- CVE-2026-27456: mount(8) TOCTOU symlink attack via loop device. The SUID
mount follows symlinks when resolving loop backing file paths. On
systems where non-root users are permitted to mount loop devices (via
'user' option in fstab), this allows access to arbitrary files.
For more information, see:
- https://nvd.nist.gov/vuln/detail/CVE-2026-27456
- 5e390467b2
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
(cherry picked from commit cd52acda14)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
The host-mkpasswd package needs the <crypt.h> header file installed
on the host. This file was initially provided by the libc headers,
as part of its libcrypt implementation. libcrypt was removed in
glibc 2.39. Most Linux distributions continued to ship a libcrypt
replacement such as libxcrypt in their default installation.
Some mainstream Linux distribution (such as Debian testing) are
now starting to remove this libcrypt and <crypt.h> header from their
default image (of development files).
In that case, host-mkpasswd is failing to build with the error:
/buildroot/output/build/host-mkpasswd/mkpasswd.c:54:10: fatal error: crypt.h: No such file or directory
The issue can be reproduced by adapting the Buildroot Docker file,
by setting "FROM debian:testing-20260316" in [1] and removing
the build of tar.
Note: debian 13 (for example Docker tag trixie-20260316) still
includes the libcrypt-dev package by default, which provides crypt.h.
In Buildroot, there is already the host-heimdal and host-systemd
packages requiring host-libxcrypt. Also, the host-libxcrypt is a
small standalone package. For those reasons, instead of adding
the Buildroot prerequisite on the user host to provide a libcrypt
implementation, this commit simply add the package dependency to
host-libxcrypt, to make sure it will work in all situations.
[1] https://gitlab.com/buildroot.org/buildroot/-/blob/2026.02/support/docker/Dockerfile#L11
Signed-off-by: Julien Olivain <ju.o@free.fr>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
(cherry picked from commit cf26c77280)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
https://github.com/libexif/libexif/blob/v0.6.26/NEWS
* Security issues fixed:
* CVE-2026-40386: An unsigned integer underflow in Fuji and Olympus
makernote handling
* CVE-2026-40385: An unsigned integer overflow on 32bit systems in
Nikon makernote handling
* CVE-2026-32775: A buffer overwrite via integer underflow in makernote
handling
Switched to xz tarball and tarball hash provided by upstream.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
(cherry picked from commit 3f9bcf42c3)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
This fixes the following vulnerability:
- CVE-2026-25075:
strongSwan versions 4.5.0 prior to 6.0.5 contain an integer underflow
vulnerability in the EAP-TTLS AVP parser that allows unauthenticated
remote attackers to cause a denial of service by sending crafted AVP
data with invalid length fields during IKEv2 authentication. Attackers
can exploit the failure to validate AVP length fields before
subtraction to trigger excessive memory allocation or NULL pointer
dereference, crashing the charon IKE daemon.
https://www.cve.org/CVERecord?id=CVE-2026-25075
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
(cherry picked from commit cc7c20d817)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Using this defconfig
BR2_x86_64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_PER_PACKAGE_DIRECTORIES=y
BR2_PACKAGE_STRONGSWAN=y
BR2_PACKAGE_STRONGSWAN_WOLFSSL=y
applied after 'git reset 10a70b1af6 --hard'
replicates the build error
http://autobuild.buildroot.net/results/278b3f74c48c858ae368d59069752adb69c05246
which was fixed in 2024 by
89d512729c
Note: The git tree was reset to the commit right before the fix.
Instead of using the fix mentioned above the build error is also fixed
when the buildroot commit 6c18375434 is
cherry-picked onto 10a70b1af6.
This means the build error mentioned above is really fixed by enabling
BR2_PACKAGE_WOLFSSL_ALL. Therefore we can remove the content of commit
89d512729c.
This commit also prevents a build error in strongswan
wolfssl_rsa_private_key.c: In function 'create_empty':
wolfssl_rsa_private_key.c:450:18: error: 'RsaKey' has no member named 'rng'
450 | this->rsa.rng = &this->rng;
caused by the upcoming bump of wolfSSL from 5.7.2 to 5.9.1 due to
upstream commit
cd88a8ae88
which was first included in wolfSSL 5.9.0.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
(cherry picked from commit 5d6378db9d)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Fixes the following security vulnerability:
- CVE-2026-23865:
An integer overflow in the tt_var_load_item_variation_store function
of the Freetype library in versions 2.13.2 and 2.13.3 may allow for an
out of bounds read operation when parsing HVAR/VVAR/MVAR tables in
OpenType variable fonts. This issue is fixed in version 2.14.2.
For more information, see
- https://www.cve.org/CVERecord?id=CVE-2026-23865
- fc85a25584.patch
(cherry picked from commit 6c3933d14b)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
- CVE-2026-34743:
XZ Utils provide a general-purpose data-compression library plus
command-line tools. Prior to version 5.8.3, if lzma_index_decoder()
was used to decode an Index that contained no Records, the resulting
lzma_index was left in a state where where a subsequent
lzma_index_append() would allocate too little memory, and a buffer
overflow would occur. This issue has been patched in version 5.8.3.
