Previous refactoring of the script moved the LDSO detection into a
file-not-exists condition, causing onyl the very first executable to
get bundled.
Solve the problem by unconditionally checking for LDSO again.
Fixes: 9030a78a71 ("scripts: bundle-libraries: prevent loading host locales")
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
The current make-ras.sh image generation script for the ZyXEL NBG6617
has portability issues with bash. Because of this, factory images are
currently not built correctly by the OpenWRT buildbots.
This commit replaces the make-ras.sh by C-written mkrasimage.
The new mkrasimage is also compatible with other ZyXEL devices using
the ras image-format.
This is not tested with the NBG6616 but it correctly builds the
header for ZyXEL factory image.
Signed-off-by: David Bauer <mail@david-bauer.net>
Binary patch the bundled glibc library to inhibit loading of host locale
archives in order to avoid triggering internal libc assertions when
invoking shipped, bundled executables.
The problem has been solved with upstream Glibc commit
0062ace229 ("Gracefully handle incompatible locale data") but we still
need to deal with older Glibc binaries for some time to come.
Fixes FS#1803
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Use Time::HiRes when available and fallback to raw syscall interface
when not. If that fails too, simply report 0, 0 as real time.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Replace the GNU time program invocation with a simple Perl script reporting
the timing values. Since we require Perl anyway for the build system, we can
as well use that instead of requiring a random GNU utility rarely installed
by default.
Fixes: ff6e62b288 ("build: log time taken by each packages/steps")
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Compiler is producing the warning:
warning: format not a string literal and no format arguments
[-Wformat-security]
This patch makes the format a literal string in printf statements.
This with: gcc version 7.3.0 (Ubuntu 7.3.0-16ubuntu3) in Ubuntu 18.04
Signed-off-by: Paul Schulz <paul@mawsonlakes.org>
Allow enabling/commenting/disabling each feed individually by using a
tristate config symbol.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
The src-dummy method does not actually obtain any feed, but it can be used
to insert addtional entries into the opkg distfeeds.conf. This is useful to
make package feeds available to users without requiring the corresponding
source feeds to be available during build.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
- Make the code more GitHub-specific
- Requires mirror hash to work with .gitattributes
- Use different API depending on whether PKG_SOURCE_VERSION is a
complete commit id or other ref types like tags
- Fix removing symbolic link
- pre-clean dir_untar for possible leftovers from previous run
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Move SCAN_DEPS to scan.mk to eliminate redundancy with scripts/feeds
Add image/*.mk to SCAN_DEPS for targets to pick up newly added devices
Signed-off-by: Felix Fietkau <nbd@nbd.name>
A new python script scripts/download.py is added to fetch tarballs using
GitHub archive API [1], then repack in a reproducible way same as the
current DownloadMethod/git
GitHub imposes a 60 reqs/hour rate limit on unauthenticated API
access[2]. This affects fetching commit date for feeding tar --mtime=
argument. However, observation indicates that archive download is NOT
subject to this limit at the moment. In the rare cases where download
fails because of this, we will falback to using DownloadMethod/git
The missing piece in the GitHub API is that it cannot provide in the
tarball dependent submodules's source code. In that case, the
implementation will also fallback to using DownloadMethod/git
[1] Get archive link, https://developer.github.com/v3/repos/contents/#get-archive-link
[2] Rate limiting, https://developer.github.com/v3/#rate-limiting
v2 <- v1:
- allow passing multiple urls with --urls argument
- add commit ts cache. can be helpful on retry
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
This patch adds support for ZyXEL NBG6617
Hardware highlights:
SOC: IPQ4018 / QCA Dakota
CPU: Quad-Core ARMv7 Processor rev 5 (v7l) Cortex-A7
DRAM: 256 MiB DDR3L-1600/1866 Nanya NT5CC128M16IP-DI @ 537 MHz
NOR: 32 MiB Macronix MX25L25635F
ETH: Qualcomm Atheros QCA8075 Gigabit Switch (4 x LAN, 1 x WAN)
USB: 1 x 3.0 (via Synopsys DesignWare DWC3 controller in the SoC)
WLAN1: Qualcomm Atheros QCA4018 2.4GHz 802.11bgn 2:2x2
WLAN2: Qualcomm Atheros QCA4018 5GHz 802.11a/n/ac 2:2x2
INPUT: RESET Button, WIFI/Rfkill Togglebutton, WPS Button
LEDS: Power, WAN, LAN 1-4, WLAN 2.4GHz, WLAN 5GHz, USB, WPS
Serial:
WARNING: The serial port needs a TTL/RS-232 3.3v level converter!
The Serial setting is 115200-8-N-1. The 1x4 .1" header comes
pre-soldered. Pinout:
1. 3v3 (Label printed on the PCB), 2. RX, 3. GND, 4. TX
first install / debricking / restore stock:
0. Have a PC running a tftp-server @ 192.168.1.99/24
1. connect the PC to any LAN-Ports
2. put the openwrt...-factory.bin (or V1.00(ABCT.X).bin for stock) file
into the tftp-server root directory and rename it to just "ras.bin".
3. power-cycle the router and hold down the the WPS button (for 30sek)
4. Wait (for a long time - the serial console provides some progress
reports. The u-boot says it best: "Please be patient".
5. Once the power LED starts to flashes slowly and the USB + WPS LEDs
flashes fast at the same time. You have to reboot the device and
it should then come right up.
Installation via Web-UI:
0. Connect a PC to the powered-on router. It will assign your PC a
IP-address via DHCP
1. Access the Web-UI at 192.168.1.1 (Default Passwort: 1234)
2. Go to the "Expert Mode"
3. Under "Maintenance", select "Firmware-Upgrade"
4. Upload the OpenWRT factory image
5. Wait for the Device to finish.
It will reboot into OpenWRT without any additional actions needed.
