Richard Purdie [Thu, 1 Sep 2016 21:10:21 +0000 (22:10 +0100)]
runqemu: Enable virtio RNG for all platforms
We have problems where systems simply stop booting and hang. This is due
to a lack of entropy which means ssh keys and networking can't be brought
up. Adding in the virtio-rng passthrough support allows host entropy to
pass into the guess and avoids these hangs.
This is particularly problematic after the gnutls upgrade which starts
using /dev/random instead of /dev/urandom but was an issue we'd occasionally
seem before that.
It particualrly affected x86 and ppc machines for some reason.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Tue, 6 Sep 2016 13:46:10 +0000 (14:46 +0100)]
runqemu: Allow unique network interface MAC addresses
Current qemu instances all share the same MAC address. This shouldn't be an
issue as they are all on separate network interfaces, however on the slight
chance this is causing problems, its easy enough to ensure we use unique
MAC addresses based on the IP numbers we assign.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
runqemu: fix run from testimage with non-standard DEPLOY_DIR_IMAGE
testimage.bbclass uses runqemu to execute runtime tests on a qemu
target, this means that bitbake is already running and `bitbake -e`
can't be called to obtain bitbake variables.
runqemu tries to work around being unable to read values for
bitbake variables by inferring the MACHINE from the
DEPLOY_DIR_IMAGE setting, however if a user sets that variable in
a manner which doesn't follow the systems expectations (i.e. if
running `bitbake -c testimage` against a directory of pre-generated
images in a user-specified path) the inferring of the MACHINE name
from the DEPLOY_DIR_IMAGE location will fail.
It's possible that check_arg_machine() shouldn't cause runqemu to
fail and that runqemu should proceed with the user-supplied value
even if it can't be verified. This patch simply ensures that a
workflow where the user sets DEPLOY_DIR_IMAGE continues to work
without changing too much of the runqemu code.
[YOCTO #10238]
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
runqemu: fixes for when invoked during a bitbake run
When runqemu is invoked from a running bitbake instance it will be
unable to call `bitbake -e` due to the lock held by the calling
bitbake instance.
Our test code sets an OE_TMPDIR environment variable from which we
can infer/guess paths. Add code to do so when self.bitbake_e can't
be set, much as the sh version of runqemu did.
[YOCTO #10240]
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
runqemu: better handle running on a host with different paths
If the STAGING_*_NATIVE directories from the config file don't exist
and we're in a sourced OE build directory try to extract the paths
from `bitbake -e`
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
runqemu: assume artefacts are relative to *.qemuboot.conf
When runqemu is started with a *.qemuboot.conf arg assume that image
artefacts are relative to that file, rather than in whatever
directory the DEPLOY_DIR_IMAGE variable in the conf file points to.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Thu, 25 Aug 2016 14:41:05 +0000 (07:41 -0700)]
runqemu: refactor it and remove machine knowledge
Previously, runqemu had hard coded machine knowledge, which limited its
usage, for example, qemu can boot genericx86, but runqemu can't, we need
edit runqemu/runqemu-internal a lot if we want to boot genericx86.
Now bsp conf files can set vars to make it can be boot by runqemu, and
qemuboot.bbclass will save these info to DEPLOY_DIR_IMAGE/qemuboot.conf.
Please see qemuboot.bbclass' comments on how to set the vars.
* Re-write it in python3, which can reduce lines from 1239 to about 750
lines
* All the machine knowledges are gone
* All of the TUN_ARCH knowledge are gone
* All the previous options are preserved, and there is a new way to run
runqemu: (it doesn't need run "bitake -e" in such a case)
$ runqemu tmp/deploy/images/qemux86
or:
$ runqemu tmp/deploy/images/qemuarm/<image>.ext4
or:
$ runqemu tmp/deploy/images/qemuarm/qemuboot.conf
* Fixed audio support, not limited on x86 or x86_64
* Fix SLIRP mode, add help message, avoid mixing with tap
* Fix NFS boot, it will extract <image>.tar.bz2 or tar.gz to
DEPLOY_DIR_IMAGE/<image>-nfsroot when no NFS_DIR, and remove it after
stop.
