Skip to contentSkip to conversation

corosync: patch CVE-2026-35091

PATCH New Gyorgy Sarvari 1 participant 6 messages Apr 13

lore
Share & tools
lore
#0 lore raw

PATCH 1/6 corosync: patch CVE-2026-35091

Details: https://nvd.nist.gov/vuln/detail/CVE-2026-35091

Pick the patch that mentions the CVE ID explicitly (it was identified by Debian also as the fix[1])

[1]: https://security-tracker.debian.org/tracker/CVE-2026-35091

---
 .../corosync/corosync/CVE-2026-35091.patch    | 47 +++++++++++++++++++
 .../corosync/corosync_3.1.10.bb               |  1 +
 2 files changed, 48 insertions(+)
 create mode 100644 meta-networking/recipes-extended/corosync/corosync/CVE-2026-35091.patch
meta-networking/recipes-extended/corosync/corosync/CVE-2026-35091.patch new file +47 -0
new file mode 100644
index 0000000000..8afa5d6841
@@ -0,0 +1,47 @@
+From b9cb461121c8721c94a94309eb345a3c2f9ee9b4 Mon Sep 17 00:00:00 2001
+From: Jan Friesse <[email protected]>
+Date: Thu, 2 Apr 2026 09:00:39 +0200
+Subject: [PATCH] totemsrp: Return error if sanity check fails
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+Previously, the check_memb_commit_token_sanity function correctly
+checked the minimum message length. However, if the message was too
+short, it incorrectly returned a success code (0) instead of the
+expected failure code (-1).
+
+This commit ensures the appropriate error code is returned when the
+message length sanity check fails.
+
+Fixes: CVE-2026-35091
+
+Reported-by: Sebastián Alba Vives (@Sebasteuo / 0xS4bb1) <[email protected]>
+Signed-off-by: Jan Friesse <[email protected]>
+Also-proposed-by: nicholasyang <[email protected]>
+Reviewed-by: Christine Caulfield <[email protected]>
+
+CVE: CVE-2026-35091
+Upstream-Status: Backport [https://github.com/corosync/corosync/commit/a16614accfdb3481264d7281843fadf439d9ab1b]
+Signed-off-by: Gyorgy Sarvari <[email protected]>
+---
+ exec/totemsrp.c | 4 ++--
+ 1 file changed, 2 insertions(+), 2 deletions(-)
+
+diff --git a/exec/totemsrp.c b/exec/totemsrp.c
+index 35bf971..94d6c21 100644
+--- a/exec/totemsrp.c
++++ b/exec/totemsrp.c
+@@ -3811,10 +3811,10 @@ static int check_memb_commit_token_sanity(
+ 		log_printf (instance->totemsrp_log_level_security,
+ 		    "Received memb_commit_token message is too short...  ignoring.");
+ 
+-		return (0);
++		return (-1);
+ 	}
+ 
+-	addr_entries= mct_msg->addr_entries;
++	addr_entries = mct_msg->addr_entries;
+ 	if (endian_conversion_needed) {
+ 		addr_entries = swab32(addr_entries);
+ 	}
meta-networking/recipes-extended/corosync/corosync_3.1.10.bb +1 -0
index 07d9333ec8..7ccccefed5 100644
@@ -9,6 +9,7 @@ inherit autotools pkgconfig systemd github-releases
 
 SRC_URI = "${GITHUB_BASE_URI}/download/v${PV}/${BP}.tar.gz \
            file://corosync.conf \
+           file://CVE-2026-35091.patch \
            "
 SRC_URI[sha256sum] = "be361c827f99b215b3bd3fa2fb071c03dac6831c2a351963d938caef62604bc8"
 UPSTREAM_CHECK_GITTAGREGEX = "v(?P<pver>\d+(\.\d+)+)"
#1 lore raw

PATCH 2/6 corosync: patch CVE-2026-35092

Details: https://nvd.nist.gov/vuln/detail/CVE-2026-35092

Pick the patch that mentions the CVE ID explicitly (the same commit was identified by Debian also[1])

