qt6-base.git
4 months agoqDecodeDataUrl(): fix precondition violation in call to QByteArrayView::at()
Debian Qt/KDE Maintainers [Fri, 18 Jul 2025 13:28:20 +0000 (15:28 +0200)]
qDecodeDataUrl(): fix precondition violation in call to QByteArrayView::at()

Origin: upstream, https://download.qt.io/official_releases/qt/6.8/CVE-2025-5455-qtbase-6.8.patch
Last-Update: 2025-06-29

It is a precondition violation to call QByteArrayView::at() with
size() as argument. The code used that, though, as an implicit
end-of-string check, assuming == ' ' and == '=' would both fail for
null bytes. Besides, QByteArrays (but most certainly QByteArrayViews)
need not be null-terminated, so this could read even past size().

To fix, use higher-level API (startsWith()), consuming parsed tokens
along the way.

Gbp-Pq: Name upstream_cve-2025-5455_fix_data_assertion_error.diff

4 months ago[PATCH] QTextMarkdownImporter: Fix heap-buffer-overflow
Shawn Rutledge [Thu, 27 Mar 2025 14:17:21 +0000 (15:17 +0100)]
[PATCH] QTextMarkdownImporter: Fix heap-buffer-overflow

After finding the end marker `---`, the code expected more characters
beyond: typically at least a trailing newline. But QStringView::sliced()
crashes if asked for a substring that starts at or beyond the end.

Now it's restructured into a separate splitFrontMatter() function, and
we're stricter, tolerating only `---\n` or `---\r\n` as marker lines.
So the code is easier to prove correct, and we don't need to check
characters between the end of the marker and the end of the line
(to allow inadvertent whitespace, for example). If the markers are
not valid, the Markdown parser will see them as thematic breaks,
as it would have done if we were not extracting the Front Matter
beforehand.

Amends e10c9b5c0f8f194a79ce12dcf9b6b5cb19976942 and
bffddc6a993c4b6b64922e8d327bdf32e0d4975a

Credit to OSS-Fuzz which found this as issue 42533775.

[ChangeLog][QtGui][Text] Fixed a heap buffer overflow in
QTextMarkdownImporter. The first marker for Front Matter
must begin at the first character of a Markdown document,
and both markers must be exactly ---\n or ---\r\n.

Done-with: Marc Mutz <marc.mutz@qt.io>
Fixes: QTBUG-135284
Change-Id: I66412d21ecc0c4eabde443d70865ed2abad86d89
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
(cherry picked from commit 25986746947798e1a22d0830d3bcb11a55fcd3ae)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit eced22d7250fc7ba4dbafa1694bf149c2259d9ea)
(cherry picked from commit 9e59a924a04606c386b970ee6c9c7819cdd7ae1a)

Gbp-Pq: Name upstream_cve-2025-3512_fix_heap_buffer_overflow.diff

4 months ago[PATCH] QStorageInfo/Unix: fix declaration of 'mnt' on Hurd
Pino Toscano [Sun, 6 Oct 2024 07:40:14 +0000 (09:40 +0200)]
[PATCH] QStorageInfo/Unix: fix declaration of 'mnt' on Hurd

Directly use the right type for it, after the Linux code was dropped.
Fixes commit 543ae6e6a43519b9fca6758c4a8c78625fcb2c86.

Change-Id: I1417853153d2851262f40713b3318f0dae09c391
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Gbp-Pq: Name upstream_QStorageInfo-Unix-fix-declaration-of-mnt-on-Hurd.patch

4 months ago[PATCH] IPC: add PATH_MAX-less fallback definition for MAX_PATH
Pino Toscano [Sat, 22 Jun 2024 17:55:15 +0000 (19:55 +0200)]
[PATCH] IPC: add PATH_MAX-less fallback definition for MAX_PATH

Define MAX_PATH also when PATH_MAX is not defined (e.g on GNU/Hurd).
MAX_PATH is Windows constant, and it is used in this file only in a
code path for Windows; because of this, the static fallback define
should be good enough.

