The ImageBuilder `make manifest` prints all installed packages. This
function can be used to create a list of package and corresponding
package versions before attempting image creation.
When called with `--strip-abi` OPKG can automatically strip attached
ABIVersions from package names. Make this function accessible for the
ImageBuilder by adding a `STRIP_ABI` variable.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Invoke bundle-libraries.sh with any buildroot related directory entries
removed from $PATH to avoid picking up cross versions of utilities like
ldd which will not properly work when used against host executables.
This should fix executable bundling for glibc-target imagebuilders.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
When building images with the imagebuilder, the partition signature
never changes. The signature is generated by hashing SOURCE_DATE_EPOCH
and LINUX_VERMAGIC which are undefined. Prepopulate these variables, as
done by the SDK.
Signed-off-by: Matthew Gyurgyik <matthew@gyurgyik.io>
The ImageBuilder downloads pre-built packages and adds them to images.
This process uses `opkg` which has the capability to verify package list
signatures via `usign`, as enabled per default on running OpenWrt
devices.
Until now this was disabled for ImageBuilders because neither the `opkg`
keys nor the `opkg-add` script was present during first packagelist
update.
To harden the ImageBuilder against *drive-by-download-attacks* both keys
and verification script are added to the ImageBuilder allowing `opkg` to
verify downloaded package indices.
This commit adds `opkg-add` to the ImageBuilder scripts folder. The keys
folder is added to ImageBuilder $TOPDIR to have an obvious place for users to
store their own keys. The `option check_signature` is appended to the
repositories.conf file. All of the above only happens if the Buildbot
runs with the SIGNATURE_CHECK option.
The keys stored in the ImageBuilder keys/ are the same as included in
the openwrt-keyring package. To avoid the chicken-egg problem of
downloading and verifying a package, containing signing keys, the keys
are added during the ImageBuilder generation. They are same as in
shipped images (stored at `/etc/opkg/keys/`).
To allow a local package feed in which the user can add additional
packages, a local set of `usign` and `ucert` keys is generated, same as
building OpenWrt from source. The private key signs the local repository
inside the packages/ folder. The local public key is added to the keys/
folder to be considered by `opkg` when updating repositories. This way a
local package feed can be modified while requiring `opkg` to check
signatures for remote feed, making HTTPS optional.
The new option `ADD_LOCAL_KEY` allows to add the local key inside the
created images, adding the advantage that sysupgrades can validate the
ImageBuilders local key.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Without an absolute path to staging_dir/host/bin/sstrip the Makefile
tries to run a host installed version of sstrip, which is likely not
available.
Signed-off-by: Paul Spooren <mail@aparcar.org>
With the fix of external kmod feeds it is possible to ship the
ImageBuilder without any packages except the pseudo packages kernel and
libc. Therefore the local package feeds becomes optional.
This commit adds a check to the package_reload function to only run if
the local feed is existing.
Signed-off-by: Paul Spooren <mail@aparcar.org>
The buildbots generate a kmod archive which should be used instead of a
local copy. This is possible due to the introduction of a kernelversion
specific feed.
This commit adds the ability of using only signed package feeds.
Signed-off-by: Paul Spooren <mail@aparcar.org>
The `libfakeroot` files are currently missing in the ImageBuilder. As
`fakeroot` is always built, copy those files unconditionally.
Signed-off-by: Paul Spooren <mail@aparcar.org>
The package manager `opkg` offers the function `whatdepends` to print
packages that depend on a specific package.
This feature is useful when used in a CI to not only build an upgraded
package but all packages with a dependency.
Usage:
make whatdepends PACKAGE=libipset
The resulting list can be fed into a SDK building all packages and warn
if anything fails.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Both IB and SDK now use the same logic for packing.
This commit add reproducible multithread compression to the SDK and
corrects the file mtime for both. Previously all files where just copied
over from the build system, generating random mtimes.
Signed-off-by: Paul Spooren <mail@aparcar.org>
The folder `json_info_files` contains multiple JSON files which describe
created firmware images. The folder is not removed between builds as the
ImageBuilder does not use `image.mk`.
Not removing the JSON files result in a merged `profiles.json` file
containing entries for outdated or non-existing images.
This commit adds the `json_info_files/` cleanup step to the ImageBuilder
Makefile.
Signed-off-by: Paul Spooren <mail@aparcar.org>
The patch 4a1a58a3 build, imagebuilder: Do not require libncurses-dev
was supposed to remove libncurses as a requirement for the ImageBuilder.
However as the IB=1 is only exported during building, not for checking
requirements, it did never actually work.
This commit export IB=1 to the requirement check.
