| Age | Commit message (Collapse) | Author | Files |
|---|
|
When using a defconfig, using yes '' | make oldconfig may not result in
the correct configuration being set. For example:
$ ARCH=mips make qi_lb60_defconfig
#
# configuration written to .config
#
$ grep USB_ETH .config
CONFIG_USB_ETH=y
# CONFIG_USB_ETH_RNDIS is not set
# CONFIG_USB_ETH_EEM is not set
$ cp arch/mips/configs/qi_lb60_defconfig .config
$ yes '' | make ARCH=mips oldconfig
[...]
$ grep USB_ETH .config
CONFIG_USB_ETH=m
# CONFIG_USB_ETH_RNDIS is not set
# CONFIG_USB_ETH_EEM is not set
Using make olddefconfig solves that but we'll use the oldnoconfig alias
for backward compatibility with older kernels.
$ cp arch/mips/configs/qi_lb60_defconfig .config
$ make ARCH=mips oldnoconfig
scripts/kconfig/conf --olddefconfig Kconfig
#
# configuration written to .config
#
$ grep USB_ETH .config
CONFIG_USB_ETH=y
# CONFIG_USB_ETH_RNDIS is not set
# CONFIG_USB_ETH_EEM is not set
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Removing files from the source tree via find, exec and rm is not the
most efficient operation, due to (among other things) the many forked
processes.
If we use -delete, it saves a significant amount of time. But -delete
does not work with -prune (since it forces -depth). To maintain the
lib, tools and scripts source files, we can hide them temporarily,
skip their hidden directories and then finally restore them.
Time for install before this change:
real 2m48.563s
user 0m35.220s
sys 0m33.036s
Time for install after this change:
real 1m21.301s
user 0m33.160s
sys 0m28.388s
We could further speed this up by using inline perl to delete the files,
but that complexity is avoided for now.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If a kernel is built without any external modules (aka no CONFIG_*=m),
then during a modules_install of the kernel an empty directory is
created at /lib/modules/${KERNEL_VERIONS}/kernel. This is behaviour of
the kernel infrastructure, the directory would normally be populated
with the modules that were built.
However because of the expectations of kernel-modules-split, no packages
are created when there are no modules and an empty directory lingers.
This raises QA issues as warning or errors (depending on the distro).
The following patch changes the kernel_do_install task to check if the
directory is empty and if so removes it.
Signed-off-by: Nathan Rossi <nathan.rossi@xilinx.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
Signed-off-by: Koen Kooi <koen.kooi@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This class has been empty since 2010, so we shouldn't need it anymore. A
check of common layers suggests there shouldn't be any references to it
outside of OE-Core.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This reverts commit 6a6735cb98d529c8c5376aedb5c951339f54fe96.
The module class already ensures the scripts are rebuilt correctly. Running
this at sstate installation time is problematic since it can require the
cross compiler. Adding such a dependency would cause issues of its own.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The new do_bundle_initramfs task introduced in
609d5a9ab9e58bb1c2bcc2145399fbc8b701b85a defeats using the sstate
cache. The kernel is resurrected from the sstate cache but ends up being
built again since do_bundle_initramfs depends on do_compile.
The task is no longer nostamp to avoid causing unnecessary rebuilds. The
sstate checksum stamps should know when to rebuild.
The task now runs before do_deploy and part of the work has been moved to
do_deploy where it now writes to ${DEPLOYDIR} rather than
${DEPLOY_DIR_IMAGE} so that the files end up in sstate.
The task can also race against do_install since both call into the kernel
build system. This is fixed by making do_bundle_initramfs run after
do_install (which therefore also fixes the problem that
3baa63b4d588c3262254528b406ede265dd117bf was addressing.)
Signed-off-by: Mike Crowe <mac@mcrowe.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This reverts commit 3baa63b4d588c3262254528b406ede265dd117bf. It broke
builds that aren't using kernel-yocto.
Signed-off-by: Mike Crowe <mac@mcrowe.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
${KERNEL_OUTPUT} is being renamed/restored in bundle_initramfs task, so we
must ensure bundle_initramfs run after kernel_link_vmlinux where the link
of vmlinux is created as the bootable image.
