4.4 (2024-10-10)
tools: Use getopt
Implement and document option -h
eeprog: Use force option when data comes from a pipe
i2cdetect: Display more functionality bits with option -F
i2cdump: Remove support for SMBus block mode
i2cget: Document SMBus block mode
Fix the return code of option -h
i2cset: Fix the return code of option -h
i2ctransfer: Sort command line options and add to help text
Add an option to print binary data
Drop redundant variable arg_idx
py-smbus: Install in the defined prefix
Use setuptools instead of distutils
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Changes from 5.3.1 to 5.3.2
---------------------------
1. The pretty printer now produces fewer spurious newlines; at the
outermost level it now adds newlines between block comments and
the block or function that follows them. The extra final newline
is no longer produced.
2. OpenVMS 9.2-2 x86_64 is now supported.
3. On Linux and macos systems, the -no-pie linker flag is no longer required.
PMA now works on macos systems with Apple silicon, and not just
Intel systems.
4. Still more subtle issues related to uninitialized array elements have
been fixed.
5. Associative arrays should now not grow quite as fast as they used to.
6. The code and documentation are now consistent with each other with
respect to path searching and adding .awk to the filename. Both
are always done, even with --posix and --traditional.
7. As usual, there have been several minor code cleanups and bug fixes.
See the ChangeLog for details.
Changes from 5.3.0 to 5.3.1
---------------------------
1. More subtle issues related to uninitialized array elements have
been fixed.
2. A number of bugs in the debugger related to handling of arrays
have been fixed.
3. Some subtle bugs in the API have been fixed.
4. Use of MPFR is now possible again on 32-bit Power PC Mac systems.
5. Race conditions around broken pipes for system() and read and write
pipes should now be closed off.
6. Support for OSF/1 has been removed.
7. The never-documented --nostalgia option has been removed. It was
causing bug reports.
8. The implementation of printf/sprintf has been thoroughly reworked
in order to make the code more maintainable and to fix a goodly
number of corner cases.
9. As usual, there have been several minor code cleanups and bug fixes.
See the ChangeLog for details.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Breaking Changes
The web terminal is readonly by default now, to make it writable,
use -W option.
Changelog 1.7.7
Bump to 1.7.7
cmake: add static version back
Changelog 1.7.6
workflow: update actions packages
cmake: get version from git ref
Changelog 1.7.5
html: allow overriding client options with URL query
server: add --exit-no-conn option
html: add safari mobile full screen support
html: Add Unicode 11 xterm.js addon
html: upgrade to xterm 5.4.0
add trzszDragInitTimeout option
Changelog 1.7.4
Leverage Docker Layer System
html: update dependencies
build(deps): update zlib version 1.2.13 to 1.3 in cross-build.sh
server: print working directory on start
server: replace --readonly with --writable
server: fix -U option not work
server: do not expose port to public by default
Add -y flag for easier installation
improve support for trzsz
info.max_http_header_data = 65535;
Add Docker pulls badge
fix flow control
Fix typo, OPTOINS -> OPTIONS
Bump CMake version to 2.8.12
Signed-off-by: Paul Donald <newtwen+github@gmail.com>
I modified the makefile, so that it will build two packages: jq and
jq-full. The former will remain unchanged and the latter will have a
dependency to the oniguruma library, so jq-full will have regex functions
enabled.
Signed-off-by: Maximilian Trüpschuch <milanx@gmx.net>
When there is an error building packages other than coremark
and re-run with make -j1 V=s, the coremark package will report error.
the root cause is the folder was already created in the first run,
and not removed before the second run.
To fix this, use 'mkdir -p' instead of 'mkdir'.
Co-authored-by: Jonas Gorski <jonas.gorski@gmail.com>
Signed-off-by: Rye Sears <xlighting@gmail.com>
* realtek-poe was only being built for realtek/rtl838x with mips-4kec
* needed by targets in realtek/rtl839x with mips-24kc
Signed-off-by: Stephen Howell <howels@allthatwemight.be>
Added upstream patch to fix starting containers with complex hooks.
No existing patches needed to be rebased/simple version bump to 6.0.4.
Build system: x86/64
Build-tested: x86/64
Run-tested: x86/64
Signed-off-by: John Audia <therealgraysky@proton.me>
Update nano edit to version 8.4.
Add configure flags to overcome upstream bug 66978, where the newly added
gnulib module for strcasecmp always fails in the configure step when
cross-compiling.
https://savannah.gnu.org/bugs/?66978b2927d1b1f
The added flags disable the strcasecmp function from gnulib.