To open the ZyXEL NBG6617:
0. remove the four rubber feet glued on the backside
1. remove the four philips screws and pry open the top cover
(by applying force between the plastic top housing from the
backside/lan-port side)
Access the real u-boot shell:
ZyXEL uses a proprietary loader/shell on top of u-boot: "ZyXEL zloader v2.02"
When the device is starting up, the user can enter the the loader shell
by simply pressing a key within the 3 seconds once the following string
appears on the serial console:
| Hit any key to stop autoboot: 3
The user is then dropped to a locked shell.
|NBG6617> HELP
|ATEN x[,y] set BootExtension Debug Flag (y=password)
|ATSE x show the seed of password generator
|ATSH dump manufacturer related data in ROM
|ATRT [x,y,z,u] RAM read/write test (x=level, y=start addr, z=end addr, u=iterations)
|ATGO boot up whole system
|ATUR x upgrade RAS image (filename)
|NBG6617>
In order to escape/unlock a password challenge has to be passed.
Note: the value is dynamic! you have to calculate your own!
First use ATSE $MODELNAME (MODELNAME is the hostname in u-boot env)
to get the challange value/seed.
|NBG6617> ATSE NBG6617
|012345678901
This seed/value can be converted to the password with the help of this
bash script (Thanks to http://www.adslayuda.com/Zyxel650-9.html authors):
- tool.sh -
ror32() {
echo $(( ($1 >> $2) | (($1 << (32 - $2) & (2**32-1)) ) ))
}
v="0x$1"
a="0x${v:2:6}"
b=$(( $a + 0x10F0A563))
c=$(( 0x${v:12:14} & 7 ))
p=$(( $(ror32 $b $c) ^ $a ))
printf "ATEN 1,%X\n" $p
- end of tool.sh -
|# bash ./tool.sh 012345678901
|
|ATEN 1,879C711
copy and paste the result into the shell to unlock zloader.
|NBG6617> ATEN 1,0046B0017430
If the entered code was correct the shell will change to
use the ATGU command to enter the real u-boot shell.
|NBG6617> ATGU
|NBG6617#
Co-authored-by: David Bauer <mail@david-bauer.net>
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: David Bauer <mail@david-bauer.net>
When sending script/env diff's output to a file or pipeline, you
don't want escape characters for ANSI color sequences fouling the
byte stream.
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
SourceForge has supported HTTPS for its downloads for a long time now.
I have not been able to see any failures resulting from this change.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
This allegedly fixes compilation of the library bundler preload library on
Apple OS X. The resulting executables have not been runtime tested due to a
lack of suitable test hardware.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
* QCA IPQ4019
* 256 MB of RAM
* 32 MB of SPI NOR flash (s25fl256s1)
- 2x 15 MB available; but one of the 15 MB regions is the recovery image
* 2T2R 2.4 GHz
- QCA4019 hw1.0 (SoC)
- requires special BDF in QCA4019/hw1.0/board-2.bin with
bus=ahb,bmi-chip-id=0,bmi-board-id=20,variant=OM-A62
* 2T2R 5 GHz (channel 36-64)
- QCA9888 hw2.0 (PCI)
- requires special BDF in QCA9888/hw2.0/board-2.bin
bus=pci,bmi-chip-id=0,bmi-board-id=16,variant=OM-A62
* 2T2R 5 GHz (channel 100-165)
- QCA4019 hw1.0 (SoC)
- requires special BDF in QCA4019/hw1.0/board-2.bin with
bus=ahb,bmi-chip-id=0,bmi-board-id=21,variant=OM-A62
* multi-color LED (controlled via red/green/blue GPIOs)
* 1x button (reset; kmod-input-gpio-keys compatible)
* external watchdog
- triggered GPIO
* 1x USB (xHCI)
* TTL pins are on board (arrow points to VCC, then follows: GND, TX, RX)
* 2x gigabit ethernet
- phy@mdio3:
+ Label: Ethernet 1
+ gmac0 (ethaddr) in original firmware
+ 802.3at POE+
- phy@mdio4:
+ Label: Ethernet 2
+ gmac1 (eth1addr) in original firmware
+ 18-24V passive POE (mode B)
* powered only via POE
The tool ap51-flash (https://github.com/ap51-flash/ap51-flash) should be
used to transfer the factory image to the u-boot when the device boots up.
The initramfs image can be started using
setenv bootargs 'loglevel=8 earlycon=msm_serial_dm,0x78af000 console=ttyMSM0,115200 mtdparts=spi0.0:256k(0:SBL1),128k(0:MIBIB),384k(0:QSEE),64k(0:CDT),64k(0:DDRPARAMS),64k(0:APPSBLENV),512k(0:APPSBL),64k(0:ART),64k(0:custom),64k(0:KEYS),15552k(inactive),15552k(inactive2)'
tftpboot 0x84000000 openwrt-ipq40xx-openmesh_a62-initramfs-fit-uImage.itb
set fdt_high 0x85000000
bootm 0x84000000
Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
Use the EVA bootloader to load a small linux system into the ram and boot
it from there:
./scripts/flashing/eva_ramboot.py 192.168.178.1 path/to/initramfs-kernel.bin
Signed-off-by: Valentin Spreckels <Valentin.Spreckels@Informatik.Uni-Oldenburg.DE>
Acked-by: John Crispin <john@phrozen.org>
[reworded commit message]
Signed-off-by: Mathias Kresin <dev@kresin.me>
Some devices only boot when a special config is found in the image and
completely ignore the default entry during the selection. These devices can
now use the variable DEVICE_DTS_CONFIG in their device image definition.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
/lib/functions.sh can deal with Require-User specifications that only
contain a group, but no user. Adjust metadata.pm to allow such
specifications as well.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
``make xconfig`` toplevel target will invoke ``make qconf`` inside
./scripts/config directory, which results a ``qconf`` executable.
