A patch was added to the Linux kernel in 5.1.0-rc3 which adds a
requirement that the host build environment include pkg-config. Add the
correct host-pkgconf dependency and environment variables to ensure
Linux picks up the correct libraries.
Move the existing LINUX_MAKE_ENV assignment earlier, to simplify the
append-assignment in the libelf conditional block.
Fixes: #11761
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
Suggested-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Stuart Summers <stuart.summers@intel.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[Arnout: extend commit message as suggested by Yann]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
(cherry picked from commit 615e2edc2a)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
With the arrival of linux v5.0, we need yet another condition to set
_SITE correctly. Instead of continuing this madness, solve the problem
generically: use v2.6 for 2.6.*, and use the number before the first dot
in the other cases.
While we're at it, remove the comment which has been incorrect since
80d7b68167 (7 years ago).
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Jan Kundrát <jan.kundrat@cesnet.cz>
Tested-by: Adam Duskett <aduskett@gmail.com>
Reviewed-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 4ed7246a59)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
We have virtually no way to know the hashes for user-supplied patches,
so we should just ignore them.
Reported-by: Simon van der Veldt <simon.vanderveldt@gmail.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <peter@korsgaard.com>
Tested-by: Simon van der Veldt <simon.vanderveldt@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
This patch bumps the Linux CIP version to v4.4.171-cip30 and updates the
download url to the new official one.
Signed-off-by: Angelo Compagnucci <angelo@amarulasolutions.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Add an option to compile device trees in Linux with symbol generation
such that device tree overlays can be loaded on the target system
Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
[Arnout: remove "default n" and move setting of LINUX_MAKE_ENV to the
place where the rest is set.]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
linux is a bit different than other kconfig-package, because it has
"toolchain" in KCONFIG_DEPENDENCIES. Thanks to this, host-ccache *is*
ready by the time kconfig invocations are made, so we could use
$(HOSTCC) as the host compiler for kconfig related operations.
However, for consistency with other kconfig-package packages, we chose
to use $(HOSTCC_NOCCACHE) as well.
We cannot rely on the default value of HOSTCC passed by the
kconfig-package infrastructure, because $(LINUX_MAKE_FLAGS) also
contains a HOSTCC definition that would override the one passed by the
kconfig-package infrastructure.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Processor Counter Monitor (PCM) is an application programming interface
(API) and a set of tools based on the API to monitor performance and
energy metrics of Intel(R) Core(TM), Xeon(R), Atom(TM) and Xeon Phi(TM)
processors.
This package contains a patch on the pmu-query.py script to look for the
pcm-core program at the default path. It's not nice to have a Buildroot
specific patch but let's use one while we look for a solution that is
acceptable upstream.
Signed-off-by: Carlos Santos <casantos@datacom.ind.br>
[Peter: Needs C++, force X86_MSR on in linux]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When using an initramfs, on the first-pass build, we create a dummy cpio
so that the build succeeeds. The real cpio will come later, and we'll do
a second-pass build to use the actual cpio.
However, when we touch that dummy cpio, the images/ directory may not
yet exist, since commit d0f4f95e39 (Makefile: rework main directory
creation logic) removed its creation at the begining of the build, to
only at the moment we need it, i.e. during the *_INSTALL_IMAGES_CMDS
steps.
However, the linux build is not a _INSTALL_IMAGES_CMDS step, so there is
no guarantee that images/ already exist at that time.
Fix that by explicitly creating images/ before touching the dummy cpio.
Reported-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Matthew Weber <matthew.weber@rockwellcollins.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Reviewed-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Checking for the existence of the dtc binary built by the
non-dependent dtc package may cause instable behaviour when giving more
freedom on the order of how the packages are built (parallelization).
In addidion, when moving to per-package host/target method, the check
would always trigger in the isolated host, leading to linux-dtc always
being installed as dtc.
This in turn may lead to undesired overwriting of the real host-dtc binary
when finally assembling the global host dir.
