The previous approach was to uncompress N times a big tarball (638 MB)
where N=130 is the number of supported languages. Each iteration would
only extract a single file, but it still needs to uncompress the whole
tarball. This is of course completely inefficient.
Now, we uncompress the tarball only once to extract all relevant files,
and then iterate N times to copy the file needed for each language.
This massively speeds up builds, at the expense of temporarily requiring
more build space (about 1 GB more)
Signed-off-by: Baptiste Jonglez <git@bitsofnetworks.org>
This fixes support for x86, auc now selects the right combined image
depending on the system being booted in EFI mode or not.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
The fix requires the use of the generated dnsmasq `lancache.conf` file.
So I moved configuration of the hosts directory out of the UCI and into
the generated dnsmasq configuration file to simplify the script.
Signed-off-by: Gerard Ryan <G.M0N3Y.2503@gmail.com>
Running podman as users other than root seems to require that those
users can read /usr/share/containers/seccomp.json. This change sets the
permissions on that file to match those used on Fedora.
Signed-off-by: W. Michael Petullo <mike@flyn.org>
No longer fail in case a package is part of multiple lists (which may
happen temporarily in case it becomes 'nonshared').
Bump version to prepare for sync accross branches.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
* don't segfault if server reply is incomplete
* always request 'generic' image on x86/generic and x86/64
* make SNAPSHOT branches sticky, eg. stay on 21.02-SNAPSHOT and do
not 'upgrade' to 21.02.0-rc2.
* output requested server URL when debugging
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
squashfs-tools aren't of much use without xz compression support
on OpenWrt, as we build kernel with squashfs supporting only xz.
Enabled support for xz compression by default.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
* quotes around all variables
* always use 'read -r' instead of 'read'
* some more minor shellcheck fixes
* reorder LVM ops for 'up' call to allow it to succeed
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This is a security release that fixes a single bug:
- tighten up plugin-finding logic (#811)
Users of libcni are strongly encouraged to update.
Added me to list of maintainers as requested by @dangowrt.
Signed-off-by: Oskari Rauta <oskari.rauta@gmail.com>
Patch fixing segfaults on nulls was removed due to patch's content being now included in conmon's source since containers/conmon@355dbf1
Signed-off-by: Oskari Rauta <oskari.rauta@gmail.com>
The sysroot has to point to the toolchain directory. The directory
includes libraries like libm. However, the STAGING_DIR needs to be
added to the CFLAGS.
Fixes#15773
[As suggested by jow]
Signed-off-by: Nick Hainke <vincent@systemli.org>
Added a small patch to fix compilation.
Renamed patch files to have a .patch suffix. Makes them easier to read
with vim and probably IDEs.
Remove UCLIBC depend since it's no longer present.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
gpsd-utils include management tools for GPS daemon - and this commmit
particularly introduces gpsdctl to add/remove device from the GPS
daemon's device list over control socket.
Signed-off-by: ekathva advaita <ekathva@hotmail.com>
Running podman as users other than root seems to require that those
users can read the configuration files in /etc/containers. This change
sets the permissions of /etc/containers and its contents to match those
used on Fedora.
Signed-off-by: W. Michael Petullo <mike@flyn.org>
Miscellaneous:
- Remove prompt to use docker compose in the up command
- Bump py to 1.10.0 in requirements-indirect.txt
Signed-off-by: Javier Marcet <javier@marcet.info>
* Use $(COMMITCOUNT)
* Use SPDX
* Use CA (ucert) public key
* Update repo link
* Update maintainer email
* Format description
Signed-off-by: Paul Spooren <mail@aparcar.org>
Using strcmp() to compare a version string doesn't work well.
Use verrevcmp() function from opkg instead.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Use sfdisk to get GPT partition by name as partition names are not
known by the kernel if added via partx.
Make sure physical volume names are unique, if possible correlate
with the disks serial number and/or card's cid.
mkf2fs apparently returns 134 even in case format succeeded, so don't
fail in that case (this fixes rw volumes large enough for F2FS to be
selected by the lvm scripts of uvol).
