* -image and -base change PKG to:
PKG:${KERNEL_PACKAGE_NAME}-image = "${KERNEL_PACKAGE_NAME}-image-${@legitimize_package_name(d.getVar(KERNEL_VERSION))}"
PKG:${KERNEL_PACKAGE_NAME}-base = "${KERNEL_PACKAGE_NAME}-${@legitimize_package_name(d.getVar(KERNEL_VERSION))}"
* but only when debian.bbclass is inheritted they add the original package name into RPROVIDES by:
https://git.openembedded.org/openembedded-core/commit/?id=3409c4379559afbb1d1d29045582995147a33bbc
* fixes the build if some packagegroup or something RDEPENDS on kernel-image or kernel-base
and the DISTRO doesn't inherit debian.bbclass
* as shown in pkgdata:
linux-raspberrypi $ egrep "^(PKG:)|(RPRO)" 6.6.36+git-*/pkgdata/runtime/kernel-image
6.6.36+git-debian/pkgdata/runtime/kernel-image:RPROVIDES:kernel-image: kernel-image (=6.6.36+git@PRSERV_PV_AUTOINC@+733366844f_769634f344)
6.6.36+git-debian/pkgdata/runtime/kernel-image:PKG:kernel-image: kernel-image-6.6.36-v8
6.6.36+git-without-debian/pkgdata/runtime/kernel-image:PKG:kernel-image: kernel-image-6.6.36-v8
linux-raspberrypi $ egrep "^(PKG:)|(RPRO)" 6.6.36+git-*/pkgdata/runtime/kernel-image-image
6.6.36+git-debian/pkgdata/runtime/kernel-image-image:RPROVIDES:kernel-image-image: kernel-image-image (=6.6.36+git@PRSERV_PV_AUTOINC@+733366844f_769634f344)
6.6.36+git-debian/pkgdata/runtime/kernel-image-image:PKG:kernel-image-image: kernel-image-image-6.6.36-v8
6.6.36+git-without-debian/pkgdata/runtime/kernel-image-image:PKG:kernel-image-image: kernel-image-image-6.6.36-v8
(From OE-Core rev: 05498781657a3f8b38b000f91594ecd78850ce47)
(From OE-Core rev: 350513959f6800eef6579153c2ae95960ca24ea7)
Signed-off-by: Martin Jansa <martin.jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit 9cb954884b)
Signed-off-by: Jörg Sommer <joerg.sommer@navimatix.de>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
If the kernel folder does not exist, find will result in an error.
This can occur if the kernel has no modules but, for example, custom modules are created.
Add check before deleting.
(From OE-Core rev: 4e4681b26e0d88ad219d72b75e598e6b81b430fc)
Signed-off-by: Heiko Thole <heiko.thole@entwicklung.eq-3.de>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit 7ef767d84d56b25498e45db83bb8f9d9caebeaf9)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
[Backport cd2072e5d953 from the master without HOSTPKG_CONFIG setting]
The pkg-config workaround has been applied for kernel image building, but
not for module building. So pkg-config variables are different between
do_compile and do_compile_kernelmodules tasks. It may unnecessary trigger
rebuilding of a few host tools at the later task.
Especially when CONFIG_DEBUG_INFO_BTF is enabled in the kernel, it may even
trigger rebuilding vmlinux at do_compile_kernelmodules due to the rebuilt
host tools such as certs/extract-cert or objtool (on x86). This eventually
creates an inconsistent set of kernel binaries.
Here is the repro steps:
- Check out nanbield on x86
- The unexpected rebuild happens on kirkstone or possibly earlier
- Ensure that pahole is available (e.g. via meta-oe)
- Set KERNEL_DEBUG to "True" to properly set up PAHOLE
e.g.
$ export KERNEL_DEBUG="True"
$ export BB_ENV_PASSTHROUGH_ADDITIONS="${BB_ENV_PASSTHROUGH_ADDITIONS} KERNEL_DEBUG"
- Enable CONFIG_DEBUG_INFO_BTF=y
e.g.
$ bitbake -c menuconfig virtual/kernel
-> Kernel hacking
-> Compile-time checks and compiler options
-> Generate BTF typeinfo
- Build the kernel
e.g.
