On Wed Nov 4, 2026 at 6:53 AM CET, Darsh Kelaiya -X (dkelaiya - E INFOCHIPS PRIVATE LIMITED at Cisco) via openembedded-core wrote:
From: Darsh Kelaiya <dkelaiya@cisco.com>Hello,
Backport the upstream fixes for CVE-2026-54369 and CVE-2026-54370 to
ACL 2.3.2.
CVE-2026-54369 adds descriptor-relative ACL APIs so callers can control
symlink traversal. CVE-2026-54370 uses those APIs to harden getfacl,
setfacl, setfacl --restore and chacl against path replacement races and
unsafe symlink traversal.
Retain ACL 2.3.2 on the stable Wrynose branch and backport the security
fixes with the required _at APIs and helper dependencies. This follows
OE-Core's focus on security and bug fixes for stable branches [1],
avoiding a general recipe upgrade to ACL 2.4.0. Existing pathname API
behavior is preserved.
The new public acl_*_at declarations conflict with private helpers
using the same names in GNU tar 1.35. Patch 1 backports the upstream
tar rename before patches 2 and 3 introduce and consume
the new ACL APIs.
[1] https://docs.yoctoproject.org/6.0/ref-manual/release-process.html#stable-release-process
Darsh Kelaiya (3):
tar: avoid name conflicts with new libacl APIs
acl: fix CVE-2026-54369
acl: fix CVE-2026-54370
.../tar/tar/Avoid-acl_-prefix-for-functions.patch | 103 ++
meta/recipes-extended/tar/tar_1.35.bb | 1 +
.../attr/acl/CVE-2026-54369-dependent_p1.patch | 69 ++
.../attr/acl/CVE-2026-54369-dependent_p2.patch | 200 ++++
.../attr/acl/CVE-2026-54369-dependent_p3.patch | 712 +++++++++++++
meta/recipes-support/attr/acl/CVE-2026-54369_p1.patch | 719 +++++++++++++
meta/recipes-support/attr/acl/CVE-2026-54369_p2.patch | 957 ++++++++++++++++++
meta/recipes-support/attr/acl/CVE-2026-54369_p3.patch | 109 ++
.../attr/acl/CVE-2026-54370-dependent_p1.patch | 168 +++
.../attr/acl/CVE-2026-54370-dependent_p2.patch | 761 ++++++++++++++
.../attr/acl/CVE-2026-54370-dependent_p3.patch | 439 ++++++++
.../attr/acl/CVE-2026-54370-dependent_p4.patch | 140 +++
meta/recipes-support/attr/acl/CVE-2026-54370_p1.patch | 299 ++++++
meta/recipes-support/attr/acl/CVE-2026-54370_p2.patch | 403 ++++++++
meta/recipes-support/attr/acl/CVE-2026-54370_p3.patch | 603 +++++++++++
meta/recipes-support/attr/acl/CVE-2026-54370_p4.patch | 213 ++++
meta/recipes-support/attr/acl_2.3.2.bb | 14 +
17 files changed, 5910 insertions(+)
create mode 100644 meta/recipes-extended/tar/tar/Avoid-acl_-prefix-for-functions.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54369-dependent_p1.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54369-dependent_p2.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54369-dependent_p3.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54369_p1.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54369_p2.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54369_p3.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370-dependent_p1.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370-dependent_p2.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370-dependent_p3.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370-dependent_p4.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370_p1.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370_p2.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370_p3.patch
create mode 100644 meta/recipes-support/attr/acl/CVE-2026-54370_p4.patch
This is quite a big series ("5910 insertions(+)"). Looking at it, I think
this might be just backporting most of the risky parts of the 2.4.0
upgrade, but then, we might just be better off simply upgrading to
2.4.0. Do you have thoughts about that? Are there definite breakage on a
potential 2.4.0 upgrade?
Obviously, a 2.4.0 upgrade would need serious thoughts/discussion but
this might be the way forward.
Regards,
--
Yoann Congal
Smile ECS