dgit.raspbian.org Git - libreoffice.git/log

Merge version 4:25.2.3-2+rpi1+deb13u5 and 4:25.2.3-2+deb13u6 to produce 4:25.2.3-2+rpi1+deb13u6

Merge version 4:25.2.3-2+rpi1+deb13u3 and 4:25.2.3-2+deb13u5 to produce 4:25.2.3-2+rpi1+deb13u5

Merge libreoffice (4:25.2.3-2+deb13u6) import into refs/heads/workingbranch

[PATCH] check for hb_shape_full failure

Signed-off-by: Caolán McNamara <caolan.mcnamara@collabora.com>
Change-Id: I5fafcb7d592fb4c9091cb2bab64cb6ac47462d71
Reviewed-on: https://gerrit.collaboraoffice.com/c/online/+/2102
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
(cherry picked from commit 22c19be8fae977a90f83a82d8acccd0b46c7cf20)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/204946
Tested-by: Jenkins
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
Gbp-Pq: Name check-for-hb_shape_full-failure.diff

[PATCH] drop malformed duplicate-id calc change track actions

Signed-off-by: Caolán McNamara <caolan.mcnamara@collabora.com>
Change-Id: I0d532eb84d122e39ce71e94b4760b94cd636eeea
Reviewed-on: https://gerrit.collaboraoffice.com/c/online/+/1691
Tested-by: Jenkins CPCI <releng@collaboraoffice.com>
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
(cherry picked from commit b5b56a2239369e45d7fac2ea965558a1f337b602)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/205091
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
Tested-by: Jenkins
(cherry picked from commit a20ae7b5f80a45e49a47ce22f5e92749cd9816a0)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/205101

Gbp-Pq: Name CVE-2026-8358.diff

[PATCH] A formula of length L, composed entirely of open tokens, needs L+1 slots

Signed-off-by: Caolán McNamara <caolan.mcnamara@collabora.com>
Change-Id: I7cf1fd4fdaca780a8f0cf0d0b9d10def9b3af1ef
Reviewed-on: https://gerrit.collaboraoffice.com/c/online/+/1690
Tested-by: Jenkins CPCI <releng@collaboraoffice.com>
(cherry picked from commit 0b7a9149d7c7b5e044bb4f02668297bdf41b64d6)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/205090
Tested-by: Jenkins
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
(cherry picked from commit a28cead4b7796642d946173b479281a6272117f2)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/205102

Gbp-Pq: Name CVE-2026-8357.diff

[PATCH] SdrEscherImport::RecolorGraphic reads but doesn't use FillColors

it never did, but at:

commit 025bfa7e510bdab3ef93ad45a731fc25085ba3cc
Date: Thu Apr 23 10:03:58 2015 +0100

cppcheck: unreadVariable

the wrong choice was made wrt the dead store, the dead loop should have
been fixed rather than let the unused fill colors accumulate in the
globalcolors buffer.

This function doesn't need to consume the unused data, all callers don't
expect the entire record to be read so just drop the loop.

Signed-off-by: Caolán McNamara <caolan.mcnamara@collabora.com>
Change-Id: I02bce4dd78dc566f18edc3d45603534c3f724cc3
Reviewed-on: https://gerrit.collaboraoffice.com/c/online/+/1689
(cherry picked from commit 07faeadace3d8d93225f9dd72c4f911e331070e2)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/205089
Tested-by: Jenkins
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
(cherry picked from commit 9ba100c0d2131862c9356a86906921abc80d8f98)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/205097

Gbp-Pq: Name CVE-2026-8356.diff

CVE-2026-6045

CVE-2026-6045 EMF+ Heap-buffer-overflow in EMFPBrush::Read

CVE-2026-6045 EMF+ Heap-buffer-overflow in EMFPBrush::Read
A nested format problem tucked away in the rendering path, probably
need to add something dedicated to the simpler fuzzer to force that
render path to be exercised.

