tools/quilt requires GNU diffutils to compile. Failure can be simulated
by installing Alpine Linux without diffutils.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Speed goes from:
Executed in 178.08 secs fish external
usr time 20.16 mins 509.00 micros 20.16 mins
sys time 2.88 mins 39.00 micros 2.88 mins
To:
Executed in 175.90 secs fish external
usr time 20.19 mins 0.00 micros 20.19 mins
sys time 2.85 mins 497.00 micros 2.85 mins
Tested with "time make -j 12" on AMD Ryzen 3600
When building individual packages, the build time difference is often
significantly bigger than that.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Rosen Penev <rosenp@gmail.com>
ninja is faster at building cmake packages than make, and according to reports
also more reliable at handling parallel builds
This commit includes a patch that adds GNU make jobserver support, in order to
allow more precise control over the number of parallel tasks
Enable parallel build by default for packages using ninja
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Don't attempt to copy initramfs images for devices which do not output
an initramfs image.
This was breaking builds for mpc85xx-p1010 since mid-march.
Signed-off-by: David Bauer <mail@david-bauer.net>
The variable was missing in the definition of DEFAULT_DEVICE_VARS which
caused it to contain wrong values, messing up the resulting JSON files.
This patch adds the variable DEVICE_PACKAGES to DEFAULT_DEVICE_VARS.
Suggested-by: Baptiste Jonglez <git@bitsofnetworks.org>
Signed-off-by: Paul Spooren <mail@aparcar.org>
This fixes a regression introduced with commit
5ed1e5140a ("build: build kernel image
before building modules/packages").
Before this commit the make target would always include "modules",
resulting in a MODPOST and a complete Module.symvers file. Since this
commit a MODPOST of the kernel modules is not guaranteed for kernels <
5.10. This results in some broken SDKs in which external packages that
depend on exported symbols from kernel modules fail to compile.
Adding "modules" back to the calls to the CompileImage defines fixes the
regression. For kernels > 5.10 this is not needed, but it doesn't cause
any harm either.
Tested with kernels 5.4.x and 5.10.x.
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
Remove \n that mangles output, and fix inconsistent version name check.
Example before:
Build dependency: Please install the GNU C++ Compiler (g++) 6 or later
Build dependency: \nPlease reinstall the GNU C++ Compiler (4.8 or later) - it appears to be broken
Build dependency: Please install ncurses. (Missing libncurses.so or ncurses.h)
Signed-off-by: Karl Palsson <karlp@tweak.net.au>
Use an 'if' so the absence of $(LINUX_DIR)/user_headers doesn't make the
line evaluate to false and cause the build to fail.
Signed-off-by: Matt Merhar <mattmerhar@protonmail.com>
Removed upstreamed:
generic/pending-5.4/770-02-net-ethernet-mtk_eth_soc-fix-rx-vlan-offload.patch
All other patches automatically rebased.
Build system: x86_64
Build-tested: ipq806x/R7800
Note that since I rebased the previous commit, I removed my Run-tested line
although I confirm building the image successfully.
Signed-off-by: John Audia <graysky@archlinux.us>
Removed upstreamed:
generic/backport-5.4/050-gro-fix-napi_gro_frags-Fast-GRO-breakage-due-to-IP-a.patch
bcm63xx/patches-5.4/434-nand-brcmnand-fix-OOB-R-W-with-Hamming-ECC.patch*
Removed/code was included upstream and therefore redundant:
ramips/patches-5.4/999-fix-pci-init-mt7620.patch
All other patches automatically rebased.
* update_kernel.sh did not flag this yet it was included in 5.4.119[1], as a
result of the rebase, I removed my testing lines since I did not go back to
test built or to run test 5.4.119 with the removed patch present.
1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v5.4.119&id=e5b3e69eb36ac1178a7a2392616fd29afd288c4e
Signed-off-by: John Audia <graysky@archlinux.us>
Before this commit, it was assumed that mkhash is in the PATH. While
this was fine for the normal build workflow, this led to some issues if
make TOPDIR="$(pwd)" -C "$pkgdir" compile
was called manually. In most of the cases, I just saw warnings like this:
make: Entering directory '/home/.../package/gluon-status-page'
bash: line 1: mkhash: command not found
bash: line 1: mkhash: command not found
bash: line 1: mkhash: command not found
bash: line 1: mkhash: command not found
bash: line 1: mkhash: command not found
bash: line 1: mkhash: command not found
bash: line 1: mkhash: command not found
bash: line 1: mkhash: command not found
[...]