This commit removes leftover ``qconf`` executable during ``make
config-clean``.
Signed-off-by: Alif M. Ahmad <alive4ever@live.com>
Remove references to /etc/, /lib/ and /usr/ from the bundled ld.so
interpreter using simple binary patching.
This is needed to prevent loading host system libraries such as
libnss_compat.so.2 on foreign systems, which may result in ld.so
inconsistency assertions.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Prevent emitting broken dependency statements when resolving references
to non existing packages to an empty provider list.
Fixes: 47d6b05ad3 ("metadata: always resolve dependencies through provides list")
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
The logic for choice between source and binary packages was reversed.
Fixes: 52719c2b67 "metadata: scripts/feeds: distinguish between source
and binary packages, resolve virtual dependencies"
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Package "features" seem to be unused for some time. In any case, custom
Config.in snippets and package PROVIDES are a much more flexible way to
express similar options.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Properly resolve build depends to source packages and runtime depends to
binary packages. Dependencies on virtual packages are resolved to the first
provider now.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Instead of adding virtual packages to the normal package list, keep a
separate list for provides, make each package provide itself, and resolve
all dependencies through this list. This allows to use PROVIDES to replace
existing packages.
Fixes FS#837.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Runtime depends cannot have a buildtype suffix, and they never refer to
source package names. In addition, this adds warnings about unsatisfiable
dependencies.
Furthermore, this change fixes the generation of conditional build
dependencies for virtual packages provided by different source packages.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Target build depends are similar to host build depends in that they refer
to source packages rather than binary packages. Therefore, it makes sense
to handle them together, rather than putting them in a list together with
runtime depends and trying to figure out if the entries refer to source or
to binary packages afterwards.
This does lead to PKG_BUILD_DEPENDS entries referring to binary package
names not working anymore, which requires some fixes in the package
repositories.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
This feature has been unused for years, and its scope is too limited to be
actually useful.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Every single reference to subdir was concatenated with the source package
name, so it makes sense to store the concatenated value instead.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
We often want to access fields of a source packages through pkg->{src}.
Allow accessing them directly instead of resolving the source hash through
srcpackages.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
All build dependencies are between source packages. Interating over source
rather than binary packages simplifies parts of the code and prepares
further improvement.
As a side effect, this changes the implicit default variant of a few
packages (the first defined is used now instead of the lexicographically
first).
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Build types are a property of source rather than binary packages. This is a
preparation for followup cleanup.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Turn the srcpackage values into hashes to allow storing more information
than just binary package names.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Nothing explicitly depends on base-files, and even if it would, it would
not cause any problems. Remove the unused special case.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
The new build commands operate on the input image and use it again as
output image. This conflicts with the way combined-ext-image.sh was
operating. It required that input and output files are different files and
and that it can write freely to the output file.
This can be avoided when all intermediate build steps by
combined-ext-image.sh are done in a temporary directory. The output file is
then only overwritten in the last step.
Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com>
A self-dependency is not an error worth a warning; rather, it is very
common: whenever there are dependencies between different binary packages
originating from the same source package, such dependencies occur. Not
actually generating dependency rules is correct, but already handled a few
lines below.
A typo prevented this redundant rule from working, which is the reason the
warning was not actually printed.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
When calling a download target, hash verification is now completely
skipped if we set PKG_HASH=skip.
This allows to easily bump package version:
$ make package/<mypackage>/download PKG_HASH=skip V=s
$ make package/<mypackage>/check FIXUP=1 V=s
This will download the new version of the package, and then automatically
update PKG_HASH with the hash of the new version. Of course, it is still
the responsibility of the packager to ensure that the new tarball is
legitimate, because it is downloaded from a possibly untrusted source.
Fixes: b30ba14e ("scripts/download.pl: fail loudly if provided hash is unsupported")
Signed-off-by: Baptiste Jonglez <git@bitsofnetworks.org>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Acked-by: Stijn Tintel <stijn@linux-ipv6.be>
Signed-off-by: John Crispin <john@phrozen.org>
This commit adds qconf related files to .gitignore.
The files to be tracked are qconf.cc, qconf.h, and images.c.
The files to be ignored are qconf*.o, qconf*, qconf.moc, and
.tmp_qtcheck.
Signed-off-by: Alif M. Ahmad <alive4ever@live.com>
qconf is kconfig UI utilizing QT toolkit. This makes it possible to use
graphical interface interaction to configure LEDE build target.
This commit adds qconf target to ./script/config/Makefile to be used by
toplevel ``make xconfig`` later.
Signed-off-by: Alif M. Ahmad <alive4ever@live.com>
Fix multiple syntax errors in shelscripts (of packages only)
These errors were causing many conditions to not working properly
Signed-off-by: Lorenzo Santina <lorenzo.santina@edu.unito.it>
[increase PKG_RELEASE, drop command substitution from directip.sh]
Signed-off-by: Mathias Kresin <dev@kresin.em>
Currently, if the provided hash is unsupported (length different from 32
or 64 bytes), we happily download the requested file without any kind of
checksum verification.
This is quite dangerous and may provide a false sense of security, because
a single typo in the hash (e.g. one character deleted by mistake) may skip
checksum verification entirely.