From 279c7ea61829efe5cabc5832d06e1833ad28379f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Caol=C3=A1n=20McNamara?= <caolan.mcnamara@collabora.com>
Date: Thu, 9 Apr 2026 20:00:38 +0100
Subject: [PATCH] check that the file can provide the claimed data

and make sure we initialize these locals

Change-Id: Ifa899e36f678216574364e5206037ab57b2d19d9
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/203577
Tested-by: Jenkins
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/203628
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Gbp-Pq: Name CVE-2026-6045.diff

CVE-2026-6040

CVE-2026-6040: ODT use-after-free in lcl_InsertBlankWidthChars

oss-fuzz efforts might not have found this because the fuzzer
dictionary was based on OpenDocument-v1.3-schema.rng and the
loext:blank-width-char isn't in that schema, adding in the extra
extension schema might help for the future.

From 997ef5c01cedc4a4f8b966310d4a79906009735e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Caol=C3=A1n=20McNamara?= <caolan.mcnamara@collabora.com>
Date: Thu, 9 Apr 2026 17:47:09 +0100
Subject: [PATCH] process loext:blank-width-char better

Change-Id: Iea005facd85443091c5144a0a0f8f15fa995dbf3
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/203576
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
Tested-by: Jenkins
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/203627
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Gbp-Pq: Name CVE-2026-6040.diff

CVE-2026-6039

CVE-2026-6039: DXF heap-buffer-overflow in DrawLWPolyLineEntity

It looks like our oss-fuzz efforts didn't find this because we capped
the dxffuzzer.options size at 64KB which is too small to capture this
issue.

From cf825c7608ded2cb1b7c846bde15d0ebcf6a5c44 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Caol=C3=A1n=20McNamara?= <caolan.mcnamara@collabora.com>
Date: Thu, 9 Apr 2026 17:23:39 +0100
Subject: [PATCH] stay within max Polygon points

Change-Id: I02e34bb413e6332d8c5683504a63a591a33d7730
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/203575
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
Tested-by: Jenkins
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/203626
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Gbp-Pq: Name CVE-2026-6039.diff

[PATCH] Conform AlignEngine parsing to what section 2.3.4.10 of the spec has

Change-Id: Ibb9162b1ce7993ef74665ec0329c95b423fa8174
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/201944
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Tested-by: Jenkins
Gbp-Pq: Name Conform-AlignEngine-parsing-to-spec.diff

[PATCH] default to EUR for Bulgaria

Bulgaria will switch to EUR 2026-01-01.

Change-Id: I0ed47f1681c406af2e92cdd2fdfd0bdace7f8216

Gbp-Pq: Name default-to-Euro-for-Bulgaria.diff

[PATCH] qt: Consolidate to one toOUString helper

Instead of having one for vcl in vcl/inc/qt5/QtTools.hxx
and one in avmedia/source/qt6/QtFrameGrabber.cxx,
move the existing implementation to
include/vcl/qt/QtUtils.hxx and use it everywhere.

Change-Id: I8576ab0fe2fd12bb25ae262e59acb8412ef8be0c
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/184057
Tested-by: Jenkins
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>
(cherry picked from commit d023035acf83ee1b61dfc03333bfc6e612bb58f6)

Gbp-Pq: Name qt-Consolidate-to-one-toOUString-helper.diff

tdf#166055 avmedia qt: Use GStreamer frame grabber by default

As described in tdf#166055, QtFrameGrabber currently
still causes issues (freezes or crashes).

The QtMultimedia API does not provide a simple
way to synchronously retrieve a video frame, so
the current approach is to connect to the
QVideoSink::videoFrameChanged signal and start
playing the video until the first frame arrives.

There are various QtMultimedia plugins/backends
(at least GStreamer and ffmpeg for Linux). Some
use multiple threads internally. Some logic
needs to be run in the main thread however, so
it's not possible to move things to a separate
thread/event loop to decouple it from the main
thread.
As a consequence, there is a need to trigger
event processing while waiting for the frame,
as QVideoSink::videoFrameChanged would otherwise
never be called.

Triggering event processing can have bad side-effects
however, as seen in tdf#166055, e.g. leading to
endless recursion or crashes when processing LO events.