Signed-off-by: Ming Liu <ming.liu@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
If kernel_do_compile() encounters a .lzma image for the initramfs, it
attempts to decompress it using the little-known "lzmash" utility. This
may not be present, which will cause the build to fail.
Fortunately, it appears that the more mainstream "lzma" tool is
compatible so we can use that one instead.
Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
The commit [kernel.bbclass: test for hardlinks before installing] doesn't
work on all build machines without a space between the ! and the test for
hardlink equivalance. The build continues, while the test fails and no
copy is made into the sysroot. Eventually tasks like build_hdimg will fail
with the missing kernel with the root cause being less than obvious.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
commit df564c4 [kernel: Use hardlinks for do_populate_sysroot for speed]
does indeed speed up the build, but it also means that the KERNEL_OUPUT
and KERNEL_IMAGETYPE may be hardlinks.
If they are hardlinks, install complains that they are actually the same
file and stops the build. We can easily test and avoid the copy if the
file is already in place.
[YOCTO #5527]
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The kernel tree is large and doesn't need to be copied. Override
the default sysroot handling function to use a hardlink copying
function in python.
This commit also drops the copying of the /lib directory which
just contains the kernel modules. We never use those in the sysroot
so there is little point in carrying those around.
For linux-yocto this takes the do_populate_sysroot time 24s -> 14s.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Whilst the comment says we can't do this, its incorrect, install operates
on WORKDIR and its hard for different parts of WORKDIR to be on different
filesystems. Hardlinking instead of copying is therefore a nice performance
gain.
Also, completely skip the Documentation directory (adding a dummy Makefile
to keep make happy) and tweak the other cp commands to link since if we
don't we'd get "this is the same file" type errors from cp for some kernel
versions.
For do_install on linux-yocto, this takes it from 227s -> 84s.
(From OE-Core rev: 0727e510f8533216518563b1533e804a1dc44402)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Since kernel will build twice when we are trying to bundle kernel
and initramfs together after commit 609d5a9ab("kernel.bbclass,
image.bbclass: Implement kernel INITRAMFS dependency and bundling"),
thus, the second building for kernel would fail if rm_work is done
previously.
To fix this problem, we need to make do_bundle_initramfs task run
before do_rm_work task.
[YOCTO #5416]
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Xufeng Zhang <xufeng.zhang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
When building against the sysroot, out of tree modules can require modpost
and other utilities normally found in the kernel's scripts directory. For
the kernel source in the staging dir, these scripts have been removed to
avoid mixing archiectures when packaging kernel-dev (among other things).
Rather than further complicate the kernel's install rule, or its packaging,
we can restore the scripts by building them in the kernel staging directory
after the sstate is installed, making them available to packages that need them.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When user runs devshell in kernel recipe and compiles kernel,
this will fail by unrecognized option of LD.
----
$ make
...
arm-poky-linux-gnueabi-ld: unrecognized option '-Wl,-O1'
arm-poky-linux-gnueabi-ld: use the --help option for usage information
----
This set to empty LDFLAGS when user runs the devshell, and solve this problem.
Signed-off-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Activating the INITRAMFS_TASK can cause circular dependencies, but
that is up to the end user to resolve in recipes. The INITRAMFS_TASK
should also trigger immediate linking of the cpio task in the first
compile pass. This was a subtle regression introduced by: 609d5a9ab
(kernel.bbclass, image.bbclass: Implement kernel INITRAMFS dependency
and bundling).
This patch restores the previous behavior and only affects the
INITRAMFS_TASK which is not set by any of the default build profiles
in oe-core.
Reviewed-by: Andrea Adami <andrea.adami@gmail.com>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
bundling
This patch aims to fix the following two cases for the INITRAMFS generation.
1) Allow an image recipe to specify a paired INITRAMFS recipe such
as core-image-minimal-initramfs. This allows building a base
image which always generates the needed initramfs image in one step
2) Allow building a single binary which contains a kernel and
the initramfs.
A key requirement of the initramfs is to be able to add kernel
modules. The current implementation of the INITRAMFS_IMAGE variable
has a circular dependency when using kernel modules in the initramfs
image.bb file that is caused by kernel.bbclass trying to build the
initramfs before the kernel's do_install rule.