Instead, fail immediately if we don't support the provided hash.
In particular, if an external package repository decides to change the
hash algorithm one day, we will now fail loudly instead of skipping
checksum verification without complaints.
Note: if some users of scripts/download.pl knowingly provide an empty hash
because they don't need checksum verification, this change will break
them. This does not seem to be the case currently, but if this feature is
ever needed, an option should be added to download.pl instead of relying
on the hash being empty.
Fixes: eaa4eba10a ("scripts/download.pl: add SHA-256 support")
Signed-off-by: Baptiste Jonglez <git@bitsofnetworks.org>
If CONFIG_DOWNLOAD_FOLDER is set to for example "~/dl", the download
script fails to create the .hash and .dl files with the following
errors:
Cannot create file ~/dl/dropbear-2017.75.tar.bz2.dl: No such file or directory
sh: 1: cannot create ~/dl/dropbear-2017.75.tar.bz2.hash: Directory nonexistent
If the tarball already exists in the ~/dl dir, it's properly found and
used, so this issue only affects the download.pl script.
This patch calls glob() on the target dir parameter, which will expand `~`.
Signed-off-by: Zoltan Gyarmati <mr.zoltan.gyarmati@gmail.com>
When a package declares a PKG_BUILD_DEPENDENCY or HOST_BUILD_DEPENDENCY on
a not existing build type, the metadata script will emit a reference to an
unresolvable build target in tmp/.packagedeps, causing the make process to
fail hard in a way not catchable by the IGNORE_ERRORS mechanism.
In a situation where a package "test-a" declares a build dependency
"PKG_BUILD_DEPENDS:=test-b/host" while the Makefile of "test-b" does not
implement a HostBuild, make fails with an unrecoverable error in the form:
make[1]: Entering directory '...'
make[1]: *** No rule to make target 'package/test-b/host/compile',
needed by 'package/test-a/compile'. Stop.
make[1]: Leaving directory '...'
.../toplevel.mk:200: recipe for target 'package/test-a/compile' failed
make: *** [package/test-a/compile] Error 2
Extend the metadata generation script to catch such unresolved references
and emit a visable warning upon detection.
After this change, the script will emit a warning similar to:
WARNING: Makefile "package/test-a/Makefile" has a build dependency on
"test-b/host" but "package/test-b/Makefile" does not implement a
"host" build type
Fixes a global build cluster outage which occured after the "python-cffi"
feed package removed its HostBuild which the "python-cryptography" package
build-depended on.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Only mangle argv[0] of the first executed process and leave the argument
vector of subsequent invocations as-is to allow child programs to properly
discover resources relative to their binary locations.
Fixes "cc1" discovery when executing the host gcc through the bundled
"ccache" executable.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Internet2 isn't considered a trusted issuer meaning that https links to
rit.edu will fail.
The host mirror.csclub.uwaterloo.ca has a trusted SSL cert and peering
is good so it can replace rit.edu without performance issues.
Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
[Jo-Philipp Wich: rewrapped commit message]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Commit 72d751cba9 "build: rework library bundling" introduced a new helper
binary "runas" whose sole purpose was mangling the argv vector passed to
the actual called ELF image so that the renamed executable could obtain the
proper name from argv[0].
This approach, however totally defeated the purpose of calling bundled ELF
executables through the shipped ld.so loader since the execv() invocation
performed by "runas" would cause the kernel the interprete the final program
image through the system ELF loader again.
To solve the problem, use an alternative approach of shipping a shared object
"runas.so" which uses an ELF ".init_array" function pointer to obtain the
argv[] vector of the to-be-executed main() function and mangle it in-place.
The actual argv[0] value to use is communicated out-of-band using an
environment variable "RUNAS_ARG0" by the shell wrapper script. The wrapper
script also takes care of setting LD_PRELOAD to instruct the shipped ELF
loader to preload the actual ELF program image with the "runas.so" helper
library.
Fixes FS#909.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Commit 6f5f328003 removed freebsd.sh
because it was outdated and bad practice. Let's be consistent and remove
openbsd.sh as well.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
The script will now detect uid/gid collision and can generate a table of
current allocation
./scripts/package-metadata.pl usergroup tmp/.packageinfo \
| sort -k 1,1r -k 3,3n \
| column -t
This should ensure that no collision will happen for each single build
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
libthread-db is a package that can be configured for external
toolchains, so let's have the script probe for it.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
The script can be used to quickly spin up QEMU virtual machines with lan
and wan network. Please read the initial part of the script for
instructions about how to configure host machine
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Without this change the code checked if the string was contained in the
feature option and not if the string matches the complete word. This only
removes the nand option from the omap24xx target, the other changes are
only removing options which were added twice.
Signed-off-by: Hauke Mehrtens <hauke.mehrtens@intel.com>
The feeds script sets value of TOPDIR in a way that is inconsistent
with how toplevel Makefile sets it. The inconsistency manifests when I
use a "build directory" with symlinks to LEDE source (see below).
When make is invoked in such a directory, make's TOPDIR variable is
set to that directory, whereas scripts/feeds sets TOPDIR to the top of
LEDE source, which results in creating feeds directory inside the LEDE
source instead of in the build directory.
This patch changes the script so that it reuses the TOPDIR value form
the environment if it exists. The result is that 'make
package/symlinks' correctly fetches feeds to the build directory
instead in the source.