See also

    commit 697405b533a8ae5b6a8f5bd184b9344a96f71c69
    Author: Michael Weghorn <m.weghorn@posteo.de>
    Date:   Wed Apr 9 09:42:45 2025 +0200

        tdf#166055 qt avmedia: Provide media/player size right away

for more background and some thoughts.

For now, avoid the problem by using
the GStreamer based frame grabber even with
QtPlayer by default.

This means that the QtMultimedia
framework is still used for video playback with the
qt6 VCL plugin, where LO's GStreamer implementation
doesn't work properly, see

    commit 441d8ed9be0e7f831b455a69b8688dcb79a8bc00
    Author: Michael Weghorn <m.weghorn@posteo.de>
    Date:   Mon May 20 16:25:09 2024 +0200

        tdf#145735 avmedia qt: Use QtMultimedia for Qt 6 media playback

. However, in order to preview images (that are e.g. shown
in Impress when not in presentation mode) the LO GStreamer
backend is used now.
(That works fine, no UI interaction is needed
for that.)

Adjust QtPlayer::createFrameGrabber accordingly
to return the LO default platform player/frame grabber
implementation (which is the GStreamer one on Linux), see also
MediaWindowImpl::createPlayer. Use the service to
avoid having to link GStreamer libraries.

However, allow to force the use of QtFrameGrabber
by setting environment variable
SAL_VCL_QT_USE_QT_FRAME_GRABBER.

Change-Id: Ibe5c42adf6e8f699f33e5f089bd2f6076c3de4f7
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/184912
Tested-by: Jenkins
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>
(cherry picked from commit 08533ca4e2526644b803c40c0c3d3c96f43762af)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/184918
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Gbp-Pq: Name avmedia-qt-use-gstreamer-frame-grabber-by-default.diff

[PATCH] add EUR for Bulgaria/Lew

Bulgaria will join Eurozone 2026-01-01 at which time around the default
needs to be switched

see https://www.ecb.europa.eu/press/pr/date/2025/html/ecb.pr250708~b9676a9fa8.de.html for the conversion rate

This is the < 26.2 version including the euro wizard (cf.
b1a2f727ca99ecd3402d4b051b99cbfd24266e59)

Change-Id: I840ddda4308549703c53f998812848888dd830f6
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/189279
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins
From 147f2fe0e46ef030ccee10581b97ffdce770b79f Mon Sep 17 00:00:00 2001
From: Xisco Fauli <xiscofauli@libreoffice.org>
Date: Fri, 28 Nov 2025 10:25:42 +0100
Subject: fix typo: BLN -> BGN
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

After
commit 37d1cc071ba0cc7f9922c4409d0a014a722ebbce
Author: Rene Engelhard <rene@rene-engelhard.de>
Date: Sat Aug 9 19:05:15 2025 +0200

add EUR for Bulgaria/Lew

Change-Id: I6679404cba0be0219bf7ede06297f9b1c990e8b4
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/194760
Reviewed-by: Eike Rathke <lo@erack.de>
Tested-by: Jenkins
Reviewed-by: René Engelhard <rene@debian.org>
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/194766

Gbp-Pq: Name add-EUR-for-Bulgaria-Lew.diff

[PATCH] implement --with-system-colamd (for internal lpsolve)

using suitesparse. Tested with both suitesparse 5.12.0 (Debian stable,
without pc) and 7.9.0 (Debian unstable, with .pc)

Change-Id: I6e0d014455ea2dec111aa3380bd49bd586dfad0a

Gbp-Pq: Name system-colamd.diff

disable-uitest-xmlsecurity-gpg

Gbp-Pq: Name disable-uitest-xmlsecurity-gpg.diff

fix-32bits-test-build

Gbp-Pq: Name fix-32bits-test-build.diff

lo-xlate-lang-add-tl

Gbp-Pq: Name lo-xlate-lang-add-tl.diff

[PATCH] sunjre: check all four possible OpenJDK variants for libjvm.so

Since OpenJDK 21, the "minimal" and "zero" variants of JVM no longer
reside in the "server" directory in $JAVA_HOME/lib, making the loader
unable to find libjvm.so. This regresses LO builds on platforms with
only those variants of JVM (i.e. without a HotSpot port).

