The package Makefile contains an unnecessary PKG_BUILD_DEPENDS
line that builds gettext and libiconv even in cases where the
build has gettext-full and libiconv-full selected.
This behaviour can and does mask errors in other package Makefiles
that are dependent on libiconv-full if it is compiled before
these packages by causing libiconv-stub to be created and put in
the staging tree and potentially linked against by these
dependent packages when they omit to specify an appropriate
PKG_BUILD_DEPENDS line.
procps-ng does not require gettext or libiconv to be built in
order to compile correctly.
Signed-off-by: Ian Cooper <iancooper@hotmail.com>
This enables a system-wide .bashrc file (/etc/bash.bashrc), as well as
some other options related to startup files, and sources /etc/shinit for
interactive shells.
Fixes https://bugs.openwrt.org/index.php?do=details&task_id=3019.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
The package Makefile was not taking into consideration that the build
may be using BUILD_NLS with libintl-full and libiconv-full and was
trying to link the wrong versions of these libraries in this case.
The necessary flags are added by nls.mk to TARGET_LDFLAGS and can be
passed to irqbalance's configure script for setting the GLIB2_LIBS
variable instead of the explicit static link to the libiconv stub.
The PKG_BUILD_DEPENDS line should be modified so as to add to and not
override the definition set by nls.mk, which will ensure the right
version of libiconv and libintl is built beforehand.
A DEPENDS:= line should be added to the package definition using the
variables defined in nls.mk, which will add the appropriate version
of libintl and libiconv (vanilla or -full versions)
If USE_GLIBC is true, then libpthread needs to be explicitly passed
to the configure script in the GLIB2_LIBS variable for linking.
Signed-off-by: Ian Cooper <iancooper@hotmail.com>
The former can be implemented as a meson argument.
The latter is already default.
No compiled difference, therefore no PKG_RELEASE bump.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
ntfs-3g does not include nor install a file like this.
Cleaned up Makefile for consistency between packages.
Added PKG_BUILD_PARALLEL for faster compilation.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
So far, all output created by scripts run with micrond has been
discarded. Since there is no reason for that and it also does not
match the expected behavior, this enables both stdout and stderr
output for the service.
If not desired, a user can still use >/dev/null or similar in his/her
micrond jobs to disable output easily and similar to what it would be
on other systems.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This package can not be installed if you have installed less from
busybox.
Collected errors:
* check_data_file_clashes: Package less wants to install file /bin/less
But that file is already provided by package * busybox
* opkg_install_cmd: Cannot install package less.
To avoid this error, I moved it from /bin/less to /usr/bin/less.
If you install it now, it changes symlink from busybox to /usr/bin/less
/bin/less -> busybox*
/bin/less -> /usr/bin/less*
When you remove it, it changes symlink back to busybox.
That's why postint and postrm scripts are no longer needed.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
This removes Python-related build variants, and adds
PYTHON3_PKG_BUILD:=0 and minor build adjustments (where appropriate),
for non-Python packages. There should be no changes to build output.
This also updates some include paths for python3-package.mk and/or
python3-host.mk to be relative to the package Makefile.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
All patches are not needed anymore. squashfs-tools has implemented them in
one way or another.
Converted to download release tarballs. This should work better to get
notified of a newer release.
Also adds support for ZSTD compression:
https://facebook.github.io/zstd/
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Add new PCI and USB drivers to flashrom.mk to disable them in other
build variant.s
Fix DefineConfig calls to use new config symbols, and have it only
append NEED_FEATURE=yes to MAKE_FLAGS. Appending NEED_FEATURE=no would
still enable the feature because upstream compares against an empty
string instead of comparing to yes.
Drop libusb-compat dependency as all drivers use libusb-1 now.
Remove upstreamed patch 010-add-arc.patch.
Rework 0001-fix_internal_bitbang.patch, give it a correct prefix and
more descriptive name.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
libwrap is fairly useless. It's also legacy and should not be used/
dbus is a big dependency. It's fairly useless in the OpenWrt context.
I don't know how useful netlink is.
Disabling BSD behavior results in a slightly smaller size.
Disable LDAP. No size difference but I have a feeling it pulls in
another dependency...
