The patch carried only a subject line and a placeholder commit id, so
nothing recorded why marking struct top_hash_entry as PACKED fixes a
bus error, or whether it can be dropped.
Restore the description and the sign-off from the commit that
introduced it in this feed, keeping the original author's wording,
and record the upstream status. Verified that BIRD 2.19.2 still
declares the structure without any alignment attribute, so the patch
is still needed.
The header keeps the "---" separator the sibling patch uses, and
orders Signed-off-by last among the trailers.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The script was a single bare command with no shebang and a trailing
space. Every other test-version.sh in the OpenWrt feeds starts with
#!/bin/sh and dispatches on $PKG_NAME, so that a sub-package added
later fails loudly instead of being probed with a command that
happens to work for a different package.
Also use grep -F, so the dots in the version are not treated as
regular expression wildcards, and record why the daemon is checked
rather than birdc/birdcl, which do not report a version of their
own.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The script was a single bare command with no shebang and a trailing
space. Every other test-version.sh in the OpenWrt feeds starts with
#!/bin/sh and dispatches on $PKG_NAME, so that a sub-package added
later fails loudly instead of being probed with a command that
happens to work for a different package.
Also use grep -F, so the dots in the version are not treated as
regular expression wildcards, and record why the daemon is checked
rather than birdc/birdcl, which do not report a version of their
own.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
/etc/bird4.conf and /etc/bird6.conf are left over from the bird1
packages, which shipped a separate configuration file per address
family. bird3 is a single daemon handling both and installs
/etc/bird.conf only, so the two entries refer to files this package
never creates.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
/etc/bird4.conf and /etc/bird6.conf are left over from the bird1
packages, which shipped a separate configuration file per address
family. bird2 is a single daemon handling both and installs
/etc/bird.conf only, so the two entries refer to files this package
never creates.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The bird3cl package was titled as "(v2)" client although it is built
from BIRD 3. Fix the copy-paste leftover from the bird2 Makefile.
Also add PKG_CPE_ID, the same product identifier as used by bird2,
so published CVEs can be matched against this package.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Replace the free-form license boilerplate with an SPDX license
identifier, in line with the other packages in this feed.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
get_interface_address() calls io.open() and dereferences the result
without checking it. /sys/class/net does not contain interfaces
only: with the bonding module loaded it also holds the plain file
bonding_masters, which the `ls -1` in get_interfaces_names() returns
and which has no address below it. io.open() then returns nil and
the script dies with "attempt to index a nil value", so no bat-hosts
data is published at all on such a node.
Return nil when the file cannot be opened and skip those entries in
the caller, which would otherwise index the interface table with
nil. get_hostname() gets the same guard.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
- Install binaries with INSTALL_BIN instead of `cp -fpR`, which is
the standard way to install executables with defined permissions.
- Replace the `[ "x$(CONFIG_...)" == "xy" ] && ... ; true` shell
constructs with make-level $(if ...) conditionals. The `==`
operator is a bashism and the shell round-trip is not needed to
decide at install time whether an optional file is packaged.
- Drop the explicit dependency on libc, which every package depends
on implicitly.
- List the optionally installed /etc/alfred/bat-hosts.lua in
conffiles, so local modifications survive upgrades.
- Indent MAKE_FLAGS consistently with tabs.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The comment claimed the version check has to be skipped because
running the script may trigger a reboot. That is not what happens:
reboot_now() is only reached once naywatch has seen a neighbour
(ACTIVE=1) and then lost it for more than MIN_KICK intervals, and
the watchdog branch is not taken either, because the probe leaves
use_watchdog empty.
What does happen is that naywatch takes positional arguments - check
interval, watchdog timeout, and so on - rather than options, so a
probe flag ends up as the check interval and the script enters its
monitoring loop instead of reporting anything, until the probe
timeout kills it.