If upstream (gnulib?) fixes things later, the flags should be removed.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Commit 9fc79e2e2622 ("download: don't overwrite VERSION variable")
changed the variable for direct download call from VERSION to
SOURCE_VERSION.
This cause the dl_github_archive script to pass empty value for
--version arg making it always clone HEAD.
Correctly update the variable to SOURCE_VERSION to actually clone the
expected commit HASH.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Bump to latest upstream release.
Removed upstreamed: 020-replace_uint16_t
Manually rebased: 010-gcc14.patch
To fix the following build error, added two commits from upstream:
020-nvme-print-add-fallback-for-.patch
021-pligins-netapp-add-include-of-libgen.h-for-basename.patch
Tested the resulting binary and it seems fine:
% nvme smart-log /dev/nvme0 | grep "temperature"
temperature : 42 °C (315 K)
Link to upstream issue: https://github.com/linux-nvme/nvme-cli/issues/2743
Build error:
../nvme-print.c: In function 'is_temperature_fahrenheit':
../nvme-print.c:805:19: error: 'LC_MEASUREMENT' undeclared (first use in this function)
805 | setlocale(LC_MEASUREMENT, "");
| ^~~~~~~~~~~~~~
../nvme-print.c:805:19: note: each undeclared identifier is reported only once for each function it appears in
Build system: x86/64
Build-tested: x86/64
Run-tested: x86/64
Signed-off-by: John Audia <therealgraysky@proton.me>
PKG_FIXUP:=autoreconf introduced in this commit[1] to fix builds with GCC 14
does not play well with GCC 13. Apply it conditionally.
I build some coreutils packages under GCC 13 and again under GCC 14 and both
completed successfully.
Build system: x86/64
Build-tested: x86/64
Fixes https://github.com/openwrt/packages/issues/26175
1. b1a648e1ff
Signed-off-by: John Audia <therealgraysky@proton.me>
This fixes the build on GCC 14 and solves issue https://github.com/openwrt/packages/issues/26175
Maintainer: @hnyman
Compile tested: x86/64, QEMU Standard PC (Q35 + ICH9, 2009), r29064-696ad7b1aa09
Compile tested: ath79/generic, TP-Link Archer C7 v4, r29064-696ad7b1aa09
Compile tested: realtek/rtl838x, Netgear GS108T v3, r29064-696ad7b1aa09
Run tested: x86/64, QEMU Standard PC (Q35 + ICH9, 2009), r29064-696ad7b1aa09, booted and used for 7h without issues
Run tested: ath79/generic, TP-Link Archer C7 v4, r29064-696ad7b1aa09, booted and used for 7h without issues
Run tested: realtek/rtl838x, Netgear GS108T v3, r29064-696ad7b1aa09, booted and used for 7h without issues
Signed-off-by: Pascal Ernster <git@hardfalcon.net>
As documented by "man git-rev-parse", the "--short" option shortens
commit sha1sums to at least "length" characters, equal to core.abbrev if
that is specified in ~/.gitconfig.
The development processes of some other open source projects require
having a
[core]
abbrev = 12
in the .gitconfig, which is incompatible with the way in which docker
wants PKG_GIT_SHORT_COMMIT.
On my system, I get these errors:
make[3]: Entering directory 'feeds/packages/utils/dockerd'
(...)
# Verify CLI is the same version
( CLI_MAKEFILE="../docker/Makefile"; CLI_VERSION=$( grep --only-matching --perl-regexp '(?<=PKG_VERSION:=)(.*)' "${CLI_MAKEFILE}" ); if [ "${CLI_VERSION}" != "27.3.1" ]; then echo "ERROR: Expected 'PKG_VERSION:=27.3.1' in '${CLI_MAKEFILE}', found 'PKG_VERSION:=${CLI_VERSION}'"; exit 1; fi )
# Verify PKG_GIT_SHORT_COMMIT
( EXPECTED_PKG_GIT_SHORT_COMMIT=$( feeds/packages/utils/dockerd/git-short-commit.sh 'github.com/moby/moby' 'v27.3.1' 'tmp/git-short-commit/dockerd-27.3.1' ); if [ "${EXPECTED_PKG_GIT_SHORT_COMMIT}" != "41ca978" ]; then echo "ERROR: Expected 'PKG_GIT_SHORT_COMMIT:=${EXPECTED_PKG_GIT_SHORT_COMMIT}', found 'PKG_GIT_SHORT_COMMIT:=41ca978'"; exit 1; fi )
Trying remote 'github.com/moby/moby'
fatal: 'github.com/moby/moby' does not appear to be a git repository
fatal: Could not read from remote repository.