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Remove "prompt" command, dropped in kconfig-v5.6, from
Package/sane-backends/config, replacing it with "tristate". This does
not affect the generated package.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
Update nano editor to version 4.9.1
2020.03.31 - GNU nano 4.9.1 "Sapperdeflap"
Two bugs introduced in version 4.9 are fixed: the cursor
getting misplaced when undoing line cuts, and filtering
of the whole buffer to a new buffer not working.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
This reverts commit 97dbc515e5.
libgd has been been fixed to provide correct pkgconfig info,
so that nut again compiles in buildbot. Thus the nut plugin
can be re-enabled in collectd.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Disable nut plugin for now, as nut is currently
failing to build in buildbot (maybe due to libgd),
and the nut plugin prevents compiling collectd itself.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
This version of yunbridge will cease to function once Python 2 is
removed from the feed. It does not appear[1] that upstream will be
updated with a version that uses Python 3.
This package will be added to the abandoned packages feed.
[1]: https://github.com/arduino/YunBridge/issues/31
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
Add fallback zero value definition for MAP_SYNC etc. even when building
for CONFIG_LINUX.
Fixesopenwrt/packages#11067
Reported-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
It's causing issues with newer glib2:
../libgammu/libGammu.so.8.1.41.0: undefined reference to `libiconv'
../libgammu/libGammu.so.8.1.41.0: undefined reference to `libiconv_open'
../libgammu/libGammu.so.8.1.41.0: undefined reference to `libiconv_close'
It's because the libiconv-stub does not include a shared library that
gammu can link to.
Removed most patches since they don't seem to be needed.
Ran init script through shellcheck.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Run the service under an unprivileged
user account
The following parameters are now configurable
niceness
max concurrency (defaults to number of CPUs)
user
Added flags "-no-browser"
Disabled in-place upgrades (disabled in the
build already)
Redirected stderr/stdout to syslog
Added support for "reload_config"
Increased "term_timeout" to 15s to give it
plenty of time to shut down gracefully
Properly handled non-existing directories
Removed a softlink that assumes a specific naming
convention in syncthing
Added a comment that using external storage is a
recommend configration
Signed-off-by: Marc Benoit <marcb62185@gmail.com>
[increased package release]
Signed-off-by: Paul Spooren <mail@aparcar.org>
Since upstream commit dc92994 the LDFLAGS used for settings the version
(and build host and user) changed resulting in "unknown" versions in
syncthing.
Correct version detection is important for syncthing to assure
compatibility with other running instances. The "unknown" version fails
to communicate and sync with correctly compiled instances.
This patch updates the syncthing Makefile to inject the correct
variables and thereby fixes the version detection in syncthing.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Drop upstreamed patches.
The file appversion.default was removed from Domoticz, causing the hacks
to inject APPVERSION, APPDATE and APPHASH to fail. As the appversion.h
is generated during compile time, implementing a new way to inject these
defines is non-trivial, so simply drop them.
As the minor version for this release is no longer based on the number
of commits, the package versioning needs to be revised if we want to
build a git snapshot instead of stable release. Leave this for another
day and drop that logic for now.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Explicitly disable the new dpdk_telemetry module to fix
compilation in buildbot.
(other dpdk modules are aleady disabled)
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Life changes, I'm no longer avaiable to be an sufficiently active
OpenWrt developer. Therefore removing myself from maintainer from
all packages I maintained in OpenWrt.
All the best for those who keep up the good work!
Signed-off-by: Daniel F. Dickinson <cshored@thecshore.com>
iptables 1.8.4 drops support for a combined libiptc, providing split
libip4tc & libip6tc instead. Adjust collectd dependencies in advance of
iptables bump from 1.8.3 to 1.8.4
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
This is intended as a match with standard urngd. They serve same purpose
and urngd starts as first with 00. Starting haveged later can create
issues if you replace urngd with it. The example problem is if
uci-defaults script decides to generate certificate. Haveged can supply
entropy but it is started later and to mitigate this urngd would still
have to be installed. This means that haveget can't serve as replacement
without moving it to match start order of urngd.
Signed-off-by: Karel Kočí <karel.koci@nic.cz>
If libacl is built, gnu sed finds it during configuration and enables support
linking in libacl. This results in build failures due to the missing dependency.
Consequently, use CONFIGURE_ARGS to disable acl support.
Reported-by: Rosen Penev rosenp@gmail.com
Signed-off-by: Russell Senior <russell@personaltelco.net>
The previous implementation always succeeded so no scan was performed.
This now fixes that and it correctly scans for BTRFS devices if BTRFS
support is in kernel.
Signed-off-by: Karel Kočí <karel.koci@nic.cz>
The original patch[1] this package was based on has been merged into
OpenWrt master, making this package unnecessary.
[1]: https://patchwork.ozlabs.org/patch/1094493/
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
Add PKG_BUILD_PARALLEL for faster compilation.
Replace ' ' with TAB for consistency between packages.
Refresh patches and remove outdated ones.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Use PKG_INSTALL for consistency between packages.
