APT 2.8.4+: Promote weak key warnings to errors

[Julian Andres Klode]

(2.8.3 update references this bug as previous uploads tried to fix it and have not been fully reverted, per discussion with rbasak we agreed to just ignore this bug for SRU verification; as in mark it verification-done, and then re-open the bug after the package migrated

For the discussion, see https://irclogs.ubuntu.com/2024/10/16/%23ubuntu-devel.html
)

⚠️ Only land this in the release/updates pocket after PPAs have been resigned

(This bumps the apt version to 2.8.0. APT uses the odd/even number system, with 2.7.x being the development series for 2.8, and this is the only change left for the 2.8 release, save for some minor translation/test suite improvements)

(Please also look at bug 2073126 for the follow-up changes to mitigate regression potential).

[Impact]
APT is currently just warning about keys that it should be rejecting to give Launchpad time to resign PPAs. This needs to be bumped to an error such that the crypto policy is fully implemented and we only trust keys that are still being trusted. #2055193

A warning provides some help right now to third-parties to fix their repositories, but it's not *safe*: A repository could have multiple signing keys and be signed by a good key now, then later, a previous key still in trusted.gpg.d could be revoked and we'd degrade to warnings, which, given that we update in the background automatically, the user may not see.

Other fixes:
- The test suite has been made less flaky in two places
- Documentation translation has been unfuzzied for URL changes in 2.7.14

[Test plan]
The vast regression test suite prevents regression in other components.

Test Case A: Existing noble system (warning)

0. Update an existing noble container to the new APT
1. Observe/etc/apt/apt.conf.d/00-temporary-rsa1024 being created
2. Add a PPA with an old 1024-bit signing key
3. Run apt update
4. Observe that the PPA is updated with a warning

Test Case B: New noble system (error)

0. Bootstrap a new noble system including apt from proposed (using e.g. mmdebstrap)
1. Observe NO /etc/apt/apt.conf.d/00-temporary-rsa1024
2. Add a PPA with an old 1024-bit signing key
3. Run apt update
4. Observe that the PPA is not updated, but the other repositories are

Test Case C: mantic -> noble (error)

0. Upgrade mantic to noble w/ apt from proposed, observe behavior as in B

Test Case D: jammy -> noble (error)

0. Upgrade jammy to noble w/ apt from proposed, observe behavior as in B

[Where problems could occur]
apt will start to fail updates of repositories with weak signing keys, but it will have warned users about that before. Given that it is still early in the cycle, and we only enable updates from 22.04.1 for 24.04.1, this seems the right tradeoff for future security.

[Steve Langasek (vorlon)]

> ⚠️ Only land this in the release pocket after PPAs have been resigned

I am marking this incomplete, as it is unclear to me what the intended timeline is for the resigning, or if your comment about the release pocket should also apply to -updates.

[Julian Andres Klode (juliank)]

The same caveat applies to -updates, but there is a question of whether we should ship 2.8.0 as this or make 2.8.0 different, I did not push a tag for it yet.

i.e. given that this is a stable release update that will break PPAs users currently have warnings for, it might make sense to make it break that a couple months down the road after we have a transition period, i.e. we can "timebomb" things by making apt treat the weak keys as expiring in August (August because we really want this sorted out by the point release when the big wave of 22.04 users upgrades).

We could also introduce a new version of software-properties-common that adds PPA key refresh. We should then trigger that by apt postinst, or in the software-properties-common postinst. I do not believe we need to enforce a strict ordering relationship here, so 2.8.0 as is should technically be good to go.

It's understandable that breaking existing repositories in a stable release is not optimal, however the warnings don't work as a security mechanism - we do not show you which weak keys are trusted, just which weak keys were used to sign the repository:

Hence if you have a 1024R key and a 4096R that can sign a repository, but it's signed by the 4096R key now, you don't see the 1024R key, and an attacker could resign the repository with it and silently attack you.

