das Ver=
zeichnis ist nicht leer
dpkg: warning - unable to delete old file `/usr/share/zoneinfo': Das Verz=
eichnis ist nicht leer
Setting up libc6 (2.0.7.19981211-6) ...

Kappa:/home/sleske#=20
**********************
As you can see, while dpkg warns me about downgrading in general, it fail=
s to notice
that mount_2.10f-2.deb has a pre-dependency which the downgrading breaks.

Proposed fix: Fix the dependency checking!

I hope I made no mistake here, but to me this seems like a grave bug.
Or is dependency checking deliberately disabled for downgrading?

Greetings,

Sebastian Leske































=20
---------------------------------------
Received: (at 61156-close) by bugs.debian.org; 16 Sep 2003 18:23:44 +0000
>From katie@auric.debian.org Tue Sep 16 13:23:44 2003
Return-path: <katie@auric.debian.org>
Received: from auric.debian.org [206.246.226.45] 
	by master.debian.org with esmtp (Exim 3.35 1 (Debian))
	id 19zKUJ-00029y-00; Tue, 16 Sep 2003 13:23:43 -0500
Received: from katie by auric.debian.org with local (Exim 3.35 1 (Debian))
	id 19zKNy-0002NS-00; Tue, 16 Sep 2003 14:17:10 -0400
From: Adam Heath <doogie@debian.org>
To: 61156-close@bugs.debian.org
X-Katie: $Revision: 1.35 $
Subject: Bug#61156: fixed in dpkg 1.10.11
Message-Id: <E19zKNy-0002NS-00@auric.debian.org>
Sender: Archive Administrator <katie@auric.debian.org>
Date: Tue, 16 Sep 2003 14:17:10 -0400
Delivered-To: 61156-close@bugs.debian.org

Source: dpkg
Source-Version: 1.10.11

We believe that the bug you reported is fixed in the latest version of
dpkg, which is due to be installed in the Debian FTP archive:

dpkg-dev_1.10.11_all.deb
  to pool/main/d/dpkg/dpkg-dev_1.10.11_all.deb
dpkg-doc_1.10.11_all.deb
  to pool/main/d/dpkg/dpkg-doc_1.10.11_all.deb
dpkg_1.10.11.dsc
  to pool/main/d/dpkg/dpkg_1.10.11.dsc
dpkg_1.10.11.tar.gz
  to pool/main/d/dpkg/dpkg_1.10.11.tar.gz
dpkg_1.10.11_i386.deb
  to pool/main/d/dpkg/dpkg_1.10.11_i386.deb
dselect_1.10.11_i386.deb
  to pool/main/d/dpkg/dselect_1.10.11_i386.deb



A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed.  If you
have further comments please address them to 61156@bugs.debian.org,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
Adam Heath <doogie@debian.org> (supplier of updated dpkg package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing ftpmaster@debian.org)