Add PKG_BUILD_PARALLEL for faster compilation.
Fix wrong locking path. First discovered here:
https://forum.openwrt.org/t/cannot-setup-dm-crypt/56836
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Added support for reload_config
This service does not need to start so early (even
before the netwrok is up). Start it after
the device is mostly up and operational.
Compile-tested on: ipq806x
Runtime-tested on: ipq806x
Signed-off-by: Marc Benoit <marcb62185@gmail.com>
Added the ability to enable/disable via a config file;
the default is enabled for backwards compatibility.
This is an additional service, so running it the
same priority as dnsmasq, etc does not make sense.
Added USE_PROCD=1
"done" starts at "START=95", so this should start earlier
Added STOP=10 to stop syncthing early
Compile-tested on: ipq806x, ipq40xx
Runtime-tested on: ipq806x, ipq40xx
Signed-off-by: Marc Benoit <marcb62185@gmail.com>
Added the ability to change sampling interval
and provide a list of IRQ's to ignore
via /etc/config/irqbalance
Signed-off-by: Marc Benoit <marcb62185@gmail.com>
Upgraded from 1.2.0 to 1.6.0 to pick up
all the latest fixes and improvements
irqbalance made glib2 mandatory since 1.3.0, but is
using very little of it. glib2 linked statically
to minimize the impact.
1.2.0 size: 33,619
1.6.0 size: 99,539
~# irqbalance --version
irqbalance version 1.6.0
~# ldd /usr/sbin/irqbalance
/lib/ld-musl-armhf.so.1 (0xb6f5f000)
libgcc_s.so.1 => /lib/libgcc_s.so.1 (0xb6f43000)
libc.so => /lib/ld-musl-armhf.so.1 (0xb6f5f000)
~# opkg info irqbalance
Package: irqbalance
Version: 1.6.0-6
Depends: libc
Status: install user installed
Architecture: arm_cortex-a7_neon-vfpv4
Installed-Time: 1582839249
Compile-tested: ipq806x/R7800 & ipq40xx/GL-B1300
Run-tested: ipq806x/R7800 & ipq40xx/GL-B1300
Signed-off-by: Marc Benoit <marcb62185@gmail.com>
Currently it only cleans up binaries when executing
Build/InstallDev without deleting .built stamp file.
This leads to wrong information about existence of
built runc binaries and causes error when executing
package/containerd/install twice.
Signed-off-by: Texot Qi <tete1030@gmail.com>
Currently it only cleans up binaries when executing
Build/InstallDev without deleting .built stamp file.
This leads to wrong information about existence of
built runc binaries and causes error when executing
package/runc/install twice.
Signed-off-by: Texot Qi <tete1030@gmail.com>
Infozip (earlier name "zip") has 12 year old source code that
would require work to make the binary actually work.
> root@router1:/# zip
> zip error: Not supported (uzoff_t not same size as zoff_t)
The package was rename from zip to infozip a few weeks ago
due to buildbot's relutance to compile zip in master and 19.07
ever since host zip was added as a build tool.
Reference to #10985 and #11089 as well as
ad8c2d6099
But as the binary does not work, lets remove the package to
avoid confusion for the end-users.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Convert bigclown packages to use pypi.mk and pypi in general instead of
github.
This allows us to remove version replacement as that is already done in
pypi archive.
Updated dependencies to also contain proxy dependencies that are listed
in upstream.
Signed-off-by: Karel Kočí <karel.koci@nic.cz>
Fix compilation error due to multiple definition
x86_64-openwrt-linux-musl/bin/ld: lib/libcoreutils.a(strtold.o):(.data+0x0):
multiple definition of `minus_zero'; lib/libcoreutils.a(strtod.o):(.data+0x0):
first defined here
collect2: error: ld returned 1 exit status
Signed-off-by: DENG Qingfang <dengqf6@mail2.sysu.edu.cn>
Since the brcm2708 target has been renamed to bcm27xx in openwrt
repo, update the dependency to the target in this package.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Removed CMAKE_INSTALL as there's no need for InstallDev.
Added PKG_BUILD_PARALLEL for faster compilation.
Small cleanups.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
* Move more environment variables into GoPackage/Environment
* Split GoPackage/Environment into target and build sections
* Do not set GOROOT_FINAL for Go packages (setting it should only affect
the Go compiler and not Go packages)
* Set CGO_LDFLAGS to $(TARGET_LDFLAGS)
* Move GO_TARGET_* variables from golang-values.mk, and GO_VERSION_*
variables from golang-version.mk, into golang/Makefile
This also updates runc, containerd, and docker-ce to reflect the changes
in GoPackage/Environment.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
Today CVE-2020-7221 was publicly discussed on oss-sec [1]. MariaDB
upstream had not mentioned this CVE in their last release notes. The CVE
is related to auth-pam and the possibility of a local mariadb to root
user exploit in the mysql_install_db script.
