Previous commits installed non-elf files into /lib/modules/$VER/.
COLLECT_KERNEL_DEBUG tries to strip all files and these two files
break the build.
Fix it by copying only kernel modules for debug info collection.
Fixes: e1d8e57614 ("kernel: include modinfo for built-in modules")
Fixes: 29f6da4340 ("kernel: include built-in module list")
Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Partially revert changes to verbose logging that break the 'check' target
dependencies and trigger many runtime warnings like:
/home/kodidev/openwrt-project/include/toplevel.mk:213: *** mixed implicit and normal rules: deprecated syntax
Fixes: e4a43cda0 ("build: allow var.% targets to bypass the prepare steps")
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
In order to pass a status message at runtime,
which is usually listing subtargets
of a Makefile target or an error message,
from a child invocation of Make (submake)
through the parent process to the terminal,
the file descriptors 8 and 9 are opened to be used
by the functions MESSAGE and ERROR_MESSAGE.
However, there are situations where those functions
can be called while not in a submake or a subshell
or a child process which results in a shell error:
/bin/bash: 8: Bad file descriptor
Commit aee3594ffc
("verbose.mk: print ERROR messages in non-verbose")
has exposed this issue to more cases, but it is not the root cause.
To solve this, use the exit code of the first printf attempt
to the alternative file descriptors in order to tell whether
the standard file descriptors need to be used instead.
In order to get rid of the "Bad file descriptor" error, stderr is
redirected to null after grouping the two printf alternatives
into one command to combine outputs.
For ERROR_MESSAGE, the real message is redirected to stderr
after redirecting the error from the attempted printing to null.
For MESSAGE, without redirection, the Make function "shell"
will absorb the actual message from stdout and input the value into the Makefile,
therefore the dummy variable "_NULL", previously used merely for causing
a call to the MESSAGE function to trigger without writing target rules,
now has and a real value when defined, so rename it to "_MESSAGE"
as a placeholder for the real message when the output should be stdout.
When "_MESSAGE" has a value, use Make function "info" to
finally bring it from the Makefile to the terminal.
This also fixes what is likely a typo, in that
while file descriptor 9 is meant to redirect to stderr
for use in error messages like in the function ERROR_MESSAGE,
that function has printf redirecting to file descriptor 8 instead.
Fixes: a4c8d4e37 ("build: make the color of the 'configuration out of sync' warning red")
Signed-off-by: Michael Pratt <mcpratt@pm.me>
These targets are used to input variable values from the Make
context into other things like python scripts, so log messages
should be silenced and build prerequisites should be skipped.
The same thing is done for the other variable print target "val.%".
While at it, combine identical target rules into one definition.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Many NAND devices use a build recipe with "append-ubi | check-size" to
ensure factory images don't exceed the target flash partition size.
However, UBI reserves space for bad block handling and other operational
overhead, and thus 'check-size' can overestimate the space available by
several MB. In practice, this means a failed check is definitely a failure,
while a passing check is only probably a pass.
Improve the situation by teaching 'Build/append-ubi' to check image sizes
while accounting for UBI reserved blocks. Add new device variable NAND_SIZE
and use with existing IMAGE_SIZE to derate the available space. Each UBI
device reserves 20 PEBs per 1024 PEBs of the entire NAND device for bad
blocks, plus an additional 4 PEBs overhead.
Many devices can transparently enable this check by setting NAND_SIZE based
on their flash storage, and may then remove any unneeded 'check-size'.
Link: http://www.linux-mtd.infradead.org/doc/ubi.html#L_overhead
Suggested-by: Shiji Yang <yangshiji66@qq.com>
Suggested-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
Current factory image sizes for Linksys devices are 256-byte aligned. This
is not an issue writing factory images from the OpenWrt or Linksys GUIs,
but can lead to failures using a TFTP client from the Linksys bootloader:
NAND write: device 1 offset 0x2800000, size 0xc00100
Attempt to write to non page aligned data
NAND write to offset 2800000 failed -22
0 bytes written: ERROR
Simplify Linksys footer creation by migrating to a makefile build recipe,
and pre-pad the footer (with 0xFF) to ensure the final image is $(PAGESIZE)
aligned. Finally, remove the old linksys-image.sh script no longer needed.
Linksys footer details are given below for future reference. The 256-byte
footer is appended to factory images and tested by both the Linksys
Upgrader (observed in EA6350v3) and OpenWrt sysupgrade.