The override stays, only its justification is corrected.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The patch carried the raw `git format-patch` output including the
"diff --git"/"index" lines and the trailing git version signature,
which the refresh check of the CI strips, so any PR touching this
package failed with:
Dirty patches detected, please refresh and review the diff
Normalize the patch to the refreshed form. No content change.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
prince requires a configuration file and does not provide any option
to report its version, so the generic version check of the CI
runtime tests always fails for this package. Skip the version check.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The package version is derived from PKG_SOURCE_DATE and the git
hash, which the binary does not report, so the generic version check
of the CI runtime tests always fails for this package. Skip the
version check.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
The package version is derived from PKG_SOURCE_DATE and the git
hash, which neither ohybridproxy nor zonestitcher report, so the
generic version check of the CI runtime tests always fails for these
packages. Skip the version check.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
naywatch is a watchdog shell script which does not provide any
option to report its version, so the generic version check of the CI
runtime tests fails for this package. On top of that, the generic
check executes each installed executable with several flags to probe
for the version, which must never happen for a watchdog script that
may trigger a reboot of the test environment.
Providing a test-version.sh both skips the version probing of the
binary and overrides the version check.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
babel-pinger does not provide any option to report its version, so
the generic version check of the CI runtime tests always fails for
this package. Skip the version check.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
ahcpd does not provide any option to report its version, so the
generic version check of the CI runtime tests always fails for this
package. Override it by checking that the binary starts and prints
its usage instead.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
- Enable bot comments to provide contributors with immediate feedback
- Enable stale bot
- Execute the feeds package test build only when the formalities check succeeds
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Version 3.3.1 (2026-06-09)
o BGP: Fix crash when incoming connection for disabled protocol arrives
o BGP: Fix parsing labelled NLRIs with no next hop
o RCU: Catch leaks sooner
Version 3.3.0 (2026-05-25)
o BGP: Export memory consumption optimization
o BGP: Fix hostentry handling of MPLS and EVPN routes
o BGP: Block connections on explicitly disabled instances
o Proto/CLI: Lock show-commands
o Proto: Loop persists through down state
o Nest/CLI: Show only the longest prefix match in 'show route for'
o Nest: Lockless attribute cache
o Removed locking in random number generation
o ASPA: Fix downstream validation
o BMP: Fix route sending
o BGP: Fix route refresh after restart
o BGP: Fix dynamic peer connection
o Filters: Fix string attributes
o Filters: Fix ROA check autoreload reconfiguration
o Logging: Fix error handling
o Kernel: Fix graceful recovery
o Pipe: Fix rare collision bug
o Config: Allow keyword redefinition
o Merged 2.19
Signed-off-by: Alan Feng <fthasdd@largerworld.cc>
This release fix problems below:
o BGP/MPLS Ethernet VPNs using VXLAN tunnels
o BGP: PMSI tunnel attribute
o Linux Netlink implementation for bridge interfaces
o BGP: Automatic peering based on discovered neighbors
o RAdv: Router discovery based on incoming Router Advertisments
o Nest: Add message when attempting to reload protocol that is not UP
o BMP: Fix off-by-one buffer overflow
o OSPF: Fix infinite loop in OSPF Graceful Restart
o ASPA: Fix downstream validation
o Filters: Fix string attributes
o Logging: Fix error handling
o Minor filter improvements
o Various documentation fixes
Signed-off-by: Bradford Zhang <zyc@zyc.name>
Linux 6.18.33 stable backported kmalloc_obj/kmalloc_objs/kzalloc_obj,
causing a redefinition conflict with compat-hacks.h which also defines
these for kernels before 7.0.0. Exclude the 6.18.33-6.18.x stable
range from the backport definitions.
Signed-off-by: Janusz Dziedzic <janusz.dziedzic@gmail.com>
The EXTRA_CFLAGS for batman-adv were not applied with Linux 6.18. As
result, C files for a specific features were build but the datastructures
were missing the members for this specific feature.
This can be avoided by switching to a more modern way of setting up the
make command (like in this case "mt76"). Besides simplifying and cleaning
up the code, it is now also directly providing the CONFIG preprocessor
variables as part of NOSTDINC_FLAGS. This works on the current kernel
versions 6.12 (default) and 6.18 (testing).
Signed-off-by: Sven Eckelmann <sven@narfation.org>