$ bitbake virtual/kernel
The BTF information in the resulting bzImage and kernel modules are
inconsistent, because the module's BTF information is generated using the
"second" vmlinux that doesn't have the identical BTF to the "first" vmlinux.
These modules can't be loaded at runtime due to the BTF mismatch.
This also leads to a build-id mismatch between the installed bzImage and
vmlinux since the bzImage is created from the first vmlinux, but the
installed vmlinux is the second one.
$ eu-readelf -n tmp/work/qemux86_64-poky-linux/linux-yocto/6.5.13+git/image/boot/{bzImage*,vmlinux*} | grep "Build ID"
Build ID: 4a0d62ee7fef0244950f0f604253729875bea493
Build ID: fb99b3d91399dbe42bf67ddee59e0f5a0c7f74d9
To avoid the unexpected rebuilding that results in such inconsistency, set
the same pkg-config variables when building kernel and modules. For kernel
5.19 and above, simply set the HOSTPKG_CONFIG in the make command line.
(From OE-Core rev: d63af11e92094487d6e358f27283e5385937e7a8)
Signed-off-by: Munehisa Kamata <kamatam@amazon.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
The initial fix for localversion setting in 6.3+ broke older
recipes and also broke recipes setting localversion in a kernel
recipe, as make-mod-scripts (and other locations) can trigger
a regeneration of files and don't have access to the variable.
Moving the setting of this variable to the global namespace
doesn't make sense, so we follow the example of the kernel-abiversion
and save a kernel-localversion to the build artifacts.
Recipes that may regenerate scripts/dynamic files, must
depend on the do_shared_workedir of the kernel and use the helper
function to read the file storing the localversion.
(From OE-Core rev: cca0971a7d92d823cc0c2b16cf14a7b2ed8ecb61)
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
cherry-picked from master b378eec156998eea55ba61e59103cb34fab0d07c
Signed-off-by: Andreas Helbech Kleist <andreaskleist@gmail.com>
Acked-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
During testing of the v6.4 reference kernel, it was noticed that
on-target modules no longer matched the magic value of the running
kernel.
This was due to a different localversion in the cross built kernel
and the scripts / resources created on target.
This was due to changes in the setlocalversion script introduced
in the v6.3 series.
The .scmversion file is no longer used (or packaged) to inhibit
the addition of a "+" (through querying of the git status of the
kernel) or the setting of a local version.
We recently introduced the KERNEL_LOCALVERSION variable to allow
recipes to place a value in .scmversion, so we extend the use of
that variable to kernel-arch.bbclass and use it to set the
exported variable LOCALVERSION.
We must do it at the kernel-arch level, as the variable must be
exported in any kernel build to ensure that setlocalversion always
correctly sets the localversion.
(From OE-Core rev: 74897e505db19a23a5b864a48a4fa97d657605c8)
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
cherry-picked from master 765b13b7305c8d2f222cfc66d77c02e6a088c691
Signed-off-by: Andreas Helbech Kleist <andreaskleist@gmail.com>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Just like UBOOT_LOCALVERSION, an end user can set KERNEL_LOCALVERSION
to append a string to the name of the local version of the kernel
image.
(From OE-Core rev: 29a0d2a49df8e24f3948a2ddf03743f0cb5d09dd)
Signed-off-by: Ming Liu <liu.ming50@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
cherry-picked from master 229435a52f36ddec5f85fb6d5ccd42044b688397
Signed-off-by: Andreas Helbech Kleist <andreaskleist@gmail.com>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
This reverts commit d9273edae80978c34f8426f34f991b9598828aa9.
This commit is causing breakage for some vendor kernel builds.
(From OE-Core rev: 59cc2e75c15f8c6371a4c4a3b7bd2e6c3f145fbc)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
During testing of the v6.4 reference kernel, it was noticed that
on-target modules no longer matched the magic value of the running
kernel.
This was due to a different localversion in the cross built kernel
and the scripts / resources created on target.
This was due to changes in the setlocalversion script introduced
in the v6.3 series.
The .scmversion file is no longer used (or packaged) to inhibit
the addition of a "+" (through querying of the git status of the
kernel) or the setting of a local version.