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Format: 1.7
Date: Tue, 16 Sep 2003 12:52:11 -0500
Source: dpkg
Binary: dpkg-doc dpkg dselect dpkg-dev dpkg-static
Architecture: source all i386
Version: 1.10.11
Distribution: unstable
Urgency: low
Maintainer: Dpkg Development <debian-dpkg@lists.debian.org>
Changed-By: Adam Heath <doogie@debian.org>
Description: 
 dpkg       - Package maintenance system for Debian
 dpkg-dev   - Package building tools for Debian
 dpkg-doc   - Dpkg Internals Documentation
 dselect    - a user tool to manage Debian packages
Closes: 8241 37254 61156 91348 106793 111562 112386 115759 126621 127342 131893 133745 137313 150642 151845 153096 154277 154633 155822 157973 160447 161819 163621 165099 167426 169359 169619 170575 173618 174013 174971 175236 175363 175770 177753 179385 183840 191261 191810 196361 198218 199360 200351 203097 203557 203782 204123 204894 206164 206416 207259 208014 208894 210285
Changes: 
 dpkg (1.10.11) unstable; urgency=low
 .
   * dpkg no longer provides any BYHAND files.  If you need the source
     tarball, use the orig.tar.gz file available in the pool.  The
     precompiled tars were hold-overs from when debs were not standard
     ar/tar files.
   * Install /usr/share/doc/dpkg/pseudo-tags, which talks about the bug
     titling we use to mark bugs for filtering in the Debian bts.
   * Updated ca.po.
   * Teach dpkg-source that ~ is legal in version numbers
   * Add build-depends on sgml-base, and increase gettext build-depends to
     0.12.1-3.
   * Updated config.sub and config.guess in cvs.
   * Change use of strncpy in parsedb to memcpy.  Closes: #160447
   * No longer generate helpmsgs.{cc,h}.
   * several controllib changes:
     * Moved dependency parsing logic from dpkg-checkbuilddeps.
       The dpkg-checkbuilddeps parser didn't support empty fields, while
       the controllib parser did.
     * All dependency fields parsed by controllib.pl support [arch]
       specifiers.  If the arch specifier doesn't apply for the current
       arch, then the item will not be added to the internal list structure.
     * Comment lines(that begin with '#') are removed during parsing.
     * Store the original casing of field names, for use in error reporting.
     * Store the original field's casing with 'o:' prepended in %fi.
       * ballombe@merulo.debian.org:
         dpkg-checkbuilddeps now handles empty fields.  Closes: #111562
       * Branden Robinson <branden@debian.org>:
         support comments(first char on line begins with '#') in
         debian/control.  Closes: #191810
       * Bastian Blank <waldi@debian.org>:
         Allow for per-arch generated dependency fields.  Closes: #170575
       * Branden Robinson <branden@debian.org>:
         Report the original casing of field names when they are unknown.
         Closes: #177753
   * Patch dpkg-buildpackage to call debian/rules -qn build-arch, and if
     it's available, modify -B handling appropriately.  If build-arch is not
     available, then when -B was called, do *not* pass -B on to
     dpkg-checkbuilddeps.  Closes: #203097
   * Frank Lichtenheld <frank@lichtenheld.de>:
     Apply patch, to handle missing infodir.bak, and empty files.  Closes:
     #196361
   * Change standard_startup and standard_shutdown into macros.  Closes:
     #200351.
   * Andrew Suffield <asuffield@debian.org>:
     Fix inverted test of eof in parsedb, while looking for eof after
     field name.  Closes: #198218
   * Yann Dirson <dirson@debian.org>
     Add set -e to dpkg-doc.postinst.  Closes: #191261
   * Michael Weber <michaelw@debian.org>:
     dpkg --print-architecture now does gcc -dumpmachine instead of
     --print-libgcc-file-name.  Closes: #131893, #8241, #106793, #210285
   * Daniel Silverstone <dsilvers@digital-scurf.org>:
     Apply patch to change hashing function, and increase bin size, for
     the package database.  Closes: #206416
   * Paul Eggert <eggert@CS.UCLA.EDU>:
     Apply patch to make dpkg work with POSIX 1003.1-2001.  Closes:
     #204894
   * Joey Hess <joeyh@debian.org>:
     Fix "stripping trailing /" message from dpkg-statoverride.  Closes:
     #203782
   * Sami Liedes <sliedes@cc.hut.fi>:
     Increase filesdb bin size, and alter hashing function.  Closes:
     #179385
   * Anthony Towns <aj@azure.humbug.org.au>:
     Retry removing empty dirs during purge.  Closes: #112386
   * Darren Salt <linux@youmustbejoking.demon.co.uk>:
     Add .cvsignore to dpkg-source's default diff ignore regex.  Also fix
     missing \ escape for .deps.  Closes: #174013
   * Koblinger Egmont <egmont@uhulinux.hu>:
     Fix segfault with --status-fd.  Closes: #173618
