To create packages the `ipkg-build` script is used which double packs
`control.tar.gz` and `data.tar.gz` to a single package. By default it's
using a verbose username instead of a numeric value for files.
Official OpenWrt images (artifacts) are created within docker containers
which do not seem to contain those verbose usernames and instead
defaults to numeric values.
This becomes a problem when rebuilding public artifacts because other
build environments may offer verbose usernames and there the created
packages is different from the official ones.
With this commit `ipkg-build` always uses numeric values for user/group
and thereby making it easier to reproduce official artifacts.
Signed-off-by: Paul Spooren <mail@aparcar.org>
CRC32 is available in a standard library. It seems reasonable
to defer to that rather than run a custom implementation.
Signed-off-by: Doug Kerr <dek3rr@gmail.com>
mtools recursive copy (mcopy -s ...) is using READDIR(3) to iterate
over the directory entries, hence they end up in the FAT filesystem in
traversal order which breaks reproducibility (rather than being added
to the FAT filesystem in a reproducible order). Implement recursive
copy in gen_image_generic.sh in Shell code instead, as in that way we
can force files to be copied in reproducible order.
Fixes: aece8f5ae8 ("scripts/gen_image_generic.sh: generate reproducible EFI filesystem")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Generate FAT filesystem for EFI boot in a reproducible way:
* use '--invariant' option of mkfs.fat
* set timestamps of all files to SOURCE_DATE_EPOCH
* make sure files are ordered locale-independent
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Set LC_ALL=C environment variable when calling 'sort' as the sort
order otherwise depends on the locale set.
Fixes: 56ce110b73 ("scripts: make sure conffiles are sorted")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
It may happen that conffiles are in different order on different builds.
Make sure they have the same order by sorting them.
FIX: #9612
Signed-off-by: Paul Spooren <mail@aparcar.org>
Introduce `sha256_unsigned` which is a checksum of the image _before_ a
signature is attached. This is helpful to compare image reproducibility.
Since the `.sha256sum` file is located in the $(KDIR) folder, switch
$(BIN_DIR) with $(KDIR) to simplify the code. The value of $(BIN_DIR)
itself is not stored inside the resulting JSON file, so it can be
replaced.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Set fixed timestamp for kernel other files in /boot filesystem.
This should help making x86 *combined* images reproducible.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
be64enc, be16dec, and be32dec are declared on FreeBSD 13.0, in
/usr/include/sys/endian.h so we should not declare them.
Fixes the following error during feeds update:
staging_dir/host/bin/mkhash: No such file or directory
gcc scripts/mkhash.c
scripts/mkhash.c:111:1: error: redefinition of 'be64enc'
111 | be64enc(void *buf, uint64_t u)
Signed-off-by: Georgi Valkov <gvalkov@abv.bg>
It seems, that there are currently some unhandled corner cases in which
`.toolchain_build_ver` results in empty file and thus forcing rebuilds,
even if the toolchain was build correctly just a few moments ago. Until
proper fix is found, workaround that by checking for this corner case
and simply populate `.toolchain_build_ver` file.
While at it, improve the UX and display version mismatch, so it's more
clear what has forced the rebuild:
"Toolchain build version changed (11.2.0-1 != ), running make targetclean"
References: https://gitlab.com/ynezz/openwrt/-/jobs/2133332533/raw
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Fixes following complaints and suggestions:
In scripts/check-toolchain-clean.sh line 2:
eval `grep CONFIG_GCC_VERSION .config`
^-- SC2046 (warning): Quote this to prevent word splitting.
^-- SC2006 (style): Use $(...) notation instead of legacy backticks `...`.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
diffconfig.sh runs ./scripts/config/conf, but it does not get built
with 'make {menu,x,n}config. Call 'make ./scripts/config/conf' to
ensure it's been built before running it, aborting in case of failure.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>[removed Fixes: due revert]
diffconfig.sh runs ./scripts/config/conf, but it does not get built
with 'make {menu,x,n}config. Call 'make ./scripts/config/conf' to
ensure it's been built befpre running it.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
Functional Changes
---------- -------
- make 'imply' not impose any restrictions: allow symbols implied by y
to become m
- change "modules" from sub-option to first-level attribute
Bugfixes
--------
- nconf: fix core dump when searching in empty menu
- nconf: stop endless search loops
- xconfig: fix content of the main widget
- xconfig: fix support for the split view mode
Other Changes
----- -------
- highlight xconfig 'comment' lines with '***'
- xconfig: navigate menus on hyperlinks
- xconfig: drop support for Qt4
- improve host ncurses detection
Update the 'option modules' usage to just 'modules' in Config.in.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
Both $(AUTORELEASE) and $(PKG_SRC_VERSION) (from luci.git) use the Git
log to determine releases and package timestamps.
Feeds are shallow cloned by default, resulting in an incomplete Git log
and therefore different local package versions than offered upstream.
This commits sets the default feeds to use `src-git-full` to solve that.
Add fixes from "2b1d92f: scripts/feeds: silence git warning by selecting
pull style" to `src-git-full`
Signed-off-by: Paul Spooren <mail@aparcar.org>
The value is retreived from a env variable which defaults to be read as
a string. However the SOURCE_DATE_EPOCH is a unix timestamp aka integer.
Fix this to allow downstream tools to parse the value directly.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Ubuntu started to flag which as deprecated and it
seems which is not really standard and may vary
across Distro.
Drop the use of which and use the standard 'command -v'
for this simple task.
Which is still present in the prereq if some package/script
still use which.
A utility script called command_all.sh is implemented that
will just mimic the output of which -a.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
This commits adds the ability to print Kernel versions of all
targets/subtargets. If a testing Kernel is set print that version as
well.
Example output:
apm821xx/nand 5.10
apm821xx/sata 5.10
arc770/generic 5.4
archs38/generic 5.4
armvirt/32 5.10
armvirt/64 5.10
at91/sam9x 5.10
at91/sama5 5.10
ath25/generic 5.4
ath79/generic 5.4 5.10
ath79/mikrotik 5.4 5.10
--- %< ---
This should help to get a quick update on the state of Kernels.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Adds generic support for sysupgrading on eMMC-based devices.
Provide function emmc_do_upgrade and emmc_copy_config to be used in
/lib/upgrade/platform.sh instead of redundantly implementing the same
logic over and over again.
Similar to generic sysupgrade on NAND, use environment variables
CI_KERNPART, CI_ROOTPART and newly introduce CI_DATAPART to indicate
GPT partition names to be used. On devices with more than one MMC
block device, CI_ROOTDEV can be used to specify the MMC device for
partition name lookups.
Also allow to select block devices directly using EMMC_KERN_DEV,
EMMC_ROOT_DEV and EMMC_DATA_DEV, as using GPT partition names is not
always an option (e.g. when forced to use MBR).
To easily handle writing kernel and rootfs make use of sysupgrade.tar
format convention which is also already used for generic NAND support.
Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
CC: Li Zhang <li.zhang@gl-inet.com>
CC: TruongSinh Tran-Nguyen <i@truongsinh.pro>
Currently ubinize-image script always expects the
rootfs image to be passed and a volume for it created.
So, to allow only ubinizing a kernel for example which
the MikroTik hAP ac3 and other new NAND devices from
MikroTik require make rootfs an optional parameter like
kernel.
Signed-off-by: Robert Marko <robimarko@gmail.com>