Thus rework the linux-dtc install condition to be defined by configuration
rather than compile time order.
Signed-off-by: Andreas Naumann <anaumann@ultratronik.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
We already turn on kernel features for several packages, so let's do it
for intel-microcode too, otherwise it's impossible to load the microcode
(by means of iucode-tools).
Signed-off-by: Carlos Santos <casantos@datacom.com.br>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
We already turn on kernel features for several packages, so let's do it
for audit too, since the daemon is useless and fails to load otherwise.
Notice that we also turn NET on, since AUDIT depends on NET, like we do
for the wireguard package.
Signed-off-by: Carlos Santos <casantos@datacom.com.br>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
This patch adds a simple way to change the linux bootup logo.
The patch was kept purposely simple to support only the use cause
where a user needs a color linux boot up logo.
Signed-off-by: Angelo Compagnucci <angelo@amarulasolutions.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[Peter: clarify/reword option/help text, automatically enable framebuffer
support, fix convert path]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Starting with linux-4.18, the kconfig from the kernel can call
to the compiler to test its capabilities; see:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/scripts/Kconfig.include
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Jan Kundrát <jan.kundrat@cesnet.cz>
Tested-by: Jan Kundrát <jan.kundrat@cesnet.cz>
Reviewed-by: Jan Kundrát <jan.kundrat@cesnet.cz>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
host-{flex,bison} are only needed to generate the dtc parser, so we
don't need them if the kernel does not have support for device tree.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Rely on the system provided ones if avalable, and only resort to use our
owns if the sytem does not provide them.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Jan Kundrát <jan.kundrat@cesnet.cz>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
The commit that added the dependency on host-{bison,flex} did
so because the pre-generated kconfig parser source files were
removed from the kernel tree, in linux-4.16.
But then, in linux-4.17, the pre-generated dtc parser source
files were in turn removed as well.
So, document the two reasons why they are needed, so we don't
accidentally remove them when we (soon) introduce the kconfig
dependencies.
(Also fix the first assignment to LINUX_DEPENDENCIES to be a
simple assignement, not an append-assignment.)
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
This commit replaces the loop copying out-of-tree DTS into the kernel
tree by a make foreach loop instead of a shell for loop. This allows
to error out if one of the DTS file cannot be copied (for example if
it doesn't exist).
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
gcc-8 started warning about function aliases that have a non-matching
prototype. This seems rather useful in general, but it causes tons of
warnings in the Linux kernel, where we rely on abusing those aliases
for system call entry points, in order to sanitze the arguments passed
from user space in registers.
See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82435
Disable the attribute-alias warning introduced by gcc-8 by adding
-Wno-attribute-alias to KCFLAGS.
Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Perf profiling cannot be used if CONFIG_PERF_EVENTS is not enabled in the
kernel configuration. Similar to other tools, like ktap, we can enable the
right options automatically.
Signed-off-by: Jan Heylen <jan.heylen@nokia.com>
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Kconfig uses either pkg-config or hard-coded /usr/include paths to find
the ncurses or ncursesw library. If ncursesw is found, it will include
<ncursesw.h>. Since Buildroot's host-ncurses doesn't install a .pc file,
and linux.mk anyway doesn't pass the pkg-config options to find the host
pkg-config files, Kconfig will always find the system's ncursesw.h.
However, since commit dde090c299 (linux: fix passing of host CFLAGS and
LDFLAGS) HOST_LDFLAGS is passed to the linux build system. Thus, if
host-ncurses was already built before 'make linux-menuconfig' is called,
the build will pick up libncurses from the host directory, which is NOT
widechar. Thus, two different ncurses configurations are mixed into the
final mconf program. This will result in serious breakage in the
rendering of the menus (lots of @ and question mark characters).