For more information, see:
- https://www.cve.org/CVERecord?id=CVE-2026-34743
- https://security-tracker.debian.org/tracker/CVE-2026-34743
- c8c22869e7
(cherry picked from commit 7246352273)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Commit [1] introduced a set of patches to fix CVE-2025-31115.
Since [2] the security patches neeed to reference the vulnerability with
the `CVE: ` trailer in the patch header.
This set of patch is no longer present on master branch with xz bump [3]
but hasn't been cherry picked to 2025.02.x branch.
[1] 06c6c49fe8 package/xz: add security patches fixing CVE-2025-31115
[2] 1167d0ff3d docs/manual: mention CVE trailer
[3] aaa50b0e61 package/xz: bump to version 5.8.1
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
See the release notes: https://www.python.org/downloads/release/python-31213/
This fixes a handful of bugs and the following vulnerability:
- CVE-2024-6923:
There is a MEDIUM severity vulnerability affecting CPython. The
email module didn’t properly quote newlines for email headers when
serializing an email message allowing for header injection when an
email is serialized.
https://www.cve.org/CVERecord?id=CVE-2024-6923
This also includes a mitigation for a libexpat vulnerability:
- CVE-2025-59375:
libexpat in Expat before 2.7.2 allows attackers to trigger large
dynamic memory allocations via a small document that is submitted for
parsing.
https://www.cve.org/CVERecord?id=CVE-2025-59375
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
(cherry picked from commit f862711b10)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
From the release notes:
https://github.com/redis/redis/blob/7.2.13/00-RELEASENOTES
================================================================================
Redis 7.2.13 Released Mon 23 Feb 2026 10:00:00 IST
================================================================================
SECURITY: There is a security fix in the release
* A user can manipulate data read by a connection by injecting \r\n sequences
into a Redis error reply
Signed-off-by: Titouan Christophe <titouan.christophe@mind.be>
(cherry picked from commit 29a4927157)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
Those target configurations marked as "secondary" will be tested with a
lower probability in autobuilders. The intent is to focus the Buildroot
development (and bugfixes) on mainstream configurations, while still
keeping some other, less common, configurations.
This commit marks the following CPU architectures as secondary:
BR2_arcle (ARC little endian)
BR2_armeb (ARM big endian)
BR2_aarch64_be (AArch64 big endian)
BR2_hppa (PA-RISC)
BR2_loongarch64
BR2_m68k
BR2_microblazeel (Microblaze AXI, little endian)
BR2_microblazebe (Microblaze non-AXI, big endian)
BR2_mips (MIPS, big endian)
BR2_mipsel (MIPS, little endian)
BR2_mips64 (MIPS64 big endian)
BR2_or1k (OpenRISC)
BR2_powerpc (PowerPC, big endian)
BR2_powerpc64 (PowerPC64, big endian)
BR2_s390x
BR2_sh (SuperH)
BR2_sparc
BR2_sparc64
BR2_xtensa
In the following CPU architecture, only specific configurations are
marked secondary:
In BR2_armeb (ARM little endian),
- armv4 cores (arm920t, arm922t, fa526/626, strongarm sa110/sa1100)
In BR2_riscv:
- all 32-bit configurations
- 64-bit no-MMU
In BR2_i386 (x86 32-bit),
- i486, i586 and X1000 CPU
There was no strict rule established to build this list of secondary
configurations. This list was built mainly from the observation of
which architecture/CPUs are still widely used in the field
(disregarding its age), the quality of their upstream support and
the general relevance in the Buildroot project.
Signed-off-by: Julien Olivain <ju.o@free.fr>
[Arnout: use !BR2_HIDE_SECONDARY_TARGET_OPTIONS instead of
BR2_SHOW_SECONDARY_TARGET_OPTIONS]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
(cherry picked from commit ddce0814a7)
[thomas: kept x1000 & i586 as primary]
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
The option is currently unused, which leads to a check-symbol warning.
Suggested-by: Arnout Vandecappelle <arnout@rnout.be>
Signed-off-by: Julien Olivain <ju.o@free.fr>
[Arnout: squash two patches]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
(cherry picked from commit 4642f903cd)
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
This avoids having to run b4 prep --set-prefixes 2025.02.x to add the
2025.02.x prefix to the patch subject.
You do need b4 0.11+ to make use of that.
Signed-off-by: Quentin Schulz <quentin.schulz@cherry.de>
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
The master branch bumped libspdm in commit [1] that fixes the build
error that appeared with openssl bump [2].
The 2025.02.x branch remained on openssl 3.4 and didn't need a backport.
With the EOL of openssl 3.4 and 2025.02.x is moving to 3.5 this backport
is now necessary to build libspdm.
[1] a01c07f572 package/libspdm: bump version to 3.8.0
[2] 9868ca6ec9 package/libopenssl: bump version to 3.5.0
Signed-off-by: Thomas Perale <thomas.perale@mind.be>
