Thanks for the review!

AppArmor has been disabled due to too many false positives for many
years, tracking in https://bugs.debian.org/cgi-
bin/bugreport.cgi?bug=875890 as noted above.

Perl dependencies review tracking in https://bugs.debian.org/cgi-
bin/bugreport.cgi?bug=1115678. The similar MySQL client package is
smaller and does not have any of the Perl scripts, so indeed for MariaDB
the libconfig-inifiles-perl might need to also have a MIR filed, but I
will do more research before doing it.

MariaDB 11.8 is a long-term supported release with security fixes and
source code releases until spring of 2030, and thus chosen for Debian
and Ubuntu. The newer 12.x are short-term releases that we intentionally
don't import to Debian/Ubuntu (this is similar to MySQL 8.4 vs 9.x
series).

The libmariadbd is the embedded server and it does not have symbols
tracking as it is not a traditional library, but the whole server built
into a single binary to be embedded in other apps as a whole. There are
no packages in Debian/Ubuntu that build with libmariadbd-dev. Anyone
using it would have their own custom setup.

The Lintian error/warnings are mostly due to special constructs in the
server how it creates plugins for itself. The server build has all
hardening flags already, but in some plugins the internal ABI structure
is partial, and Lintian does not detect the hardening properly. They
need a separate deep-dive and should be documented in Lintian overrides.
The executable bit issues are minor, and they along with other minor
Lintian issues have been reported upstream in e.g. MDEV-6153, MDEV-21869
and MDEV-33837, but haven't gotten full attention due to being
insignificant. I will ping upstream to consider fixing them as after
all, the executable bits are trivial to fix.

Lintian warning about orig-tarball-missing-upstream-signature is likely
an artifact of how the package was downloaded. Upstream publishes
signatures for all releases and they are always imported to
Debian/Ubuntu, see e.g.
http://archive.ubuntu.com/ubuntu/pool/universe/m/mariadb/mariadb_11.8.3.orig.tar.gz.asc

For the highlighted cases of potential malloc() misuse I don't have any
immediate comments, they need to be researched. I do however know that
upstream is running Valgrind, ASAN, MSAN, UBSAN and TSAN in their CI to
catch and fix memory issues. For mysys/my_setuser.c I don't have any
immediate comments either, needs further research. This is however a
piece of software that has had extensive and ongoing security audits and
testing, and the security posture should be pretty good. All findings
need to be double-checked of course, and documented (e.g.
mysys/my_setuser.c is currently lacking inline comments explaining why
setuid is used). I am also considering adding some memory sanitizer
checking or other security scanning in the CI for the Debian packaging
if that could provide more value beyond what upstream is already doing.

I will report back here when there are changes in the package or further
research results/documentation is available.


** Bug watch added: Debian Bug tracker #1115678
   https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1115678

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2122095

Title:
  [MIR] mariadb

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/mariadb/+bug/2122095/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to