]> code.ossystems Code Review - openembedded-core.git/log
openembedded-core.git
4 years agoclasses/sanity: Bump minimum python version to 3.5
Joshua Watt [Mon, 5 Oct 2020 15:38:51 +0000 (10:38 -0500)]
classes/sanity: Bump minimum python version to 3.5

Bumps the minimum python version to 3.5 to match bitbake and the test
matrix

Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agolibtools-cross/shadow-sysroot: Use nopackages inherit
Richard Purdie [Thu, 1 Oct 2020 15:50:25 +0000 (16:50 +0100)]
libtools-cross/shadow-sysroot: Use nopackages inherit

When testing pseudo changes I realised these recipes have packaging tasks
but don't generate packages. Drop the packages tasks for cleanliness.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoglibc: do_stash_locale must not delete files from ${D}
Richard Purdie [Wed, 30 Sep 2020 15:34:08 +0000 (16:34 +0100)]
glibc: do_stash_locale must not delete files from ${D}

do_stash_locale doesn't run in fakeroot context, do_install does. We therefore
shouldn't delete files that do_install has added or it leaves potentially
problemtic entries in the fakeroot database. Leaving the files around
doesn't change or break anything else.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosystemd: Use ROOTPREFIX without suffixed slash in systemd.pc.in
Khem Raj [Wed, 30 Sep 2020 04:54:09 +0000 (21:54 -0700)]
systemd: Use ROOTPREFIX without suffixed slash in systemd.pc.in

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoweston-init: Add environment file support for systemd unit file
Fabio Berton [Wed, 30 Sep 2020 21:16:44 +0000 (18:16 -0300)]
weston-init: Add environment file support for systemd unit file

The /etc/default/weston file is already installed and it can be used
as systemd EnvironmentFile.

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>
4 years agoparted: improve ptest
Ross Burton [Thu, 1 Oct 2020 16:11:11 +0000 (17:11 +0100)]
parted: improve ptest

Add a RRECOMMENDS on kernel-module-scsi-debug as this module is needed
for many of the tests.

Create the udev mount blacklist directory before writing into it, as it
doesn't always exist.

Delete any existing log files before calling make, as otherwise the
tests cannot be repeated.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agocoreutils: improve coreutils-ptest RDEPENDS
Ross Burton [Thu, 1 Oct 2020 16:11:10 +0000 (17:11 +0100)]
coreutils: improve coreutils-ptest RDEPENDS

Add perl-modules to the RDEPENDS to avoid having to list all perl
modules that are required, as this package isn't going to be installed
outside of testing builds.

Remove libmodule-build-perl from RDEPENDS, it appears that this isn't
needed anymore. With and without this package the test results on my
image are the same: 619 tests, 462 pass, 154 skip, 3 fail.  This
*drastically* reduces the build impact of enabling ptest as
packagegroup-core-build-essential (thus target gcc) is in the dependency
chain.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobjam-native: don't do debug builds
Ross Burton [Tue, 29 Sep 2020 22:16:37 +0000 (23:16 +0100)]
bjam-native: don't do debug builds

Previously this recipe was changed to do debug builds because otherwise insane
warns that the binary is already stripped.

However, debug builds for boost.build also pass -O0.  It turns out that given
how large Boost is (or, how bad boost.build is) doing a release build with -O3
knocks a third off the walltime for a Boost package in my test, mainly by reducing
how long it spends deciding that nothing needs to be rebuilt in do_install:

  PKG      TASK        ABSDIFF  RELDIFF  WALLTIME1 -> WALLTIME2
  boost    do_install  -330.7s   -69.2%     477.6s -> 146.9s
  boost    do_compile    -7.1s    -2.7%     269.3s -> 262.2s

Replace debug mode with INSANE_SKIP=already-stripped.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokernel-fitimage: generate openssl RSA keys for signing fitimage
Usama Arif [Wed, 30 Sep 2020 10:48:00 +0000 (11:48 +0100)]
kernel-fitimage: generate openssl RSA keys for signing fitimage

The keys are only generated if they dont exist. The key
generation can be turned off by setting FIT_GENERATE_KEYS to "0".
The default key length for private keys is 2048 and the default
format for public key certificate is x.509.