* More bsps can be boot, such as genericx86 and genericx86-64.
* The patch for qemuzynq, qemuzynqmp, qemumicroblaze has been sent to
meta-xilinx' mailing list.
* I can't find any qemush4 bsp or how to build it, so it is not
considered atm.
Robert Yang [Tue, 6 Sep 2016 06:15:05 +0000 (23:15 -0700)]
qemuarm.conf: set PREFERRED_VERSION_linux-yocto
The base_version_less_or_equal() will raise errors if
PREFERRED_VERSION_linux-yocto is None. For example, when we build
DISTRO = "nodistro", PREFERRED_VERSION_linux-yocto is not be defined
since it is defined in poky.conf, and then bitbake will
choose the higher version which is 4.8 currently, so set
PREFERRED_VERSION_linux-yocto to 4.8, otherwise, runqemu can't boot it.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Tue, 30 Aug 2016 22:56:38 +0000 (23:56 +0100)]
qemuarm: Add DTB file new kernel
For kernels after 4.7, we need to ensure the DTB file for the kernel is
used by runqemu. Doing this conditionally based upon the kernel verison
being built seems to be the only way forward for this.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
ERROR: qemu-2.7.0-r1 do_install_ptest_base: Function failed:
do_install_ptest_base
...
cp: cannot stat '...tmp/work/core2-64-poky-linux/qemu/2.7.0-r1
/qemu-2.7.0/tests/Makefile: No such file or directory
...
Commit 46e7b70699d8bf4db08c8bb5111974318dd5416d in qemu project
renamed tests/Makefile to tests/Makefile.include, we apply the
same change in recipe accordingly to fix this issue.
Fixes [YOCTO #10245]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Thu, 8 Sep 2016 10:26:48 +0000 (03:26 -0700)]
qemu: 2.6.0 -> 2.7.0
This upgrade can fix a qemuppc + openssh bug, the ssh connection maybe
refused or closed randomly, and it's not easy to reproduce. RP pointed
that this upgrade can fix the problem, and it does work in my local
testing.
* Update add-ptest-in-makefile.patch
* Drop backported patch 0001-configure-support-vte-2.91.patch
Here is the Changlog:
http://wiki.qemu.org/ChangeLog/2.7
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fabio Berton [Fri, 26 Aug 2016 19:55:02 +0000 (16:55 -0300)]
watchdog-config: Add recipe
Provides configuration files for watchdog.
Add watchdog-config as a runtime dependence of watchdog and remove
watchdog.conf file from watchdog installation.
Signed-off-by: Fabio Berton <fabio.berton@ossystems.com.br> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fabio Berton [Fri, 26 Aug 2016 19:55:01 +0000 (16:55 -0300)]
watchdog: Add wd_keepalive package
This is a simplified version of the watchdog daemon. It only opens
/dev/watchdog, and keeps writing to it often enough to keep the kernel
from resetting, at least once per minute. Each write delays the reboot
time another minute. After a minute of inactivity the watchdog hardware
will cause a reset. In the case of the software watchdog the ability to
reboot will depend on the state of the machines and interrupts.
Installs wd_keepalive binary and enable initscript.
Signed-off-by: Fabio Berton <fabio.berton@ossystems.com.br> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Thu, 8 Sep 2016 18:48:35 +0000 (14:48 -0400)]
linux-yocto: update LINUX_VERSION to -rc5
The SRCREVs were previously updated to -rc5, but the LINUX_VERSION
was missed. As such, we are building and booting -rc5, but all the
packaging says -rc4.
Worth a quick update while we wait for -rc6
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Jussi Kukkonen [Thu, 1 Sep 2016 06:53:04 +0000 (09:53 +0300)]
xserver-nodm-init: Deprecate /etc/X11/Xserver
This commit should provide the same functionality as before, but
should make meta-oe xserver-nodm-init-2.0 obsolete as well as
keep systemd and sysvinit startup better in sync.