Signed-off-by: Paul Spooren <mail@aparcar.org>
JSON info files contain machine readable information of built profiles
and resulting images. These files were added in commit 881ed09ee6
("build: create JSON files containing image info").
They are useful for firmware wizards and script checking for
reproducibility.
Currently all JSON files are stored next to the built images, resulting
in up to 168 individual files for the ath79/generic target.
This patch refactors the JSON creation to store individual per image
(not per profile) files in $(BUILD_DIR)/json_info_files and create an
single overview file called `profiles.json` in the target directory.
Storing per image files and not per profile solves the problem of
parallel file writes. If a profiles sysupgrade and factory image are
finished at the same time both processes would write to the same JSON
file, resulting in randomly broken outputs.
Some target like x86/64 do not use the image code yet, resulting in
missing JSON files. If no JSON info files were created, no
`profiles.json` files is created as it would be empty anyway.
As before, this creation is enabled by default only if `BUILDBOT` is set.
Tested via buildroot & ImageBuilder on ath79/generic, imx6 and x86/64.
Signed-off-by: Paul Spooren <mail@aparcar.org>
[json_info_files dir handling in Make, if case refactoring]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
This speeds up the packing of the imagebuilder a lot:
imagebuilder-T0.tar.xz real 0m25.199s user 2m45.967s sys 0m1.218s
imagebuilder-T1.tar.xz real 2m02.543s user 2m02.418s sys 0m1.653s
imagebuilder-T2.tar.xz real 1m03.684s user 1m59.931s sys 0m0.587s
imagebuilder-T3.tar.xz real 0m48.033s user 2m02.904s sys 0m0.637s
imagebuilder-T4.tar.xz real 0m38.963s user 2m15.521s sys 0m0.783s
imagebuilder-T5.tar.xz real 0m37.994s user 2m21.461s sys 0m0.919s
imagebuilder-T6.tar.xz real 0m39.524s user 2m48.115s sys 0m1.279s
imagebuilder-T7.tar.xz real 0m34.061s user 2m45.097s sys 0m1.174s
imagebuilder-T8.tar.xz real 0m27.286s user 2m55.449s sys 0m1.329s
imagebuilder-T9.tar.xz real 0m25.205s user 2m44.894s sys 0m1.208s
To keep the output reproducible in any case, we enforce a minimum amount
of 2 threads.
Signed-off-by: Martin Schiller <ms@dev.tdt.de>
[refactored into reusable NPROC var, more verbose commit message]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
For x86/64 (maybe more) target the SUPPORTED_DEVICES variable is empty
which causes the `&&` junction to fail, producing a non zero exit code.
Tested-by: Paul Spooren <mail@aparcar.org>
Fixed-by: Jo-Philipp Wich <jo@mein.io>
Signed-off-by: Paul Spooren <mail@aparcar.org>
Adds a new variable DISABLED_SERVICES to ImageBuilder Makefile, which
defines a list of services (installed as /etc/init.d/*) to be disabled
during the build of a custom image (normally all are enabled).
It comes handy when a particular service should not be run under normal
circumstances, but should be ready in the image for situations when it
might be needed.
Signed-off-by: Richard Musil <risa2000x@gmail.com>
This is useful in for the attendedsyupsgrade server (asu) to
distinguish between snapshot version. Currently asu can't tell devices
requesting a snapshot build if the same build is already installed.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Instead of showing a slightly more readable target like
"ar71xx (Generic)" print the more generic format "ar71xx/genric"
Signed-off-by: Paul Spooren <mail@aparcar.org>
Having image metadata (and signature) appended is a condition for
semi-automated sysupgrade, hence IB needs to be able to tell which
images will end up with metadata.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This really simplifies debugging, if a package is not found or a feed is
not reachable, a proper stderr is printed. Currently it would only say
`_call_manifest` failed.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Tested with 18.06.0-rc2/ar71xx/generic/tl-wdr4300-v1, image & list
This PR is based on the work of @fewckert[1] with slight improvements.
Add function `manifest` to show the manifest of the produced image,
before actually building it. The manifest contains an orderd list of
package name and version.
This is usefull to check package dependencies but also determine a
unique and reproducible image name before building the package. The
sysupgrade server[2] builds images on request with individual package
selection. To distignish between created images which contain differnt
packages, the EXTRA_IMAGE_NAME is set to a shortend hash of the
manifest's content. So far the image was renamed afterwards as the
manifests content was unknown, however this corrupts the signed
sha256sums. This patch allows a clean solution as to dtermine the
manifest in advance and set the EXTRA_IMAGE_NAME accordingly.