I use the following commands to create the build directory:
ln -s $SRC/config config
ln -s $SRC/Config.in Config.in
ln -s $SRC/feeds.conf.default feeds.conf.default
ln -s $SRC/include include
ln -s $SRC/Makefile Makefile
mkdir package
ln -s $SRC/package/base-files package/base-files
ln -s $SRC/package/boot package/boot
ln -s $SRC/package/devel package/devel
ln -s $SRC/package/firmware package/firmware
ln -s $SRC/package/kernel package/kernel
ln -s $SRC/package/libs package/libs
ln -s $SRC/package/Makefile package/Makefile
ln -s $SRC/package/network package/network
ln -s $SRC/package/system package/system
ln -s $SRC/package/utils package/utils
ln -s $SRC/rules.mk rules.mk
ln -s $SRC/scripts scripts
ln -s $SRC/target target
ln -s $SRC/toolchain toolchain
ln -s $SRC/tools tools
This allows me to easily test changes in LEDE on multiple targets.
Signed-off-by: Michal Sojka <sojkam1@fel.cvut.cz>
Implement a new flag "-f" for the feeds update command which causes the
script to fall back to a more agressive git update strategy in case there
are locally modified files in the feeds directory.
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>
Add a fallback case to get_source_date_epoch.sh which reports the modification
time of the script itself in case there is no SCM information available, e.g.
when downloading .tar.gz or .zip tarballs produced by Github.
Also fix the mercurial case while we're at it.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
The short git hash suffix printed by getver.sh is taken from the
latest local commit, change this to use the hash from latest
upstream commit if available. This is considered the intended
behavior based on commit message a642a11fac,
introducing getver.sh.
Signed-off-by: Magnus Kroken <mkroken@gmail.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Conditional dependencies use the '(!cond) || dep' syntax, whereas
conditional select uses 'dep if cond'.
Add an extra check to suppress emitting a conditional if an equal
conditional select already exists.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
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>
This will be used to simplify the build system code for checking hashes.
Instead of using various variants of md5sum / openssl, use one simple
utility for all of them
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Because wget doesn't know how to do Negotiate authentication with a proxy
and curl does, use curl if it's present. The user is expected to have a
~/.curlrc that sets the options necessary for any proxy authentication.
A ~/.curlrc is completely optional however and curl will work in exactly
the same manner as wget without one.
Signed-off-by: Brian J. Murrell <brian@interlinx.bc.ca>
[Jo-Philipp Wich: Rework code to detect curl usability by checking --version,
Use vararg style open() to bypass the shell when downloading,
Use Text::ParseWords to decompose env vars into arguments]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
The external script used to generate the package lists for the
LEDE wiki's table of packages [1] and package indexes [2] requires
a "Source:" field in the package lists to find package makefiles.
The package makefiles are used to read the package's Category and Submenu.
The "Source:" field was removed in commit
b4aa3c899c
to reduce package list sizes and lessen opkg issues in low ram devices.
Add a separate package list file with full data to be used by the wiki's script.
It's called Packages.manifest and isn't compressed as it's not necessary.
1. https://lede-project.org/packages/start
2. https://lede-project.org/packages/index/start
Signed-off-by: Alberto Bursi <alberto.bursi@outlook.it>
This will attempt to automatically fix common mistakes like using MD5
instead of SHA256, using the MD5SUM variable instead of HASH, or even a
missing mirror file hash.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
In the build system, flock will prevent multiple concurrent downloads
for the same file. However, if one download request for the same file is
waiting for another one to finish, it will result in downloading the
same file twice consecutively.
Prevent this issue by exiting immediately if the file has already been
downloaded
Signed-off-by: Felix Fietkau <nbd@nbd.name>
We have switched opkg to sha256 a long time ago, and shrinking package
lists is useful for systems that are running low on RAM
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Change getver.sh to append a short Git commit hash to the end of the artifical
revision number. This way we still have order- and comparable commit numbers
but also a direct relation to the Git commit.
The new output format will look like "r2400+2-882472e" for dirty trees or like
"r2402-882472e" for clean ones.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
The SDK Makefile still trys to copy the docs folder which was removed
with 882f4d2d63. This causes an SDK build
error.
All other removals are just cleanup.
Signed-off-by: Mathias Kresin <dev@kresin.me>
If something goes wrong and script can't find upstream revision it will
return something like:
r2220
which looks like a valid upstream revision 2220. We cant' distinguish it
from e.g. 2200 upstream commits and 20 local ones.
The new format still provides revision number but also points clearly
that is may be not the upstream one:
r0+2220
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Acked-by: John Crispin < john@phrozen.org>
So far we were displaying "local" which could be misinterpreted. It
wasn't possible e.g. to say if src-link feed was initialized or not.
Hopefully "X" makes (a bit) more sense.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
It's always hard to find a reasonable width that will make everyone
happy. This one at least makes "telephony" (one of default feeds) name
fit the column and hopefully isn't too big.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
So far, package-metadata.pl always considered the first provider of a virtual
package to be the default variant which might deviate from what buildroot
considers to be the default.
Change the Kconfig dependency / select code generation for virtual package
providers to consider the DEFAULT_VARIANT to be the primary provider and only
fall back to the first provider if no default variant was explicitely tagged.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Currently the code emitting dependencies for provide candidates is overwriting
the specification calculated by the previous conditional dependency handling
code, rendering dependencies on virtual PROVIDES packages in conjunction with
conditional dependencies unusable.
Instead of overwriting, append the PROVIDES dependency spec in order to fix
using DEPENDS on virtual provider packages in conjunction with conditions.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
This is outdated and bad practice, general dependencies should be documented and leave the rest up to the user.
Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
When selecting devices from the Target Devices menu, the brand choices
of naming makes it confusing to find particular devices by name, as the
sorting is case sensitve. AirTight came after ALFA, and devolo and
jjPlus both came after Zyxel.