Footer format:
.LINKSYS. Checked by Linksys upgrader before continuing. (9 bytes)
<VERSION> Upgrade version number, unchecked so arbitrary. (8 bytes)
<TYPE> Model of device, space padded (0x20). (15 bytes)
<CRC> CRC checksum of factory image to flash. (8 bytes)
<padding> Padding ('0' + 0x20 * 7) (8 bytes)
<signature> Signature of signer, unchecked so arbitrary. (16 bytes)
<padding> Padding with nulls (0x00) (192 bytes)
Link: https://github.com/openwrt/openwrt/pull/11405#issuecomment-1358510123
Link: https://github.com/openwrt/openwrt/pull/11405#issuecomment-1587517739
Reported-by: Stijn Segers <foss@volatilesystems.org>
Reported-by: Wyatt Martin <wawowl@gmail.com>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
Add the make function 'exp_units' for helping evaluate k/m/g size units in
expressions, and use this to consistently replace many ad hoc substitutions
like '$(subst k,* 1024,$(subst m, * 1024k,$(IMAGE_SIZE)))' in makefiles.
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
Some patched u-boots may have problems with parallel build.
Do not enforce parallel build here so one can set PKG_BUILD_PARALLEL:=0
in the specific u-boot Makefile also before including the u-boot.mk.
Signed-off-by: Martin Schiller <ms@dev.tdt.de>
This reverts commit 7ceb76ca3a.
Python 3.12 removes the distutils package and is therefore not
compatible. We have to check downstream what relies on distutils before
adding actual support for Python 3.12. Sorry for the noise.
With this in-place, the macOS CI job fails and turns things red, revert
for now.
Signed-off-by: Paul Spooren <mail@aparcar.org>
creates SGE encrypted factory images
to use via the D-Link web interface
rename the old factory unencrypted images to recovery
for use in the recovery console when recovery is needed
DIR-1935-A1 , DIR-853-A1 , DIR-853-A3 , DIR-867-A1 ,
DIR-878-A1 and DIR-882-A1
Signed-off-by: Alan Luck <luckyhome2008@gmail.com>
Changes introduced in commit d604a07225 ("build: add CycloneDX SBOM
JSON support") broke ImageBuilder:
Cannot open '/openwrt-imagebuilder-ath79-generic.Linux-x86_64/tmp/.packageinfo': No such file or directory
So lets fix it by wrapping the BOM generation behind condition of IB
feature check.
Fixes: #13881
Fixes: d604a07225 ("build: add CycloneDX SBOM JSON support")
Signed-off-by: Petr Štetiar <ynezz@true.cz>
`true` might be a shell built-in, or simply not accessible in the hardcoded locations.
Replace it with a custom script that does nothing.
Signed-off-by: Ilya Katsnelson <me@0upti.me>
CycloneDX is an open source standard developed by the OWASP foundation.
It supports a wide range of development ecosystems, a comprehensive set
of use cases, and focuses on automation, ease of adoption, and
progressive enhancement of SBOMs (Software Bill Of Materials) throughout
build pipelines.
So lets add support for CycloneDX SBOM for packages and images
manifests.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
There is no standard for ABI versioning, so its not possible to find out
from `libext2fs2`, `libiwinfo20230701` or `libss2` package names if
thats just package name or package name with ABI version included. To
help with the decision, lets make ABI version aviable in package index.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Common Platform Enumeration (CPE) is a structured naming scheme for
information technology systems, software, and packages.
In order for the information to be processed further, it should also be
available in package index files.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Kernel module packages compiling is not cached (e.g. mac80211)
even with CONFIG_CCACHE on.
CC should be set to KERNEL_CC in KERNEL_MAKE_FLAGS at kernel.mk
to allow kernel module packages using ccache.
Signed-off-by: Zeyu Dong <dzy201415@gmail.com>
Make use of new toolchain define. TOOLCHAIN_DIR should be used only for
toolchain related packages and for everything else TOOLCHAIN_ROOT_DIR
and other define should be used instead.
Switch to new entry where possible.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The current code fails if we have package or host tools with no patches
to apply. The error printend is the following: (taking ubus as an
example)
make[2]: Entering directory '/home/ansuel/openwrt-ansuel/openwrt/scripts/config'
make[2]: 'conf' is up to date.
make[2]: Leaving directory '/home/ansuel/openwrt-ansuel/openwrt/scripts/config'
make[1]: Entering directory '/home/ansuel/openwrt-ansuel/openwrt'
make[2]: Entering directory '/home/ansuel/openwrt-ansuel/openwrt/package/system/ubus'
The source directory contains no quilt patches.
make[2]: *** [Makefile:81: quilt-check] Error 1
make[2]: Leaving directory '/home/ansuel/openwrt-ansuel/openwrt/package/system/ubus'
time: package/system/ubus/refresh#0.06#0.00#0.07
ERROR: package/system/ubus failed to build.
make[1]: *** [package/Makefile:120: package/system/ubus/refresh] Error 1
make[1]: Leaving directory '/home/ansuel/openwrt-ansuel/openwrt'
make: *** [/home/ansuel/openwrt-ansuel/openwrt/include/toplevel.mk:232: package/ubus/refresh] Error 2
We exit 1 after saying that there are no patches because later in the
function quilt pop fails to execute.