/etc/X11/Xserver is not called anymore: it is provided by both
x11-common and xserver-common with no useful differences (but some
annoying ones). Instead xserver-nodm-init provides
/etc/xserver-nodm/Xserver as the startup script and
/etc/default/xserver-nodm as the default settings file. These are
used by both init systems.
The Xserver script could be completely removed (with sysv and
systemd calling xinit directly), but to keep compatibility with
meta-oes xserver-nodm-init-2.0 the Xserver script sources
/etc/X11/xserver-common if one exists -- and systemd EnvironmentFile
cannot do that.
x11-common used to have a packageconfig to easily control screen
blanking. Move this to xserver-nodm-init.
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
André Draszik [Thu, 18 Aug 2016 07:56:25 +0000 (08:56 +0100)]
kernel-module-split.bbclass: generate dependencies across recipes
The information retrieved via depmod is incomplete with
regards to kernel modules that are dependencies, in
particular where two kernel modules are built from
different source trees / recipes, which leads to incomplete
dependency information for packages created.
So far, our packages created didn't contain dependencies on
packages created by other recipes, as we solely use depmod
for that, and depmod can only work well after *all* kernel
modules have been copied into one place - it doesn't work
well in a staged approach.
Now that all .ko have correct dependency information at packaging
time, we can use that information to properly track dependencies
across recipies, and can combine the information from the
.modinfo elf section with the information from depmod.
Signed-off-by: André Draszik <git@andred.net> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
André Draszik [Thu, 18 Aug 2016 07:56:24 +0000 (08:56 +0100)]
module.bbclass: use Module.symvers for dependants
When compiling multiple external kernel modules, where one
depends on the other, there are two problems at the
moment:
1) we get compile time warnings from the kernel build
system due to missing symbols (from modpost).
2) Any modules generated are missing dependency
information (in the .modinfo elf section) for any
dependencies outside the current source tree and
outside the kernel itself.
This is expected, but the kernel build system has a way to
deal with this - the dependent module is expected to
specify KBUILD_EXTRA_SYMBOLS (as a space-separated list)
to point to any and all Module.symvers of kernel modules
that are dependencies.
While 1) by itself is not really a big issue, 2) prevents
the packaging process from generating cross-source tree
package dependencies.
As a first step to solve the missing dependencies in
packages created, we:
1) install Module.symvers of all external kernel module
builds (into a location that is automatically packaged
into the -dev package)
2) make use of KBUILD_EXTRA_SYMBOLS and pass the location
of all Module.symvers of all kernel-module-* packages
we depend on
This solves both problems mentioned above.
Signed-off-by: André Draszik <git@andred.net> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Thu, 8 Sep 2016 01:08:49 +0000 (21:08 -0400)]
linux-yocto/4.4/4.8: uvesafb: provide option to specify timeout for task completion
Integrating the following patch:
[
We try to make this change a generic extension, but it is
actually for a corner case. When a VM (qemu) gets a very limited
cpu bandwidth from host, which could be under a heavy load, the
existing 5000 ms timeout could occur and trigger error messages
in the task function's callers.
This change adds a new timeout parameter so that we can tweak
the value as a workaround or for troubleshooting purposes. In
the infinite wait case, A warning message is printed at 5000ms
interval.
In real world, the current 5 sec is generous enough for a video
request in my opinion, so this change could not be very useful.
Bruce Ashfield [Thu, 8 Sep 2016 01:08:47 +0000 (21:08 -0400)]
kernel-yocto: restore missing configuration meta data
Some of the meta-data from the 4.4 kernel was missing from the 4.8
branch. This resulted in some functionality drops and also a size/time
increase in the kernel build (due to debug being turned on).