As a workaround (suggested by Yann), don't pass HOST_CFLAGS and
HOST_LDFLAGS when running kconfig commands. For kconfig, we should never
need host packages anyway. This way, the kconfig calls will always use
the system's ncurses and never our host-ncurses.
Note that the same problem could pop up for other kconfig packages as
well if we ever pass HOST_CFLAGS/HOST_LDFLAGS to them. We could force
HOSTCC=$(HOSTCC) directly in kconfig-package. However, for now there
are no other packages that exhibit this problem, so this can be
revisited when they do.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: David De Grave <david.degrave@essensium.com>
Cc: Scott Fan <fancp2007@gmail.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
Tested-by: Matt Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Fixes https://gitlab.com/buildroot.org/buildroot/-/jobs/66561794
LINUX_DTS_NAME may end up with a leading space because of the += logic, and
may contain multiple dts files - Neither of which works when we construct
the {cu,simple}Image.$(LINUX_DTS_NAME) make target name.
Fix it by using the first word in the variable.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
These three symbols:
KERNEL_ARCH_PATH
KERNEL_DTBS
KERNEL_DTS_NAME
are defined and used only inside this file, so use the LINUX_ namespace
for them instead of KERNEL_.
Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Use only one space before backslash.
Remove consecutive empty line.
Indent with tabs.
Signed-off-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Custom dts files are still conditionally copied based on non existing
boolean. So it is currently not possible to use custom dts file(s) at all.
List of dts files is now iterated and files are copied into dedicated kernel arch dir.
Signed-off-by: Rafal Susz <rafal.susz@gmail.com>
Tested-by: Jan Kundrát <jan.kundrat@cesnet.cz>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Just use whatever the user specified in the list. An empty list means no
DTS was specified.
No need to add legacy option, as the behaviour does not change.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Simon van der Veldt <simon.vanderveldt@gmail.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Tested-by: Joshua Henderson <joshua.henderson@microchip.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Some Linux kernel configuration options (such as CONFIG_UNWINDER_ORC)
require building a host program that needs libelf.
Users who have libelf installed on their system won't see a problem,
but users who don't have libelf installed will get a build
failure. Therefore, this commit adds an option that allows a user to
indicate that his Linux kernel configuration requires libelf. When
this option is enabled, we add host-elfutils to the dependencies of
the linux package (host-elfutils provides the libelf library).
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Some Linux kernel configuration options (such as
CONFIG_SYSTEM_TRUSTED_KEYRING) require building a host program called
extract-cert, which itself needs OpenSSL.
Users having OpenSSL installed on their system won't see a problem,
but users who don't have OpenSSL installed will get a build
failure. This commit adds a new option that allows users to indicate
that their Linux configuration requires building host-openssl.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
We were passing HOSTCFLAGS="$(HOSTCFLAGS)" to Linux. However:
- HOSTCFLAGS in Buildroot doesn't exist, and is empty, so this
assignment never did anything. The name of the variable in
Buildroot in HOST_CFLAGS.
- HOSTCFLAGS in Linux isn't used everywhere, and passing it overrides
the default HOSTCFLAGS value defined in the main Linux kernel
Makefile.
In addition, there is no way to pass additional host LDFLAGS in the
Linux kernel build system.
Therefore, we simply shoehorn our HOST_CFLAGS and HOST_LDFLAGS while
passing HOSTCC to the Linux kernel build system. This has been tested
to work fine with host OpenSSL and host libelf only available in
$(HOST_DIR).
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Tested-by: Frank Hunleth <fhunleth@troodon-software.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
For some boards, for example the Raspberry Pi, it's necessary to build
in-tree dts files as well as custom/out of tree dts-files (dt-blob.bin).
The existing logic made these two options exclusive, this commit changes
that to allow both in-tree as well as custom sources for dts files.
Signed-off-by: Simon van der Veldt <simon.vanderveldt@gmail.com>
[Arnout: re-wrap help, add extra empty line, change = into +=]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Add allwinner mali kernel driver package. Used in combination with
userspace Allwinner openGL libraries, it gives possibility to use 3D openGL
SoC acceleration.