   * Colin Watson <cjwatson@debian.org>,
     Andrew Suffield <asuffield@debian.org>:
     Add .svn, {arch}, and .arch-ids to dpkg-source -i default regex.
     Closes: #169359
   * Falk Hueffner <falk@debian.org>:
     Added alphaev68-linux-gnu to archtable.  Closes: #199360
   * Elie Rosenblum <fnord@debian.org>:
     Apply patch to support OpenBSD.  Closes: #154277
   * Geoff Richards <qef@laxan.com>:
     Add --config to action section, and add as EXAMPLES section.  Closes:
     #208014, #115759
   * Yann Dirson <dirson@debian.org>:
     install-info --version and --help now display on stdout.  Closes:
     #154633
   * Dan Jacobson <jidanni@dman.ddts.net>:
     Mention what dpkg and dpkg-query -S will not find.  Closes: #153096
   * Sebastian Leske <sleske@enterprise.mathematik.uni-essen.de>:
     Mention in dpkg(8) that downgrading a package can have serious
     consequences.  Closes: #61156
   * Dan Jacobson <jidanni@dman.ddts.net>:
     In dpkg-scanpackages(8), including small note that other tools might
     use the generated Packages file.  Closes: #208894
   * Warren Turkal <wturkal@cbu.edu>:
     Fix some typos in dpkg-source(1).  Closes: #207259, #204123, #167426
   * Dan Jacobson <jidanni@dman.ddts.net>:
     Alter text in dselect(8), implying that other interfaces are available
     for installing packages.  Closes: #206164.
   * Daniel Bonniot <bonniot@users.sourceforge.net>:
     Note in start-stop-daemon(8) that --nicelevel takes an argument.
   * Ruben Porras <nahoo82@telefonica.net>:
     Remove duplicate 'can' in deb-control(5).  Closes: #183840
   * Dan Jacobson <jidanni@dman.ddts.net>:
     Fix short description in dpkg-scansources(8).  Closes: #175770
   * Zefram <zefram@fysh.org>:
     In dpkg(8), move dpkg-query options to their own subsection, like
     dpkg-deb.  Closes: #174971
   * Thomas Hood <jdthood@yahoo.co.uk>
     Add --dry-run and --simulate as synonyms to --no-act.  Closes:
     #165099
   * Junichi Uekawa <dancer@netfort.gr.jp>:
     Fix wrong reference to DEB_BUILD_GNU_SYSTEM on --host in
     dpkg-architecture(1).  Closes: #163621
   * Michael Klein <michael.klein@puffin.lb.shuttle.de>:
     s/Usuaully/Usually/ in deb-control.5.  Closes: #161819
   * Joey Hess <joeyh@debian.org>:
     Removed reference in dselect(8) about recommends not being handled
     well.  Closes: #157973
   * Andrew Suffield <aps100@doc.ic.ac.uk>:
     s/dpkg-source/dpkg-parsechangelog/ in dpkg-parsechangelog -h.
     Closes: #155822
   * Matej Vela <vela@debian.org>:
     Unhighlight "May be repeated multiple times." in dpkg-source(1).
     Closes: #151845
   * Jordi Mallach <jordi@debian.org>:
     Make all actions in dpkg-deb -h lower case.  Closes: #150642
   * Alexander Hvostov <root@aoi.dyndns.org>:
     Add -g|--group to start-stop-daemon, and document properly.  Closes:
     #127342
   * Szerb Tamas <toma@mlf.linux.rulez.org>:
     s/THANKS/THANKS.gz/ in dpkg(8).  Closes: #91348
   * "James R. Van Zandt" <jrv@vanzandt-pc.mitre.org>:
     In dpkg(8), mention running dselect, install after --set-selections.
     Closes: #175236
   * Yann Dirson <dirson@debian.org>:
     Remove disabled --command-fd from help.  Closes: #169619
   * Tobias Burnus <tburnus@physik.fu-berlin.de>:
     Apply patch to update-alternatives that provides --set and --remove-all
     commands.  Closes: #133745.
   * Osamu Aoki <debian@aokiconsulting.com>:
     Add --all to update-alternatives, that calls --config on all
     alternatives.  Closes: #126621
   * Dan Jacobson <jidanni@jidanni.org>:
     Alter --config prompt, to be more clear, that one must 'press' enter to
     accept the default.  Closes: #203557
   * Peter Moulder <pmoulder@bowman.csse.monash.edu.au>:
     Fix some -w issues in update-alternatives.  Closes: #137313
   * Eduard Bloch <edi@gmx.de>:
     Apply patch to check whether the target path exists on removing.
     Closes: #37254
   * Colin Walters <walters@debian.org>:
     After opening files, set binmode.  Closes: #175363
Files: 
 c7e30e4e0bba75b3ec78ab93bfb0d9be 636 base required dpkg_1.10.11.dsc
 223521ddf95001272e4277a3e0c88099 1550456 base required dpkg_1.10.11.tar.gz
 5e7875129eb802280a39bb1e52de6d9b 1127996 base required dpkg_1.10.11_i386.deb
 7109eeca641fe77778b3f09e56b8f83f 94706 base required dselect_1.10.11_i386.deb
 4543b320c5c1812655cf3f21bd0fc1d2 112472 utils standard dpkg-dev_1.10.11_all.deb
 4f180e7a2a5745f9bae370e085a04a08 10638 doc optional dpkg-doc_1.10.11_all.deb

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQE/Z1JSiix9wovUpIkRAhceAJ0TNxmtszpRPDU/0qqrQsYY0+QEvwCgidTM
qem1H1YD1kM6loTFOaMtkyc=
=l3Sv
-----END PGP SIGNATURE-----



Reply to: