Backport upstream patch, that was released in GLib 2.84.4 [1],
such that we can apply it onto GLib 2.82 in Buildroot LTS
This fixes the following vulnerability:
- CVE-2025-7039:
A flaw was found in glib. An integer overflow during temporary file
creation leads to an out-of-bounds memory access, allowing an attacker
to potentially perform path traversal or access private temporary file
content by creating symbolic links. This vulnerability allows a local
attacker to manipulate file paths and access unauthorized data. The
core issue stems from insufficient validation of file path lengths
during temporary file operations.
https://www.cve.org/CVERecord?id=CVE-2025-7039
[1] https://gitlab.gnome.org/GNOME/glib/-/releases/2.84.4
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
(cherry picked from commit 3252f45279)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Buildroot LTS 2025.02.x utilises libglib2 version 2.82.5 which is prone
to CVE-2025-6052 [1].
0001-gstring-Fix-overflow-check-when-expanding-the-string.patch is
included in the package/libglib2 directory. This patch mitigates
CVE-2025-6052 as seen in [1], by backporting to libglib2 2.82.5.
libglib2 shares part of its infrastructure with libglib2-bootstrap,
by adding the patch in package/libglib2 the patch isn't applied to
libglib2-bootstrap.
When building libglib2, the patch was correctly applied in
output/build/host-libglib2-2.82.5/glib/gstring.c . When building
libglib2-bootstrap it was verified that the patch was not applied here.
Signed-off-by: Tim Soubry <tim.soubry@mind.be>
[Arnout: fix checkpackage error]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
There are a lot of interconnected changes in these releases:
Since 2.79.0, the introspection (i11n) data for libglib2 was moved out
of gobject-introspection (GOI) and into libglib2 itself, thus building
libglib2 with i11n needs tools provided by GOI [0].
However, GOI needs libglib2, but it is content with a libglib2 that
does not have i11n. So we introduce a new package, libglib2-bootstrap,
that provides a minimalist libglib2 that is enough to build GOI, build
GOI, then build the final, complete libglib2 with i11n.
However, not every package needs i11n, so we make that optional: if GOI
is enabled, libglib2 will have i11n, otherwise it won't.
This means the dependency ordering has changed: GOI used to
build-depend on libglib2, but now that's the other way around. This
will cause issues with packages that build-depend only on GOI and then
expect a libglib2 with i11n; those will only get a basic libglib2
without i11n at build time, but a full one at runtime. Those packages
will have to be fixed to also include a build dependency on libglib2.
The version bump between libglib2 and GOI must be done in lock-step,
because newer GOI does not support an older libglib2 and, as seen
above, newer libglib2 has the data previously provided by older GOI.
As a consequence of now generating its own i11n data, libglib2 needs to
be able to run target programs during the build, to extract data layout
and generate i11n data, a bit like GOI does [1]. So we provide a qemu
wrapper similar to the one in nodejs-src, and use that as meson's
exe_wrapper.
Except for our first patch, the other libglib2 patches were backports,
and are present upstream already, so we can drop them; the first patch
needs being refreshed.
We also move the remaining patch to a versioned sub-directory, so that
it is easier to share the patches between libglib2-bootstrap and
libglib2, should we need to add more patches in the future.
Drop the "giscanner: remove dependency on distutils.msvccompile"
patch, included upstream.
Updated hash for giscanner/scannerlexer.l is due to code changes, no
change in license, see [2] (two commits in 2024).
[0] https://gitlab.gnome.org/GNOME/glib/-/blob/2.79.0/NEWS?ref_type=tags#L8-17
[1] technically, this should not be needed if libglib2 only uses the
wrappers provided by GOI: g-ir-scanner et al. However, its
meson.build explicitly checks that it can run generated binaries
through the exe_wrapper. Investigating if that's really needed is
left as an exercise for the interested party: in practice, we will
have to be able to run via host-qemu, so this does not add any build
time overhead.
[2] https://gitlab.gnome.org/GNOME/gobject-introspection/-/commits/1.82.0/giscanner/scannerlexer.l?ref_type=tags
Signed-off-by: Fiona Klute (WIWA) <fiona.klute@gmx.de>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Tested-by: Marcus Hoffmann <buildroot@bubu1.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
This commit allows the package patches to be applied with fuzz factor 0.
The fuzz factor specifies how many lines of the patch can be inexactly
matched, so the value 0 requires all lines to be exactly matched.
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Arnout Vandecappelle <arnout@mind.be>
In upstream commit e71ecc8771a4f13bc6046438ab0845944831b9a6 ("build:
Remove deprecated -Diconv option"), merged since glib 2.75.1, the
meson -Diconv option was removed.
In Buildroot, this means that the build of libglib2 has been broken
since commit 3f9622fe3d, which bumped
libglib2 from 2.72.3 to 2.76.1 for configurations that have libiconv
enabled, causing this build failure:
../output-1/build/libglib2-2.76.1/meson.build:1:0: ERROR: Unknown options: "iconv"
iconv is now automatically detected by Meson machinery, and so the
option was considered as no longer being needed. This commit fixes
that by dropping the useless -Diconv=external.