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
- Add support for AppArmor
- Gracefully stop containers and pods on shutdown
I found out that If you change location of containers to persistent storage instead of tmpfs, starting them will fail unless they have been stopped. If this is the case that reboot has occurred before pods and containers have been stopped, they cannot be started, they have to be removed and re-created. Change in initscript tries to avoid that. Even if containers are running at tmpfs, this won't hurt. Still, if something happens and system hangs/reboots/etc, script won't save you from that. It's just a attempt to make things better.
I also enabled AppArmor support for future possibilities.
Signed-off-by: Oskari Rauta <oskari.rauta@gmail.com>
Currently, this package can not be installed while using standard path
of busybox, because binary killall wants to be installed on the same
location as busybox.
Collision:
• /usr/bin/killall: busybox (new-file), psmisc (existing-file)
Many of these binaries, which provides alternatives were moved to
folder /usr/libexec like wget, sed, findutils, less.
So I moved killall to /usr/libexec and others leave in touch and added
ALTERNATIVES for it, because preinstall script is no longer necessary.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Emmit ubus events when volumes come up/down.
Make sure volume state is always well defined by introducing
additional state 'write-prepare' (wp) during mkfs.
Add init scripts to bring up volumes at boot.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
HAVE_LIBPCIACCESS that is currently passed through MAKE_VARS to disable
building with libpciaccess can't be set through the environment.
Instead, use CONFIG_CON_PCI, which can be passed through the environment
and will disable libpciaccess.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
Switch to AUTORELEASE for simplicity.
Use autotools build.
Remove upstream patch.
Minor cleanups for consistency between packages.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Make sure filesystem is ready when volume becomes available.
Use 'write-once' as initial state for read-only volumes, only allow
writing to volumes in that state and transision to 'read-only' once
write has completed.
Also fix a typo which prevented 'list' command from working with LVM.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Bugs:
- Fix for invalid handler warning on Windows builds
- Fix config hash to trigger container recreation on IPC mode updates
- Fix conversion map for placement.max_replicas_per_node
- Remove extra scan suggestion on build
Signed-off-by: Javier Marcet <javier@marcet.info>
Features:
- Add profile filter to docker-compose config
- Add a depends_on condition to wait for successful service completion
Miscellaneous:
- Add image scan message on build
- Update warning message for --no-ansi to mention --ansi never as alternative
- Bump docker-py to 5.0.0
- Bump PyYAML to 5.4.1
- Bump python-dotenv to 0.17.0
Signed-off-by: Javier Marcet <javier@marcet.info>
The package needs libseccomp, which does not currently support arc.
In order to avoid a circular dependency, we must avoid arc here as well.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
modbus-utils was not intended to be added at this stage. Remove it.
Fixes: 312594f86 ("uvol: add new package")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
uvol is a wrapper-script which allows automated handling of storage
volumes. uvol currently comes with backend support for LVM2 and UBI,
covering practically all options for storage large enough to be
managed (NAND, SPI-NAND, eMMC, SATA, NVME, virtio-blk, ...).
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This was probably a work-around for an issue with dependencies which
was fixed by
988ed00802
Remove it as all other packages with `-selinux` variants do provide
a non-SELinux-variant without any suffix and that works now, see
procd vs. procd-selinux
busybox vs. busybox-selinux
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Instead of just appending the driver serial including the '0x' prefix,
use '-' prefix instead to make it more readable.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
The 'autopart' package is intended for devices with rather large
block device storage (ie. SATA or MMC).
It automatically allocates the free space on the block device used
for booting into an LVM2 physical volume.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
crun is the prefered container run-time of podman, it's faster than
runc and has a much lower memory footprint.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Oskari Rauta <oskari.rauta@gmail.com>
* Updated the .ga TLD server.
* Removed the .cd and cf TLD servers.
* Removed 72 new gTLDs which are no longer active.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Ariadne Conill (1):
database: do not chroot(".") unless actually necessary
Martin Vahlensieck (1):
Use correct port when redirected
Timo Teräs (5):
db: fix control character check to use uint8_t
libfetch: send Proxy-Authorization also for https connect
del: report correctly package's provides names
Signed-off-by: Paul Spooren <mail@aparcar.org>
faster to compile.