While these were only warnings and the package still compiled sucessfully,
I also observed that some package even fail to build because of this.
After applying this commit, the variable $(MKHASH) is introduced. This
variable points to $(STAGING_DIR_HOST)/bin/mkhash, which is always the
correct path.
Signed-off-by: Leonardo Mörlein <me@irrelefant.net>
When building for MikroTik devices the kernel2minor tool will sometimes
fail with:
Can't get lstat from kernel file!: No such file or directory.
This is because kernel2minor expects paths no longer than 250 chars.
To work around this the include/image-commands.mk has been modified
to copy the kernel to a temporary file (/tmp/tmp.XXXXXXXXXX) before
calling kernel2minor.
Signed-off-by: François Chavant <francois@chavant.info>
OpenWRT requires a number of Perl modules to be installed. It wasn't checking on all of them.
This patch adds checks for Perl FindBin, File::Copy, File::Compare and Thread::Queue modules.
Failing to install these, will have the build break at some point. By adding these to the
prereq-build.mk script, they are checked on forehand.
Tested on a Fedora 33 and 34 (beta) that was freshly installed. Fedora appears to
break up Perl modules into small packages that need to be installed for the build to succeed.
Signed-off-by: Bas Mevissen <abuse@basmevissen.nl>
These have long been obsolete. For reference, here's the Linux version where
each symbol has been dropped:
CONFIG_IP6_NF_QUEUE - 3.5
CONFIG_IP6_NF_TARGET_LOG - 3.4
CONFIG_IP_NF_MATCH_DSCP - 2.6.19
CONFIG_NF_CONNTRACK_IPV4 - 4.19
CONFIG_NF_CONNTRACK_IPV6 - 4.19
CONFIG_NF_CONNTRACK_RTCACHE - out-of-tree, superseded by flow offloading
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
Manually rebased*
generic/backport-5.4/700-v5.5-net-core-allow-fast-GRO-for-skbs-with-Ethernet-heade.patch
Added new backport*
generic/backport-5.4/050-gro-fix-napi_gro_frags-Fast-GRO-breakage-due-to-IP-a.patch
All others updated automatically.
The new backport was included based on this[1] upstream commit that will be
mainlined soon. This change is needed because Eric Dumazet's check for
NET_IP_ALIGN (landed in 5.4.114) causes huge slowdowns on drivers which use
napi_gro_frags().
Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800
No dmesg regressions, everything functional
*Credit to Alexander Lobakin
1. https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=7ad18ff6449cbd6beb26b53128ddf56d2685aa93
Signed-off-by: John Audia <graysky@archlinux.us>
The 'append-image-stage' command doesn't work when setting the
EXTRA_IMAGE_NAME option of the ImageBuilder as in that case
DEVICE_IMG_PREFIX is modified and no longer matches the value it had in
buildroot. Choose a filename independent of DEVICE_IMG_PREFIX for
images staged using 'append-image-stage' to fix that.
Fixes: de4b29dab9 ("image: introduce 'append-image-stage' build command")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Similar to 'append-image' this new command appends an existing binary.
'append-image-stage' also makes a copy of that binary and keeps it in
$(STAGING_DIR_IMAGE). When called from within the ImageBuilder, this
copy is used instead of expecting the binary to be present.
This is useful for artifacts which include the initramfs/recovery image
which is usually not included in the ImageBuilder.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Ran update_kernel.sh in a fresh clone without any existing toolchains. No
manual intervention needed.
Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800
No dmesg regressions, everything functional
Signed-off-by: John Audia <graysky@archlinux.us>
Ran update_kernel.sh in a fresh clone without any existing toolchains.
Removed upstreamed:
mvebu/patches-5.4/319-ARM-dts-turris-omnia-configure-LED-2--INTn-pin-as-interrupt-pin.patch
Build system : x86_64
Build-tested : ipq806x/R7800
Run-tested : ipq806x/R7800
No dmesg regressions, everything functional
Signed-off-by: John Audia <graysky@archlinux.us>
Ran update_kernel.sh in a fresh clone without any existing toolchains.