[1]: https://security-tracker.debian.org/tracker/CVE-2026-35092

---
 .../corosync/corosync/CVE-2026-35092.patch    | 57 +++++++++++++++++++
 .../corosync/corosync_3.1.10.bb               |  1 +
 2 files changed, 58 insertions(+)
 create mode 100644 meta-networking/recipes-extended/corosync/corosync/CVE-2026-35092.patch
meta-networking/recipes-extended/corosync/corosync/CVE-2026-35092.patch new file +57 -0
new file mode 100644
index 0000000000..8182647840
@@ -0,0 +1,57 @@
+From 8f8a4747a0223b8897deda9a40a8a099c61fa80f Mon Sep 17 00:00:00 2001
+From: Jan Friesse <[email protected]>
+Date: Thu, 2 Apr 2026 09:44:06 +0200
+Subject: [PATCH] totemsrp: Fix integer overflow in memb_join_sanity
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+This commit addresses an integer overflow (wraparound) vulnerability
+in the check_memb_join_sanity function.
+
+Previously, the 32-bit unsigned network values proc_list_entries and
+failed_list_entries were added together before being promoted to
+size_t. This allowed the addition to wrap around in 32-bit arithmetic
+(e.g., 0x80000000 + 0x80000000 = 0), resulting in a required_len
+calculation that was incorrectly small.
+
+The solution is to cast the list entries to size_t and verify that
+neither exceeds the maximum allowed value before the addition occurs.
+
+Fixes: CVE-2026-35092
+
+Reported-by: Sebastián Alba Vives (@Sebasteuo / 0xS4bb1) <[email protected]>
+Signed-off-by: Jan Friesse <[email protected]>
+Also-proposed-by: nicholasyang <[email protected]>
+Reviewed-by: Christine Caulfield <[email protected]>
+
+CVE: CVE-2026-35092
+Upstream-Status: Backport [https://github.com/corosync/corosync/commit/4082294f5094a7591e4e00658c5a605f05d644f1]
+Signed-off-by: Gyorgy Sarvari <[email protected]>
+---
+ exec/totemsrp.c | 12 +++++++++++-
+ 1 file changed, 11 insertions(+), 1 deletion(-)
+
+diff --git a/exec/totemsrp.c b/exec/totemsrp.c
+index 94d6c21..6845cec 100644
+--- a/exec/totemsrp.c
++++ b/exec/totemsrp.c
+@@ -3786,7 +3786,17 @@ static int check_memb_join_sanity(
+ 		failed_list_entries = swab32(failed_list_entries);
+ 	}
+ 
+-	required_len = sizeof(struct memb_join) + ((proc_list_entries + failed_list_entries) * sizeof(struct srp_addr));
++	if (proc_list_entries > PROCESSOR_COUNT_MAX ||
++	    failed_list_entries > PROCESSOR_COUNT_MAX) {
++		log_printf (instance->totemsrp_log_level_security,
++		    "Received memb_join message list_entries exceeds the maximum "
++		    "allowed value...  ignoring.");
++
++		return (-1);
++	}
++
++	required_len = sizeof(struct memb_join) +
++	  (((size_t)proc_list_entries + (size_t)failed_list_entries) * sizeof(struct srp_addr));
+ 	if (msg_len < required_len) {
+ 		log_printf (instance->totemsrp_log_level_security,
+ 		    "Received memb_join message is too short...  ignoring.");
meta-networking/recipes-extended/corosync/corosync_3.1.10.bb +1 -0
index 7ccccefed5..77dea16e98 100644
@@ -10,6 +10,7 @@ inherit autotools pkgconfig systemd github-releases
 SRC_URI = "${GITHUB_BASE_URI}/download/v${PV}/${BP}.tar.gz \
            file://corosync.conf \
            file://CVE-2026-35091.patch \
+           file://CVE-2026-35092.patch \
            "
 SRC_URI[sha256sum] = "be361c827f99b215b3bd3fa2fb071c03dac6831c2a351963d938caef62604bc8"
 UPSTREAM_CHECK_GITTAGREGEX = "v(?P<pver>\d+(\.\d+)+)"
#2 lore raw

PATCH 3/6 ez-ipupdate: add CVE tag to CVE-fixing patch

In reply to Gyorgy Sarvari

An already existing patch fixes a CVE (CVE-2004-0980), but it since the patch didn't have the CVE tag, the cve checker did not pick it up.