Another related change done is remove the double addition of libiconv
into the <pkg>_DEPENDENCIES variable: libiconv can only be enabled
when BR2_ENABLE_LOCALE is disabled, and libglib2/Config.in selects
BR2_PACKAGE_LIBICONV when !BR2_ENABLE_LOCALE. So testing
BR2_ENABLE_LOCALE!=y and BR2_PACKAGE_LIBICONV=y is exactly the same
thing, causing libiconv to be added twice to the dependencies.
Fixes:
http://autobuild.buildroot.net/results/d2da03f7558f3b6ee59c813bb64115702e52704c/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
If rst2html5 is installed on a host system, meson will attempt to build
html5 documentation from the docs directory, resulting in the following
error:
/usr/local/bin/rst2html5.py
Traceback (most recent call last):
File "/usr/local/bin/rst2html5.py", line 27, in <module>
from docutils.core import publish_cmdline, default_description
ModuleNotFoundError: No module named 'docutils'
Backport an upstream that guards the build of the html documentation
behind the existing ds_doc condition, which we already disable in
Buildroot.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
[yann.morin.1998@free.fr:
- as pointed out by James, backport an upstream commit rather than
carry an non-upstreamable patch
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
- glib2 has moved from PCRE to PCRE2 [1].
- Drop `fam` option since it no longer available [2].
- Drop "0001-fix-compile-time-atomic-detection.patch" since upstream
patch was dropped and new atomic detection implementation has been
applied in this version.
- Drop "0002-remove-cpp-requirement.patch" since upstream has modified
its Meson scripts to have any C++ usage as optional.
- Drop "0003-Add-Wno-format-nonliteral-to-compiler-arguments.patch"
since change is available in this version.
- Add new patch to address new Werror compilation errors [3].
- Renamed "0004-*.patch" -> "0001-*.patch" since it should still be
applicable for its mentioned build case.
[1]: https://bugzilla.redhat.com/show_bug.cgi?id=1938974
[2]: https://gitlab.gnome.org/GNOME/glib/-/merge_requests/2570
[3]: https://gitlab.gnome.org/GNOME/glib/-/merge_requests/3390
Signed-off-by: James Knight <james.d.knight@live.com>
[yann.morin.1998@free.fr:
- update upstream status on new patch
- regenerate .checkpackageignore
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Providing an upstream status for the remaining libglib2 patch from a
recent version bump. This patch was introduced in Buildroot around
November 2021 to help deal with build errors where `GObject-2.0.gir`
could not be found (tied to Buildroot changes made in
`gobject-introspection.mk`). The discussion thread mentions this could
be submitted upstream, but looks like it never was. Reflecting that this
package has not been submitted upstream.
Signed-off-by: James Knight <james.d.knight@live.com>
[yann.morin.1998@free.fr: update .checkpackageignore]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Update fourth patch to add girdir in gio-2.0.pc. This will fix the
following build failure with libnice raised since commit
aade2fd293:
/home/giuliobenetti/autobuild/run/instance-3/output-1/host/riscv32-buildroot-linux-gnu/sysroot/usr/bin/g-ir-compiler nice/Nice-0.1.gir --output nice/Nice-0.1.typelib --includedir=/usr/share/gir-1.0
Could not find GIR file 'GObject-2.0.gir'; check XDG_DATA_DIRS or use --includedir
Fixes:
- http://autobuild.buildroot.org/results/3088ef32b03e0bb984291b1227b187f1ff816eb7
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Xattrs is required for SELinux. Explicitly enable the feature when the
libselinux package is selected.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Other changes:
- Drop 0002-allow-explicit-disabling-of-tests.patch as this patch has been
upstreamed and the option is now simply "tests."
- Add -Dglib_debug=disabled and -Dlibelf=disabled as defaults to both
HOST_LIBGLIB2_CONF_OPTS and LIBGLIB2_CONF_OPTS
- Refactor existing patches to apply to 2.68.1
Tested with test-pkg -p libglib2 -a.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
- Fix CVE-2021-27218: An issue was discovered in GNOME GLib before
2.66.7 and 2.67.x before 2.67.4. If g_byte_array_new_take() was called
with a buffer of 4GB or more on a 64-bit platform, the length would be
truncated modulo 2**32, causing unintended length truncation.
- Fix CVE-2021-27219: An issue was discovered in GNOME GLib before
2.66.6 and 2.67.x before 2.67.3. The function g_bytes_new has an
integer overflow on 64-bit platforms due to an implicit cast from 64
bits to 32 bits. The overflow could potentially lead to memory
corruption.
https://gitlab.gnome.org/GNOME/glib/-/blob/2.66.7/NEWS
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Patch '0001-fix-compile-time-atomic-detection.patch' claims to be Merged but
this is not true. The linked issue is closed with 'Needs information', and
the code itself is effectively not merged.
Clarify the 'Upstream-status' line to make this more clear.