Signed-off-by: Usama Arif <usama.arif@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agooeqa/selftest/devtool: Add sync call to test teardown
Richard Purdie [Tue, 29 Sep 2020 16:56:49 +0000 (17:56 +0100)]
oeqa/selftest/devtool: Add sync call to test teardown

Devtool tests are heavy on IO and if bitbake can't write out its caches, we see
timeouts. Call "sync" around the tests to ensure the IO queue doesn't get too
large, taking any IO hit here rather than in bitbake shutdown.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agotestsdk.py: remove workspace/sources to avoid failure in case of multilib
Chen Qi [Tue, 29 Sep 2020 03:22:15 +0000 (11:22 +0800)]
testsdk.py: remove workspace/sources to avoid failure in case of multilib

When multilib is enabled, there are multiple environment scripts, and the
test cases for eSDK are executed for each environment script.

And we will have the following problem when executing test cases for the
second environment script.

  ERROR: Source tree path /.../workspace/sources/librdfa already exists and is not empty

So after executing test cases for one environment, we clean up the sources
diretory to avoid such failure.

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agopopulate_sdk_ext.bbclass: add ESDK_MANIFEST_EXCLUDES
Chen Qi [Tue, 29 Sep 2020 03:22:14 +0000 (11:22 +0800)]
populate_sdk_ext.bbclass: add ESDK_MANIFEST_EXCLUDES

Add ESDK_MANIFEST_EXCLUDES to enable excluding items in sdk-conf-manifest.

By default, files under conf/ are all added to sdk-conf-manifest, as the
manifest file is set to 'conf/*'.

However, there are situations where some configuration files under conf/
directory are not intended to be added to sdk-conf-manifest, thus adding
ESDK_MANIFEST_EXCLUDES to enable users to do this.

This variable takes the form of glob matching.
e.g.
ESDK_MANIFEST_EXCLUDES = "conf/autogen*"
This would exclude all files under conf/ starting with 'autogen' from
sdk-conf-manifest.

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agooeqa/selftest/cases/devtool.py: avoid .pyc race
Tim Orling [Tue, 29 Sep 2020 04:49:05 +0000 (21:49 -0700)]
oeqa/selftest/cases/devtool.py: avoid .pyc race

In certain conditions, most likely under heavy load on the
AutoBuilder, the prebuilt .pyc files are attempting to be
executed before they have been completely copied. Avoid
this by not copying the .pyc files (nor the __pycache__
directory). The impact of python3-native recreating the .pyc
files should hopefully be negligible.

YOCTO#13421
YOCTO#13803

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agodebianutils: update the debian snapshot version
Mingli Yu [Tue, 29 Sep 2020 06:19:20 +0000 (14:19 +0800)]
debianutils: update the debian snapshot version

After debianutils upgrades to 4.11.1 in [1], there is no
4.11.1 source in the old debian snapshot. Update the snapshot
version to fix the gap.

[1] https://git.openembedded.org/openembedded-core/commit/?id=0c492a0768cd15ff40db35f459853e69c55f8cc6

Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agocve-check: add CVE_CHECK_REPORT_PATCHED variable to suppress reporting of patched...
Chris Laplante [Tue, 29 Sep 2020 15:57:46 +0000 (11:57 -0400)]
cve-check: add CVE_CHECK_REPORT_PATCHED variable to suppress reporting of patched CVEs

Default behavior is not changed. To suppress patched CVEs, set:

        CVE_CHECK_REPORT_PATCHED = ""

Signed-off-by: Chris Laplante <chris.laplante@agilent.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agocve-check: introduce CVE_CHECK_RECIPE_FILE variable to allow changing of per-recipe...
Chris Laplante [Tue, 29 Sep 2020 15:57:45 +0000 (11:57 -0400)]
cve-check: introduce CVE_CHECK_RECIPE_FILE variable to allow changing of per-recipe check file

The addition of this variable also makes it possible to change the
output suffix of the check files, e.g. in local.conf:

CVE_CHECK_MANIFEST_append = ".txt"
CVE_CHECK_RECIPE_FILE_append = ".txt"

Signed-off-by: Chris Laplante <chris.laplante@agilent.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoboost: move the build directory outside of S
Ross Burton [Tue, 29 Sep 2020 22:16:38 +0000 (23:16 +0100)]
boost: move the build directory outside of S

Instead of manually creating and deleting build directories, follow the idioms
by setting B to WORKDIR/build, setting do_configure[cleandirs], and using ${B}
where appropriate.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosystemd: Drop 0023-Fix-field-efi_loader_entry_one_shot_stat-has-incompl.patch
Khem Raj [Tue, 29 Sep 2020 20:14:30 +0000 (13:14 -0700)]
systemd: Drop 0023-Fix-field-efi_loader_entry_one_shot_stat-has-incompl.patch

This patch is no longer needed as upstream has added including part of
this include list, perhaps thats all is what was needed to make it
portable

Refresh 0001-Handle-missing-gshadow.patch

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoMap license names PSF and PSFv2 to PSF-2.0
Mark Jonas [Tue, 29 Sep 2020 18:06:59 +0000 (20:06 +0200)]
Map license names PSF and PSFv2 to PSF-2.0

According to SDPX.org "Python License 2.0" is the overall Python
license, which is comprised of several licenses. "Python Software
Foundation License 2.0" is part of the complete Python license, but is
also used independently by some projects.