With this resync, we now have the missing config restored.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Thu, 8 Sep 2016 01:08:46 +0000 (21:08 -0400)]
kernel-yocto: allow --allnoconfig and --alldefconfig as KCONFIG_MODES
Previously merge_config.sh was wrapped by the configme script, configme
took the different KCONFIG_MODES as options, and used --allnoconfig
or --alldefconfig.
With the switch to merge_config.sh no longer being wrapped, the new
processing wasn't matching the existing values and only supported
allnoconfig or alldefconfig.
To avoid breaking existing layers, and also keep any working that
have already switched, we can make the case statement match both.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Thu, 8 Sep 2016 01:08:45 +0000 (21:08 -0400)]
kernel-yocto: restore kernel-meta data detection for SRC_URI elements
Before the kernel tools were simplified and streamlined, there was code
which not only migrated a patch/cfg/scc to the kernel build tree, it
also migrated any subdirectories of those patches.
The effect of this data migration was that any other meta data in
a patch's directory structure would be available for processing.
While we don't want to do this migration anymore, it is possible to
check the path of any SRC_URI patches, and if they include a "kernel-meta"
subdirectory add it to the search path.
This restores the functionality without the old complexity.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Mon, 5 Sep 2016 23:24:44 +0000 (11:24 +1200)]
recipetool: create: avoid extra blank lines in output recipe
If we output extra blank lines (because of some automated editing) then
it makes the output recipe look a bit untidy. You could argue that we
should simply have the editing code not do that, but sometimes we don't
have enough context there for that to be practical. It's simple enough
to just filter out the extra blank lines when writing the file, so just
do it that way.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Paul Eggleton [Mon, 5 Sep 2016 03:58:01 +0000 (15:58 +1200)]
recipetool: create: support node.js code outside of npm
If you have your own node.js application you may not publish it (or at
least not immediately) in an npm registry - it might just be in a
repository on github or on your local machine. Add support to recipetool
create for creating recipes to build such applications - extract their
dependencies, fetch them, and add corresponding npm:// URLs to SRC_URI,
and ensure that LICENSE / LIC_FILES_CHKSUM are updated to match. For
example, you can now run:
(I had to borrow some code from bitbake/lib/bb/fetch2/npm.py to
implement this functionality; this should be refactored out but now
isn't the time to do that refactoring.)
Part of the fix for [YOCTO #9537].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Paul Eggleton [Mon, 5 Sep 2016 22:06:01 +0000 (10:06 +1200)]
recipetool: create: allow license variable handling to be rerun
If you make adjustments to the source tree (as create_npm.py will be)
then you will need to re-run the license variable handling code at the
end so that we get all of the files that should go into
LIC_FILES_CHKSUM if nothing else. Split out the license variable
handling to a separate function in order to allow this.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Paul Eggleton [Thu, 1 Sep 2016 02:25:04 +0000 (14:25 +1200)]
recipetool: create: fix mapping python dependencies to python-dbg package
When trying to map python module dependencies to the packages that
provide them, if we're looking for .so files that satisfy
dependencies then we need to exclude files found under the .debug
directory, otherwise the dependency will get mapped to the python-dbg
package which isn't correct.
For example, this fixes creating a recipe for pyserial and not getting
python-fcntl in RDEPENDS_${PN}, leading to errors when trying to use the
serial module on the target.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Paul Eggleton [Wed, 31 Aug 2016 23:38:46 +0000 (11:38 +1200)]
devtool: update-recipe: support files with subdir=
It's rare but there are recipes that have individual files (as opposed
to archives) in SRC_URI using subdir= to put them under the source tree,
the examples in OE-Core being bzip2 and openssl. This broke devtool
update-recipe (and devtool finish) because the file wasn't unpacked into
the oe-local-files directory and thus when it came time to update the
recipe, the file was assumed to have been deleted by the user and thus
the file was erroneously removed. Add logic to handle these properly so
that this doesn't happen.