Please make sure you have the correct access rights
and the repository exists.
Trying remote 'https://github.com/moby/moby'
remote: Enumerating objects: 11117, done.
From https://github.com/moby/moby
* tag v27.3.1 -> FETCH_HEAD
HEAD is now at 41ca978a0a54 Merge pull request #48525 from thaJeztah/27.x_backport_govulncheck_permissions
ERROR: Expected 'PKG_GIT_SHORT_COMMIT:=41ca978a0a54', found 'PKG_GIT_SHORT_COMMIT:=41ca978'
make[3]: *** [Makefile:198: build_dir/target-aarch64_generic_glibc/dockerd-27.3.1/.prepared_d76b59f2eb81424899b1fbb9e44f77e2_6664517399ebbbc92a37c5bb081b5c53] Error 1
make[3]: Leaving directory 'feeds/packages/utils/dockerd'
time: package/feeds/packages/dockerd/compile#1.71#1.18#5.38
ERROR: package/feeds/packages/dockerd failed to build.
Since --short supports a length argument, use that to break the
dependency on the system .gitconfig.
Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
Also fixes MIPS builds.
Seems there is some inline assembly that won't work with MIPS16
instructions.
Signed-off-by: Alexandru Ardelean <alex@shruggie.ro>
Adjust the dependency to the virtual coreutils main package in
each app to be selective. Otherwise you need to first select the
main coreutils before the actuall apps can be selected. That has
prevented other applications from depending on just one individual
coreutils app, as they have needed to depend also on the empty main
coreutils package.
Reference to discussion in:
https://github.com/openwrt/luci/issues/7605
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Upstream is preparing the migration to a new website. As part of this, they
will be dropping the `www` prefix. Also, the package source is updated to use
mc's official OSU OSL mirror over HTTPS.
Signed-off-by: Yury V. Zaytsev <yury@shurup.com>
Users running unprivileged containers will need to create
/etc/subgid and /etc/subuid and want to have them preserved
across updates. This commit adds them to the default backup set.
Signed-off-by: John Audia <therealgraysky@proton.me>
Co-authored-by: Tianling Shen <cnsztl@gmail.com>
Docker's backend storage driver can be configurable for certain
filesystems. The default is the overlay storage driver, but if you run
openwrt on a system with btrfs, this will allow you to override the
default configuration by settings the storage_driver in uci in dockerd's
global section. This value will be used in the created dockerd.json
file.
Signed-off-by: Keith T. Garner <kgarner@kgarner.com>
More than one device can be added to the gpsd at startup. Currently the gpsd
service start script in OpenWrt treats this as an option with only one value.
To allow multiple devices to be specified, the uci 'option device' must be
removed and a new uci 'list devices' is added.
This change means that several devices can be specified at gpsd start with
the new uci 'list devices' configuration.
Running configurations in the field are migrated by a migration script.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
The 'readonly' option tells the GPSD that it is not allowed to write to the
GPS TTY, only reading the NMEA data stream is allowed. This option of the
GPSD must be set for my mobile radio device MV31, otherwise the GPSD can not
read the NMEA data stream on the GPS TTY because the GPSD blocks during
plugin probing.
gpsd:PROG: CORE: Probing "Garmin USB binary" driver...
gpsd:PROG: CORE: Probe not found "Garmin USB binary" driver...
gpsd:PROG: CORE: Probing "GeoStar" driver...
gpsd:PROG: Sent GeoStar packet id 0xc1
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
The 'gpsd' offers the possibility to call the script '/etc/gpsd/device-hook'
when a GPS source was added or removed via gpsdctl.
In addition to the '/etc/gpsd/device-hook' call an event is now triggered
too after the 'gpsd' has started. This allows scripts to configurre 'gpsd'
receivers.
The following events are available for '/etc/hotplug.d/gpsd' scripts:
* ACTIVATE via '/etc/gpsd/device-hook'
* DEACTIVATE via '/etc/gpsd/device-hook'
* STARTED via '/etc/init.d/gpsd'
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
This change makes it possible to add new gps device without restarting the
'gpsd' service by using the 'gpsdctl' command.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Modified 025-remove-unsupported-option.patch to both remove
the bsdtar command as it ends in errors, see below, and to
circumvent an error when extracting to overlayfs[1].