This does _not_ apply to the Target Profile list, as that includes
"Default - all profiles" inside the profile list.
Signed-off-by: Karl Palsson <karlp@etactica.com>
TARGET_MULTI_PROFILE and TARGET_PER_DEVICE_ROOTFS get some help text to
try and clarify their behaviour.
Signed-off-by: Karl Palsson <karlp@etactica.com>
If core packages are overridden, CONFIG_OVERRIDE_PKGS is set
based on the scan order of packages, which eventually causes
that config value to be modified on each build and with
that causes the build process to warn for configuration
being out of sync.
This commit changes the CONFIG_OVERRIDE_PKGS to be sorted
and prevents that false warning.
Signed-off-by: Zefir Kurtisi <zefir.kurtisi@neratec.com>
Add a new option to each device in multi-profile mode, allowing to provide
a list of packages to add or remove. In case of added packages, the user
must take care that these are selected to be built.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Dependencies on purely virtual packages (satisfied by PROVIDES) that were
not using "selects" ("+" flag) would be prepended with the prefix
"PACKAGE_" twice, breaking the first alternative.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Move the "which svn" and "which git" calls next to the timestamp commands
using those tools to not prematurely fail on systems where svn or git are
not present.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
9pfs is used by kvm to share files between host and guest,
add proper config option to enable it.
Signed-off-by: Matteo Croce <matteo.croce@canonical.com>
Fix the scripts/remote-gdb script when CONFIG_BUILD_SUFFIX is set.
CONFIG_BUILD_SUFFIX extends the name of the folder build_dir/target*
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Now that the "sysupgrade-nand" step is used by non-NAND targets as well,
rename it to "sysupgrade-tar" to make it more generic.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Older git versions seem output the original argument to stdout if there
is no upstream, presumably because they try to do things with it
internally. This can be prevented by passing --verify to it, which
should be safe on newer git versions.
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
Instead of assuming master is the current branch and origin the right
upstream, try to get both dynamically. If the current branch is not
tracking any upstream, use the origin of the master branch.
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
This is not a valid option in older git version, used in e.g. RHEL6.
Reported-by: Steven Haigh <netwiz@crc.id.au>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
Change the revision output to r<upstream-revision>+<local commits> so
it is easier to get the base revision (and see if there are local
commits).
Example:
$ ./scripts/getver.sh
r794+3
$
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
Instead of ignoring all metadata for package/kernel/linux, process it
and only suppress emitting config data to tmp/.config-package.in
This ensures that packages that select kmod-* packages can inherit their
depdendencies.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This patch adds support of Mikrotik yaffs2 filesystem image for kernel file
and tools/kernel2minor package.
We neede this to boot kernel through RouterBoot on new Mikrotik NOR flash devices.
Signed-off-by: Sergey Sergeev <adron@yapic.net>
When using scripts/feeds upgrade the .config needs to be
updated but the code to do so was also autocreating a
.config if one didn't exist. This is counter-productive
when you have not yet used menuconfig (or other config targets)
because things like selecting or deselecting CONFIG_ALL
(to build all package by default) only works if the
package selection has not already been done via an
existing .config selection.
Signed-off-by: Daniel Dickinson <lede@daniel.thecshore.com>
Provide HTTPS URL when possible, try to keep 8 mirrors per entry and spread
over several locations of the world. Since most active contributors are in
US/CA and/or EU prioritize mirrors that are within those regions if possible.
Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
Apparently symlink-tree has not been used in quite some time as it
fails to symlink the always required config dir
Also, if we pulled from git but .git is missing we get
many error messages on the symlinked tree without this
patch (which symlinks .git, if present)
Signed-off-by: Daniel Dickinson <openwrt@daniel.thecshore.com>
According to gzip 1.7 release note:
The GZIP environment variable is now obsolescent; gzip now warns if
it is used, and rejects attempts to use dangerous options or operands.
You can use an alias or script instead.
Fix this warning by using pipe instead
Signed-off-by: Syrone Wong <wong.syrone@gmail.com>
Introduce a new symbol ALL_NONSHARED which selects all non-sharable packages
by default. This option is mainly intented for buildbot setups to build the
target dependant software subset only.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Add a new string symbol "CONFIG_TARGET_SUBTARGET" which contains the name of
the chosen subtarget or "generic" if there are no subtargets available.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Path to the Git repository directory can be overriden by using the '$GIT_DIR'
environment variable. This patch improves detection of Git repository by using
'git-rev-parse', which respects '$GIT_DIR' environment variable, instead of just
checking the existence of '.git' directory.
Signed-off-by: Vasilis Tsiligiannis <acinonyx@openwrt.gr>
SVN-Revision: 49165
Future Open Mesh u-boot versions are changing the check of the image files
(vmlinux, rootfs) from md5 to sha256. Having both in them should be enough
to ensure backward and forward compatibility.
Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com>
SVN-Revision: 49140
The --mtime argument to 'tar' sets the modification time for all files within
the archive, which determines the timestamp files will get when they are
extracted. In this case, rootfs and other tarballs will get mtimes which
correspond to the last commit timestamp of the build system, as reported by
git/subversion.
This is a step towards reproducible image builds.
Signed-off-by: bryan newbold <bnewbold@robocracy.org>
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 48586
SOURCE_DATE_EPOCH is the date of the last modified file using git/svn
as date source.
See https://reproducible-builds.org/specs/source-date-epoch/
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 48584
date -r $unix_timestamp on bsd is equal to date -d @$unix_timestamp on
linux. To support reproducible builds and not loosing every timestamp
it's required to convert a unix timestamp into human readable timestamp
./scripts/portable_date $unix_timstamp +%T
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
SVN-Revision: 48583
This is useful to just use the kmods from an official build while supplying
base packages from a custom feed or the other way around; for just overriding
the kmods with a local repo while using official repos for the rest.