[Peter: Rename to sunxi-mali-mainline-driver. Use revision selection from
sunxi-mali-mainline package. Depend on that package and default to
y if dependencies are met. Tweak Linux config]
Signed-off-by: Giulio Benetti <giulio.benetti@micronovasrl.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Currently, the rule to rebuild the Linux kernel with an initramfs
directly depends on the path of the file of the intermediate cpio image.
This is inherently "bad" from a purity point of view; linux.mk should
not have to delve into the fs internals.
Rather, make it directly depend on the "frontal" rule that generates the
cpio image.
Drop the comment for linux-rebuild-with-initramfs, it was misleading
(talking about generating "the initramfs list of files", which is not
what was done, since we use a cpio as source of initramfs, not a list of
files).
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The LINUX_KCONFIG_FIXUP_CMDS are meant to deselect any compression
option that are not selected in the buildroot configuration. But it only
deselects the last one in the list instead of all of them because it
overwrites the LINUX_COMPRESSION_OPT_ variable instead of appending to
it. Only the last option set to that variable gets deselected.
This produces the warning:
.config:2216:warning: override: KERNEL_GZIP changes choice state
is emitted when buildroot runs olddefconfig when buildroot configures a
kernel with a custom config that has a different kernel compression
option set to what is configured in buildroot.
Accumulate all the deselected compression options instead of overwriting
them to ensure all non-selected options get deselected..
Signed-off-by: Cam Hutchison <camh@xdna.net>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Starting with 4.12-rc1, tarballs are generated by cgit directly from
Linus's tree. This also implies that no .tar.xz can be used for them.
This method also applies to older release candidates.
Signed-off-by: Luis Araneda <luaraneda@gmail.com>
[Arnout: added comment in the code]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Xtensa core configuration must be added to linux before it can be
built for that xtensa CPU variant. Extract configuration files from the
xtensa overlay as is done for other packages that need to be configured
for a specific xtensa core.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Since things are no longer installed in $(HOST_DIR)/usr, the callers
should also not refer to it.
This is a mechanical change with
git grep -l '$(HOST_DIR)/usr/bin' | xargs sed -i 's%$(HOST_DIR)/usr/bin%$(HOST_DIR)/bin%g'
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
We don't have a hash for any linux version. We currently also don't
have a hash for the latest version, but if we ever add a hash, it
will only be for the latest version.
This simplifies the code a little because soon we will add hash checks
for git as well, so we also need an exclusion in that case. It's much
easier to always exclude except in the single case where we do have a
hash.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
In 4c10eedc1 (systemd: enable required kernel features), we added
setting a few required kernel features to ensure systemd works.
However, there was a typo for one of the variables: CONFIG_TMPFS_XATTR
was written as CONFIG_TMPFS_POSIX_XATTR, which does not exist (and never
ever existed, at least not since 2.6.12)...
Reported-by: Michael Heinemann <posted@heine.so>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Cc: Michael Heinemann <posted@heine.so>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
A long time ago, the blind config option BR2_LINUX_KERNEL_UBOOT_IMAGE
was introduced to be able to trigger the linux -> host-uboot-tools
dependency. Back in those days, there was no user-configurable
BR2_PACKAGE_HOST_UBOOT_TOOLS.
Now, however, it is possible to select a custom kernel image name that
needs uboot-tools, and manually enable BR2_PACKAGE_HOST_UBOOT_TOOLS. In
this case, however, the linux -> host-uboot-tools is missed and the
build is not reproducible. An example of such a situation is the
upcoming CI40 defconfig.
As a solution, remove BR2_LINUX_KERNEL_UBOOT_IMAGE entirely. Instead,
just select BR2_PACKAGE_HOST_UBOOT_TOOLS and add the dependency if it
is selected.