The solution for this problem is to have the kernel's
do_bundle_initramfs_image task depend on the do_rootfs from the
INITRAMFS_IMAGE and not some intermediate point. The image.bbclass
will also sets up dependencies to make the initramfs creation task run
last.
The code to bundle the kernel and initramfs together has been added.
At a high level, all it is doing is invoking a second compilation of
the kernel but changing the value of CONFIG_INITRAMFS_SOURCE to point
to the generated initramfs from the image recipe.
[YOCTO #4072]
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Acked-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
Fixes [YOCTO #4991]
The kernel image is installed as part of the kernel-image package, but
the symlink creation/removal via alternatives is being done in
pkg_post(inst|rm)_kernel-base.
Move the postinst alternatives logic into the kernel-image functions.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
The modules-${MACHINE}.tgz should link to ${MODULE_TARBALL_BASE_NAME}.
But now it links to an invalid file ${MODULE_TARBALL_BASE_NAME}.bin.
Remove the extra .bin suffix.
[YOCTO #4595]
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
* DEPLOYDIR is used as target for every other file deployed by do_deploy,
this was just unnecessary switching CWD to create relative symlink and I
have no idea why rm -f was used together with ln -sf.
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
Compiling the Linux kernel requires binutils; kernel.bbclass uses
INHIBIT_DEFAULT_DEPS so it had better depend on binutils explicitly.
(The lack of this dependency isn't always a problem because binutils
is required to build gcc-cross but if gcc-cross is reconstructed from
the sstate cache then gcc-cross's dependency on binutils-cross is
ignored due to being in the safe dependency list in
setscene_depvalid.)
Signed-off-by: Mike Crowe <mac@mcrowe.com>
Acked-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
Fix QA warnings seen when using 'traditional' kernel recipes e.g.:
WARNING: QA Issue: .../recipes-kernel/linux/linux_3.0.18.bb: Variable
FILES is set as not being package specific, please fix this.
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently the names used for the kernel in deploy will contain "AUTOINC"
instead of the final incremental numbering. This fixes the problem by
ensuring data is obtained from the PR service and using the PKG* variables
instead of PE/PV/PR directly.
[YOCTO #4293]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
when we have multilib and a 64bit machine and initramfs then image/lib
directory will not be created and it will end up with errors like
| DEBUG: Executing python function sstate_task_prefunc
| DEBUG: Python function sstate_task_prefunc finished
| DEBUG: Executing shell function do_deploy
| tar: lib: Cannot stat: No such file or directory
| tar: Exiting with failure status due to previous errors
So in order to overcome this shortcoming lets mkdir -p the
lib directory so the modules can be happily installed.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Allow recipes to specify sections to be stripped from the kernel output
using KERNEL_IMAGE_STRIP_EXTRA_SECTIONS. For example:
KERNEL_IMAGE_STRIP_EXTRA_SECTIONS = ".comment .unwanted"
The kernel output is stripped in place.
Since the toolchain does not give indication when the specified sections
are absent, we read the sections first and make this report by issuing a
warning to the developer.
The toolchain by default strips the image with the -s option (even
when -s is not specified):
-s --strip-all Remove all symbol and relocation information
For example, these sections are always removed:
.debug_aranges
.debug_info
.debug_abbrev
.debug_line
.debug_frame
.debug_str
.debug_loc
.debug_ranges
.symtab
.strtab
In addition to these, the sections listed in
KERNEL_IMAGE_STRIP_EXTRA_SECTIONS will also be removed.
Only stripping of vmlinux (elf) is supported at this time. A warning
will be given if the image type is not vmlinux.
Stripping the image could also be done in the kernel, but that would
only work for linux-yocto based kernels, so it's not the route we
decided to go.
[YOCTO 3515]
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Michel Thebeau <michel.thebeau@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
do_sizecheck has a few issues especially with vmlinux image type.
It breaks because KERNEL_OUTPUT is a path relative to ${B}. When
do_sizecheck runs it does not find the file (because the working
directory is elsewhere) and does not fail.
Also, the image file referenced by KERNEL_OUTPUT may be a link.
Finally, when do_sizecheck deletes the oversized kernel image it leaves
the previously run do_compile task with inaccurate status.