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 48475
The Apache Software Foundation offers diverse download mirros.
For packaging Apache software a new alias @APACHE is defined.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
SVN-Revision: 48270
Cleanup the @GNOME source download location definitions:
* remove dead and stale mirrors
* adjust to changes at directory structure
* add one new working mirror
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 47825
Use similar approach as r47461 to reduce output for busybox config.
Since r39435 CONFIG_BUSYBOX_CUSTOM=y has caused the addition of
all ~260 busybox config options to the diffconfig output even if
only one option has been changed.
This patch eliminates the unchanged busybox config options from the output
and leaves only the actually modified options.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 47463
These are from today's master branch of:
http://git.savannah.gnu.org/gitweb/?p=config.git;a=tree
In particular it adds support for ARC architecture plus some more
improvements and fixes.
This patch is built-tested against NetGear WNDR3800.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Florian Fainelli <florian@openwrt.org>
Cc: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 47360
This change adds support for specifying a build ID for kernel modules.
This is done by setting PKG_BUILD_ID to a hexadecimal string, which will
then be passed to the kernel linker. In addition, when this flag is set,
the build ID debug symbol (.note.gnu.build-id) will not be stripped from
the kernel module. This symbol is exported in sysfs by the kernel (if
the kernel is compiled with CONFIG_KALLSYMS) and so can be used to
uniquely identify a version of a kernel module in a running kernel. This
is useful for keeping track of different versions of a module when doing
experiments and development.
Modules that specify the build ID will be ~100 bytes larger (depending
on the length of the build ID specified). There is no size difference
for kernel modules that do not set this variable.
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
SVN-Revision: 47290
The u-boot on Open Mesh devices checks the whole transfered image against a
md5sum. This is stored inside the option filemd5sum inside the
fwupgrade.cfg. The bootloader will not check it when this setting is
missing and could therefore write invalid images to the flash.
Signed-off-by: Sven Eckelmann <sven@open-mesh.com>
SVN-Revision: 46925
kernel.org now suggests a different mirror address. this one also
support IPv6 connections and was faster for me.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 46875
several packages reported different file order between builds
make binutils, kmod-sched reproducible
Signed-off-by: Dirk Neukirchen <dirkneukirchen@web.de>
SVN-Revision: 46859
Add a switch to scripts/feeds that allows listing the feeds with their
currently checked out revisions in feeds.conf compatible format.
This allows providing a feeds.conf for public builds to make replication
of the build easier.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46706
This tool avoids generating the redboot script directly from Makefile, which
can get really complicated when adapting image/Makefile to the new IB system.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46488
This patch is a modified version of the patch being discussed at here:
https://patchwork.ozlabs.org/patch/471303/
from Martin Strbacka <martin.strbacka@nic.cz>
My version modifies scripts/feeds so that an error with one of the feeds just
raises the error flag, but script continues and tries to update the other
feeds. After all feeds have been updated, the script returns 1 if at least
one feed failed, and 0 on success with all feeds. The user can then utilise
the status in his build script, if he wants.
signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 46374
As a follow-up to r46026, this commit sets the modification
times of files inside the tarballs to deterministic values.
It uses the date of the last git or svn commit and falls
back to the old behavior if those are not available.
Signed-off-by: Reiner Herrmann <reiner@reiner-h.de>
SVN-Revision: 46360
lookup_target was trampling the $feed variable, resulting in the -p flag
no longer preferentially installing from the named feed.
Make sure to use a local variable for this instead.
Signed-off-by: Karl Palsson <karlp@remake.is>
SVN-Revision: 46120
When creating control.tar.gz, data.tar.gz and $package.ipk
tar compresses them with default gzip options.
This includes build timestamps in their gzip headers.
By setting the GZIP env variable to -n, gzip no longer embeds
them, which is a step towards reproducibility.
See also [1] for more detailed motivations.
[1]: https://wiki.debian.org/ReproducibleBuilds/About
Signed-off-by: Reiner Herrmann <reiner@reiner-h.de>
SVN-Revision: 46026
In r45785 ixp4xx was changed to use a common image prefix also for
kernels, but the script creating them was not updated to generate the
expected filenames.
Fix this by letting arm-magic.sh respect $IMG_PREFIX for the output
kernel file names as well.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 45803
Hello,
if you run:
./scripts/feeds clean
It removes ./feeds folder but not ./package/feeds/ which is full of dangling links then. This patch fixes it.
Best Regards,
Martin Strbačka
Signed-off-by: Martin Strbacka <martin.strbacka@nic.cz>
SVN-Revision: 45738
mkits.sh help currently shows dtb file as optional, but generates an
invalid its file when this option is unspecified.
We're fixing this problem by removing the use of the sed command and
just including variables instead.
Signed-off-by: Mathieu Olivari <mathieu@qca.qualcomm.com>
SVN-Revision: 45659
Split out kmods from ALL to make it easier to create local builds that
are compatible kmod-wise with releases.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 44830
Git internals are referenced by .git which isn't necessarily a
directory. It may also be a file that references the actual .git
directory using the gitdir directive.
If .git is assumed to be a directory the build will not be able to get
the correct version when openwrt is included as a git submodule because
when used as a submodule .git will actually be a file referencing to a
subdirectory in the parent's git dir.
When the correct version is not detected some image generation tools
will fail because the OpenWrt string will be 'OpenWrtunknown' which is
too long for some header formats.