Example error message is like this:

> exception occurred: Could not create Java implementation loader at
> ./stoc/source/javaloader/javaloader.cxx:551

Fix it by probing all four possible variant directories for libjvm.so.
This is tested on Debian unstable (loong64 port) to fix the packaging.

Link: https://bugs.openjdk.org/browse/JDK-8273494
Link: https://github.com/openjdk/jdk/pull/5440
Link: https://github.com/openjdk/jdk/commit/8fbcc8239a3fc04e56ebbd287c7bb5db731977b7
Change-Id: I3545f93054d7a3af0181e4c739c82efaaf8aeaed

Gbp-Pq: Name jdk-minimal-and-zero-paths.diff

pdfium-ports

Gbp-Pq: Name pdfium-ports.diff

adapt-for-new-carlito

Gbp-Pq: Name adapt-for-new-carlito.diff

we-dont-have-the-needed-fonts

Gbp-Pq: Name we-dont-have-the-needed-fonts.diff

do-not-abort-on-NON_APPLICATION_FONT_USE

Gbp-Pq: Name do-not-abort-on-NON_APPLICATION_FONT_USE.diff

moreIconsDialog-accesses-internet

Gbp-Pq: Name moreIconsDialog-accesses-internet.diff

Explicitly set fpu mode on armhf

Gbp-Pq: Name arm-fp-mode.diff

Support tofu+pgp trust model in GnuPG

Bug-Debian: https://bugs.debian.org/955271
Forwarded: no

GnuPG supports a trust-on-first-use layer that sits on top of the
standard PGP trust model. If this is enabled, 'gpg --list-keys' needs
write and lock permissions on the TOFU database to return any useful
data. Allow this access through AppArmor.

Gbp-Pq: Name apparmor-gnupg-tofu.diff

apparmor-updates

Gbp-Pq: Name apparmor-updates.diff

[PATCH] Resolves: rhbz#1432468 disable opencl by default

Change-Id: Ie037fcabdd219f195425979dd721501fb5527573

Gbp-Pq: Name no-opencl-per-default.diff

disable-shortcuts_tab_navigation-uitest

Gbp-Pq: Name disable-shortcuts_tab_navigation-uitest.diff

add-access2base-doc

Gbp-Pq: Name add-access2base-doc.diff

Add safer float comparisons to bridgetest equals()

Bug-Ubuntu: https://launchpad.net/bugs/1832360

Gbp-Pq: Name fix-flaky-bridgetest.diff

apparmor-opencl

apparmor: Add opencl support

AppArmor in Debian Buster now has OpenCL abstractions.

Include OpenCL abstractions to fix OpenCL usage in Calc.

Gbp-Pq: Name apparmor-opencl.diff

[PATCH] mariadb

Gbp-Pq: Name use-mariadb-java-instead-of-mysql-java.diff

disableClassPathURLCheck

Gbp-Pq: Name disableClassPathURLCheck.diff

apparmor-mesa

Gbp-Pq: Name apparmor-mesa.diff

[PATCH] apparmor: use dri-enumerate abstraction

Remove backported rule and use new dri-enumerate abstraction instead.
dri-enumerate is available in AppArmor 2.13, which recently migrated
into Debian Buster.

Change-Id: I64919edc1882f7bc1e65cfb94686464c5350f699

Gbp-Pq: Name apparmor-cleanups.diff

apparmor-allow-java.security

Gbp-Pq: Name apparmor-allow-java.security.diff

do-not-hide-test-output

Gbp-Pq: Name do-not-hide-test-output.diff

disable-java-in-odk-build-examples-on-zero-vm

Gbp-Pq: Name disable-java-in-odk-build-examples-on-zero-vm.diff

appstream-ignore-startcenter

Gbp-Pq: Name appstream-ignore-startcenter.diff

Hide startcenter and math from the shell

Bug-Ubuntu: https://launchpad.net/bugs/1696250
Forwarded: not-needed

Gbp-Pq: Name hide-math-desktop-file.patch

apparmor-complain

Gbp-Pq: Name apparmor-complain.diff

cppunit-optional

Gbp-Pq: Name cppunit-optional.diff

system-officeotron-and-odfvalidator

Gbp-Pq: Name system-officeotron-and-odfvalidator.diff

no-packagekit-per-default

Gbp-Pq: Name no-packagekit-per-default.diff

javadoc-optional

Gemeinsame Unterverzeichnisse: odk-old/config und odk/config.
Gemeinsame Unterverzeichnisse: odk-old/docs und odk/docs.
Gemeinsame Unterverzeichnisse: odk-old/examples und odk/examples.