So far the license names PSF and PSFv2 found in LICENSE are mapped to
to Python-2.0. This patch maps PSF and PSFv2 to PSF-2.0 und thus
corrects the impression that Python-2.0 and PSF-2.0 are synonymous.

Signed-off-by: Mark Jonas <toertel@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoAdd license text for PSF-2.0
Mark Jonas [Tue, 29 Sep 2020 18:06:58 +0000 (20:06 +0200)]
Add license text for PSF-2.0

Signed-off-by: Mark Jonas <toertel@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agotarget/ssh.py: Add dump_target support
Saul Wold [Mon, 28 Sep 2020 21:23:13 +0000 (14:23 -0700)]
target/ssh.py: Add dump_target support

This adds the dump_target support when the ssh command fails with
a 'No route to host'. This is will provide additional data when a
Qemu target fails to respond during autobuilder testing. This does
not fix 14002 [0], but may help track down why qemu looses networking

[0] https://bugzilla.yoctoproject.org/show_bug.cgi?id=14002

Signed-off-by: Saul Wold <saul.wold@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agotestimage: Add testimage_dump_target to kwargs
Saul Wold [Mon, 28 Sep 2020 21:23:12 +0000 (14:23 -0700)]
testimage: Add testimage_dump_target to kwargs

This passes the list of commands to run on the OEQemuTarget when
the TargetDumper needs to run in a test context due to a failure
on the target.

This is added here as a kwargs because the 'd' dictionary is not
available in the staticmethod getTarget in the
OERuntimeTestContextExecutor class. The OEQemuTarget is different
from the QemuTarget which already uses the list of commands from
testimage_dump_target from 'd'. The create_dir() is needed to
initialize the TargetDumper's dump_dir variable.

Signed-off-by: Saul Wold <saul.wold@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agolibdrm: fix build failure
akuster [Mon, 28 Sep 2020 19:10:51 +0000 (12:10 -0700)]
libdrm: fix build failure

Failure seen on my CentOS7 build host

Signed-off-by: akuster <akuster@mvista.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agorpm: add PACKAGECONFIG for the systemd inhibit plugin
Ross Burton [Mon, 28 Sep 2020 18:43:28 +0000 (19:43 +0100)]
rpm: add PACKAGECONFIG for the systemd inhibit plugin

RPM ships a systemd inhibit plugin, that will tell systemd to inhibit a
reboot or sleep during a package upgrades.

For native RPM this is entirely useless, and for target it's only useful
if you're using systemd+logind+rpm.

This plugin uses DBus which means it depends on expat -> cmake-native ->
libarchive-native curl-native, which is quite a dependency tree to need
in early build (required to build packages via rpm-native).

It was previously forcibly disabled for native packages but the build
dependency on DBus remained.  Add a PACKAGECONFIG for the plugin that is
only enabled for target builds with systemd and explicitly disabled for
native/nativesdk builds, but also keep the explicit disabling of all
plugins as the prioreset plugin also behaves badly inside a build.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosudo: upgrade 1.9.2 -> 1.9.3
zangrc [Fri, 25 Sep 2020 17:30:03 +0000 (01:30 +0800)]
sudo: upgrade 1.9.2 -> 1.9.3

Signed-off-by: Zang Ruochen <zangrc.fnst@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agopango: upgrade 1.46.1 -> 1.46.2
zangrc [Fri, 25 Sep 2020 17:29:53 +0000 (01:29 +0800)]
pango: upgrade 1.46.1 -> 1.46.2

Signed-off-by: Zang Ruochen <zangrc.fnst@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agostress-ng: upgrade 0.11.19 -> 0.11.21
zangrc [Fri, 25 Sep 2020 17:29:31 +0000 (01:29 +0800)]
stress-ng: upgrade 0.11.19 -> 0.11.21

Signed-off-by: Zang Ruochen <zangrc.fnst@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobind: upgrade 9.16.5 -> 9.16.7
zangrc [Sun, 27 Sep 2020 13:13:07 +0000 (21:13 +0800)]
bind: upgrade 9.16.5 -> 9.16.7