We recently introduced the KERNEL_LOCALVERSION variable to allow
recipes to place a value in .scmversion, so we extend the use of
that variable to kernel-arch.bbclass and use it to set the
exported variable LOCALVERSION.
We must do it at the kernel-arch level, as the variable must be
exported in any kernel build to ensure that setlocalversion always
correctly sets the localversion.
(From OE-Core rev: d9273edae80978c34f8426f34f991b9598828aa9)
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Cherry-picked from master 765b13b7305c8d2f222cfc66d77c02e6a088c691
Signed-off-by: Andreas Helbech Kleist <andreaskleist@gmail.com>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
The latest 6.5 kernels do not appear to create the source file in
${D}${nonarch_base_libdir}/modules/${KERNEL_VERSION}/source so the
recipe errors out when trying to remove it. Simple fix is to add the
-f (force) flag to the call.
(From OE-Core rev: 2e669bf797b15d803e7d6a700e449bdc467a4bcc)
(From OE-Core rev: 7e177848f97eb9958619c28b5e5dadee12f67507)
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Alexander Sverdlin <alexander.sverdlin@siemens.com>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Due to an oversight in the do_symlink_kernsrc function, the path
comparison between "S" and "STAGING_KERNEL_DIR" is broken. The code
obtains both variables, but modifies the local copy of "S" before
comparing them, causing the comparison to always return false.
This can cause the build to fail when the EXTERNALSRC flag is enabled,
since the code will try to create a symlink even if one already exists.
This patch resolves the issue by comparing the variables before they are
modified.
(From OE-Core rev: cf2267f80ec44b24c627347df7efbd492a07dcfa)
Signed-off-by: Staffan Rydén <staffan.ryden@axis.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
(cherry picked from commit afd2038ef8a66a5e6433be31a14e1eb0d9f9a1d3)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Since explicit debug package creation via ${KERNEL_PACKAGE_NAME}-dbg has
been added to kernel, it has to cover all PACKAGE_DEBUG_SPLIT_STYLE
options. For ex. when the variable "debug-file-directory" package search
path has to be set explicitly, otherwise it will not find any files.
(From OE-Core rev: 9adbda8450c57f49edf85e3b3433304e8ac8267e)
Signed-off-by: Andrej Valek <andrej.valek@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit 9c39da147683dcaaa244b3ddc4531c4408ad5c9e)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
If a specific kernel provider or configuration wants to enable BTF
and pahole analysis, it isn't currently possible due to the explicit
definition to false in the base kernel build arguments.
pahole is now detected by the kernel built itself, so unless
pahole-native is enabled, the result is the same.
If a kernel does require an explicit disable of pahole, it is better
to carry PAHOLE=false in those specific recipes.
(From OE-Core rev: 44b75c7442b05a2fedcc0d73982566c2bef84357)
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit b1e4851a36ed47ce6ba880a49264b9a57c78cf4f)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Use the notation suggested by Martin Jansa to avoid a bashism. Also
switch KERNEL_DTBVENDORED to the more common 0/1 notation instead of
true/false.
Fixes: 04ab57d200 (kernel-devicetree: allow specification of dtb
directory, 2023-05-02)
(From OE-Core rev: b7d6fc07462e6fca09d0db7f43a62920250ef053)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Fedora/Redhat and Arch are somewhat standardized on their dtb directory
structure. Let's add some flags to configure yocto to mimic that
behavior.
Add the following variables to the kernel class:
- KERNEL_DTBDEST (controls the destination directory for dtbs)
- KERNEL_DTBVENDORED (controls if vendor subdirectories are to
be respected)
Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST
and KERNEL_DTBVENDORED is expected to be "true"/"false". This only
applies to the package directory structure. The deploydir structure is
purposely left untouched for compatibility with existing recipes.
By default this is configured to behave the same as the current recipe
and produce a flat dtb directory at KERNEL_IMAGEDEST.
(From OE-Core rev: 77d6118e20751f68ad7104edec8f14dbb5ba6ec1)
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Steve Sakoman <steve@sakoman.com>
This is a partial fix for bugzilla 15059 [https://bugzilla.yoctoproject.org/show_bug.cgi?id=15059]
It has been noted by several people that when an initramfs is bundled:
- a lot of the kernel is rebuilt
- it takes a really long time
When looking at the logs, the second kernel compilation (that performs
the bundle) is not using the parallel make settings, and builds with
-j1.