(We still have another potential problem in that these files become part
of the initial commit from upstream, which could be confusing because
they didn't come from there - but that's a separate issue and not one
that is trivially solved.)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Paul Eggleton [Wed, 31 Aug 2016 03:28:46 +0000 (15:28 +1200)]
lib/oe/patch: handle non-UTF8 encoding when reading patches
When extracting patches from a git repository with PATCHTOOL = "git" we
cannot assume that all patches will be UTF-8 formatted, so as with other
places in this module, try latin-1 if utf-8 fails.
This fixes UnicodeDecodeError running devtool update-recipe or devtool
finish on the openssl recipe.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Jack Mitchell [Wed, 7 Sep 2016 11:20:05 +0000 (12:20 +0100)]
file: build with c std as c99
when using a toolchain not shipped by OE core such as linaro we
can't be sure what the std will be set to. Set to compile as c99
which is the lowest version supported.
Signed-off-by: Jack Mitchell <jack@embed.me.uk> Signed-off-by: Ross Burton <ross.burton@intel.com>
Robert Yang [Wed, 7 Sep 2016 08:28:36 +0000 (01:28 -0700)]
pciutils: fix PACKAGECONFIG
The PACKAGECONFIG's value doens't go into EXTRA_OECONF, but
PACKAGECONFIG_CONFARGS.
Fixed:
pciutils-3.5.1: libpci rdepends on libudev, but it isn't a build dependency, missing eudev in DEPENDS or PACKAGECONFIG? [build-deps]
pciutils-3.5.1: pciutils rdepends on libudev, but it isn't a build dependency, missing eudev in DEPENDS or PACKAGECONFIG? [build-deps]
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Markus Lehtonen [Wed, 7 Sep 2016 07:28:45 +0000 (10:28 +0300)]
oeqa.buildperf: try harder when splitting 'nevr' string
Try to be more intelligent when splitting out recipe name, epoch,
version and revision from the buildstat directory name. Previous
assumption was that package versions never contain a dash but obviously
that is not necessarily true. The new assumption is that the package
version starts with a number.
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
systemd_230.bb: Set journal RuntimeMaxSize to 64M as default
At this time systemd journald uses the /run tmpfs to store logs
by default systemd uses 15% of available space [1] of the /run
partition, when the space runs out journald starts to vaccum/store
the logs into /var/log [1].
It causes two problems one of them is timeout dev-ttySN.device's
when enable debug and use journal as systemd.log_target [2] the other
is related to don't find syslog entries into the journal log [3].
This problems are now more evident because i recently enabled the
systemd debug option in testimage [4].
One area of improvement will be add support in systemd journald to
read these parameters from the kernel cmdline like systemd.log_target,
if the support exists we could add that parameter at level of testimage.
Stefan Agner [Sun, 4 Sep 2016 20:25:55 +0000 (13:25 -0700)]
busybox: avoid circular dependency when using initramfs
The kernel does not automatically mount devtmpfs when using initramfs
based booting (even when using CONFIG_DEVTMPFS_MOUNT). If the rootfs
is built with USE_DEVFS=1 (which is the default), the system ends up
with a completely empty /dev to begin with.
Busybox uses the first entry in inittab slightly different than
other init systems:
<id>: WARNING: This field has a non-traditional meaning for BusyBox init!
The id field is used by BusyBox init to specify the controlling tty for
the specified process to run on. The contents of this field are
appended to "/dev/" and used as-is.
Since /dev/null is not there yet, Busybox throws errors instead of
executing the commands, and hence never mounts devtmpfs:
init started: BusyBox v1.24.1 (2016-09-04 11:53:14 PDT)
can't open /dev/null: No such file or directory
can't open /dev/null: No such file or directory
can't open /dev/null: No such file or directory
can't open /dev/null: No such file or directory
can't open /dev/null: No such file or directory
can't open /dev/null: No such file or directory
can't open /dev/null: No such file or directory
Avoid this circular dependency by not specifing <id>. With that
Busybox ends up using the stdio of the init process and executes
the inittab just fine.