So, do the following:
- specify that the working directory should be ${B}
- use ls -L to reference to the real file, and ensure that the link
file is created
- keep the oversized image file so the status of do_compile is valid
[YOCTO #3514]
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Michel Thebeau <michel.thebeau@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* During the base kernel_do_deploy function the directory is
changed to DEPLOYDIR in order to do some cleanup and symlinking.
However, the directory is not changed back to the original
starting directory ${S} at the end. For append functions this
means that the starting directory is not ${S} as expected but
instead ${DEPLOYDIR}.
For functions like the do_deploy_append in
recipes-kernel/linux/linux-dtb.inc there is an assumption that
you are still in the source directory and not the DEPLOYDIR.
Without this change the .dtb files are not copied because the
check for the existence of ${DTS_FILE} which is a relative
path from the ${S} directory fails. This means that the .dtb
files are not copied into the deploy directory and subsequently
the deploy/images directory.
In the log.do_deploy file you will see lines like:
Warning: arch/arm/boot/dts/xxxxx.dts is not available!
Signed-off-by: Chase Maupin <Chase.Maupin@ti.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
* this way we can reuse the same functionality also for external modules
including module_autoload_foo and module_conf_foo functionality
* MODULE_PACKAGES variable was removed (splited modules are now returned
by do_split_packages
* KERNEL_MODULES_META_PACKAGE is used to append all splitted packages
to RDEPENDS. In kernel.bbclass it's old "kernel-modules" in
module.bbclass it defaults to ${PN} for upgrade path from
single PN with all modules to PN depending on all new kernel-module-*
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
With recent kernel versions, some ARM configurations need may fail to
build with errors like this:
multiple load addresses: 0x80008000 0x80008000
This is incompatible with uImages
Specify LOADADDR on the commandline to build an uImage
We cannot pass this information in EXTRA_OEMAKE, as
"meta/classes/kernel.bbclass" explicitly ignores all EXTRA_OEMAKE
settings. So add KERNEL_EXTRA_ARGS parameter so affected boards
can add for example
KERNEL_EXTRA_ARGS = "LOADADDR=0x80008000"
to their configuration. We make this general enough so other needed
command line options can be passed as well when compiling the kernel.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Previously the build path to STAGING_KERNEL_DIR was being embedded into the
package post install scripts. We avoid this behavior by generating a special
depmodwrapper script. This script contains that hard-coded path, ensuring
that re-use of the sstate-cache (and/or packages) will always run through the
wrapper generated by the current build with a checksum that includes
STAGING_KERNEL_DIR.
[ YOCTO #3962 ]
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When running depmod, from kmod, warnings about missing modules.order
and modules.builtins occur. Looking back it is not clear why these
files were being ignored, other then nobody knew what to do with them.
Add the files into the kernel-base package, if they exist, ensuring
they will always get installed if at least one module is installed.
[ YOCTO #3963 ]
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
Prepending to populate_packages is rather ugly and means its hard to trace
errors and also profiling informaiton is summed together in one function.
This patch starts to split out the prepends to become separate functions
to avoid these issues. This is generally a neater way to write functions
than prepending to where there can sometimes be variable scope issues
and we've been bitten by whitespace issues in the past.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Small performance tweaks for populate_packages_prepend:
* Compile the regexps once at the start
* Don't keep importing a module which is already imported
* No need to check PKG is set, we'd have failed long before now if it wasn't
* Don't export PATH, bitbake takes care of this at the task level
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
kernel_do_install() populates $kerneldir with files needed to build
external modules. To accomplish this there are several copy commands
to get source from the kernel source tree and build trees after which
a 'clean' is performed. Since we are copying from the build tree we
get about 1G of .o and .cmd files copied over only to have them
removed when we clean. This adds additional IO overhead as well as
overhead caused by pseudo. By avoiding copying these files in the
first place we get multiple gains:
* avoid initial copy
* avoid file deletes during clean
* reduce pseudo overhead
Additionally we are making use of cpio vs cp which tends to be
significantly faster at performing copies.
With these changes I observe a 15-30% decrease in the time to complete
the do_install() operation on the kernel.