We are already explicitly passing PARALLEL_MAKE when building kernel
modules, and by extending that explicit use to the main kernel
compilation, we ensure that we always get a parallel build.
Build times chnaged from more than 30 minutes for the bundle, to
3 minutes in local testing.
The question of whether or not too much is rebuilding during the
bundle step is still an open question, but with this tweak, at least
the build time is back in the realm of acceptable.
(From OE-Core rev: a2a889b760785474dbc04e3ec11521f6da90161d)
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
(cherry picked from commit 88fd394ecf0f2174b792075d409d87046896426b)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
pkgconfig is being required to find dependencies for building kernel
native tools, move "inherit pkgconfig" to kernel.bbclass so BSP kernel
recipes can also benefit from it.
(From OE-Core rev: e09c50e01d0f5cf45b9622ab0ed33df12bdeb7ee)
Signed-off-by: Ming Liu <liu.ming50@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
(cherry picked from commit 8a84bd98e3fbc16c782f83064801e469d086911e)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
With the 6.1 kernel we've seen reproducibility/determinism issues where the kernel
seems to rebuild headers referencing the hostname or local user. kernel-devsrc
building after the kernel seemed to trigger it in some cases.
Moving the definitions to the bbclass used by all the kernel recipe code including
kernel-devsrc seems to be the best way to ensure this doesn't happen.
(From OE-Core rev: 4c302827365903b99a517651c2a14661ce78f79b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit 60681baa97daf4f3856453c34d6be08b6771a81b)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently, allyesconfig test runs for x86_64 fail with:
ERROR: linux-yocto-5.19.17+gitAUTOINC+0cba9aa404_aaf4490d18-r0 do_package:
QA Issue: linux-yocto: Files/directories were installed but not shipped in any
package:
/lib/modules/5.19.17/kernel/drivers/nvdimm
With CONFIG_NVDIMM_TEST_BUILD=m, an empty nvdimm directory is created during
modules_install, which triggers the QA issue.
Extend kernel_do_install() to also remove inner empty directories that might
get created by modules_install.
(From OE-Core rev: 2f2abcb6d2394cdfd3985499ed882f43a5d2a299)
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
(cherry picked from commit 7120b09a33af4c9a18063c0f2e51fb598697e39c)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently, the KERNEL_DEBUG_TIMESTAMPS is not working as expected
at rebuild. That is, even if we set it to "1", the kernel build time
is not changed. The problem could be reproduced by the following steps.
1. bitbake core-image-minimal; start image and check `uname -a` output.
2. set in local.conf: KERNEL_DEBUG_TIMESTAMPS = "1"
3. bitbake core-image-minimal; start image and check `uname -a` output.
It's expected that after enabling KERNEL_DEBUG_TIMESTAMPS, the kernel
build time will be set to current date. But it's not. This is because
the compile.h was not re-generated when do_compile task was re-executed.
In mkcompile_h, we have:
"""
# Only replace the real compile.h if the new one is different,
# in order to preserve the timestamp and avoid unnecessary
# recompilations.
# We don't consider the file changed if only the date/time changed,
# unless KBUILD_BUILD_TIMESTAMP was explicitly set (e.g. for
# reproducible builds with that value referring to a commit timestamp).
# A kernel config change will increase the generation number, thus
# causing compile.h to be updated (including date/time) due to the
# changed comment in the
# first line.
"""
It has made it very clear that it will not be re-generated unless
we have KBUILD_BUILD_TIMESTAMP set explicitly. So we set this variable
explicitly in do_compile to fix this issue.
(From OE-Core rev: 640ac18b2daed698adbf849a5aef55f5de9e5db5)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
(cherry picked from commit 1b68c2d2d385013a1c535ef81172494302a36d74)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When building with CONFIG_MODVERSIONS=y and CONFIG_RANDSTRUCT=y we need
to copy the build assets generated for the randstrutc seed to
STAGING_KERNEL_BUILDDIR, otherwise the out-of-tree modules build will
generate those assets which will result in a different
RANDSTRUCT_HASHED_SEED.
