Backport upstream fix for CVE-2026-58013 from MR !5170. The fix
prevents a memcmp() from reading past the end of the GString
buffer when using a multi-byte line terminator in GIOChannel.
The patch also includes a unit test for the fix. An additional
#include <stdint.h> was added to the test file to ensure
compatibility with glib 2.80.4 headers.
CVE: CVE-2026-58013
Upstream patches:
- https://gitlab.gnome.org/GNOME/glib/-/merge_requests/5170.patch
Resolves: RHEL-212239
This commit was backported by Ymir, a Red Hat Enterprise Linux software maintenance AI agent.
Assisted-by: Ymir
Backport fix for CVE-2026-58016 from upstream commit
c9da977c178fbfc0e4caf99f9fdf5dc433d6fcc2 (MR !5156).
The fix corrects broken XML parser state handling for
<node> element nesting in gio/gdbusintrospection.c, and
adds unit tests for invalid introspection XML parsing.
CVE: CVE-2026-58016
Upstream patches:
- c9da977c17.patch
Resolves: RHEL-190613
This commit was backported by Ymir, a Red Hat Enterprise Linux software maintenance AI agent.
Assisted-by: Ymir
A patch has been backported to fix an issue where GDBusConnection could send a
message with an invalid serial of zero after the serial counter overflowed.
This could happen in very long-running processes that use a single
GDBusConnection to send a large number of messages.
Upstream fix: b94b44407a.patch
Resolves: RHEL-114857
This commit was backported by Jotnar, a Red Hat Enterprise Linux software maintenance AI agent.
Assisted-by: Jotnar
This is a part of approved Fedora change:
https://fedoraproject.org/wiki/Changes/Separate_dtrace_package
systemtap-sdt-devel will stop requiring systemtap-sdt-dtrace
and that would break the build of this package.
(cherry picked from commit 42c424fc227ddd3a74d8b80617164f0883a2c62e)
Also, rename an old patch because I've soured on my own strategy for
naming patches after merge request IDs, which doesn't tell you what the
patch actually does.
Resolves: RHEL-84440
FIPS mode is currently broken in both Fedora rawhide and RHEL 10. This
test is no longer useful because it's now testing whether GnuTLS works
rather than whether GLib works.
Related: RHEL-44610
They pass at build time, but fail during gating, presumably because
crypto policy is different and blocks SHA-1. Let's just not run these
tests, same as we already do for MD-5. There is plenty of test coverage
for the other HMAC algorithms.
Related: RHEL-44610
Not doing this in RHEL.
This is likely going to break if the container system's runc is
outdated. Consider updating the container host.
Related: RHEL-30209
- Rebase gnutls-hmac.patch
- Add gir and typelib files that moved here from gobject-introspection
- Explicitly conflict with older gobject-introspection
- Switch to building docs with gi-docgen instead of gtk-doc