Note that this may introduce a redundant dependency in case the user
selected BR2_PACKAGE_HOST_UBOOT_TOOLS for some other reason (e.g. to
be able to generate a U-Boot environment to include in the image, while
the kernel is built as a zImage). However, the redundant dependency
shouldn't hurt much.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Abhimanyu Vishwakarma <abhimanyu.v@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This patch adds an easy way to select the CIP project SLTS
kernel within the kernel menu.
Signed-off-by: Angelo Compagnucci <angelo.compagnucci@gmail.com>
[Thomas: remove BR2_CIP_KERNEL_REPO_URL option.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
We want to use SPDX identifier for license strings as much as possible.
SPDX short identifier for GPLv2/GPLv2+ is GPL-2.0/GPL-2.0+.
This change is done by using following command.
find . -name "*.mk" | xargs sed -ri '/LICENSE( )?[\+:]?=/s/\<GPLv2\>/GPL-2.0/g'
Signed-off-by: Rahul Bedarkar <rahulbedarkar89@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Whitespaces were searched using the following regex:
[ ]{1,}\t
and then manually removed in most of the cases. For
xserver_xorg-server.mk, tabs before backslashes were removed.
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
If not set the system will use an empty string which will result in
download errors for 'linux-.tar.gz' packages.
This patch makes it obvious to the user that the variable needs to be
set.
Signed-off-by: Christian Kellermann <christian.kellermann@solectrix.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Fix kernel reproducible build if a non-C locale is used on the host
system.
When building the Linux kernel, scripts/gen_initramfs_list.sh does 'date
-d"$KBUILD_BUILD_TIMESTAMP" +%s'. In linux.mk, Buildroot sets
KBUILD_BUILD_TIMESTAMP to "$(shell date -d @$(SOURCE_DATE_EPOCH))".
For example, if LANG=fr_FR.UTF-8 is defined in the host system, it does
not work:
- LC_ALL=C date -d"$(LC_ALL=C date)" : ok
- LC_ALL=C date -d"$(LC_ALL=fr_FR.UTF-8 date)" : error
LANG/LC_ALL variables exported in the main Makefiles are not passed in
the $(shell ...) sub-shells.
Signed-off-by: Jean-Baptiste Trédez <jean-baptiste.tredez@basystemes.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Linux kernel include a few information about build environment in its binary.
This feature is incompatible with BR2_REPRODUCIBLE. This patch overload build
information when BR2_REPRODUCIBLE is enabled.
Note that usage of KBUILD_BUILD_TIMESTAMP is not mandatory since Buildroot
use `fakedate'. However, native solution is prefered when upstream
provide one.
Signed-off-by: Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Jérôme Pouiller <jezz@sysmic.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The kernel source tree also contains the sources for various userland
tools, of which cpupower, perf or selftests.
Currently, we have support for building those tools as part of the
kernel build procedure. This looked the correct thing to do so far,
because, well, they *are* part of the kernel source tree and some
really have to be the same version as the kernel that will run.
However, this is causing quite a non-trivial-to-break circular
dependency in some configurations. For example, this defconfig fails to
build (similar to the one reported by Paul):
BR2_arm=y
BR2_cortex_a7=y
BR2_ARM_FPU_NEON_VFPV4=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_INIT_SYSTEMD=y
BR2_LINUX_KERNEL=y
BR2_LINUX_KERNEL_CUSTOM_GIT=y
BR2_LINUX_KERNEL_CUSTOM_REPO_URL="https://github.com/raspberrypi/linux.git"
BR2_LINUX_KERNEL_CUSTOM_REPO_VERSION="26f3b72a9c049be10e6af196252283e1f6ab9d1f"
BR2_LINUX_KERNEL_DEFCONFIG="bcm2709"
BR2_PACKAGE_LINUX_TOOLS_CPUPOWER=y
BR2_PACKAGE_CRYPTODEV=y
BR2_PACKAGE_OPENSSL=y
BR2_PACKAGE_LIBCURL=y
This causes a circular dependency, as explained by Thomas:
- When libcurl is enabled, systemd depends on it
- When OpenSSL is enabled, obviously, will use it for SSL support
- When cryptodev-linux is enabled, OpenSSL will depend on it to use
crypto accelerators supported in the kernel via cryptodev-linux.