Having no patches for a package and calling refresh should not be
a critical error and the function should just do nothing.
To handle this improve quilt.mk with the following addition.
- If we don't have any patch for the package, we print a warning and we
create an empty series. This is useful to trick quilt and make it do
nothing.
We also create a status file .quilt_no_patch to detect in the other
function that we don't have patches to handle.
- In refresh makefile target, we check if .quilt_no_patch exist and
we skip quilt cleanup if this exist.
- In RefreshDir function we change the logic and now we delete the
patches directory and not only the content. This is done as a cleanup
to clean case with empty patches directory.
- In RefreshDir we check if .quilt_no_patch exist and we skip creating
the patches directory and copying the refreshed patches.
- In RefreshDir we delete at the end any trace of .quilt_no_patch if
present.
This is needed to support run like package/refresh that will run the
refresh process on any package present in the buildroot.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
To better reference them for diagnostic use, reference the PATCH_DIR and
FILES_DIR with the absolute path instead of using ./ and reference by
the relative location.
No behaviour change intended.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
When cmake is invoked to build a package it usually reports a warning
about unused variables passed to it. This is caused by openwrt passing
all supported variables to cmake, even if they are not all required by
the package being compiled.
To reduce clutter when compiling such packages these warnings are now
suppressed.
Approved-by: Rosen Penev <rosenp@gmail.com>
Reviewed-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Tjalling Hattink <t.hattink@fugro.com>
Currently, the same information is stored at the Packages.manifest in
the 'Package:' variable and also additionally in the 'SourceName:' variable.
So we have for Packages.manifest for strongswan-charon-cmd:
```
Package: strongswan-charon-cmd
Version: 5.9.11-1
SourceName: strongswan-charon-cmd
License: GPL-2.0-or-later
Section: net
```
This is not correct. Several installable packages are built from the same
strongswan source. Therefore it makes more sense that the source name is
really the source name. In this case the it is 'strongswan'.
After this change the Packages.manifest for strongswan-charon-cmd:
```
Package: strongswan-charon-cmd
Version: 5.9.11-1
SourceName: strongswan
License: GPL-2.0-or-later
Section: net
```
In summary. The 'Package' name is the name of the package to be installed
on the target system. The 'SourceName' is the compile unit from which the
package was build from. This must be the same for all installable
packages built from the same compile unit. This commit fixes that.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.1.49
All no patches automatically rebased, just update to checksum for new version.
Build system: x86/64
Build-tested: x86/64/AMD Cezanne
Run-tested: x86/64/AMD Cezanne
Signed-off-by: John Audia <therealgraysky@proton.me>
Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.1.48
No patches changed in this bump, only update was to checksum.
Build system: x86/64
Build-tested: x86/64/AMD Cezanne, filogic/xiaomi_redmi-router-ax6000-ubootmod
Run-tested: x86/64/AMD Cezanne, filogic/xiaomi_redmi-router-ax6000-ubootmod
Signed-off-by: John Audia <therealgraysky@proton.me>
Now that most cases do the same thing in SetupHostCommand, merge them
together into one. To allow moving the generic symlink check, invert the
check and let it check for relative links by matching on link targets
that do not start with a slash.
This then allows us to also drop the absolute link case, shortening the
case statement further.
This reorders the check to
* if it is not a symlink, do not change it
* if it is a symlink and it points to the found command, do not change it
* if it is a symlink with a relative path, do not change it
* else, update/replace it
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
To avoid replacing host built binaries with symlinks again, a check for
an appropriate stamp was added in 729909c07f ("prereq-build: do not
replace binaries with symlinks"). Unfortunately the stamp directory does
not exist in the SDK, so the fix was ineffective there.
This caused the packages builders to e.g. use the host tar again, which
in turn made the tarballs created different since it may lack
reproducibility fixes, or implement these differently, causing spurious
hash failures on source repository based packages.
Fix this by dropping the stamp dir check, and just check that the file
is usable.
Fixes: 729909c07f ("prereq-build: do not replace binaries with symlinks")
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
The current minimum OS requirement for OpenWrt is Ubuntu 18.04, which
includes 7 and 8. 8 is necessary for ccache.
gcc and g+++ are now symlinked to staging_dir, similar to Python.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Currently the git protocol downloads all submodules of the target
repository. This can be unwieldy for repositories with a lot of submodules
where only a subset are required in the context of the OpenWrt build.
This change adds a PKG_SOURCE_SUBMODULES variable to configure this
behavior. It takes a space-separated list of submodule paths, or the word
"skip" to disable submodule downloads entirely. The default is to download
all submodules, i.e. preserving current behavior.
Signed-off-by: Karsten Sperling <ksperling@apple.com>
Stable kernel git log:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/log/?h=v6.1.38
No patches needed to be rebased. Just updated checksum.