Signed-off-by: Stefan Agner <stefan@agner.ch> Signed-off-by: Ross Burton <ross.burton@intel.com>
Marek Vasut [Sun, 4 Sep 2016 16:20:20 +0000 (18:20 +0200)]
libcap: Replace EXTRA_OECONF with PACKAGECONFIG_CONFARGS
When building libcap and DISTRO_FEATURES does not contain pam,
the build will fail on missing pam headers. This is because the
bits from EXTRA_OECONF moved to PACKAGECONFIG_CONFARGS and thus
the necessary options are not propagated to oe_runmake anymore.
Replace EXTRA_OECONF with PACKAGECONFIG_CONFARGS to fix this.
Add a simple test to validate that the number of files in the
destination matches the number of files in the source after the
copyhardlinktree() has been performed.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The change to preserve extended attributes in copytree() and
copyhardlinktree() (e591d69103a40ec4f76d1132a6039d9cb1555103)
resulted in an incorrect cp invocation in copyhardlinktree() when
the source directory contained hidden files.
This was because the passed src was modified in place but some code
paths expected it to remain unmodified from the passed value.
Resolve the issue by constructing a new source string, rather than
modifying the passed in string.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Zhixiong Chi [Mon, 5 Sep 2016 04:08:13 +0000 (12:08 +0800)]
rpm: ensure rpm2cpio call rpm relocation code
We need to call rpmcliInit to ensure the rpm relocation code is called.
when we allow rpm2cpio to be relocatable, The adjusted path used to find
the macro files was being built into the binary and this path was valid
for the machine it was built on and some of our other build machines,
but invalid on some others, and was not being properly overridden at
runtime.
when we export the wrsdk and source the sdk, then execute rpm2cpio xxx.rpm|cpio -t.
we will get the following error :
"rpm-5.4.14/rpmdb/dbconfig.c:493:
db3New: Assertion `dbOpts != ((void *)0) && *dbOpts != '\0'' failed.
Signed-off-by: Zhixiong Chi <Zhixiong.Chi@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
André Draszik [Fri, 2 Sep 2016 22:04:41 +0000 (23:04 +0100)]
ofono: RRECOMMENDS tun.ko & APN database
- kernel-module-tun is needed so that ofono can create the
ppp network interface
- mobile-broadband-provider-info is needed as an explicit
dependency even though it is in DEPENDS, because it's
just an xml database, and the DEPENDS simply allows
ofono to figure out its location in the file system
(using pkg-config during configure). But there is no
shared library dependency or so for bitbake to figure
out this runtime dependency.
We make it a recommendation only, so that it can still
be removed from filesystem images in case people build
images that don't need the provider database (and e.g.
hard-code APNs for specific use-cases)
Signed-off-by: André Draszik <git@andred.net> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
While comparing what were supposed to be similar
filesystems from different build machines, some issues
have been noticed in the e2fsprogs recipe, in
particular with the compile_et and mk_cmds utilities.
1) target:
move compile_et and mk_cmds into the -dev package
Both are development tools, from the man pages:
compile_et - error table compiler
compile_et converts a table listing error-code names
and associated messages into a C source file suitable
for use with the com_err(3) library.
mk_cmds - error table compiler
mk_cmds converts a table listing command names and
associated help messages into a C source file suitable
for use with the ss(3) library.
2) native/nativesdk
Also apply cleaning of host path (build directory) here,
so that only the sysroot directory remains, which is
properly adjusted by the sstate handling.
3) make cleaning of host path actually work
The existing sed command wasn't working, in particular
for compile_et; we fix up the sed command so that
removal of references to the local build directory
really works. Do the same changes for mk_cmds, for
consistency.