(From OE-Core rev: d6cb9dce1ffb14f9db497e9bb0cb7265ea4064ec)
Signed-off-by: Diego Sueiro <diego.sueiro@arm.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
(cherry picked from commit b36303158b2e0273ff415bdedefb379f680b30fc)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Replacing sysroot_stage_all by a no-op recipe makes it difficult for
bbappends to stage files intentionally. Instead, just clear
SYSROOT_DIRS, allowing other bbappends to easily add new directories.
(From OE-Core rev: 45ed53ec09e24956ce3d7d008e254bc759cf5f85)
Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
(cherry picked from commit 849791e7086463a4c7c53c2c1ed9603a6c3a080d)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We're currently only passing in a subset of the kernel make flags to
menuconfig. Fix this to be consistent with all the other kernel
operations since these are becomming increasingly reliant on host
compilers and flags and target toolchains as well.
(From OE-Core rev: e36f3c3c7de052945edbb62bb2d6de7639360c47)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit 8c616bc090d1834a21073a33209323220c05d2e5)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It as been bothering me a bit that we don't set CC and LD consistently
for the kernel make calls and this can lead to interesting bugs as the
kernel increases in complexity. Add them to EXTRA_OEMAKE so they're
always passed in. This makes everything slightly more consistent and
less likely to break in future.
(From OE-Core rev: 0872e11ede2469a7d176c791395b139158bd22a3)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit afe8c318843e4033dcc07e4f10198df241d8e4f6)
Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The BB variable can't be referenced directly in a python function,
this misusage of the variable reference causes strip function to be
always skipped.
Fixed: b9c3db4953e4 ("kernel.bbclass: Use KERNEL_IMAGEDEST instead of hardcoded boot path")
(From OE-Core rev: c405cb4f421a8c964fa59123ee41879ebd5f829c)
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The behaviour doesn't change because KERNEL_IMAGEDEST is defined as
'boot' in the same file (line 215).
(From OE-Core rev: b9c3db4953e4e7b423ba9ec5b618fd990c378bc1)
Signed-off-by: Alessio Igor Bogani <alessio.bogani@elettra.eu>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
CONFIG_PAHOLE_HAS_SPLIT_BTF shows up in the config only when pahole is
installed on the host system. As a result, the config changes depending on
whether pahole is installed or not.
Set PAHOLE=false to ensure that it is never found.
If this is actually needed in the future, then we can add an option for
it and create a host package for pahole.
(From OE-Core rev: 43b0eca2f3fc41a0d2ba7c7679687a3e0eb54b45)
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
[afa: ported from PTXdist 0c0cec2288 to OE-core]
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
As kernel modules depend on the base kernel package, and the base kernel
recipe depends on the kernel image, it's impossible to build file system
images which contain kernel modules but not the kernel itself, such as
an initramfs.
Change the RDEPENDS to RRECOMMENDS so that the disk images can set
PACKAGE_EXCLUDE = "kernel-image-*" to remove the kernel.
(From OE-Core rev: 1c90b27d2c65cfb4f9debf0272820b6a95942f76)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There may be a case where we want to build an initramfs image
that doesnt inherit the same DISTRO_FEATURES (or others) from
the main image being built.
For example we may want our initramfs not to inherit a certain
conf or feature, say we want to use musl for a smaller footprint,
but if we are using TCLIBC=glibc for our DISTRO (and inherently
our main image), the initramfs image would inherit that conf and
be forced to use glibc, growing in size as a side effect, currently
avoiding this is not supported.
Allow the kernel class to create a multiconfig dependency
(mcdepends) vs depends for do_bundle_initramfs and define
our INITRAMFS_IMAGE from a separate multiconfig via two
new variables:
INITRAMFS_MULTICONFIG and INITRAMFS_DEPLOY_DIR_IMAGE which
define the multiconfig where the initramfs image should be
coming from and its deploy directory respectively, these two
keep a default definition which preserves current behavior
(do_bundle_initramfs uses depends).
Example usage:
- Create and use multiconfig initramfscfg.conf and set
TCLIBC=musl there, along with its TMPDIR.