- cryptodev-linux being a kernel module, it depends on linux
- linux by itself (the kernel) does not depend on pciutils, but the
linux tool "cpupower" (managed in linux-tool-cpupower) depends on
pciutils
- pciutils depends on udev when available
- udev is provided by systemd.
And indeed, during the build, we can see that make warns (it's only
reported as a *warning*, not as an actual error):
[...]
make[1]: Circular /home/ymorin/dev/buildroot/O/build/openssl-1.0.2h/.stamp_configured
<- cryptodev-linux dependency dropped.
>>> openssl 1.0.2h Downloading
[...]
So the build fails later on, when openssl is actually built:
eng_cryptodev.c:57:31: fatal error: crypto/cryptodev.h: No such file or directory
compilation terminated.
<builtin>: recipe for target 'eng_cryptodev.o' failed
Furthermore, graph-depends also detects the circular dependency, but
treats it as a hard-error:
Recursion detected for : cryptodev-linux
which is a dependency of: openssl
which is a dependency of: libcurl
which is a dependency of: systemd
which is a dependency of: udev
which is a dependency of: pciutils
which is a dependency of: linux
which is a dependency of: cryptodev-linux
Makefile:738: recipe for target 'graph-depends' failed
Of course, there is no way to break the loop without losing
functionality in either one of the involved packages *and* keep
our infrastructure and packages as-is.
The only solution is to break the loop at the linux-tools level, by
moving them away into their own package, so that the linux package will
no longer have the opportunity to depend on another package via a
dependency of one the tools.
All three linux tools are thus moved away to their own package.
The package infrastructure only knows of three types of packages: those
in package/ , in boot/ , in toolchain/ and the one in linux/ . So we
create that new linux-tools package in package/ so that we don't have to
fiddle with yet another special case in the infra. Still, we want its
configure options to appear in the kernel's sub-menu.
So, we make it a prompt-less package, with only the tools visible as
options of that package, but without the usual dependency on their
master symbol; they only depend on the Linux kernel.
Furthermore, because the kernel is such a huge pile of code, we would
not be very happy to extract it a second time just for the sake of a few
tools. We can't extract only the tools/ sub-directory from the kernel
source either, because some tools have hard-coded path to includes from
the kernel (arch and stuff).
Instead, we just use the linux source tree as our own build tree, and
ensure the linux tree is extracted and patched before linux-tools is
configured and built.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Paul Ashford <paul.ashford@zurria.co.uk>
[Thomas:
- fix typo #(@D) -> $(@D)
- fix the inclusion of the per-tool .mk files.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
We used to do a special handling of Linux kernel modules when stripping
target binaries because there's some special precious data in modules
that we must keep for them to properly operate. This is for example true
for stack unwinding data etc.
It turned out there're cases when our existing "strip --strip-unneeded"
doesn't work well. For example this removes .debug_frame section used by
Linux on ARC for stack unwinding, refer to [1] and [2] for more details.
Now Linux kernel may strip modules as a part of "modules_install" target
if INSTALL_MOD_STRIP=1 is passed in command line. And so we'll do
allowing kernel decide how to strip modules in the best way.
Still note as of today Linux kernel strips modules uniformly for all
arches with "strip" command, so this commit alone doesn't solve
mentioned problem but it opens a possibility to add later a patch to the
kernel which will strip modules for ARC differently - and that's our
plan for mainline kernel.
[1] https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/issues/86
[2] http://lists.busybox.net/pipermail/buildroot/2016-September/172161.html
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Daniel Mentz <danielmentz@google.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>