Build system: x86_64
Build-tested: x86_64/AMD Ryzen 7
Run-tested: x86_64/AMD Ryzen 7
Signed-off-by: John Audia <therealgraysky@proton.me>
[add link to stable kernel git log]
Signed-off-by: Nick Hainke <vincent@systemli.org>
Manually rebased:
generic/hack-6.1/220-arm-gc_sections.patch
armsr/patches-6.1/221-armsr-disable_gc_sections_armv7.patch
All other patches automatically rebased.
Signed-off-by: John Audia <therealgraysky@proton.me>
All patches automatically rebased.
Acknowledgment to @john-tho for the changes to fs.mk to accommodate new paths
introduced in 29429a1f58
Build system: x86_64
Build-tested: bcm2711/RPi4B
Run-tested: bcm2711/RPi4B
Signed-off-by: John Audia <therealgraysky@proton.me>
If CONFIG_USE_MOLD is set, all target packages will use the mold linker.
Except the ones which opted-out via setting PKG_BUILD_FLAGS:=no-mold.
Signed-off-by: Andre Heider <a.heider@gmail.com>
If a kernel package was defined where all KCONFIG symbols were dynamic,
and versioned, no FILES would be installed, as the foreach evaluation was
providing the value of the variable defined by the KCONFIG symbol name
including the version test
Fix this by calling the version_filter function on the list of KCONFIG
variable names run through by foreach
Example, kernel 6.1:
KCONFIG:=CONFIG_OLD@lt6.1 CONFIG_NEW@ge6.1
filter-out any KCONFIG settings forced by package:
CONFIG_OLD@lt6.1 CONFIG_NEW@ge6.1
there are dynamic settings, so for each of them,
get the value of the make variable defined by symbol name:
CONFIG_OLD@lt6.1 is not set
CONFIG_NEW@ge6.1 is not set
versus
CONFIG_OLD is not set
CONFIG_NEW=m
test if any of these are m, or y
if yes, install files, otherwise, nothing to install
Signed-off-by: John Thomson <git@johnthomson.fastmail.com.au>
These fields are used for EAX12 and EX6250v2 series, and perhaps other
devices. Compatibility is preserved with the WAX202 and WAX206.
In addition, adds the related vars to DEVICE_VARS so that the variables
work correctly with multiple devices.
References in GPL source:
https://www.downloads.netgear.com/files/GPL/EAX12_EAX11v2_EAX15v2_GPL_V1.0.3.34_src.tar.gz
* tools/imgencoder/src/gj_enc.c
Contains code that generates the encrypted image.
Signed-off-by: Wenli Looi <wlooi@ucalgary.ca>
Device specifications:
======================
* Qualcomm/Atheros AR9344
* 128 MB of RAM
* 16 MB of SPI NOR flash
* 2x 10/100 Mbps Ethernet
* 2T2R 2.4/5 GHz Wi-Fi
* 4x GPIO-LEDs (1x wifi, 2x ethernet, 1x power)
* 1x GPIO-button (reset)
* 2x fast ethernet
- lan1
+ builtin switch port 1
+ used as WAN interface
- lan2
+ builtin switch port 2
+ used as LAN interface
* 9-30V DC
* external antennas
Flashing instructions:
======================
Log in to https://192.168.127.253/
Username: admin
Password: moxa
Open Maintenance > Firmware Upgrade and install the factory image.
Serial console access:
======================
Connect a RS232-USB converter to the maintenance port.
Pinout: (reset button left) [GND] [NC] [RX] [TX]
Firmware Recovery:
==================
When the WLAN and SYS LEDs are flashing, the device is in recovery mode.
Serial console access is required to proceed with recovery.
Download the original image from MOXA and rename it to 'awk-1137c.rom'.
Set up a TFTP server at 192.168.127.1 and connect to a lan port.
Follow the instructions on the serial console to start the recovery.
Signed-off-by: Maximilian Martin <mm@simonwunderlich.de>
Make it possible to easily customize U-Boot config options via new
`UBOOT_CUSTOMIZE_CONFIG` variable, so we don't need to patch config
files or override config step with shell hackery.
This generic approach uses `config` CLI to tweak the .config as needed,
for example:
UBOOT_CUSTOMIZE_CONFIG := \
--enable CMD_EFIDEBUG \
--enable CMD_BOOTMENU \
--enable AUTOBOOT \
--enable AUTOBOOT_MENU_SHOW \
--disable AUTOBOOT_KEYED \
--disable AUTOBOOT_USE_MENUKEY \
--disable BOOTMENU_DISABLE_UBOOT_CONSOLE \
--set-val BOOTDELAY 2
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Handle compiling device tree overlay blobs separate to allow for
overlays being compiled with different parameters, mostly to safe
space.
Allow defining DEVICE_DTC_FLAGS and DEVICE_DTCO_FLAGS as per-device
parameters to be passed to dtc. Previously some boards directly used
DTC_FLAGS in their build recipe which then also affected other boards.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Instead of generating full config nodes incl. kernel, generate minimal
config nodes for device tree overlays to be applied to the main config.