- Add the following to our DISTRO.conf:
INITRAMFS_MULTICONFIG = "initramfscfg"
and set
INITRAMFS_DEPLOY_DIR_IMAGE to the DEPLOY_DIR_IMAGE of the
initramfscfg multiconfig (hence our main kernel will be
able to grab it from there and bundle it).
This will result in our musl based initramfs bundled in our
main kernel and our main image to be glibc based.
(From OE-Core rev: 2d317b2685211f1b0d102705a63c0000df96f45f)
Signed-off-by: Alejandro Enedino Hernandez Samaniego <alhe@linux.microsoft.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Re-use the runstrip() code from oe.packaging, for the kernel
stripping process. Since runstrip() is python the kernel do_strip()
need to be converted to Python also. The stripped kernel image
will be used for deployment in do_deploy(). This will allow the
package.bbclass to split an unstripped kernel binary for the debug
information and extended packagedata. The extended package data is
used by create-spdx.
(From OE-Core rev: e8d9caede5f08154ca615fdaba676b7a4ae05b01)
Signed-off-by: Saul Wold <saul.wold@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Adding the dbg package allows the package bbclass to parse the
debug information which can then be used by the create_spdx bbclass
(From OE-Core rev: b35b68e4ec4a82ada20ab861d29f96cdcb21dec2)
Signed-off-by: Saul Wold <saul.wold@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Allow .bin suffix to be removed (or changed) in the various artifact
filenames. Removing this extension is useful when trying to remove
symlinks and present only unversioned image files (especially for the
FIT image).
(From OE-Core rev: cbecc3cf06eb7359fedf3c6af281cc72178cad18)
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Allow setting any of the *_LINK_NAME variables to empty string in order
to disable creating symlinks for kernel artifacts, as you can already
for filesystem images with IMAGE_LINK_NAME. Additionally, for the image
type named symlinks, add a KERNEL_IMAGETYPE_SYMLINK boolean variable
which you can set to 0 to disable those symlinks as well.
(From OE-Core rev: d7341f1f22c32ff6cc95d7127f26f87d7fc9c6bd)
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In 526bdd88ccd758204452579333ba188e29270bde the imageType loop in
kernel_do_deploy was changed to use KERNEL_IMAGETYPE_FOR_MAKE rather
than KERNEL_IMAGETYPES. This broke the special handling for fitImage
immediately below because KERNEL_IMAGETYPE_FOR_MAKE never contains
fitImage.
It has always been my understanding that KERNEL_IMAGETYPE_FOR_MAKE
controlled what was passed to make, but KERNEL_IMAGETYPE controlled what
was installed/deployed. When the two are different then it's the
responsibility of whoever set KERNEL_IMAGETYPE_FOR_MAKE to ensure that
whatever comes out of the kernel build system has been transformed in to
the requested form by the time of installation. This is what happens for
kernel.bbclass's own support for vmlinux.gz.
I think this means that for KERNEL_IMAGETYPE vmlinux.gz, kernel.bbclass
is responsible for generating vmlinux.gz.initramfs[1] so that
kernel_do_deploy can deploy it. This means that the change in
526bdd88ccd758204452579333ba188e29270bde can be reverted, fixing
KERNEL_IMAGETYPE = "fitImage".
In addition, it ought to be possible for recipes and other classes that
use kernel.bbclass to hook into this mechanism by setting
KERNEL_IMAGETYPE_FOR_MAKE and performing their own transformations.
do_bundle_initramfs calls kernel_do_compile and we don't want it to
transform vmlinux to vmlinux.gz at that point, since it will fight
against the careful renaming and preserving that do_bundle_initramfs
does. Let's separate the transformation out of kernel_do_compile to a
new do_transform_kernel task that can be run at the right time. This
means that it's also logical to perform the equivalent translation for
the kernel with the initramfs in a separate
do_transform_bundled_initramfs task too.
This leaves two clear customisation points for recipes and other classes
to hook into the process and perform their transformations:
do_transform_kernel and do_transform_bundled_initramfs.