Manually rebased due to movement of rx-offload.c in 5.4.110:
layerscape/patches-5.4/802-can-0002-can-rx-offload-fix-long-lines.patch
layerscape/patches-5.4/802-can-0003-can-rx-offload-can_rx_offload_compare-fix-typo.patch
layerscape/patches-5.4/802-can-0004-can-rx-offload-can_rx_offload_irq_offload_timestamp-.patch
layerscape/patches-5.4/802-can-0005-can-rx-offload-can_rx_offload_reset-remove-no-op-fun.patch
layerscape/patches-5.4/802-can-0006-can-rx-offload-Prepare-for-CAN-FD-support.patch
layerscape/patches-5.4/802-can-0018-can-flexcan-use-struct-canfd_frame-for-CAN-classic-f.patch
Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800
No dmesg regressions, everything functional
Signed-off-by: John Audia <graysky@archlinux.us>
Use update_kernel to refresh all patches, required manual updates to:
610-netfilter_match_bypass_default_checks.patch
611-netfilter_match_bypass_default_table.patch
762-net-bridge-switchdev-Refactor-br_switchdev_fdb_notif.patch
764-net-bridge-switchdev-Send-FDB-notifications-for-host.patch
Run-tested: x86_64
Nothing screamed out but any funny business with linux bridging should
suspect this update first.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Ran update_kernel.sh in a fresh clone without any existing toolchains.
Manually rebased:
pending-5.4/611-netfilter_match_bypass_default_table.patch
The upstream change affecting this patch is the revert of an earlier
kernel commit. Therefore, we just revert our corresponding changes
in [1].
Build system: x86_64
Build-tested: ipq806x/R7800
[1] 9b1b89229f ("kernel: bump 5.4 to 5.4.86")
Signed-off-by: John Audia <graysky@archlinux.us>
[adjust manually rebased patch, add explanation]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The recent removal of usbutils from core and replacement by hwdata in
packages has exposed hwdata's requirement for certain GNU options on
'install' (-T) Other packages (sqm-scripts) have openwrt specific
makefile sections to avoid GNU options but I suspect this is going to
get harder in the future.
Add GNU install as a prerequisite and link into
$STAGING_DIR/host/etc/bin as per similar GNU utils
This resolves an issue building under MacOS which would otherwise use a
non-GNU options aware version of 'install'
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
In case CONFIG_TARGET_MULTI_PROFILE is set, IMG_PREFIX cannot be
expanded. Use DEVICE_IMG_PREFIX instead and make sure it's defined.
Fixes: 8f89b1ab0f ("image: add 'append-image' build command")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Commit 7ce1d9ce09 ("build: artifacts add dependency for built images")
now makes sure that sysupgrade and initramfs images are available at
the stage that artifacts are created.
Allow making use of that with a new build command 'append-image' to
be used in artifacts.
See the next commit for an example.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Add possibility to use images and initramfs in artifacts.
Signed-off-by: Oskari Lemmela <oskari@lemmela.net>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Straightforward refresh of patches using update_kernel.
Removed (reverse-applicable):
bmips/patches-5.10/010-v5.11-net-dsa-implement-a-central-TX-reallocation-procedur.patch
Run tested: x86_64 (apu2)
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Straightforward refresh of patches using update_kernel.
Run tested: x86_64 (apu2)
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Ran update_kernel.sh in a fresh clone without any existing toolchains.
Manually rebased:
bcm27xx/950-0993-xhci-quirks-add-link-TRB-quirk-for-VL805.patch
layerscape/701-net-0231-enetc-Use-DT-protocol-information-to-set-up-the-port.patch
Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800
No dmesg regressions, everything functional
Signed-off-by: John Audia <graysky@archlinux.us>
[remove accidental whitespace edit]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
FS#3574
Adding cgroup support enables adding rules on processes
to limit resources in terms of iptable policies
Signed-off-by: Supriya Mane <sm.supriya@globaledgesoft.com>
Add new target feature 'dt-overlay' which makes DTC keep the symbol
names in the generated dtb.
Make sure additional DT overlay sources specified by the new device
variable DEVICE_DTS_OVERLAY get compiled together with the main DTS
(currently overlays got to be in the same folder). Let Build/fit pass
the generated DT overlay blobs to mkits.sh.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
With the existence of ABI versions there is no clean way to determine
the package name without an attached ABI version. The Packages index is
stored on device to know what packages are installed.
