Previously it was using killall with procd respand enabled
This was causing yggdrasil to restart after being killed
root@r3test-hap:/# service yggdrasil stop ; echo $? ; sleep 10s ; ps | grep yggdrasil
Terminated
143
6701 root 653m S /usr/sbin/yggdrasil -useconffile /tmp/yggdrasil.conf
6748 root 1308 S grep yggdrasil
Now it's just using whatever procd is using and see there, it actually stops
root@r3test-hap:/# service yggdrasil stop ; echo $? ; sleep 10s ; ps | grep yggdrasil
0
6802 root 1308 S grep yggdrasil
I assume there was some procd bug that kept it from being used properly
Signed-off-by: Maciej Krüger <mkg20001@gmail.com>
According to David Woodhouse, OpenConnect has no issues reconnecting on any
interface. Make the host dependency optional, as it can cause issues in multiple
WAN scenarios.
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
The --juniper switch has been deprecated in favour of --protocol=nc. Fix the
proto script thusly, while keeping compatibility with existing configurations.
Note that, as far as UCI is concerned, if both options juniper and vpn_protocol
are specified, the latter takes precedence.
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
Using resolveip is more robust and predictable than depending on nslookup and
awk.
This reverts commit 131ec7b3bd.
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
banIP 0.7.x is not compatible with new nft firewall (default in master and 22.03).
Mark the package as BROKEN for now.
Signed-off-by: Dirk Brenken <dev@brenken.org>
Force restart stubby if any of the trigger interfaces goes up or down.
Avoids DoT DNS lookup timeouts when default route changes, in case of multiple
upstream interfaces.
Signed-off-by: Aleksandr V. Piskunov <aleksandr.v.piskunov@gmail.com>
This commit fixes two issues on macos:
1. Added a patch to fix 'echo -n' issue with MacOS shell
(backported from upstream)
2. Redefined sys.platform='linux' for target build if build host is
MacOS (otherwise, build script tries to use MacOS logic for
OpenWrt(Linux) target build)
Signed-off-by: Sergey V. Lobanov <sergey@lobanov.in>
Use nft instead of iptables to open port 80 in the firewall when getting a
cert. Since nft doesn't allow deleting a rule by its contents, capture and
save the handle when creating the rule, and use that to delete.
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
Backport a patch in order to allow building OpenConnect against OpenSSL 1.1.x
without the need for deprecated API (further fixes will be required for OpenSSL
3.x, though).
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
On systems using seccomp, the hostapd socket files will be owned by the
'network' user/group ([source][0]). In this case, if wifi-presence is
run as root/root, then it does not have permissions to open the
hostapd socket files. This was discussed in awilliams/wifi-presence#3.
This change allows the process user/group to be specified in
/etc/config/wifi-presence. If no explicit user/group is set, then the
init script will use the owner of the socket files in /var/run/hostapd/
to determine the appropriate process user/group.
[0]: ec6293febc/package/network/services/hostapd/files/wpad.init (L35-L36)
Signed-off-by: Adam Williams <pwnfactory@gmail.com>
softflowd can filter the traffic with an optional bpf program,
specified on the command-line as a BPF expression
Signed-off-by: Jesus Fernandez Manzano <jesus.manzano@galgus.net>
1. Fixed init script would kill itself when trying to stop a service.
2. Upgrade privoxy release to 3.0.33
3. Set PKG_RELEASE to AUTORELEASE
Signed-off-by: He Xian <hexian000@outlook.com>
at least driver apcsmart-old (maybe more) allow for specifying the
type of cable used. My old UPS does will not function when cable type
is not specified.
This will add support for configuration option 'cable'
Signed-off-by: Rob J. Epping <epping@renf.us>
Now with basic support for the Array Networks SSL VPN protocol.
Also fix the OpenSSL build. OpenConnect requires support for deprecated APIs,
for the time being, so select them if compiling against OpenSSL.
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
This adds support for the child SA to be rekeyed through the byte/packet
threshold. The default is blank (which disables the byte/packet thresholds).
Signed-off-by: Joel Low <joel@joelsplace.sg>
Otherwise it will fail as follows:
failed to find a module named mdio-netlink
ERROR: mdio-netlink module not detected, and could not be loaded.
Run-tested on: ramips/mt7621
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
To fix the errors:
Sun Apr 10 14:19:41 2022 daemon.err transmission-daemon[29831]: [2022-04-10 14:19:41.098] watchdir Failed to open directory "/mnt/sda1/openwrt/transmission/watch" (2): No such file or directory (watchdir.c:358)
and
Sun Apr 10 14:20:18 2022 daemon.err transmission-daemon[30175]: [2022-04-10 14:20:18.641] Couldn't create "/mnt/sda1/openwrt/transmission/incomplete": Permission denied (file-posix.c:243)
References:
- https://github.com/openwrt/packages/issues/17674
Signed-off-by: Alexander Egorenkov <egorenar-dev@posteo.net>
* add new 'hblock' compilation source (XL, see https://hblock.molinero.dev for reference)
* print runtime/date information in ISO-8601 standard format
* minor cleanups
Signed-off-by: Dirk Brenken <dev@brenken.org>
The dependency has a PACKAGE_uacme-ualpn condition so that libev won't
be unnecessarily built if uacme-ualpn is not selected.