(I care about this because our recipes that use kernel.bbclass also set
KERNEL_IMAGETYPE_FOR_MAKE and transform vmlinux into a form suitable for
our bootloader after do_compile and do_bundle_initramfs into the format
matching KERNEL_IMAGETYPE. I'm unable to successfully bundle an
initramfs after 526bdd88ccd758204452579333ba188e29270bde, but I didn't
want to just revert that change to reintroduce the bug that it was
fixing.)
I can't say that I'm entirely happy with this change, but I'm unsure
what to do to improve it. I find the way that both the bare kernel and
the one with the initramfs both get deployed to be confusing, and a
waste of build time. I would like to not actually generate a publishable
kernel image at all during do_compile when an initramfs is in use, but I
suspect that this would affect valid use cases that I'm not aware of.
(From OE-Core rev: 10a4a132e87e835726bf5da81a60f6f509b90765)
Signed-off-by: Mike Crowe <mac@mcrowe.com>
[1] It could be argued that this should be vmlinux.initramfs.gz, but
that would require another special case in kernel_do_deploy and the
filename is only visible within this class and the recipes that use it
anyway.
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In a similar manner to cml1.bbclass, we export the pkg-config
variables to allow a direct call to pkg-config access to the
native sysroot versus the target sysroot.
The kernel doesn't use pkg-config for target configuration,
and has many explicit calls to pkg-config, without the
possibility of easy override to pkg-config-native.
The calls to pkg-config could be made cross friendly via
replacement with make variables, but until that effort is
undertaken upstream, we need a bridge approach.
In particular, this is required for dtschema validation,
which is a requirement in kernel 5.16+
(From OE-Core rev: 9172d61c57e23682c3d2c25701cbd53c84d01a27)
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since we allow the kernel package name to be varied, we should
allow those packages to provide virtua/kernel, which allows them
to be used as primary kernels.
This change drops the use of regex, since any regex would enforce
naming (and limit what can be a provider) and add complexity./
There are currently no recipes that I found, that inherit kernel
that are not kernel recipes (kernel-base, etc, provide other
facilities), so making this provides simpler shouldn't cause
problems.
Multiple kernel providers can be dealt with in the same way as
any multiple provider.
[YOCTO: #13172]
(From OE-Core rev: 365dc1471a1b67c45de58aadf29844a9ff83d30f)
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
squash with provider
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In 2017 we added changes to pass the BUILD_CFLAGS into the kernel
via BUILD_CC. This isn't really correct and the upstream kernel now has
places to pass build cflags, ldflags and more. Update our kernel
make flags to correctly use the kernel's variables. This addresses
concerns raised by kernel developers.
If this breaks some usecase please report it so we can work out how
to fix it properly.
(From OE-Core rev: 7fd06a57a1d91d8534721923f6e3951ec8220cec)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Change the "binary reproducibility" configuration within the kernel to
work off a separate variable, defaulting to reproducible builds.
This allows kernel developers wanting timestamps in their images to enable
it easily and clearly without changing the rest of the reproduciblity
code which they likely don't need to change anyway.
(From OE-Core rev: 0725ca18af7a2835aeb9616592a45ead2ee87987)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The grep pattern seems to have been wrong ever since we stopped adding
the -ffile-prefix-map via a patch in commit 20aea61385e, because the
actual upstream gcc produces
-ffile-prefix-map=<old>=<new>
and not
-ffile-prefix-map=<old=new>
Besides, these *-prefix-map options are already used when building the
kernel due to the setting of KERNEL_CC, and the kernel build system
itself also probes for and uses at least -fmacro-prefix-map (hence
taking care of __FILE__ etc., but not necessarily things that go in
.debug_info sections).
(From OE-Core rev: 5d56de8cd653d2a3a7f4c104d50c14e22d412f66)
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Enable zstd PACKAGECONFIG knob for squashfs-tools and add support for zstd
compressed initramfs and squashfs.
(From OE-Core rev: fa146049eec5f9730781830270cc9dc55a2146f7)
Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
EXTENDPKGV can contain AUTOINC so use an expanded form in the variables allowing
for later expansion during packaging to avoid race issues over the variable and
the build failures that can result.
(From OE-Core rev: 1f93d2426d22c74893e8daec9bb939133c5737c2)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the kernel configuration enables module signing but no key
is provided, then the kernel generates one during the kernel build.