So this is something we do need to target for the first point release; we want users upgrading from 22.04 to not end up in the transitional stage where they have warnings.

[Steve Langasek (vorlon)]

Hello Julian, or anyone else affected,

Accepted apt into noble-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/apt/2.8.0 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-noble to verification-done-noble. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-noble. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

[Ubuntu SRU Bot (ubuntu-sru-bot)]

All autopkgtests for the newly accepted apt (2.8.0) for noble have finished running.
The following regressions have been reported in tests triggered by the package:

apport/2.28.1-0ubuntu2 (ppc64el)
auto-apt-proxy/14.1 (arm64)
autopkgtest/5.34ubuntu2 (amd64, arm64, ppc64el)
cron/3.0pl1-184ubuntu2 (s390x)
gcc-12/12.3.0-17ubuntu1 (armhf)
sbuild/0.85.7 (amd64, armhf, i386, s390x)
update-manager/1:24.04.6 (amd64, arm64, armhf, ppc64el, s390x)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/noble/update_excuses.html#apt

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

[Julian Andres Klode (juliank)]

Removing the block-proposed tag for oracular

[Launchpad Janitor (janitor)]

This bug was fixed in the package apt - 2.9.3

---------------
apt (2.9.3) unstable; urgency=medium

  [ Julian Andres Klode ]
  * Initial implementation ("alpha") of the 3.0 solver:
    This new solver is available using the --solver 3.0 option.
    Highlights:
    - Fully backtracking solver, think DPLL without pure literal elimination
    - Manually installed packages are not offered up for removal
    - New --no-strict-pinning option allows APT to fallback to non-candidate
      versions, e.g. apt install --no-strict-pinning foo/experimental installs
      foo from experimental and will switch dependencies where needed.
    - Autoremove is more aggressive and only keeps the strongest automatically
      installed package. For example, gcc-<version> will now be offered for
      removal and no longer kept around due libtool Depends: gcc | c-compiler
      and gcc-<version> Provides: c-compiler, as `gcc` is already satisfied.
    Caveats right now:
    - Test suite is not yet passing
    - The list of automatically removable packages is not displayed
      when automatically installed packages are not removed
    - Error information gets lost on backtracking (see Debug::APT::Solver=2)
    - Error information is just rendered as A -> B implication graphs,
      with some nodes perhaps containing a "not".
    - The logic for replacing obsolete manually installed packages with
      new replacement packages (think Conflits/Replaces/Provides) is not
      yet implemented.
    - Conflict-driven clause learning is not implemented, so backtracking
      is technically pretty inefficient.
  * Solver3 integration fixes:
    - test: Ignore progress output in comparing output..
    - test-allow-scores-for-all-dependency-types: Adjust for solver3
    - EDSP: Add "solver3" alias for apt-internal-solver
  * UI work:
    - Highlight essential removals with action::remove color
    - The text of notices and audits shall not be bold
    - Separate columns by 2 spaces in lists (Closes: #1070064)
  * Support src:name shortcuts in showsrc, source, build-dep commands

  [ David Kalnischkies ]
  * Do not ignore if a cmake execute_process fails
  * Avoid figuring which kept pkgs are phased if we don't display it
  * Match version constraints before saving garbage packages
  * Do not upgrade rev-deps ear-marked for removal
  * Drop sudo-related envvars in testing framework
  * Add test for dealing with unsat Suggests promoted to Recommends
  * Allow parsing an empty Provides line (Closes: #1069874)

  [ Frans Spiesschaert ]
  * Dutch program translation update (Closes: #1070142)
  * Dutch manpages translation update (Closes: #1070143)

 -- Julian Andres Klode <email address hidden> Tue, 14 May 2024 13:01:31 +0200

[Timo Aaltonen (tjaalton)]

Hello Julian, or anyone else affected,

Accepted apt into noble-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/apt/2.8.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-noble to verification-done-noble. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-noble. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

[Timo Aaltonen (tjaalton)]

this upload is not to be accepted to -updates before the discussion on ubuntu-release@ is concluded