Remove PKG_USE_MIPS16:=0, as it is not necessary when not using the
libev that is bundled with uacme.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
The default firewall is the fw4, which uses nft. In order to not
install the legacy implementation when installing strongswan, the build
system should decide which firewall backend to use.
While we are at it, I have also added the dependency packages for IPV6.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
The host build is used to build kea-msg-compiler, which is only needed
when there are changes to .mes files. Since we're not making any changes
to such files, we do not need this.
As host build fails for Kea 2.0.2, and the git history for kea doesn't
contain any reasoning for enabling it, let's just drop it.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Using https://gitlab.freedesktop.org/mobile-broadband/ModemManager.git to download the source code.
Added compile option to compile qrtr support.
Enabled lto and additional gcc flags for perfomance and less size.
Modified to use meson as upstream has abandoned autotools.
Removed BUILD_PARALLEL options. These are default with ninja/meson.
Signed-off-by: Maxim Anisimov <maxim.anisimov.ua@gmail.com>
Adding libsctp adds IPV6 dependency to gensio, so this patch is
an attempt at working around that with the goal of getting rid of
the circular dependency error.
Signed-off-by: Nita Vesa <werecatf@outlook.com>
The updated version requires libgensio, libyaml and libpthread,
so those have been added accordingly to dependencies.
Also added arguments for the configure-script to always attempt
to build ser2net with the same settings, instead of leaving it
to guess, for consistency.
Signed-off-by: Nita Vesa <werecatf@outlook.com>
Let's move the iptables IPsec dependencies out of the strongswan package
and into the plugin package that actually depends on it,
strongswan-mod-updown. As the default updown script calls the iptables
binary, also add a dependency on the iptables-legacy package.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Change notes:
Updated Makefile package version and hash.
Added libpcre2 dependency
Removed USELIBPCRE make flag (no longer optional within sslh)
Updated patch 001 to work with new sslh Makefile
Signed-off-by: Martin Moreno <fett3270@yahoo.com>
Make sure /etc/gnunet and all its files and sub-directories are owned
by gnunet:gnunet. This is somehow necessary as file ownership otherwise
doesn't survive sysupgrade.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Enable AUTORELEASE in a separate commit so that the next commit can be
reverted without having to manually re-introduce it.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Only notable change since 0.14.0 is that pthread_mutexattr_setkind_np()
is now no longer used.
pthread_mutexattr_setkind_np() is deprecated and non-standard.
The standard version is called pthread_mutexattr_settype()
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This is a bugfix release for gnunet 0.16.1.
Noteworthy changes in 0.16.2 (since 0.16.1)
DHT: Fix path signature handling.
GNS: Fix BOX handling in zone apex.
NAMESTORE: Prevent storing under invalid labels.
Buildsystem: Fix build on *BSD and Guix.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Fixes multiple security issues:
* CVE-2022-0667 -- An assertion could occur in resume_dslookup() if the
fetch had been shut down earlier
* CVE-2022-0635 -- Lookups involving a DNAME could trigger an INSIST when
"synth-from-dnssec" was enabled
* CVE-2022-0396 -- A synchronous call to closehandle_cb() caused
isc__nm_process_sock_buffer() to be called recursively,
which in turn left TCP connections hanging in the CLOSE_WAIT
state blocking indefinitely when out-of-order processing was
disabled.
* CVE-2021-25220 -- The rules for acceptance of records into the cache
have been tightened to prevent the possibility of
poisoning if forwarders send records outside the
configured bailiwick
Signed-off-by: Noah Meyerhans <frodo@morgul.net>
As per the discussion in PR #18047, split the MTR package into
two, one with jansson enabled for JSON output, and one without.
This commit also bumps the version to 0.95. Since the MTR project
website does not seem to be updated with builds any longer, switch
to GitHub Codeload instead.
Also enable PKG_FIXUP:=autoreconf so that MTRs bootstrap.sh process
is executed properly.
Signed-off-by: Marc Egerton <marc@malloc.me>
* remove upstreamed gcc10 and cerrno patches
* disable SSO and OIDC as it needs Rust/Cargo support
Signed-off-by: Moritz Warning <moritzwarning@web.de>
Enable AUTORELEASE in a separate commit so that the next commit can be
reverted without having to manually re-introduce it.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Note that on 32-bit ARM with MUSL we don't have Unwind_GetIP() so
we need to disable backtraces.
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
Unless we're using "mktemp -u ..." (not recommended), it will
create the temp file as part of its safety checking. Thus you
should only create the name (file) if you're going to use it,
and always remove it if you have created it.
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
If named is configured to not listen on any IPv6 interfaces,
then we should run 'nsupdate' with the '-4' argument.
Also:
* cleanup RFC-1918 address detection;
* don't generate PTR records for domain entries that aren't
RFC-1918 addresses or these will generate NOTAUTH failures;
We're assuming that we're doing DNS split-horizon and that
internal addresses aren't routable.
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
To allow using gnunet on systems with firewall4, add replace direct
dependency on 'firewall' with 'uci-firewall' which is satisfied by
either 'firewall' or 'firewall4'.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