[YOCTO #3517]
Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
kernel compiler is not special and we currently have it so
we want to pass -march and -mtune options as CFLAGS to kernel
build so that compiler picks the right subarch flags when
compiling assembly files in particular. Otherwise defaults
are chosen which may not be right in many case e.g. when
compiling kernel for collie machine we should use arch=armv4
but it uses toolchain/as defaults which is armv5te
in some case e.g. thumb1 we know that kernel can not be compiled
in thumb1 mode so we can provide that information e.g. -marm
option through KERNEL_HOST_CC_ARCH variable as we do now
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Since update-modules is now obsolete, remove it from the bbclass.
[YOCTO #3598]
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* http://cgit.openembedded.org/openembedded-core/commit/?id=67e77cd948c69798695d5d5f34163ef89339b35a
moved DATETIME dependency from MODULE_TARBALL_BASE_NAME to MODULE_IMAGE_BASE_NAME
but without copying that vardepsexclude
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Define the additional variable MODULE_IMAGE_BASE_NAME since some
layers (like meta-ti) would find it handy for multi-kernel
definitions in that they wouldn't have to define their own
variable representing exactly the same thing.
Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
* kernels <2.6.30 don't have $kerneldir/tools directory
* so we got this error:
* sed: can't read ...SNIP...linux/2.6.30-r0/image/usr/src/kernel/tools/perf/Makefile: No such file or directory
* ERROR: Function failed: do_install
Signed-off-by: Marco Cavallini <m.cavallini@koansoftware.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
* without this you it's harder to see which kernel is newer version
e.g. after branch change LOCALCOUNT in SRCPV is reset to 0, so it needs PE bump for upgrade path
but then it's clear why gitr1+HASH has newer DATETIME then gitrN+HASH, so include PE in name
* it will sort alphabetically wrong in deploy dir for a while until old images are removed
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
The modules.tar.gz archive is sometimes of no value and it can take
a noticeable time to build if many modules were enabled in the kernel
configuration. The extraneous file also contributes to deploy/ clutter
and is a waste of disk space.
Allow it to be suppressed by setting MODULE_TARBALL_DEPLOY="0".
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
link to latest
* so it will work look KERNEL_IMAGE
* also we were recreating modules.tgz with every kernel build, but
overwritting the same output file
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
|
The symlink from uImage-3... to uImage is not created at image creation
time and not properly update on kernel upgrades. This is fixed by removing
the chdir. The other users of update-alternative do not change the directory
before calling it.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* kernel-modules is always added to PACKAGES later in python code and
needed to be defined as PACKAGES_DYNAMIC
* add it to PACKAGES directly and set
ALLOW_EMPTY_kernel-modules
FILES_kernel-modules
DESCRIPTION_kernel-modules
outside populate_packages_prepend like for other packages and set only
RDEPENDS_kernel-modules from python code
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* bitbake uses PACKAGES_DYNAMIC as regexp
^ could make matching faster (and it will be more clear that we're expecting regexp not glob)
* made all those last '-' optional, use .* (or nothing)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The compilation routine for the kernel has an explicit call to
build version.h, which works fine for most kernels, but the
location of it has recently changes.
commit d183e6f5 [UAPI: Move linux/version.h]
commit 10b63956 [UAPI: Plumb the UAPI Kbuilds into the user
header installation and checking]
moves the file to include/generated/linux/version.h and then to
include/generated/uapi/linux/version.h.
As a result kernel builds of 3.7 or bisection builds of intermediate
kernel commits will fail with:
make[2]: *** No rule to make target `include/linux/version.h'. Stop.
Making the explicit version.h build conditional on the version, or
via a file test would fix the problem, but it introduces some complexity
to the build.
Even without an explicit call to build version.h, it is always produced
by the kernel build, so it can simply be removed.
This extra make line was originally so that the kernel version could be
determined, so that then different instructions could be executed depending
on whether it was a 2.4 or 2.6 kernel. Since we no longer support 2.4, this
code is no longer needed.
[YOCTO: #3293]
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The update-modules mechanism is something of a historical relic and it isn't
entirely clear that it has a great deal of value nowadays. Also, it causes a
problem when building a read-only rootfs since update-modules itself refuses
to configure offline.
Allow DISTROs to circumvent this whole thing by declaring (via DISTRO_FEATURES)
that they don't wish to use update-modules. This is backfilled for existing
distributions and will have to be marked as CONSIDERED by those who actually
don't want it.
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
|