The current runtime-dependency references (with only package names
without full versions) allow mixed package installations from different
rebuilds of the same kernel version.
This creates an issue because then the modules either don't work
or taint the kernel.
Tighten RDEPENDS with the full package version, i.e. use (= ${EXTENDPKGV})
markers for inter-package dependencies.
The kernel will pull in the kernel-modules subpackage of the same
exact version automatically if KERNEL_SPLIT_MODULES="0" is set.
Otherwise the situation is the same as with the old default with
one subpackage per kernel module where they have to be upgraded
manually.
(From OE-Core rev: 56428d15803733e14d6f1faf5f77eb933ef966f3)
Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For some use cases, a monolithic kernel-modules package containing
all modules built from the kernel sources may be preferred.
For one, download time is shorter and installation time is faster.
Set KERNEL_SPLIT_MODULES="0" for this in. The default is one subpackage
per module.
Also, adapt kernel.bbclass to KERNEL_SPLIT_MODULES != "1" case
Extra RDEPENDS and other inter-package references are needed in
this case.
(From OE-Core rev: 4b13409d82e9d576e32b978f7a42a143127ab894)
Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is the result of automated script conversion:
scripts/contrib/convert-overrides.py <oe-core directory>
converting the metadata to use ":" as the override character instead of "_".
(From OE-Core rev: 42344347be29f0997cc2f7636d9603b1fe1875ae)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When packaging is disabled using the nopackages class, ensure we don't
add to PACKAGES. This fixes builds where we have an unpackaged kernel
alongside a packaged kernel.
(From OE-Core rev: 2522daf22e2c27dd9c7926feda0345978217c6c3)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The routine do_sizecheck() was historically needed by legacy
devices with limited flash memory.
The lowest extreme is probably with Zaurus collie having exactly
1024*1024 = 1048576 bytes for the kernel partition.
In the years the KERNEL_IMAGE_MAXSIZE has been converted to kilobytes
thus rounded so we have now KERNEL_IMAGE_MAXSIZE_collie = "1024".
The effect is that now the check fails because we hit curiously this
| WARNING: This kernel zImage (size=1024(K) > 1024(K)) is too big for...
even though zImage is 1047288 bytes (kernel + kexecboot-klibc-initramfs).
Fix this case using test -gt (greater) instead of -ge (greater or equal).
(From OE-Core rev: 254ca956d63b4ce6aa294213b60bb943f9f3a9e6)
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The do_bundle_initramfs() only processes kernel image
types that are found in KERNEL_IMAGETYPE_FOR_MAKE whereas
the build system can generate other types that are not
directly supported by the kernel build system. In which
case when we come to the deploy phase not all the images
mentioned in KERNEL_IMAGETYPES would have a respective
initramfs bundled image. An example is using vmlinux.gz
in KERNEL_IMAGETYPES and enabling initramfs and then we
see
install: cannot stat 'arch/arm64/boot/vmlinux.gz.initramfs': No such file or directory
So we align the deploy phase with bundle initramfs phase
and pick up relevant initramfs bundled images using
KERNEL_IMAGETYPE_FOR_MAKE instead of KERNEL_IMAGETYPES.
(From OE-Core rev: 526bdd88ccd758204452579333ba188e29270bde)
Signed-off-by: Awais Belal <awais_belal@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When including an initramfs bundle inside a FIT image, the fitImage is created
after the install task by do_assemble_fitimage_initramfs.
This happens after the generation of the initramfs bundle
(done by do_bundle_initramfs).
So, at the level of the install task we should not try to install the fitImage.
The fitImage is still not generated yet.
After the generation of the fitImage, the deploy task copies the fitImage from
the build directory to the deploy folder.
Change-Id: I3eaa6bba1412f388f710fa0f389f66631c1c4826
(From OE-Core rev: 1b67fd9ac74935fa41e960478c54e45422339138)
Signed-off-by: Abdellatif El Khlifi <abdellatif.elkhlifi@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The target system triple contains whether the target is Linux or not,
so use it to avoid situations where you can attempt to build a kernel
for systems which don't support Linux.
(From OE-Core rev: b1eeeab21a81990321468ddbdd1745ea24d1828d)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