In this way, multiple device tree overlays can be applied more easily.
While at it change filenames to upstream style, ie. use dtso and dtbo
suffix for device tree overlays.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Some individual build items install a group of programs
instead of a program matching the name of the build item.
Add support for installing stampfiles for each of the
programs installed by that build item,
which will allow more control and awareness
of what is installed by the rest of the build system,
if, for example, prereq symlink checks are looking
for the same program which is built already.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Some programs installed to staging_dir/host/bin
also install some symlinks to itself
for an alternative name.
Some of those new symlinks are overwriting
symlinks that were installed by prereq stage.
If prereq stage were to somehow be run again,
it should not be overwriting symlinks
that point to programs that are already built.
To filter that out, catch all symlinks
after first catching all symlinks
that have an absolute target
after all other cases in the case statement,
make sure it is not broken, and if so exit successfully.
Suggested-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Some programs, like bash and patch, are checked by prereq stage
and have a symlink installed, but then is later built from source.
Now that the prereq-build checks are not successful
just by finding the file alone, it is possible for
a new symlink to overwrite the installed binary.
If a normal file is found in staging_dir/host/bin,
let the check look for the associated stampfile, and if found,
skip creation of a symlink and exit successfully.
Suggested-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Signed-off-by: Michael Pratt <mcpratt@pm.me>
This reverts commit 7855378fcd.
The return "exit 1" was intentional and actually just
makes the symlink checks much more strict.
This new level of strictness added to the checks revealed
what was a confusing regression with prereq stage that
already existed but was not presenting itself
because of the simple way that checks used to be done before.
Either way, reverting to "exit 0" was a nice workaround
until the true root cause was discovered, so as to not interfere
with others' pull requests and builds in the meantime.
It turns out that this problem was the inconsistent value of $PATH
between different commands within the SetupHostCommand recipe,
now fixed in the parent commit, using the variable created
in the parent of the parent commit.
Ref: f75204036c ("prereq-build: allow host command symlinks to update")
Signed-off-by: Michael Pratt <mcpratt@pm.me>
In the recipe SetupHostCommand for checking
and creating symlinks, $PATH was only overridden
for one of several commands.
This causes the symlinks to be included
in the paths to pick a program from
when the check was repeated, because
staging_dir/host/bin was included in $PATH,
but only *sometimes*.
When the check ran again, the command succeded
with a $PATH including the symlink,
(eval "$$$$$$$$cmd")
while the path to the program was evaluated
with a $PATH NOT including the symlink,
(bin=...)
causing the symlink to be relinked incorrectly,
instead of passing as exactly the same.
Coincidentally, this was only a problem
if the symlink targeted the alternative
program with a different name.
By overriding the value of $PATH at the invocation of Make,
it will apply to the entire environment of the checks.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Add a variable that stores the original value of $PATH
in the host system's shell, before Make alters it.
This can be useful for when it is necessary
to ignore symlinks and programs made by the build system.
Define this new variable before all instances of
'export PATH:=' or similar.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
When the split was done, the case for testing kernel version wasn't
handled and only the to-be-compiled kernel version details files was
included. This cause the kernel Linux-Testing-Version output from
makefile target DUMP to report only the kernel version without the minor
version (example 6.1 instead of 6.1.29).
This value is expected with the full kernel version and this cause the
dump-target-info.pl script to not correctly identify if a target have a
testing kernel for the kernels calls.
Fix this regression by correctly including the kernel details files if
the target declare support for a testing kernel version.
Fixes: 0765466a42 ("kernel: split kernel version to dedicated files")
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
A package may run git as part of its build process, and if the package
source code is not from a git checkout, then git may traverse up the
directory tree to find buildroot's repository directory (.git).
For instance, Poetry Core, a Python build backend, will read the
contents of .gitignore for paths to exclude when creating a Python
package. If it finds buildroot's .gitignore file, then Poetry Core will
exclude all of the package's files[1].
This exports GIT_CEILING_DIRECTORIES for both package and host builds so
that git will not traverse beyond $(BUILD_DIR)/$(BUILD_DIR_HOST).
[1]: https://github.com/python-poetry/poetry/issues/5547
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
OpenSSL fails to compile in Fedora 38 container due to the following:
Can't locate IPC/Cmd.pm in @INC (you may need to install the IPC::Cmd module) (@INC contains: /openwrt/build_dir/target-x86_64_musl/openssl-3.0.8/util/perl /usr/local/lib64/perl5/5.36 /usr/local/share/perl5/5.36 /usr/lib64/perl5/vendor_perl /usr/share/perl5/vendor_perl /usr/lib64/perl5 /usr/share/perl5 /openwrt/build_dir/target-x86_64_musl/openssl-3.0.8/external/perl/Text-Template-1.56/lib) at /openwrt/build_dir/target-x86_64_musl/openssl-3.0.8/util/perl/OpenSSL/config.pm line 19.