Gemeinsame Unterverzeichnisse: odk-old/config und odk/config.
Gemeinsame Unterverzeichnisse: odk-old/docs und odk/docs.
Gemeinsame Unterverzeichnisse: odk-old/examples und odk/examples.

Gbp-Pq: Name javadoc-optional.diff

fix-internal-hsqldb-build

Gbp-Pq: Name fix-internal-hsqldb-build.diff

disable-flaky-tests

14:13 < mst__> _rene_, the toolkit unoapi tests are known to be flaky (in some
               system dependent way) e.g. on the Win@6 tinderbox it always
               crashes
14:14 < mst__> _rene_, sc.ScAccessible* tests also fail on some systems some of
               the time

Gbp-Pq: Name disable-flaky-tests.diff

debian-hardened-buildflags-no-LO-fstack-protector-strong

don't hardcode -fstack-protector-strong in configure.ac/gbuild. We get the
hardening flags from dpkg-buildflags anyway.

Gbp-Pq: Name debian-hardened-buildflags-no-LO-fstack-protector-strong.diff

debian-hardened-buildflags-CPPFLAGS

Gbp-Pq: Name debian-hardened-buildflags-CPPFLAGS.diff

mediwiki-oor-replace

Gbp-Pq: Name mediwiki-oor-replace.diff

make-package-modules-not-suck

Gbp-Pq: Name make-package-modules-not-suck.diff

jdbc-driver-classpaths

Gbp-Pq: Name jdbc-driver-classpaths.diff

reportdesign-mention-package

Gbp-Pq: Name reportdesign-mention-package.diff

sensible-lomua

===================================================================

Gbp-Pq: Name sensible-lomua.diff

help-msg-add-package-info

Gbp-Pq: Name help-msg-add-package-info.diff

mention-java-common-package

Gbp-Pq: Name mention-java-common-package.diff

install-fixes

Gbp-Pq: Name install-fixes.diff

build-against-shared-lpsolve

Gbp-Pq: Name build-against-shared-lpsolve.diff

debian-debug

Gbp-Pq: Name debian-debug.diff

splits-adapt-registry

Gbp-Pq: Name splits-adapt-registry.diff

jurt-soffice-location

commit b71107fb12e3c3125e0cb62c5a4f6636a80c6408
Author:     Bjoern Michaelsen <bjoern.michaelsen@canonical.com>
AuthorDate: Tue Jun 7 11:52:37 2011 +0200
Commit:     Bjoern Michaelsen <bjoern.michaelsen@canonical.com>
CommitDate: Tue Jun 7 11:52:37 2011 +0200

    on debian-based systems, we know where our soffice binary is

Gbp-Pq: Name jurt-soffice-location.diff

debian-opt

Gbp-Pq: Name debian-opt.diff

libreoffice (4:25.2.3-2+deb13u6) trixie; urgency=medium

  * debian/patches/check-for-hb_shape_full-failure.diff: add patch from
    libreoffice-26-2 branch to gracefully handle hb_shape_full failure,
    as can happen after the fix for CVE-2026-50593 in graphite2

[dgit import unpatched libreoffice 4:25.2.3-2+deb13u6]

Import libreoffice_25.2.3-2+deb13u6.debian.tar.xz

[dgit import tarball libreoffice 4:25.2.3-2+deb13u6 libreoffice_25.2.3-2+deb13u6.debian.tar.xz]

Merge libreoffice (4:25.2.3-2+deb13u5) import into refs/heads/workingbranch