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Commit 2acaa86ce6 replaced util-linux by
util-linux-libs but this raises the following build failure:
Makefile:585: *** util-linux-libs is in the dependency chain of libglib2 that has added it to its _DEPENDENCIES variable without selecting it or depending on it from Config.in. Stop.
To fix this build failure, use util-linux or util-linux-libs depending
on BR2_PACKAGE_UTIL_LINUX_LIBS
Fixes:
- http://autobuild.buildroot.org/results/8bf6360a343edd10c73a756ab86cdba727f9ea43
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Prior to commit 4102db0f7a ("package/libglib2: bump to version 2.60.3")
which converted libglib2 to meson, Buildroot used to set a range of
autoconf options to bypass tests that require running binaries.
The meson version of libglib2's build system has many fewer of these
checks, but there are still some and these can be fed the "correct"
answer by adding properties to cross-compilation.conf.
Add the necessary properties to indicate that we have C99 compliant
print functions to avoid pulling in the gnulib fallback.
Signed-off-by: John Keeping <john@metanate.com>
Reviewed-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Other changes:
- Change -Dlibmount from a bool to a feature.
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
This patch was missing the following section:
@@ -1,4 +1,4 @@
-project('glib', 'c', 'cpp',
+project('glib', 'c',
version : '2.62.3',
# NOTE: We keep this pinned at 0.49 because that's what Debian 10 ships
meson_version : '>= 0.49.2',
As such, the package still depended on g++.
Update the patch to add the above section.
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Changes include:
- Rename 0002-add-option-to-build-tests.patch to
0002-allow-explicit-disabling-of-tests.patch to work with 2.62.3.
Upstream now has an option called "installed_tests," which can be used to
prevent the unit tests from being built. However, the check only works for
cross-compiling and not host builds, which would mean that the tests still
build when compiling the host variant.
This conditional causes an error on older systems such as Debian8 or Centos7
because the command `objcopy --add-symbol` is used when building the test
"test_resources2," which is not available with the older version of objcopy
provided by those distributions.
Removing the conditionals and checking for just the installed_tests_enabled
option allows for a much simpler patch.
- Change -Dtests=false to -Dinstalled_tests=false in libglib2.mk as per the
above comment.
br-arm-full [1/6]: OK
br-arm-cortex-a9-glibc [2/6]: OK
br-arm-cortex-m4-full [3/6]: SKIPPED
br-x86-64-musl [4/6]: OK
br-arm-full-static [5/6]: OK
sourcery-arm [6/6]: OK
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
When updating to libglib2 2.62.0, the file 0002-add-option-to-build-tests.patch
had to be refactored to work with the new build_tests checks. In the process of
refactoring, a missing check was overlooked in glib/meson.build causing
distributions with older versions of objcopy to fail because the older version
does not contain the --add-symbol argument.
Adding the check for the build_tests argument fixes the issue.
Fixes:
http://autobuild.buildroot.net/results/dcc2352a036aba9650eae8abaaa23819fc67332d
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Fixes the following check-package warning:
package/libglib2/libglib2.mk:90: consecutive empty lines
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
libglib2 needs util-linux only for libmount, which is a configuration option.
Instead, check if util-linux-libmount is selected, and if so, set the option
to true.
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Other changes:
- Refactor 0002-add-option-to-build-tests.patch to work with the new version.
- Add the new option oss_fuzz=disabled to HOST_LIBGLIB2_CONF_OPTS and
LIBGLIB2_CONF_OPTS.
- Change -Diconv=gnu to -Diconv=external as the option has changed.
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Globally change Adam Duskett's email address to aduskett@gmail.com.
Note that one or two of the patches may have been applied upstream with
the old email address, but in that case those patches will anyway be
removed when bumping.
Signed-off-by: Adam Duskett <Aduskett@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Commit 4102db0f7 (package/libglib2: bump to version 2.60.3) did convert
libglib2 over to meson. In doing so, it left a very corner-case along.
When the target is an ARM CPU and the build is in thumb mode, then we
want to ensure that libglib2 is still built in arm mode (because of
inline asm).
But with meson, CFLAGS from the environment are passed to the host
compiler, so the build breaks, and the meson log contains:
Appending CFLAGS from environment: '-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -marm'
No LDFLAGS in the environment, not changing global flags.
No CPPFLAGS in the environment, not changing global flags.
Sanity testing C compiler: cc
Is cross compiler: False.
Sanity check compiler command line: cc -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -marm [...]/build/libglib2-2.60.4/build/meson-private/sanitycheckc.c -o [...]/build/libglib2-2.60.4/build/meson-private/sanitycheckc.exe
Sanity check compile stdout:
-----
Sanity check compile stderr:
cc: error: unrecognized command line option ‘-marm’; did you mean ‘-mabm’?
-----
meson.build:1:0: ERROR: Compiler cc can not compile programs.
Fix that by using the new per-package CFLAGS feature of the meson infra.
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Cc: Adam Duskett <aduskett@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Titouan Christophe