[Ubuntu SRU Bot (ubuntu-sru-bot)]

All autopkgtests for the newly accepted apt (2.8.1) for noble have finished running.
The following regressions have been reported in tests triggered by the package:

apport/2.28.1-0ubuntu3 (s390x)
auto-apt-proxy/14.1 (armhf, s390x)
cron/3.0pl1-184ubuntu2 (arm64)
dgit/11.8 (arm64)
gcc-10/10.5.0-4ubuntu2 (arm64)
gcc-11/11.4.0-9ubuntu1 (armhf)
gcc-13/13.2.0-23ubuntu4 (arm64, armhf)
gcc-13/unknown (s390x)
gcc-14/14-20240412-0ubuntu1 (armhf)
gcc-snapshot/1:20240117-1ubuntu1 (arm64, armhf)
ubiquity/24.04.5 (armhf)
update-notifier/unknown (s390x)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/noble/update_excuses.html#apt

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

[Robie Basak (racb)]

I discussed the plan for noble-updates and for 24.04.1 with Julian in detail yesterday. I think the plan is likely to be accepted by the SRU team. We're behind with the documentation, but I'm accepting the newest upload to noble-proposed now since time is tight. Julian will make sure that the full plan, test plan etc is documented in the near future.

[Robie Basak (racb)]

Hello Julian, or anyone else affected,

Accepted apt into noble-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/apt/2.8.2 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-noble to verification-done-noble. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-noble. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

[Ubuntu SRU Bot (ubuntu-sru-bot)]

All autopkgtests for the newly accepted apt (2.8.2) for noble have finished running.
The following regressions have been reported in tests triggered by the package:

apport/2.28.1-0ubuntu3.1 (armhf)
apt/2.8.2 (armhf, s390x)
auto-apt-proxy/14.1 (ppc64el, s390x)
cron/3.0pl1-184ubuntu2 (arm64, s390x)
dgit/11.8 (arm64, s390x)
gcc-11/11.4.0-9ubuntu1 (arm64)
gcc-12/12.3.0-17ubuntu1 (amd64, s390x)
gcc-13/13.2.0-23ubuntu4 (arm64, armhf)
gcc-14/unknown (armhf, s390x)
gcc-snapshot/1:20240117-1ubuntu1 (amd64, arm64, s390x)
open-build-service/unknown (armhf, s390x)
postgresql-debversion/unknown (s390x)
python-apt/unknown (s390x)
reportbug/unknown (s390x)
reprotest/unknown (s390x)
ubuntu-advantage-tools/unknown (s390x)
unattended-upgrades/unknown (s390x)
update-notifier/unknown (s390x)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/noble/update_excuses.html#apt

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

[Ubuntu SRU Bot (ubuntu-sru-bot)]

All autopkgtests for the newly accepted apt (2.8.2) for noble have finished running.
The following regressions have been reported in tests triggered by the package:

autopkgtest/5.34ubuntu2 (amd64, arm64, armhf)
update-manager/1:24.04.9 (amd64, arm64, armhf, i386, ppc64el, s390x)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/noble/update_excuses.html#apt

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

[Andreas Hasenack (ahasenack)]

Hello Julian, or anyone else affected,

Accepted apt into noble-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/apt/2.8.3 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-noble to verification-done-noble. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-noble. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

[Ubuntu SRU Bot (ubuntu-sru-bot)]

All autopkgtests for the newly accepted apt (2.8.3) for noble have finished running.
The following regressions have been reported in tests triggered by the package:

auto-apt-proxy/14.1 (s390x)
autopkgtest/5.38ubuntu1~24.04.1 (armhf)
ca-certificates-java/20240118 (ppc64el)
cron/3.0pl1-184ubuntu2 (ppc64el)
dgit/11.8 (ppc64el, s390x)
gcc-10/10.5.0-4ubuntu2 (ppc64el)
gcc-11/unknown (ppc64el)
gcc-12/unknown (ppc64el)
gcc-13/13.3.0-6ubuntu2~24.04 (ppc64el)
gcc-14/14.2.0-4ubuntu2~24.04 (ppc64el)
gcc-9/9.5.0-6ubuntu2 (ppc64el)
gcc-snapshot/unknown (ppc64el)
update-manager/1:24.04.9 (amd64, arm64, armhf, i386, ppc64el, s390x)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/noble/update_excuses.html#apt

