Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
58 changes: 58 additions & 0 deletions gdcm/debian/email-debian-med.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,58 @@
<!-- DRAFT - NOT SENT. Plain-text mail: send everything below the "---" line.
To: debian-med@lists.debian.org
Cc: Emmanuel Arias <eamanu@debian.org>, Étienne Mollier <emollier@debian.org>
Subject: gdcm: finishing 3.0.24-5+deb13u1 for trixie (security fixes)
Addresses come from debian/changelog (neither is listed in Uploaders).
The trixie-pu request (trixie-pu-request.md) goes after debian-med replies,
not together with this mail. -->
---

Hi,

I would like to help get the outstanding gdcm security fixes into trixie,
and I am writing first because Emmanuel has already started this work.

The debian/trixie branch on salsa has a 3.0.24-5+deb13u1 entry with the
CVE-2025-11266 fix, prepared in February and not uploaded so far. The
security tracker still lists trixie as affected by five more issues that
are fixed in unstable since 3.0.24-11: CVE-2025-48429, CVE-2025-52582,
CVE-2025-53618, CVE-2025-53619 and CVE-2026-3650. All are marked no-dsa
as minor issues, except CVE-2026-3650, which is postponed until fixed
upstream (it now is). The same set went to bullseye LTS in DLA-4652-1.

On a local copy of debian/trixie I added the four patches from unstable
for those issues, unchanged apart from two lines of upstream commit
trailer that I left out of the CVE-2026-3650 header. They apply without
fuzz on 3.0.24-5. I extended the existing changelog entry and kept
Emmanuel's part under his name. I built it in a clean trixie container.
The autopkgtest passes and lintian shows no new warnings. I also
compared the old and new binaries on 95 sample DICOM files and with
reproducers for CVE-2025-11266 and CVE-2026-3650.

That comparison turned up one regression. The frame size check from the
CVE-2025-53618/53619 patch makes gdcmconv --jpeg fail on odd-sized
images, because of the padding byte. Upstream fixed this in 9ced0f9
("Accepth odd length", released in 3.2.9), which unstable does not have
yet. I added it to the trixie update as a separate patch and reported
it for unstable, with the same patch, as #1152837.

Would it be all right if I finish debian/trixie this way and open a
merge request on salsa? If you would rather handle it yourselves, or if
there is a reason it was held back, I am happy to step back. If the MR
is accepted I would then file the trixie-pu request, and I would need a
sponsor for the upload. The proposed-updates freeze for 13.8 is in two
weeks, which is probably too soon, so I would aim for 13.9 in January.
The debdiff is ready to share.

Some background: I prepared the Ubuntu backport of the upstream
CVE-2025-11266 fix (upstream commit 5829c95) for jammy and noble, which
is how I came to this package. This would be my first contribution to
Debian, so corrections on process are very welcome.

I have also prepared the same set for bookworm on top of the
debian/bookworm branch (3.0.21-1+deb12u1, including Étienne's
CVE-2024-* patches). As bookworm is now in LTS, I would offer that one
to the LTS team separately, unless you prefer to handle it differently.

Thanks,
Valentín Torassa
Loading