BEGIN failed--compilation aborted at /openwrt/build_dir/target-x86_64_musl/openssl-3.0.8/util/perl/OpenSSL/config.pm line 19.
Compilation failed in require at ./Configure line 23.
BEGIN failed--compilation aborted at ./Configure line 23.
Seems like `dnf install -y perl-IPC-Cmd` fixes the issue.
So lets fix it by checking for Perl IPC::Cmd host module availability.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Tony has reported, that CI tools job is failing for him in macOS
container due to prereq check failure for GNU `install` utility.
Michael diagnosed it and from his traces it was clear, that the issue is
caused by a wrong return value in the success check case, so lets fix it
accordingly.
Fixes: f75204036c ("prereq-build: allow host command symlinks to update")
Reported-by: Tony Ambardar <itugrok@yahoo.com>
Diagnosed-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
New kernel version 6.1 introduced new INITRAMFS option. Add them to the
Initramfs functions to correctly compile initramfs images.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
This reverts commit 3b68fb57c9.
After refactoring build checks to update old symlinks,
and after a long time of no python 2 support,
this is no longer needed.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
This makes the prereq stage update the symlinks
installed into staging_dir/host/bin
by rearrainging the way they are verified.
Before, seeing or installing a symlink would result in
a successful exit code, and not installing a symlink
would result is a failed exit code. However,
that is not able to account for the difference
between existing good and bad links, or whether
the link would be the same if it was reinstalled,
because the check can match the program to a different path.
Instead, let a success exit code represent
identifying an existing symlink as exactly the same
as what would be installed if it did not exist,
and let a fail exit code represent
needing to install the symlink
or not having a match to the check criteria.
The failing exit code is caught by a new second attempt
for all of the check-* targets which will then indicate
to the user that there was an update by having a success
exit code when the check is run again and the link is the same.
When there is nothing to update, the checks will run only once.
This relies on the ls command to be POSIX-conformant with long format:
"path/to/link -> target/of/link"
Also, make sure the symlink is executable, not just a file,
and the directory only needs to be created once.
Fixes: #12610
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Commit 54070a1 was added to allow building proper SDKs with kernels <
5.10. Now that all targets use at least kernel 5.10 it can be reverted.
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
Some U-Boot checks for a specified uImage time and refuses to boot if
mismatched. This patch fixes it by recognizing UIMAGE_TIME parameter.
Signed-off-by: David Yang <mmyangfl@gmail.com>
Prevents subshell commands from failing to parse options
when having defined a whitespace in the VERSION_DIST.
As the called resulting images unlikely will handle
whitespace correctly, we replace them by "-".
Signed-off-by: Sven Roederer <S.Roederer@colvistec.de>
Manually adjusted before running quilt due to new location in tree:
backport-5.15/780-v5.16-bus-mhi-pci_generic-Introduce-Sierra-EM919X-support.patch
backport-5.15/781-v6.1-bus-mhi-host-always-print-detected-modem-name.patch
pending-5.15/790-bus-mhi-core-add-SBL-state-callback.patch
All other patches automatically rebased.
Build system: x86_64
Build-tested: bcm2711/RPi4B, ramips/tplink_archer-a6-v3, filogic/xiaomi_redmi-router-ax6000-ubootmod
Run-tested: bcm2711/RPi4B, ramips/tplink_archer-a6-v3, filogic/xiaomi_redmi-router-ax6000-ubootmod
Signed-off-by: John Audia <therealgraysky@proton.me>
On each generation of the archive check sum will differ, because when
checking out subversion repository, current date is used for directories
creation. Force tar to assign creation date of the last revision for all
items inside archive.
Signed-off-by: Tomasz Maciej Nowak <tmn505@gmail.com>
All targets are bumped to 5.15. Remove the old 5.10 patches, configs
and files using:
find target/linux -iname '*-5.10' -exec rm -r {} \;
Further, remove the 5.10 include.
Signed-off-by: Nick Hainke <vincent@systemli.org>
This reverts commit 91e3419a33.
Now that squashfs3-lzma generates reproducible output we can drop the
empty binary. Having a binary file in the tree is not nice and we actually
also use squashfs3-lzma for devices which expect the kernel to be loaded
from a squashfs3...
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
The filesystem is currently created on every build to trick the boot
loader of some FRITZ! devices into accepting the image. Sadly the
resulting squashfs-lzma filesystem is not reproducible. To fix this,
create a squashfs filesystem once and include it into the repository.
Creation happend as shown below
rm -rf empty_dir
mkdir empty_dir
./staging_dir/host/bin/mksquashfs-lzma \
empty_dir/ empty-squashfs-lzma \
-noappend -root-owned -be -nopad -b 65536 -fixed-time 0
Signed-off-by: Paul Spooren <mail@aparcar.org>
Some uses cases, like with autotools, need a path for 'true'
if we use them to force something to skip.