Signed-off-by: Felix Kaechele <heffer@fedoraproject.org>
SVN-Revision: 44452
Remove all rpath entries which do not point to a location below /lib or
/usr/lib and which do not begin with '$ORIGIN'.
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 44377
With this change, override information is now parsed from the metadata
and put in the %packages hash. A new hash - %overrides - is created and
exported, to be used during the .config-package.in generation.
If an override is detected, a new option CONFIG_OVERRIDE_PKGS will be
created in the .config, and will contain a space-separated list of all
the overridden packages.
Signed-off-by: Mathieu Olivari <mathieu@qca.qualcomm.com>
SVN-Revision: 44336
Currently, replacing a package available in openwrt.git requires
modifications in openwrt.git, or requires duplicating the package in a
feed but with a different name, which causes all kind of problems
related to dependencies (all packages selecting it would have to be
modified accordingly to select the new package).
With this change, if a package with the same name is present both in
feeds/ and package/ folders, the one in feeds/ can override the one
in package/, both in the menuconfig and during the build, by passing the
"-f" option to "./scripts/feeds install"
This mechanism is particularly useful for vendor tree, or in general for
application which needs to replace one particular package which exists
within openwrt.git by a custom/newer version.
Signed-off-by: Mathieu Olivari <mathieu@qca.qualcomm.com>
SVN-Revision: 44334
Do not output faulty ".ko" dependency if a scanned .ko file has an empty
"depends=" specification, also replace the `grep | sed | awk` chain with a
single sed program.
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 44206
Currently, replacing a package available in openwrt.git requires
modifications in openwrt.git, or requires duplicating the package in a
feed but with a different name, which causes all kind of problems
related to dependencies (all packages selecting it would have to be
modified accordingly to select the new package).
With this change, if a package with the same name is present both in
feeds/ and package/ folders, the one in feeds/ can override the one
in package/, both in the menuconfig and during the build, by passing the
"-f" option to "./scripts/feeds install"
This mechanism is particularly useful for vendor tree, or in general for
application which needs to replace one particular package which exists
within openwrt.git by a custom/newer version.
Signed-off-by: Mathieu Olivari <mathieu@qca.qualcomm.com>
SVN-Revision: 44076
When the initial glob() attempt on a path specified with "source" statement
yields no result, then retry the globbing on the same path with the directory
part of the currently processed file prepended.
This allows us to reference Config.in files relative to their parent instead
of relative to the top dir.
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 43652
Plain "remote" results in failure to connect using the gdb built with
the toolchain. (On atheros target at least) extended-remote also allows
"run" to restart the target process.
Signed-off-by: Karl Palsson <karlp@remake.is>
SVN-Revision: 43596
Many packages define already metadata about their license (PKG_LICENSE),
but this is only included in the ipk files.
This change allows to create the information also on the build-host,
to get an overview on the used licenses.
In the full list, also all packages without this info are shown
Signed-off-by: Thomas Langer <thomas.langer@lantiq.com>
SVN-Revision: 43070
[base-files] shell-scripting: fix wrong usage of '==' operator
normally the '==' is used for invoking a regex parser and is a bashism.
all of the fixes just want to compare a string. the used busybox-ash
will silently "ignore" this mistake, but make it portable/clean at least.
this patch does not change the behavior/logic of the scripts.
Signed-off-by: Bastian Bittorf <bittorf@bluebottle.com>
SVN-Revision: 42911
pts/feeds update -a" can fail rather silently for feeds using git, as
the script does not pause when updating a feed fails. Instead it prints the
error message and calmly continues to the next feed. It is very easy to
overlook update errors with the feeds updated first, as their text scrolls
rapidly away from the screen.
This behaviour has not been a big problem with svn feeds, as svn update stops
with a conflict message and interactively forces the user to resolve or
postpone the conflict. In any case the svn error is noticed by the user.
Majority of the feeds use now git, so this silent failure can affect users
doing private builds in an increasing amount.
Below is an example of update failing and script continuing:
perus@v1404:/Openwrt/barrier$ ./scripts/feeds update -a
Updating feed 'packages' from
'https://github.com/openwrt/packages.git;for-14.07' ...
remote: Counting objects: 17, done.
remote: Compressing objects: 100% (15/15), done.
remote: Total 17 (delta 10), reused 8 (delta 1)
Unpacking objects: 100% (17/17), done.
From https://github.com/openwrt/packages
62031da..dc26009 for-14.07 -> origin/for-14.07
Updating 62031da..dc26009
error: Your local changes to the following files would be overwritten by merge:
utils/collectd/Makefile
Please, commit your changes or stash them before you can merge.
Aborting
failed.
Updating feed 'luci' from 'http://git.openwrt.org/project/luci.git;luci-0.12' ...
Already up-to-date.
Create index file './feeds/luci.index'
Updating feed 'routing' from
'https://github.com/openwrt-routing/packages.git;for-14.07' ...
...
The script prints "failed.", but does not break the updating process. The
"update_feed" function returns an error code 1, but that value is not checked
in the "update" function, which continues to the next feed.
Return 1 as error:
https://dev.openwrt.org/browser/trunk/scripts/feeds#L547
Call to update_feed without any error monitoring:
https://dev.openwrt.org/browser/trunk/scripts/feeds#L585
The included patch makes the feeds script to stop updating after failing to
update a feed.
The script continues to the refresh_config step despite a possible failure in
updating, so the stopping action just prevents the other feeds from updating
and makes the error more clearly visible.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 42891
I defined a new download method @SAVANNAH in include/download.mk and scripts/download.pl,
and converted quilt and qemu to use that method.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 42840
