Update copyright

Update copyright

Josef Schlehofer authored 4 years ago and Jan Pavlinec committed 4 years ago

Fixes:
CMake Error at cmake/plugin.cmake:104 (MESSAGE):
Invalid value for PLUGIN_ROCKSDB

On top of it, this change is already in packages repository in mariadb's
Makefile

It is going to be part of upcoming release of OpenWrt - 20.xx/21.xx

Before this commit:
target aarch64_cortex-a53 was shown as OpenWrt array
and it was not split as it was for Turris Omnia and Turris 1.x that
there are different jobs for each version.

This solves this issue and there are two jobs for this targets with
different OpenWrt version.

Aleksandr Gumroian authored 4 years ago and Josef Schlehofer committed 4 years ago

* Fix pendingMigrationAlert when no drives connected
* Refactor storageState variables
* Update Foris JS library to v5.1.8
* NPM audit fix

Aleksandr Gumroian authored 4 years ago and Josef Schlehofer committed 4 years ago

Fixed:
* fix regression by bringing back some attributes that were removed in
latest API cleanup

Changed:
* cleanup and refactoring in module API (remove old_uuid,
  old_device_desc; add  using_external)
* more strict json schema constraints

Michal Hrusecky authored 4 years ago and Karel Koci committed 4 years ago

As part of fixes also waiting for external devices up to 5 seconds.

Michel Kraus authored 4 years ago and Karel Koci committed 4 years ago

This reenables the SATA support on the Omnia only.

This package uses 'message of the day' to inform user about limitations
of oficial Turris support.

Karel Koci authored 4 years ago and Josef Schlehofer committed 4 years ago

Changed
* `lspci` expanded to show also vendor and device numbers and names

Fixed
* `nikola` fixed to call `sentinel-nikola` instead of `nikola`

Changes:
- limit number of fowarding servers to 10
- use get_servers_ip_addresses to get forwarding servers

Changes:
- add ipcheck.py utility
- get_servers_ip_addresses to parse /tmp/resolv.conf.auto

Changes:
- use get_servers_ip_addresses to get forwarding servers

It means that we are testing just our three platforms
and one of it is even run tested in Docker. Also, it uses Docker
repositories maintainer by OpenWrt developers.

Tested platforms:
powerpc_8540 (Turris 1.x)
arm_cortex-a9_vfpv3-d16 (Turris Omnia)
aarch64_cortex-a53 (Turris MOX, Turris Shield)
- This one is run time tested in Docker

This should compile packages and upload them as artefacts, which means
that there is going to be less manual work which should lead to have
merged pull requests faster.

One minor disadvantage for now is that this compiles packages
against OpenWrt releases (stable and snapshots) without our patches, which comes
from turris-build repository and that it is tested against develop
branch from turris-os-packages.

Github Actions will be run on every push (including branches, commits)
and pull request (here on GitHub).

This comes from the OpenWrt packages repository and added a few changes,
which were not merged, yet. E.g. in packages feed, there is run test in
Docker container if there is dedicated file test.sh near Makefile, but
there is a pending change to be merged to run generic run test to check
version of package even there is no test.sh

Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>

Makes sure that mac address of eth0 is used for bridge if netbooted MOX
has multiple interfaces.

Štěpán Henek authored 4 years ago and Jan Pavlinec committed 4 years ago

* next query param fix

Vladimír Čunát authored 4 years ago and Jan Pavlinec committed 4 years ago

Apparently these "*" hostnames are frequent, and they cause annoying
log lines.  Original source and authorship:
https://forum.turris.cz/t/dhcp-host-domain-ng-py-add-lease-hostname-check-failed/13132/22

Some folders were renamed in this repository and we need to sync changes
in codeowners as well.

This patch is present in packages feed and we verify that we need to
have it as well otherwise lighttpd is not compiled against OpenWrt
master.

It fails with:
Package lighttpd-mod-cml is missing dependencies for the following libraries:
liblua5.3.so.0.0.0

Karel Koci authored 4 years ago and Jan Pavlinec committed 4 years ago

This adds option for common_passwords package to hardening list and
moves existing content to options. User now can select what exactly he
wants to harden.

Note:
mode_compress is depredecated in since lighttpd 1.4.56

Glenn Strauss authored 4 years ago and Jan Pavlinec committed 4 years ago

Signed-off-by: Glenn Strauss <gstrauss@gluelogic.com>

With recent updates of lighttpd (I'm talking about 1.4.56 and newer), there were
some changes regarding TLS and many other modules were missing.

While using meson as in OpenWrt packages feed (master branch), there is
this patch https://github.com/openwrt/packages/blob/master/net/lighttpd/patches/010-meson-lua.patch



I tried to apply it to configure, configure.ac and SConstruct, but I was
not successful. Maybe it was because I haven't do any changes in
make menuconfig to apply and save changes as I had the same issue while trying to
convert this package to use Meson and it was working once I applied
changes in make menuconfig.

It a little bit simplifies Makefile, but the biggest advantage to use
Meson and Ninja is that compilation time should be lower and we should
have builds faster. Also, we can cooperate with upstream and be in-sync
with them. It allows us to backport changes in upstream much faster.
This part is just relevant until we drop lighttpd from our feeds as I
want to have upstream lighttpd.

Co-authored-by: Josef Schlehofer <josef.schlehofer@nic.cz>
Signed-off-by: Josef Schlehofer <josef.schlehofer@nic.cz>
[added commit message]

Josef Schlehofer authored 4 years ago and Jan Pavlinec committed 4 years ago

This package failed with following output:
arm-openwrt-linux-muslgnueabi-gcc: error: /foo/build/build_dir/target-arm_cortex-a9+vfpv3-d16_musl_eabi/omnia-mcutool-0.1/mcutool/mcutool.c: No such file or directory

I found that mcutool.c is located at
build_dir/target-arm_cortex-a9+vfpv3-d16_musl_eabi/omnia-mcutool-0.1/mcutool.c

So folder mcutool was reduntant and because of it, I removed it and
package omnia-mcutool can be compiled.

As pyuci is now on PyPI, I was able to simplify Makefile
It means that we will download sources from PyPI.
- In this case, we don't need to have anymore PKG_SOURCE_* variables.
- Also, we don't need to update manually PKG_MIRROR_HASH by running
make package/foo/download and make package/foo/check

It would be enough to change just PKG_HASH and that's all.
PKG_HASH we can get on PyPI.

While at it:
- I added variable PKG_LICENSE_FILES and added source package.
- Fixed Section
- Improved Title to match title on PyPI