Signed-off-by: André Draszik <git@andred.net> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We enabled HW_RANDOM_VIRTIO for the 4.4+ kernels, but it is also needed
for 4.1 to ensure that VMs have sufficient entropy. Without this entropy
networking on qemuppc starves and triggers intermittent errors.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Wed, 31 Aug 2016 14:54:11 +0000 (10:54 -0400)]
oeqa/parselogs/qemuarm: Whitelist amba and jitter for 4.8+ kernels
With the update to the 4.8 kernel the versatile platform (and hence
qemuarm) has switched to a device tree boot.
We are using an ummodified mainline kernel versatilepb device tree,
which includes definitions of multiple amba devices. These devices
are not present in the qemu system emulation, hence throw warnings
during boot.
These warnings are not unique to oe-core, and rather than carry kernel
patches to the device tree (for now), we whitelist the known warnings
so qa testing will pass. We also can't turn amba off completely, since
it is providing valid devices (like the serial port) and AMBA is
force selected by other kconfig values.
We also have a jitterentropy warning that shows up on some hosts.
This warning is harmless, and like amba we can't turn it off in a
fragment since it is force selected by crypto (and we'd rather not
turn all crypto off). So we add it to the whitelist while investigations
continue into what is needed in the host to support this fully.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
do_kernel_configme calls merge_config.sh (installed in the sysroot by
the kern-tools-native recipe) which may invoke the compiler to complete
the configuration process.
Depending on the build (and dependencies), this may error due to sysroot poisoning [1].
The errors are similar to:
make[1]: Entering directory '4.1+gitAUTOINC+a7e53ecc27-r0/linux-x64-standard-build' HOSTCC scripts/basic/fixdep
work-shared/x64/kernel-source/scripts/basic/fixdep.c:106:23: fatal error: sys/types.h: No such file or directory
compilation terminated.
make[2]: *** [work-shared/x64/kernel-source/scripts/basic/Makefile:22: scripts/basic/x86_64-nilrt-linux-fixdep] Error 1
Adding $TOOLCHAIN_OPTIONS to $CFLAGS before calling merge_configs.sh
fixes the error because $TOOLCHAIN_OPTIONS defines the sysroot and make
uses it to correctly compile & fill all missing kernel config options.
Bruce Ashfield [Tue, 30 Aug 2016 16:49:05 +0000 (12:49 -0400)]
kernel-yocto: test for empty artifacts
With the updated kernel tools, we generate a list of sccs, patches,
configs and BSP definitions as part of the meta data generation.
It is valid if there aren't any of these artifacts found (i.e. you
are just building a branch and a default config), but invoking the
tools with no inputs isn't a good idea.
To avoid this issue, we generate a string based on the artifacts
and skip calling the tools if there's nothing to do.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Tue, 30 Aug 2016 16:49:04 +0000 (12:49 -0400)]
linux-yocto/4.8: add qemuarm device tree specification
4.7+ requires a device tree for the arm versatile family of platforms.
We add the definition to our 4.8 linux-yocto recipes so we can continue
to boot!
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Tue, 30 Aug 2016 16:49:00 +0000 (12:49 -0400)]
perf: adapt to Makefile.config
commit 4842576cd857 [perf tools: Move config/Makefile into Makefile.config]
relocated the configuration Makefile of perf. As such, we need to adapt
our fixup routines to work with the Makefile no matter where it is.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ed Bartosh [Thu, 1 Sep 2016 08:56:03 +0000 (11:56 +0300)]
toaster: fire TaskArtifacts event
Fire TaskArtifact MetaData event for deployment tasks when task
either completed or skipped. Event contains full task id
(recipe+task) and list of deployment artifacts from sstate
manifest.
This should allow Toaster to always get notified about deployment
artifacts produced by the build.
[YOCTO #9869]
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Fri, 2 Sep 2016 10:22:41 +0000 (11:22 +0100)]
populate_sdk_ext: Put populate_sdk_ext under sstate control
Adding populate_sdk task to SSTATE_TASKS should make sstate machinery
to generate manifest for deployed ext sdk artifacts and do final deployment
to SDK_DEPLOY.
This is done in a similar way to do_populate_sdk in a previous patch.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