[1]: https://github.com/lede-project/source/pull/1591
[2]: https://github.com/aparcar/attendedsysupgrade-server
Signed-off-by: Paul Spooren <mail@aparcar.org>
In addition to removing redundant code, this fixes various issues in
IB-generated images that have been fixed in prepare_rootfs before,
including better handling of CONFIG_CLEAN_IPKG and enabling of initscripts
from FILES.
We also reuse the opkg macro and remove --force-... flags that have been
removed from rootfs.mk as well.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
No longer rewrite opkg list output in package_list function, remove
the awk call in the pipe (which was intended for a single specific
use-case).
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
commit 19ac879954 (imagebuilder: add package_list function) introduced
a new function 'package_list' to the imagebuilder Makefile.
Unfortunately the package list was poluted by stdout noise of the
Makefile itself as well as opkg. Redirect those outputs to stderr to
make sure that the package_list returned doesn't contain progress
info output but really only packages.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
The imagebuilder can now list all available packages by using make
package_list. This is usefull for scripts to retrieve a list of all
packages with versions (and size)
Signed-off-by: Paul Spooren <paul@spooren.de>
[daniel@makrotopia.org: fixed commit message]
Refer to LINUX_KARCH instead of ARCH when bundling DTS files in the image
builder tarball.
While we're at it, also dereference symbolic links when copying as some
kernel architectures contain symbolic links in their DTS directories.
This fixes aarch64 imagebuilders such as brcm2708/bcm2710 ones in particular
as the kernel refers to "aarch64" as "arm64" internally.
Ref: https://forum.lede-project.org/t/lede-image-builder-problem/3680
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
This can be used to tweak the buildbot behavior without having to change
buildbot's configuration.
It will also allow us to add more aggressive clean steps (e.g. on
toolchain changes), which would break developers' workflows if enable
by default.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Use silent make invocations for sub-makes like build_image or checksum to
avoid bloating the IB output with non-status info.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Introduce a new location STAGING_DIR_IMAGE which is intended to be used by
bootloader iamges and similar image-related artifacts.
This directory is guaranteed to be persistent across kernel upgrades which
might involve a removal of KERNEL_BUILD_DIR and is guranteed to be bundled
with the image builder.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
The name "Plat'Home OpenBlocks AX3" causes the imagebuilders "make info"
command to fail with:
bash: -c: line 0: syntax error near unexpected token `('
bash: -c: line 0: `echo; [...]'
Makefile:99: recipe for target '_call_info' failed
Properly escape single quotes to avoid breaking the echo commands.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Rework the bundle-libraries.sh implementation to use a more robust approach
for executing host binaries through the shipped ELF loader and libraries.
The previous approach relied on symlinks pointing to a wrapper script which
caused various issues, especially with multicall binaries as the original
argv[0] name was not preserved through the ld.so invocation. Another down-
side was the fact that the actual binaries got moved into another directory
which caused executables to fail looking up resources with paths relative
to the executable location.
The new library wrapper implements the following improvements:
- Instead of symlinks pointing to a common wrapper, each ELF executable
is now replaced by a unqiue shell script which retains the original
program name getting called
- Instead of letting ld.so invoke the ELF executable directly, launch
the final ELF binary through a helper program which fixes up the argv[0]
argument for the target program
- Support sharing a common location for the bundled libraries instead of
having one copy in each directory containing wrapped binaries
Finally modify the SDK build to wrap the staging_dir and toolchain binaries
which allows to use the SDK on systems with a different glibc version.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
With symlink tree some directories are just symlinked which
means IB and SDK end up with a symlink instead of an actual
directory; this fixes the missing files by dereferencesing
the directories instead of copying the symlinks.
Signed-off-by: Daniel Dickinson <lede@cshore.thecshore.com>
Add a new "checksum" make target which generates an sha256sums file over the
image files produced in bin/targets/ and automatically call it during make
world after the package index generation.
The advantage of this new target is that it is guaranteed to run after the
images, the SDK and the ImageBuilder archives have been generated to ensure
that they all end up in the checksum file. Fixes FS#51.
Uses sed to postprocess the OpenSSL digest output into an sha256sum command
compatible format.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
When imagebuild sorts package lists it breaks opkg's ability to realize
that a providers for a Provides has already been installed, when the sort
results in the provider being later in the list of packages that a package
which depends on a Provides (and hence the provider is not yet installed
for opkg to realize the provider was available doesn't not handle the case
of a package that is to be installed satisfying a dependency, only one that
is already installed (or which it schedules to be installed, which in the
absence of an installed provider is whichever provider happens to be the
default)
Signed-off-by: Daniel Dickinson <openwrt@daniel.thecshore.com>