Signed-off-by: Zang Ruochen <zangrc.fnst@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosysklogd: fix parallel build issue
Changqing Li [Mon, 28 Sep 2020 05:25:17 +0000 (13:25 +0800)]
sysklogd: fix parallel build issue

fix below errors which caused by race condition:
do_package:
ERROR: dwarfsrcfiles failed with exit code 1 (cmd was ['dwarfsrcfiles', 'TOPDIR/tmp-glibc/work/ppc7400-oe-linux/sysklogd/2.1.2-r0/package/usr/lib/libsyslog.a']):
dwarfsrcfiles: TOPDIR/tmp-glibc/work/ppc7400-oe-linux/sysklogd/2.1.2-r0/package/usr/lib/libsyslog.a: not a valid ELF file

do_compile:
arm-oe-linux-gnueabi-libtool: link: (cd ".libs" && rm -f "libsyslog.so.0" && ln -s "libsyslog.so.0.0.0" "libsyslog.so.0")
arm-oe-linux-gnueabi-libtool: link: (cd ".libs" && rm -f "libsyslog.so" && ln -s "libsyslog.so.0.0.0" "libsyslog.so")
arm-oe-linux-gnueabi-libtool: link: arm-oe-linux-gnueabi-gcc-ar cru .libs/libsyslog.a  libsyslog_la-syslog.o ../lib/pidfile.o ../lib/strlcpy.o ../lib/strlcat.o
TOPDIR/tmp-glibc/work/armv7vet2hf-neon-oe-linux-gnueabi/sysklogd/2.1.2-r0/recipe-sysroot-native/usr/bin/arm-oe-linux-gnueabi/../../libexec/arm-oe-linux-gnueabi/gcc/arm-oe-linux-gnueabi/10.1.0/ar: `u' modifier ignored since `D' is the default (see `U')
TOPDIR/tmp-glibc/work/armv7vet2hf-neon-oe-linux-gnueabi/sysklogd/2.1.2-r0/recipe-sysroot-native/usr/bin/arm-oe-linux-gnueabi/../../libexec/arm-oe-linux-gnueabi/gcc/arm-oe-linux-gnueabi/10.1.0/ar: ../lib/strlcat.o: No such file or directory

Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoSpace-comma Cleanups
Jon Mason [Mon, 28 Sep 2020 13:05:46 +0000 (09:05 -0400)]
Space-comma Cleanups

Multiple files have " ," instead of ", " in expressions.  This changes
them to conform to the way the rest of them are done.

Found and corrected via:
git ls-files | xargs sed --follow-symlinks -i 's/ ,d/, d/g'

Signed-off-by: Jon Mason <jon.mason@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobitbake.conf: add name of multiconfig to BUILDCFG_HEADER when multiconfig is active
Chris Laplante [Mon, 28 Sep 2020 15:18:53 +0000 (11:18 -0400)]
bitbake.conf: add name of multiconfig to BUILDCFG_HEADER when multiconfig is active

Signed-off-by: Chris Laplante <chris.laplante@agilent.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosstate: set mode explicitly when creating directories in sstate-cache
Ross Burton [Mon, 28 Sep 2020 16:19:00 +0000 (17:19 +0100)]
sstate: set mode explicitly when creating directories in sstate-cache

When creating directories in the sstate-cache, explicitly set the mode
passed to mkdir to 0775 so that the directories are group writable, as
otherwise they cannot be shared with other users.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agorpm: disable libarchive use
Ross Burton [Mon, 28 Sep 2020 16:18:56 +0000 (17:18 +0100)]
rpm: disable libarchive use

libarchive is only needed for the rpm2archive tool, as this is of
limited use and libarchive is a non-trivial build dependency make this
optional and disabled by default.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoutils: respect scheduler affinity in cpu_count()
Ross Burton [Mon, 28 Sep 2020 16:18:55 +0000 (17:18 +0100)]
utils: respect scheduler affinity in cpu_count()

cpu_count() returns multiprocessing.cpu_count() but that is simply returns
os.cpu_count() so we could use that directly.

However this returns the number of CPUs on the host, not the number of
usable CPUs on the host. If the user is using scheduler affinity then
the number of usable CPUs may be less, so when determining how many cores
we can use check the affinity instead.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoglibc: make nscd optional
Ross Burton [Mon, 28 Sep 2020 16:18:54 +0000 (17:18 +0100)]
glibc: make nscd optional

NSCD itself is already packaged into its own package, but if it is never
going to be needed then it can be disabled and it won't be called from
glibc at all.  Add a PACKAGECONFIG for NSCD that is enabled by default.