This will work by default on Linux hosts,
and require MacOS hosts to get coreutils,
which is currently installed in the CI
for 'macos-latest' host.
In the future, prereq stage can be reworked
to search for the actual binary
instead of relying on env.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Requires: tools/libdeflate
fix consistency of executable to use
`$(STAGING_DIR_HOST)/bin/libdeflate-gzip`, and not system-installed ones
from the usual environment `PATH`;
this affects option `CONFIG_KERNEL_INITRAMFS_COMPRESSION_GZIP`
this may have worked in the past but only via side effect of having the
binaries on the host system (and whatever unpredictable version or
patchset those might be), and did not use the improved but totally
compatible libdeflate-gzip
Fixes: 330bd380e8 ("image: allow building FIT and uImage with ramdisk")
Signed-off-by: Tony Butler <spudz76@gmail.com>
Requires: tools/bzip2
fix consistency of executable to use `$(STAGING_DIR_HOST)/bin/bzip2`, and
not system-installed ones from the usual environment `PATH`;
this affects option `CONFIG_KERNEL_INITRAMFS_COMPRESSION_BZIP2`
this may have worked in the past but only via side effect of having the
binaries on the host system (and whatever unpredictable version or
patchset those might be)
Fixes: 330bd380e8 ("image: allow building FIT and uImage with ramdisk")
Signed-off-by: Tony Butler <spudz76@gmail.com>
This reverts commit a33b97dcb1.
A new version of the squashfs4 tool patch reintroduced the -Xe option.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
This adapts the engine build infrastructure to allow building providers,
and packages the legacy provider. Providers are the successors of
engines, which have been deprecated.
The legacy provider supplies OpenSSL implementations of algorithms that
have been deemed legacy, including DES, IDEA, MDC2, SEED, and Whirlpool.
Even though these algorithms are implemented in a separate package,
their removal makes the regular library smaller by 3%, so the build
options will remain to allow lean custom builds. Their defaults will
change to 'y' if not bulding for a small flash, so that the regular
legacy package will contain a complete set of algorithms.
The engine build and configuration structure was changed to accomodate
providers, and adapt to the new style of openssl.cnf in version 3.0.
There is not a clean upgrade path for the /etc/ssl/openssl.cnf file,
installed by the openssl-conf package. It is recommended to rename or
remove the old config file when flashing an image with the updated
openssl-conf package, then apply the changes manually.
An old openssl.cnf file will silently work, but new engine or provider
packages will not be enabled. Any remaining engine config files under
/etc/ssl/engines.cnf.d can be removed.
On the build side, the include file used by engine packages was renamed
to openssl-module.mk, so the engine packages in other feeds need to
adapt.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
Remove the specialized copy of libtool
which was used for linking to uClibc++, which is now removed.
Also remove references to the deprecated fixup targets
that invoked this specialized libtool, which no package uses.
Ref: 6b2ed6101 ("uclibc++: remove")
Ref: c10515db6 ("re-enable the libtool PKG_BUILD_DEPENDS for PKG_FIXUP")
Ref: 246a5b334 ("More libtool madness")
Signed-off-by: Michael Pratt <mcpratt@pm.me>
This reduces open coding and allows to easily add a knob to enable
it treewide, where chosen packages can still opt-out via "no-lto".
Some packages used LTO, but not the linker plugin. This unifies 'em
all to attempt to produce better code.
Quoting man gcc(1):
"This improves the quality of optimization by exposing more code to the
link-time optimizer."
Also use -flto=auto instead of -flto=jobserver, as it's not guaranteed
that every buildsystem uses +$(MAKE) correctly.
Signed-off-by: Andre Heider <a.heider@gmail.com>
This reduces open coding and allows to easily add a knob to
enable it treewide, where chosen packages can still opt-out via
"no-gc-sections".
Note: libnl, mbedtls and opkg only used the CFLAGS part without the
LDFLAGS counterpart. That doesn't help at all if the goal is to produce
smaller binaries. I consider that an accident, and this fixes it.
Note: there are also packages using only the LDFLAGS part. I didn't
touch those, as gc might have been disabled via CFLAGS intentionally.
Signed-off-by: Andre Heider <a.heider@gmail.com>
PKG_BUILD_FLAGS is a new variable for package Makefiles similar to
PKG_FLAGS. It's a whitespace separated list of flags to control various
aspects of how a package is build.
The build system and/or .config defines the default for each, but
every package has the means to override it. Using $flagname enables
a flag, no-$flagname disables it.
Start with PKG_IREMAP as "iremap". That's easy as no package here
nor in any package feed uses it. The default is unchanged: enabled.
