We want to use systemd for managing the lifetime of OpenVPN. Take
the debian generator (which should work with busybox ash) and the
openvpn.service (to inhibit the sysvinit script) and the target
file and install it.
On systems that have ran "update-rc.d openvpn defaults" one need
to manually execute a systemctl enable openvpn.service. This is
not done through a post-inst script and I am not sure if we should
do it. This means there is a danger of ending with a unit that
doesn't start OpenVPN automatically after upgrade!
The scripts/packages have not been tested on a device yet.
We should use a more recent version of the OpenVPN client. Import
it from the meta-networking layer like we did before. The sysvinit
script has just minor whitespace differences and wasn't imported.
We need to test opkg upgrades of OpenVPN through the VPN. We don't
want to kill the tunnel on such upgrade.
One might need to erase the environment but with a default
environment barebox will now send a LLDP during boot up. This
way we can see if a device is on a local link or not.
Fixes: SYS#488
* split out rtl-nic firmware from the big
linux-firmware package
[hfreyther: For whatever reasons the NIC works without
additional firmware as well. I couldn't see in the driver
source when that is the case and which features are missing]
Fixes: SYS#732
[hfreyther: I removed the MACHINE_ESSENTIAL_EXTRA_RDEPENDS as applications
that use the symlinks should rdepend on the package]
[hfreyther: Changed mismatch of tabs/spaces]
We want to enable the watchdog in the bootloader already and
for this to be failsafe we need to have a kernel that can boot
and get ethernet without any additional modules. Now systemd
would not be able to trigger the watchdog and the system would
reboot within 60 seconds. I am not sure how the kernel upgrade
will go and we need to try this.
Related: SYS#374
We would leak all the results for all the scans. At some point in
time the kernel OOM would kick-in and kill the process. Enable
building with -ggdb3 for debug symbols and remmeber the old list
head so we can free it.
==32240== 756 bytes in 3 blocks are definitely lost in loss record 7 of 8
==32240== at 0x40291CC: malloc (in /usr/lib/valgrind/vgpreload_memcheck-x86-linux.so)
==32240== by 0x804BCF5: iw_process_scanning_token (iwlib.c:3059)
==32240== by 0x804BCF5: iw_process_scan (iwlib.c:3255)
==32240== by 0x804BEAF: iw_scan (iwlib.c:3310)
==32240== by 0x8048FD8: scan_wifi (wifi2udp.c:72)
==32240== by 0x8048DB2: main (wifi2udp.c:143)
The notion of "task" has been deprecated for a long time because
the word is used for different things. The new word is "packagegroup"
which describes the usage of "task-*" in a more obvious way. Dora
and master builds can fully use packagegroup-* for everything and
for edison we would need to spend some time to add a provides for
packagegroup-* to the task- group. I don't intend to work on the
edison part right now.
The notion of task- has been deprecated a long time ago. Call
it packagegroup now and for "packagegroup-osmocom" provide the
compability. This has not been compile tested.
The edison build will require a bbappend to provide packagegroup
now but I will do this once I need to build for Edison again.
We need a jffs2/ubi depending on the age of the hardware and
maybe a tar.bz2 so we can easily inspect an image without
playing with mtdram. We almost never need a ubifs and certainly
not a cpio.
The DISTRO_FEATURES_INITMAN variable has been removed, use
VIRTUAL-RUNTIME_init_manager which is still present. We have
already defined the later in our poky builds
systemd-compat-units does not install any files anymore and
the machineid and other scripts are gone. We want to continue
to install the alignment file though.
This has not been tested on dora systemd yet.
In dora there is no autotools-brokensep bbclass. So let's just
set the build directory to the source directory ourselves. I have
not built the package for dora though/
Poky dizzy is more strict with -I/usr/include and the old
tcpdump could pick-up "/usr/bin/pcap-config" from the host
and fail. Just upgrade to a newer version of tcpdump and be
done with it.
* the entry "option subnet" takes the
subnet mask not the subnet address,
while the busybox dhcp-client was
fine with it, the isc-dhcp-client complained with
Error: an inet prefix is expected rather than "10.23.24.113/10.23.24.0".
WARNING: QA Issue: ortp rdepends on libcrypto, but it isn't a build dependency? [build-deps]
WARNING: QA Issue: ortp rdepends on libssl, but it isn't a build dependency? [build-deps]
We used asterisk to be able to speak SIP with the outside. LCR
has gained this capability in the past and we haven't used asterisk
in more than two years.
Asterisk doesn't seem to compile with master/dizzy and there is
little point in figuring out why.