The ABIVersion was recently removed in c921650382 "build: drop ABI
version from metadata", while ABI versions still exists. This becomes a
problem if a user tries to export installed packages via `ubus call
rpcd-sys packagelist` which would return package names including the ABI
version. Trying to find these packages in a later release with changes
ABI version is impossible.
This commits adds the `ABIVersion` field again. Knowing both the
combined (SourceName + ABIVersion) and the `ABIVersion` it is possible
to calculate the package `SourceName` without storing it in the
on-device package list.
Signed-off-by: Paul Spooren <mail@aparcar.org>
Also add a new kconfig symbol (CONFIG_KCMP) to the generic config,
disabling the SYS_kcmp syscall (it was split from
CONFIG_CHECKPOINT_RESTORE, which is disabled by default, so the
previous behaviour is kept).
Removed (upstreamed) patches:
070-net-icmp-pass-zeroed-opts-from-icmp-v6-_ndo_send-bef.patch
081-wireguard-device-do-not-generate-ICMP-for-non-IP-pac.patch
082-wireguard-queueing-get-rid-of-per-peer-ring-buffers.patch
083-wireguard-kconfig-use-arm-chacha-even-with-no-neon.patch
830-v5.12-0002-usb-serial-option-update-interface-mapping-for-ZTE-P685M.patch
Manually rebased patches:
313-helios4-dts-status-led-alias.patch
104-powerpc-mpc85xx-change-P2020RDB-dts-file-for-OpenWRT.patch
Run tested:
ath79 (TL-WDR3600)
mvebu (Turris Omnia)
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
U-boot will reject all nodes with @ since commit:
79af75f777
This will cause the OpenWrt images to fail booting,
to rectify use the config-1 as default.
Signed-off-by: Robert Marko <robert.marko@sartura.hr>
5.4.102 backported a lot of stuff that our WireGuard backport already
did, in addition to other patches we had, so those patches were
removed from that part of the series. In the process other patches were
refreshed or reworked to account for upstream changes.
This commit involved `update_kernel.sh -v -u 5.4`.
Cc: John Audia <graysky@archlinux.us>
Cc: David Bauer <mail@david-bauer.net>
Cc: Petr Štetiar <ynezz@true.cz>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
`which` utility is not shipped by default for example on recent Arch
Linux and then any steps relying on its presence fails, like for example
following Python3 prereq build check:
$ python3 --version
Python 3.9.1
$ make
/bin/sh: line 1: which: command not found
...
Checking 'python3'... failed.
So make `which` utility host build requirement.
References: PR#3820 FS#3525
Signed-off-by: Petr Štetiar <ynezz@true.cz>
This reverts commit c7aec47e5e.
The original commit replaces 'which' with 'command'. Sadly most of
them are not equivalent and for 'which -a', there is no easy
replacements that would not reimplement PATH parsing logic. Hence
revert. Keeping a dependency on which is absolutely fine.
Signed-off-by: Clemens Fruhwirth <clemens@endorphin.org>
Currently minimal GNU supported GCC version is 7 (from May 2, 2017),
buildbots are using default GCC version 6 on Debian 9 (old stable),
current Debian stable has GCC version 8.3.0.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
The previous approach of referencing artifacts in follow-up artifacts
can't work with parallel builds in the current way image.mk is built.
Refactor things so this is not needed.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Write everything needed for eMMC install into the gaps between
partitions on SD card. In that way, installation to eMMC only needs
the SD card, no additional files need to be loaded via TFTP any more.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
autotools.mk does not have any protection currently that would prevent
it from being sourced multiple times. Note that both package.mk and
host-build.mk source autotools.mk. So any package Makefile that includes
both will cause hooks to be added twice (at least twice).
This is fixed by declaring a new variable, __autotools_inc, and only
continuing if this variable doesn't equal 1. The same is done by
rules.mk already.
Also, this commit does away with an ifneq that checks PKG_FIXUP (instead
of HOST_FIXUP) for patch-libtool before adding to the host pre-configure
hook. This does not make sense.
The second ifneq is amended. The current one manually does what the
define patch_libtool_host is already doing. It can just use the define.
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
Ran update_kernel.sh in a fresh clone without any existing toolchains.
Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800
No dmesg regressions, everything functional.
Signed-off-by: John Audia <graysky@archlinux.us>
We so far had two variables IMG_PREFIX and IMAGE_PREFIX with
different content. Since these names are obviously quite
confusing, this patch renames the latter to DEVICE_IMG_PREFIX,
as it's a device-dependent variable, while IMG_PREFIX is only
(sub)target-dependent.