Rectify this ommission.

CVE details: https://nvd.nist.gov/vuln/detail/CVE-2004-0980
The same patch is used by Gentoo to mitigate this issue. Gentoo CVE advisory: https://security.gentoo.org/glsa/200411-20
Linked Gentoo bug, containing this patch: https://bugs.gentoo.org/69658

---
 .../recipes-connectivity/ez-ipupdate/files/wformat.patch   | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
meta-networking/recipes-connectivity/ez-ipupdate/files/wformat.patch +6 -1
index 3b791559d5..7463e3c535 100644
@@ -1,4 +1,9 @@
-Upstream-Status: Pending
+
+This patch is used by Gentoo to mitigate CVE-2004-0980:
+https://bugs.gentoo.org/69658
+
+CVE: CVE-2004-0980
+Upstream-Status: Inactive-Upstream [last commit: 2002]
 
 Index: ez-ipupdate-3.0.11b7/ez-ipupdate.c
 ===================================================================
#3 lore raw

PATCH 4/6 flatpak: upgrade 1.17.3 -> 1.17.6

In reply to Gyorgy Sarvari

Contains fixes for CVE-2026-34078 and CVE-2026-34079

Add explicit CVE_STATUS tags for these CVEs, because they are tracked without version info by NVD at this time.

Changelog:
17.6:
Bug fixes:
- Fix the remaining regression for Chromium based browsers by not leaking file descriptors down to wrapped command
- Fix a regression when installing extra-data without a runtime, which is the case for openh264
- Fix the remaining regression for Epiphany by ignoring unusable sandbox-expose paths for sub-sandboxes in the portal
- Fix the installed tests by allowing to add a new ref to an existing temporary ostree repo
- Avoid closing fds 0/1/2 when they are used as a bad argument to flatpak-run, and reduce duplication in handling file descriptor arguments

Enhancements:
- Disable auto-pin in flatpak-repair to preserve the pin state across re-installs
- Small improvements for the tests

17.5:
Bug fixes:
- Fix regressions caused by the sandbox escape security fix, which impact some browsers, browser-based apps and Steam (#6577, #6569, #6576, #6574)

Enhancements:
- Expand test coverage of flatpak-run features used by flatpak-portal (#6573)

17.4:
Security fixes:
- Fix a complete sandbox escape which leads to host file access and code execution in the host context (CVE-2026-34078)
- Prevent arbitrary file deletion on the host filesystem (CVE-2026-34079)
- Prevent arbitrary read-access to files in the system-helper context (GHSA-2fxp-43j9-pwvc)
- Prevent orphaning cross-user pull operations (GHSA-89xm-3m96-w3jg)

Enhancements:
- Enable ntsync unconditionally
- Automatic branch following for extensions to ensure that "no-autodownload" extensions stay functional after an update that requires a new branch
- Translation updates: eo, kk, sr, zh_CN

Bug fixes:
- Prevent CPR sequence from showing up in the terminal
- Fix a crash for apps/runtimes with multiarch permission
- Fixes for Coverity warnings
- Add test-preinstall.sh to the test matrix source
- Fix a test message to refer to "systemd-localed" instead of "located"

---
 .../flatpak/{flatpak_1.17.3.bb => flatpak_1.17.6.bb}         | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
 rename meta-oe/recipes-extended/flatpak/{flatpak_1.17.3.bb => flatpak_1.17.6.bb} (93%)
meta-oe/recipes-extended/flatpak/flatpak_1.17.6.bb renamed from meta-oe/recipes-extended/flatpak/flatpak_1.17.3.bb +4 -1
similarity index 93%
rename from meta-oe/recipes-extended/flatpak/flatpak_1.17.3.bb
rename to meta-oe/recipes-extended/flatpak/flatpak_1.17.6.bb
index cd461e2632..1512b7239f 100644
@@ -8,7 +8,7 @@ SRC_URI = " \
     file://0001-flatpak-pc-add-pc_sysrootdir.patch \
 "
 