Upstream has made amendments to the script, but according to the oss-sec
posts the folder permissions were not updated as they should have been.
In OpenWrt the script mysql_install_db is actually patched to never run
the commands in question. This has been this way since MariaDB 10.4 was
made available.
Still, the directory permissions set by the postinstall script are too
lax. To quote the discoverer of the issue, Matthias Gerstner from Suse,
they exhibit "the dangerous situation of a setuid-root binary residing
in a directory owned by an unprivileged user".
This commit fixes this by changing the permissions to the following:
root:mariadb 0750 /usr/lib/mariadb/plugin/auth_pam_tool_dir
This way the setuid-root binary is only available to root and the
mariadb user, while at the same time the mariadb user has no ownership
of the directory.
[1] https://seclists.org/oss-sec/2020/q1/55
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
This exporter exposes information of the connected stations acquired
from hostapd. These contain additional information compared to the
existing station exporter, however they require a full build of hostapd
/ wpad.
Signed-off-by: David Bauer <mail@david-bauer.net>
- Migrate libusb dependency back to libsane
(virtually all useful backends for OpenWrt would need it anyway)
- Disabled new usb-record-replay feature (avoid libxml2 dep)
- Disabled new escl backend (network-only backend are not too useful
for OpenWrt and it requires libcurl, libnetsnmp, libavahi and libxml2)
- Workaround sane-daemon/postinst installation on Imagebuild
- Enabled backends kvs40xx and mustek_usb2 (fixed upstream)
- Fix bigendian compilation
(https://gitlab.com/sane-project/backends/-/merge_requests/329)
- Fix missing std::round() for uclibc
(https://gitlab.com/sane-project/backends/issues/237)
- Fixes FS#2685: coldplug was running before usblp was loaded. Now
it grants access do usblp when a device using it is plugged.
Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
Fixed license information.
Removed patch requiring autoreconf and replaced with a configure variable.
Removed faulty patch that broke systems without a disabled crypt size hack.
Replaced with using a SED command as well as bcrypt, which works in musl.
Removed su patch and converted it to a SED command in the Makefile.
Added new shadow utilities.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Rename "zip" to "infozip" to avoid name collision, as the same
zip package has been introduces to the build tools as zip.
Buildbot does not like that.
Reference to #10985 and #11089 as well as
ad8c2d6099
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
This happens during compilation:
Enabling network
./etc/init.d/prometheus-node-exporter-lua: line 7: /lib/functions/network.sh: No such file or directory
Signed-off-by: Rosen Penev <rosenp@gmail.com>
This is conflicting with tools/zip where the HostBuild is specified.
This should allow the zip package to show up.
Several cleanups were also performed for consistency between packages.
Added PKG_BUILD_PARALLEL:=1 for faster compilation.
Remove PKG_CHECK_FORMAT_SECURITY. Patched the issue instead.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Latest fft-eval has some extra features such as ath11k support, double precision in json signal calculation, and cleanups (batctl references, minor bugfixes, etc )
This commit is a version and makefile update. Now the native project Makefile do the job (compilation and linking).
Signed-off-by: Kirill Lukonin <klukonin@gmail.com>
New patches (both for SConstruct):
* Do not import from distutils. Since Python support/modules are not
built, there is no need to import from distutils. (Importing from
distutils may prevent the package from being built on systems without
a full distutils module, e.g. Debian with python3-minimal.)
I have added back the import in places where it may be helpful to
have, if scons is run manually.
Fixes#10993.
* Do not check the size of time_t. The way this is done is to compile
and run a test program; this fails when cross-compiling[1].
This doesn't appear to affect any functionality (other than missing a
compile-time warning that things will fail in 2038 if time_t is too
small).
[1]: https://gitlab.com/gpsd/gpsd/issues/48
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
DEPENDS attribute makes canutils fail on installation procedure
Removing it makes a successful build
Signed-off-by: Paulo Machado <pffmachado@yahoo.com>
The regular Makefile is totally broken and does not pass CFLAGS. This
breaks compilation with PKG_ASLR_PIE and also does not pass -Os.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
The last OpenOCD release was 3 years ago, plenty of new functionality
was added since then. Two security-related patches went in too.
While at it, add a menuconfig option to allow building without any USB
dependencies, useful for devices counting on sysfsgpio to access
targets.
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