For consistency, also rename IMAGE_NAME to DEVICE_IMG_NAME, as
that's a device-dependent variable as well.
Cc: Paul Spooren <mail@aparcar.org>
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
That was a left-over from testing and should not have made it into the
tree. Remove it.
Fixes: 330bd380e8 ("image: allow building FIT and uImage with ramdisk")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Ran update_kernel.sh in a fresh clone without any existing toolchains.
Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800
No dmesg regressions, everything functional
Signed-off-by: John Audia <graysky@archlinux.us>
[refresh again]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Previously, build would fail for targets containing devices with not
initramfs image (such as mpc85xx-p1010). Only generate the JSON image
info for the initramfs image when we have one to avoid breaking the
builds.
Fixes commit d3140d0529 ("build/json: generate json file for initramfs")
Signed-off-by: David Bauer <mail@david-bauer.net>
Instead of embedding the initrd cpio archive into the kernel, allow
for having an external ramdisk added to the FIT or uImage.
This is useful to overcome kernel size limitations present in many
stock bootloaders, as the ramdisk is then loaded seperately and doesn't
add to the kernel size. Hence we can have larger ramdisks to host ie.
installers with all binaries to flash included (or a web-based
firmware selector).
In terms of performance and total size the differences are neglectible.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Allow for single (external-data) FIT image to hold kernel, dtb and
squashfs. In that way, the bootloader verifies the system integrity
including the rootfs, because what's the point of checking that the
hash of the kernel is correct if it won't boot in case of squashfs
being corrupted? Better allow bootloader to check everything needed
to make it at least up to failsafe mode. As a positive side effect
this change also makes the sysupgrade process on nand potentially
much easier as it is now.
In short: mkimage has a parameter '-E' which allows generating FIT
images with 'external' data rather than embedding the data into the
device-tree blob itself. In this way, the FIT structure itself remains
small and can be parsed easily (rather than having to page around
megabytes of image content). This patch makes use of that and adds
support for adding sub-images of type 'filesystem' which are used to
store the squashfs. Now U-Boot can verify the whole OS and the new
partition parsers added in the Linux kernel can detect the filesystem
sub-images, create partitions for them, and select the active rootfs
volume based on the configuration in FIT (passing configuration via
device tree could be implemented easily at a later stage).
This new FIT partition parser works for NOR flash (on top of mtdblock),
NAND flash (on top of ubiblock) as well as classic block devices
(ie. eMMC, SDcard, SATA, NVME, ...).
It could even be used to mount such FIT images via `losetup -P` on a
user PC if this patch gets included in Linux upstream one day ;)
Signed-off-by: John Crispin <john@phrozen.org>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
The update to gettext 0.21 broke packages that use autotools and
gettext because the sed line was failing with the new version. Fix with
a better sed expression.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
The initramfs images are missing from the profiles.json files.
Signed-off-by: Moritz Warning <moritzwarning@web.de>
[fix code by exporting device variables]
Signed-off-by: Paul Spooren <mail@aparcar.org>
package/openwrt-packages is left over
from what is now a legacy repository
and has no other reference in the build tree
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Use approach suggested by Adrian Schmutzler instead of introducing
another device variable.
Also revert the unnecessary white-space changes accidentally introduced
by the previous commit.
Fixed: c067b1e79b ("mediatek: move out-of-tree DTS files to dedicated dts folder")
Suggested-by: Adrian Schmutzler <mail@adrianschmutzler.de>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Use dedicated dts folder like on ramips to store device tree source
files for boards not already supported in vanilla Linux.
Doing so instead of having them in files-* has several advantages:
* we don't need to duplicate them for several kernel versions
* changes to a device tree don't trigger a complete kernel rebuild
* the files are more obvious to find
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Ran update_kernel.sh in a fresh clone without any existing toolchains.
No manual changes needed.
Build system: x86_64
Build-tested: bcm27xx/bcm2711
Signed-off-by: John Audia <graysky@archlinux.us>
GetABISuffix does not work for intra-package ABI version of provided symbols,
since ABIV_$(provided) is not set.
Fix ABI version by using $(ABIV_$(1)) directly
Signed-off-by: Felix Fietkau <nbd@nbd.name>