Packages can opt-out via:
PKG_BUILD_FLAGS:=no-iremap
(Not that any should, just to illustrate how to use it)
Signed-off-by: Andre Heider <a.heider@gmail.com>
Manually rebased:
ramips/patches-5.10/810-uvc-add-iPassion-iP2970-support.patch
All other patches automatically rebased.
Signed-off-by: John Audia <therealgraysky@proton.me>
No patches needed to be rebased.
Signed-off-by: John Audia <therealgraysky@proton.me>
Tested-by: Robert Marko <robimarko@gmail.com> #ipq807x/Dynalink WRX36
Tested-by: Stefan Lippers-Hollmann <s.l-h@gmx.de> #ipq807x/ax3600, x86_64/FW-7543B, ath79/tl-wdr3600, ipq806x/g10, ipq806x/nbg6817
No patches rebased as a function of this bump.
Build system: x86_64
Build-tested: ramips/tplink_archer-a6-v3
Run-tested: ramips/tplink_archer-a6-v3
Signed-off-by: John Audia <therealgraysky@proton.me>
autoreconf wants to use the gtkdocize tool now if a configure.ac file
defines GTK_DOC_CHECk(). OpenWrt does not ship the gtkdocize tool, just
use true instead. This fixes the build of some applications like guntls.
Fixes: 030447b8f4 ("tools/autoconf: bump to 2.71")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Recent versions of Automake
have changed dependency tracking significantly
(reference commit below)
causing breakage in some package builds
when using newer Automake with packages that need autoreconf
that were bootstrapped with an old version of Automake.
Those changes cause a great inconsistency between packages over time
where some packages may or may not use this feature,
and may or may not update the .ac and .am files
to work with the new methods.
This problem might exist in many packages
where autoreconf is not currently required,
but would cause build failure if autoreconf is used.
Fortunately, this feature is practically useless
for the purposes of Openwrt and the average developer,
so we can disable it.
GNU Automake manual states in part:
"Because dependencies are only computed as a side-effect of compilation...
no dependency information exists the first time a package is built...
dependency tracking is completely useless for one-time builds..."
A nice side-effect is that build times are slightly faster.
Ref: 6a675ef17edf7109da189f5ae70e2dc6b7665896 (automake.git)
Signed-off-by: Michael Pratt <mcpratt@pm.me>
The -Xpreset option changed format and dropped the -Xe and just require
the extreme string to be passed to the -Xpreset option.
Update the LZMA_XZ_OPTIONS to reflect this change.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The name of squashfs is confusing since in reality it's a really old
version using an old lzma library. This tools is used for old ath79
netgear target and to produde a fake squasfs3 image needed for some
specific bootloader from some OEM (AVM for example)
Rename squashfs tool to squasfs3-lzma to better describe it.
Rename the installed bin from mksquashfs-lzma to mksquashfs3-lzma.
Use tar transform to migrate the root directory in tar to the new
naming.
Drop redundant PKG_CAT variable not needed anymore.
Also update any user of this tool.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Packages in general use 4 check to trigger a recompile:
- timestamp for the build_dir
- timestamp for the staging stamp dir
- depends hash for the build_dir prepared file
- presence of package archieve in dl
If host tools are prebuilt and shipped in a container or manually
installed from an archieve, it would be ideal to skip including the
package archieve and just provide the build_dir prepared files and the
staging stamp file (and the actualy prebuilt tools).
Add some logic to skip dl download for host tools if AUTOREMOVE is
selected and checks for the presence of staging dir stamp file and build
dir stamp file.
If one of these requirements are not met, the package is redownloaded
and rebuilt.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Starting from version 2.39 binutils now warns about sections with rwx
permissions. While this is generally desirable it breaks building
ARM TrustedFirmware-A bl2 which treats warnings as errors.
Disable the warning/error for now to fix build.
Reference: 0579d9f5bc
Signed-off-by: Linhui Liu <liulinhui36@gmail.com>
To download a package the LLVM bins are not strictly needed.
Currently with an example run of make package/bridger/download V=s, the
build fail with
make[2]: Entering directory '/home/ansuel/openwrt-ansuel/openwrt/package/network/services/bridger'
bash: line 1: /home/ansuel/openwrt-ansuel/openwrt/staging_dir/host/llvm-bpf/bin/clang: No such file or directory
bash: line 1: [: : integer expression expected
/home/ansuel/openwrt-ansuel/openwrt/include/bpf.mk:71: *** ERROR: LLVM/clang version too old. Minimum required: 12, found: . Stop.
make[2]: Leaving directory '/home/ansuel/openwrt-ansuel/openwrt/package/network/services/bridger'
time: package/network/services/bridger/download#0.04#0.00#0.06
ERROR: package/network/services/bridger failed to build.
This is wrong since it may be needed to download the required packages
first and then compile them later.
Fix this by ignoring the LLVM bin check on non compile steps.
Tested-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