-SRCREV = "13b26a94a3bd6fec309a16982a3a80d83776d7ac"
+SRCREV = "9b21874f1a175a9b7c79175a221fa043e202ca73"
 
 
 inherit meson pkgconfig gettext systemd gtk-doc gobject-introspection python3native mime features_check useradd
@@ -76,3 +76,6 @@ USERADD_PACKAGES = "${PN}"
 USERADD_PARAM:${PN} = "--system --no-create-home --user-group --shell /sbin/nologin flatpak"
 
 FILES:${PN} += "${libdir} ${datadir}"
+
+CVE_STATUS[CVE-2026-34078] = "fixed-version: fixed in v1.17.4"
+CVE_STATUS[CVE-2026-34079] = "fixed-version: fixed in v1.17.4"
#4 lore raw

PATCH 5/6 libraw: mark fixed CVEs patched

In reply to Gyorgy Sarvari

These CVEs have been fixed already in the current version, however NVD tracks them with incorrect version information.

Commits that fix them:
CVE-2026-20884: https://github.com/LibRaw/LibRaw/commit/aa4458eb511daeae90676c1ce5c587106e4aaec1
CVE-2026-24450: https://github.com/LibRaw/LibRaw/commit/c911c9b9edffa5fab99f828d0fee6dd2d0f6105f

These commits were identified from the changelog of this version[1], which mentions the Talos ID of the vulnerabilities (and the Talos ID is mentioned in the NVD reports[2][3]).

[1]: https://github.com/LibRaw/LibRaw/releases/tag/0.22.1
[2]: https://nvd.nist.gov/vuln/detail/CVE-2026-24450
[3]: https://nvd.nist.gov/vuln/detail/CVE-2026-20884

---
 meta-oe/recipes-support/libraw/libraw_0.22.1.bb | 2 ++
 1 file changed, 2 insertions(+)
meta-oe/recipes-support/libraw/libraw_0.22.1.bb +2 -0
index bd0a4c0b03..2e11a7f1f9 100644
@@ -11,3 +11,5 @@ DEPENDS = "jpeg jasper lcms"
 
 CVE_STATUS[CVE-2026-5318] = "fixed-version: fixed since 0.22.1"
 CVE_STATUS[CVE-2026-5342] = "fixed-version: fixed since 0.22.1"
+CVE_STATUS[CVE-2026-20884] = "fixed-version: fixed since 0.22.1"
+CVE_STATUS[CVE-2026-24450] = "fixed-version: fixed since 0.22.1"
#5 lore raw

PATCH 6/6 minio: ignore irrelevant CVEs

In reply to Gyorgy Sarvari

Details: https://nvd.nist.gov/vuln/detail/CVE-2026-33419
https://nvd.nist.gov/vuln/detail/CVE-2026-34204

These CVEs were filed against minio server, but this recipe is for minio client tools, which is a related, but different project.

Ignore these CVEs.

---
 meta-oe/recipes-extended/minio/minio_git.bb | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
meta-oe/recipes-extended/minio/minio_git.bb +1 -1
index 511dd4d869..d08b5e754d 100644
@@ -169,4 +169,4 @@ CVE_STATUS_GROUPS += "CVE_STATUS_WRONG_CPE"
 CVE_STATUS_WRONG_CPE[status] = "cpe-incorrect: The vulnerability is in minio server, not in minio client-tools"
 CVE_STATUS_WRONG_CPE = "CVE-2018-1000538 CVE-2020-11012 CVE-2021-21287 CVE-2021-21362 \
                         CVE-2021-21390 CVE-2021-43858 CVE-2022-35919 CVE-2023-28433 \
-                        CVE-2023-28434 CVE-2024-36107"
+                        CVE-2023-28434 CVE-2024-36107 CVE-2026-33419 CVE-2026-34204"