A small selection of packages was tested going from:
Executed in 696.30 secs fish external
usr time 82.98 mins 395.00 micros 82.98 mins
sys time 9.02 mins 0.00 micros 9.02 mins
to:
Executed in 592.20 secs fish external
usr time 84.84 mins 361.00 micros 84.84 mins
sys time 8.85 mins 57.00 micros 8.85 mins
Tested by running make -j 12 and wiping staging/build_dir/target_x
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Bugs:
- Make --env-file relative to the current working directory and error out for
invalid paths. Environment file paths set with --env-file are relative to the
current working directory while the default .env file is located in the project
directory which by default is the base directory of the Compose file.
- Fix missing service property storage_opt by updating the compose schema
- Fix build extra_hosts list format
- Remove extra error message on exec
- Fix OpenSSL version mismatch error when shelling out to the ssh client
- Add missing build flags to the native builder: platform, isolation and
extra_hosts
- Remove info message on native build
- Avoid fetching logs when service logging driver is set to 'none'
Miscellaneous:
- Add compose.yml and compose.yaml to default filename list
Signed-off-by: Javier Marcet <javier@marcet.info>
Switch to AUTORELEASE for simplicity.
Add libarchive dependency since it's now used.
Build with ninja for faster compilation.
Refresh SIMD CMake patch and add musl patch.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Switch to AUTORELEASE for simplicity.
Switch to building with Ninja for faster speed.
Several cleanups for consistency between packages.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Changelog [1]:
Version 1.09
* Explicitly set the controlling TTY
Version 1.08
* Report when IP key has changed
* Scrub the environment variable for -e
Version 1.07
* Pass signals that should terminate to ssh
* Fix race around signal handling
* Report IPC errors to stderr
* Report if can't open -f password file
[1] https://sourceforge.net/p/sshpass/code/76/tree/trunk/ChangeLog
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Enables the compiler option that collectd is compiled with
debugging support. This is used at development stages to get
more messages from the collectd during development.
This option is default disabled.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
On my system the attribute DEVTYPE was not set. The plugin could not
read any data and the function call blocked forever on this function and did
not returned. By removing it, all block devices under `/sys/class/block`
were checked.
Block devices that do not support SMART were not evaluated. The
collected displays the following message.
smart plugin: checking SMART status of /dev/loop4.
smart plugin: unable to open /dev/loop4.
If you do not like this message, you could only enable device in the uci that
does support SMART.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
* include version_code in request
* include versions of selected packages in request
* add SHA256 verification via busybox sha256sum
* sort attributes in policies alphabetically
* move all API-specific string constants to precompiler macros
* set correct MIME type for JSON post request (application/json)
* output string error message if something goes wrong
* auto-generate version string
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This updates auc to use the new API of the sysupgrade image server.
Package version updates now need to be evaluated locally.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
be065c4b sepolicy: Do not try to load policy on import
8f0f0a28 selinux(8,5): Describe fcontext regular expressions
e12f71e8 python/sepolgen: allow any policy statement in if(n)def
b15dff0f python/semanage: Sort imports in alphabetical order
92e01969 python/semanage: empty stdout before exiting on BrokenPipeError
fadcc839 python/sepolicy: allow to override manpage date
2a60de8e sepolgen: sort extended rules like normal ones
9e239e55 sepolgen: print extended permissions in hexadecimal
ccd973f7 python/audit2allow: add #include <limits.h> to sepolgen-ifgen-attr-helper.c
Signed-off-by: Dominick Grift <dominick.grift@defensec.nl>
Switched to building with meson as it works better in a parallel
context.
Small Makefile adjustments for consistency.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Avoids having to use msgfmt, which happens to be broken with gettext-full.
msgfmt is used for the desktop gvim application, which is not used here.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
This fixes compiling lxc without seccomp support if libseccomp is
already installed to the staging dir. Patches were applied upstream.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
* Since the docker0 is a private network by default we can be
more accepting like the LAN is by default
Signed-off-by: Gerard Ryan <G.M0N3Y.2503@gmail.com>
* The default server is the default switch so that
queries will go through the hosts dnsmasq by default
Signed-off-by: Gerard Ryan <G.M0N3Y.2503@gmail.com>
Baptiste Jonglez