Change-Id: Ic1b9fee3b62505f86aa8ec89bbd20493bfe1f67c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Gbp-Pq: Name upstream_IPC-add-PATH_MAX-less-fallback-definition-for-MAX_PA.patch

4 months ago[PATCH] QSortFilterProxyModel: don't call index(row, 0) if there are no columns
David Faure [Mon, 3 Mar 2025 18:14:59 +0000 (19:14 +0100)]
[PATCH] QSortFilterProxyModel: don't call index(row, 0) if there are no columns

This is invalid, e.g. it asserts in
QConcatenateTablesProxyModel::index()

Fixes: QTBUG-134210
Change-Id: I21acad9497d423b0366991296e8dd498d51395ea
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
(cherry picked from commit 93694e99c214a5166fc842f92659e42260230dce)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Gbp-Pq: Name upstream_prevent_recursion_in_qsortfilterproxymodel.diff

4 months ago[PATCH] Add clamping to QColorTransferGenericFunction
Samuel Gaist [Sat, 24 May 2025 19:07:37 +0000 (21:07 +0200)]
[PATCH] Add clamping to QColorTransferGenericFunction

This ensures that the inputs are within range for the use of these
function.

Depending on the values passed, they can trigger FE_INVALID errors
and thus NaN as return values.

This can happen for example when feeding an invalid ICC profile to
QColorSpace::fromIccProfile.

Credit to OSS-Fuzz

Fixes: QTBUG-137159
Origin: upstream, https://download.qt.io/official_releases/qt/6.8/CVE-2025-5992-qtbase-6.8.patch

Gbp-Pq: Name upstream_cve-2025-5992_input_range_qcolortransformation.diff

4 months ago[PATCH] QColorTransferGeneric: fix the BT.2100 PQ EOTF
Giuseppe D'Angelo [Thu, 6 Feb 2025 20:00:55 +0000 (21:00 +0100)]
[PATCH] QColorTransferGeneric: fix the BT.2100 PQ EOTF

The PQ EOTF formula for BT.2100 [1][2] was incorrect. Fix it; while at
it, rename the variables to match the symbols used in the original
formula.

The inverse EOTF was correct, but also rename the variables there (for
the same reason).

[1] https://www.itu.int/rec/R-REC-BT.2100-2-201807-I/en
[2] https://en.wikipedia.org/wiki/Perceptual_quantizer#Technical_details

Change-Id: I6ce3a609824bee82053a16b3ff3cfc7cb396ce8f
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
(cherry picked from commit a7ff4679facb9a44dff8b63a7e461ababa6aedfb)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit 35c18eba64b6220e36ea14d32b911342e9d0921c)
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Gbp-Pq: Name upstream_QColorTransferGeneric_fix_BT2100PQEOTF.patch

4 months agoqt6-base (6.8.2+dfsg-9) unstable; urgency=medium
Patrick Franz [Fri, 18 Jul 2025 13:28:20 +0000 (15:28 +0200)]
qt6-base (6.8.2+dfsg-9) unstable; urgency=medium

  [ Patrick Franz ]
  * Backport patch to fix the PQ EOTF formula for BT.2100. This patch is
    needed to make the patch for CVE-2025-5992 applicable.
  * Backport patch to fix CVE-2025-5992 (Closes: #1109299).

[dgit import unpatched qt6-base 6.8.2+dfsg-9]

4 months agoImport qt6-base_6.8.2+dfsg-9.debian.tar.xz
Patrick Franz [Fri, 18 Jul 2025 13:28:20 +0000 (15:28 +0200)]
Import qt6-base_6.8.2+dfsg-9.debian.tar.xz

[dgit import tarball qt6-base 6.8.2+dfsg-9 qt6-base_6.8.2+dfsg-9.debian.tar.xz]

10 months agoImport qt6-base_6.8.2+dfsg.orig.tar.xz
Patrick Franz [Sun, 2 Feb 2025 20:05:30 +0000 (21:05 +0100)]
Import qt6-base_6.8.2+dfsg.orig.tar.xz

[dgit import orig qt6-base_6.8.2+dfsg.orig.tar.xz]