Error when extracting rootfs tarball with bsdtar:
tar --absolute-names --numeric-owner '--xattrs-include=*' -xpJf /var/cache/lxc//download/archlinux/current/amd64//default/rootfs.tar.xz -C /mnt/data/lxc/test/rootfs
./usr/bin/newgidmap: Cannot restore extended attributes on this system: Illegal byte sequence
lxc-create: test: ../src/lxc/lxccontainer.c: create_run_template: 1589 Failed to create container from template
lxc-create: test: ../src/lxc/tools/lxc_create.c: lxc_create_main: 318 Failed to create container test
1. https://github.com/openwrt/openwrt/issues/15888
Signed-off-by: John Audia <therealgraysky@proton.me>
Build system: x86/64
Build-tested: x86/64/AMD Cezanne
Run-tested: x86/64/AMD Cezanne
Signed-off-by: John Audia <therealgraysky@proton.me>
Add gnu-tar as a dependency for lxc-create which is needed to
properly extract the root file system tarball. Without it,
symlink target of /bin/tar is busybox and it lacks proper
support to extract with xattrs. The default gnu-tar is actually
built without this support as well, but it is able to extract
the rootfs tarballs with a warning not an error which is not the
case with busybox which ends in an error.
Signed-off-by: John Audia <therealgraysky@proton.me>
Build system: x86/64
Build-tested: x86/64/AMD Cezanne
Run-tested: x86/64/AMD Cezanne
Several deps are missing based the output of lxc-checkconfig shown below
before this commit is applied.
CONFIG_IP_NF_TARGET_MASQUERADE and CONFIG_IP6_NF_TARGET_MASQUERADE are
only needed for lxc-net which we do not package.
% lxc-checkconfig | grep missing
Cgroup device: missing
CONFIG_IP_NF_TARGET_MASQUERADE: missing
CONFIG_IP6_NF_TARGET_MASQUERADE: missing
CONFIG_NETFILTER_XT_TARGET_CHECKSUM: missing
FUSE (for use with lxcfs): missing
checkpoint restore: missing
CONFIG_UNIX_DIAG: missing
CONFIG_INET_DIAG: missing
CONFIG_PACKET_DIAG: missing
CONFIG_NETLINK_DIAG: missing
Additionally, two new patches have been added which remove checks for
options that OpenWrt currently does not package and can serve are
false positives for missing items from our kernel config, namely:
lxc-net and lxc-checkpoint
After applying this commit, below is the output showing that the kernel
config[1] should pass all tests for functionality:
% CONFIG=config ./lxc-checkconfig
LXC version 6.0.3
--- Namespaces ---
Namespaces: enabled
Utsname namespace: enabled
Ipc namespace: enabled
Pid namespace: enabled
User namespace: enabled
Network namespace: enabled
Namespace limits:
cgroup: 383849
ipc: 383849
mnt: 383849
net: 383849
pid: 383849
time: 383849
user: 383849
uts: 383849
--- Control groups ---
Cgroups: enabled
Cgroup namespace: enabled
Cgroup v1 mount points:
Cgroup v2 mount points:
- /sys/fs/cgroup
Cgroup device: enabled
Cgroup sched: enabled
Cgroup cpu account: enabled
Cgroup memory controller: enabled
Cgroup cpuset: enabled
--- Misc ---
Veth pair device: enabled, loaded
Macvlan: enabled, not loaded
Vlan: enabled, not loaded
Bridges: enabled, loaded
Advanced netfilter: enabled, loaded
CONFIG_NETFILTER_XT_TARGET_CHECKSUM: enabled, not loaded
CONFIG_NETFILTER_XT_MATCH_COMMENT: enabled, loaded
--- Checkpoint/Restore ---
checkpoint restore: missing
CONFIG_FHANDLE: enabled
CONFIG_EVENTFD: enabled
CONFIG_EPOLL: enabled
CONFIG_UNIX_DIAG: enabled
CONFIG_INET_DIAG: enabled
CONFIG_PACKET_DIAG: enabled
CONFIG_NETLINK_DIAG: enabled
File capabilities: enabled
1. Generated on 23-Feb-2025 running bcm27xx/bcm2712 on r28869+1
Signed-off-by: John Audia <therealgraysky@proton.me>
We do not package lxc-net or lxcfs so remove the lines within
lxc-checkconfig that looks forCONFIG_IP_NF_TARGET_MASQUERADE and
CONFIG_IP6_NF_TARGET_MASQUERADE which, as far as I know, is the only
part of lxc that needs them. Also remove the check for FUSE since we do
not pakcage lxcfs.
Without this commit, users will see these two as missing.
Signed-off-by: John Audia <therealgraysky@proton.me>