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

[Julian Andres Klode (juliank)]

As discussed before and documented in the bug description, we are ignoring this for verification purposes, and setting this to done. The bug will be reopened once the update has been released to be fixed properly in a future upload.

[Launchpad Janitor (janitor)]

This bug was fixed in the package apt - 2.8.3

---------------
apt (2.8.3) noble; urgency=medium

  * Revert increased key size requirements from 2.8.0-2.8.2 (LP: #2073126)
    - Revert "Only install 00-temporary-rsa1024 for >=2.7.6 and improve comment"
    - Revert "Only warn about <rsa2048 when upgrading from 2.7.x to 2.8.x"
    - Revert rsa1024 to warnings again
    This leaves the mechanisms in place and no longer warns about NIST curves.
  * Fix keeping back removals of obsolete packages; and return an error if
    ResolveByKeep() is unsuccessful (LP: #2078720)
  * Fix buffer overflow, stack overflow, exponential complexity in
    apt-ftparchive Contents generation (LP: #2083697)
    - ftparchive: Mystrdup: Add safety check and bump buffer size
    - ftparchive: contents: Avoid exponential complexity and overflows
    - test framework: Improve valgrind support
    - test: Check that apt-ftparchive handles deep paths
    - Workaround valgrind "invalid read" in ExtractTar::Go by moving large
      buffer from stack to heap. The large buffer triggered some bugs in
      valgrind stack clash protection handling.

apt (2.8.2) noble; urgency=medium

  * Only install 00-temporary-rsa1024 for >=2.7.6 and improve comment
    (follow-up for LP: #2073126)

apt (2.8.1) noble; urgency=medium

  * Only revoke weak RSA keys for now, add 'next' and 'future' levels
    (backported from 2.9.7)
    Note that the changes to warn about keys not matching the future level
    in the --audit level are not fully included, as the --audit feature
    has not yet been backported. (LP: #2073126)
  * Introduce further mitigation on upgrades from 2.7.x to allow these
    systems to continue using rsa1024 repositories with warnings
    until the 24.04.2 point release (LP: #2073126)

apt (2.8.0) noble; urgency=medium

  [ Julian Andres Klode ]
  * Revert "Temporarily downgrade key assertions to "soon worthless""
    We temporarily downgraded the errors to warnings to give the
    launchpad PPAs time to be fixed, but warnings are not safe:
    Untrusted keys could be hiding on your system, but just not
    used at the moment. Hence revert this so we get the errors we
    want. (LP: #2060721)
  * Branch off the stable 2.8.y branch for noble:
    - CI: Test in ubuntu:noble images for 2.8.y
    - debian/gbp.conf: Point at the 2.8.y branch

  [ David Kalnischkies ]
  * Test suite fixes:
    - Avoid subshell hiding failure report from testfilestats
    - Ignore umask of leftover diff_Index in failed pdiff test
  * Documentation translation fixes:
    - Fix and unfuzzy previous VCG/Graphviz URI change

 -- Julian Andres Klode <email address hidden> Tue, 22 Oct 2024 15:02:22 +0200

[Andreas Hasenack (ahasenack)]

The verification of the Stable Release Update for apt has completed successfully and the package is now being released to -updates. Subsequently, the Ubuntu Stable Release Updates Team is being unsubscribed and will not receive messages about this bug report. In the event that you encounter a regression using the package from -updates please report a new bug using ubuntu-bug and tag the bug report regression-update so we can easily find any regressions.