As a side effect if NSCD is disable, glibc with and without systemd is
binary identical, which helps sstate reuse.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agostaging: Ensure cleaned dependencies are added
Richard Purdie [Sat, 26 Sep 2020 11:00:49 +0000 (12:00 +0100)]
staging: Ensure cleaned dependencies are added

Most recipe-sysroot dependencies are handled by these prefuncs. "configure"
is special with a decidated task, prepare_recipe_sysroot which runs
beforehand.

do_prepare_recipe_sysroot does not have to be run before/after
fetch/unpack/patch, they're independent tasks. If fetch/unpack/patch
have sysroot dependencies and those tasks rerun, stale items from the
sysroot could be uninstalled and since prepare_recipe_sysroot doesn't
re-run, they could be missing by the time configure runs.

Fix this by adding the prefunc for do_configure which would ensure
the sysroot is in a good state.

This issue can be reproduced by modifying do_unpack and watching
binutils-native and vulkan-headers failing due to missing makeinfo
or ninja.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoptest-runner: Backport patch to fix inappropriate ioctl error
Khem Raj [Sat, 26 Sep 2020 14:08:26 +0000 (07:08 -0700)]
ptest-runner: Backport patch to fix inappropriate ioctl error

The srcrev bump is actually bring single commit [1] on top of 2.4.0 which fixes ptest
runs with messges like

ERROR: Unable to detach from controlling tty, Inappropriate ioctl for device

[1] https://git.yoctoproject.org/cgit/cgit.cgi/ptest-runner2/commit/?id=834670317bd3f6e427e1ac461c07ada6b8936dfd

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agooeqa/runtime : add test for RTC(Real Time Clock)
Teoh Jay Shen [Fri, 25 Sep 2020 00:25:42 +0000 (08:25 +0800)]
oeqa/runtime : add test for RTC(Real Time Clock)

This test is checking the functionality of the RTC(Real Time Clock). The Check_if_RTC_(Real_Time_Clock)_can_work_correctly manual test case from oeqa/manual/bsp-hw can be replace by this runtime test.

Signed-off-by: Teoh Jay Shen <jay.shen.teoh@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agogo-mod.bbclass: use append to add `modcacherw`
Peter A. Bigot [Sat, 26 Sep 2020 10:53:09 +0000 (05:53 -0500)]
go-mod.bbclass: use append to add `modcacherw`

This class provided default build flags but includes the base go class
which also provides default flags.  Use a different mechanism to
extend the default flags without discarding `-trimpath` from the base
class.

Signed-off-by: Peter A. Bigot <pab@pabigot.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokernel.bbclass: ensure symlink_kernsrc task gets run even with externalsrc
Rasmus Villemoes [Fri, 25 Sep 2020 14:05:44 +0000 (16:05 +0200)]
kernel.bbclass: ensure symlink_kernsrc task gets run even with externalsrc

Commit c5dfc2586b41 (kernel.bbclass: run do_symlink_kernsrc before
do_patch) fixed a race between do_symlink_kernsrc and
do_populate_lic. However, I missed the fact that when
externalsrc.bbclass is in use, the do_patch task doesn't exist,
meaning that do_symlink_kernsrc now doesn't get run at all, breaking
the build.

We need this task to run regardless of whether do_patch exists or not,
so reinstate the configure->symlink_kernsrc dependency explicitly.

Fixes: c5dfc2586b41 (kernel.bbclass: run do_symlink_kernsrc before
do_patch)
Reported-by: Chanho Park <parkch98@gmail.com>
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosstate.bbclass: Check file ownership before doing 'touch -a'
Norman Stetter [Fri, 25 Sep 2020 11:49:43 +0000 (13:49 +0200)]
sstate.bbclass: Check file ownership before doing 'touch -a'

In contrast to 'touching' a file without parameters 'touch -a' can only be
performed by the file owner. In case of a shared sstate-cache served
locally (e.g. over NFS),  where the user running bitbake is not the
owner of sstate-cache files, even if he has write access on group level,
the sstate_unpack_package task will fail.
Checking if the file is owned by the user before attempting to run
'touch -a' on it solves this.

Signed-off-by: Norman Stetter <norman.stetter@garz-fricke.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agocortex-m0.inc: Add tuning for cortex-m0
Jonathan Richardson [Wed, 23 Sep 2020 22:31:09 +0000 (15:31 -0700)]
cortex-m0.inc: Add tuning for cortex-m0

Add tuning for ARM cortex-m0.

Signed-off-by: Jonathan Richardson <jonathan.richardson@broadcom.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoopenssh: Allow enable/disable of rng-tools recommendation on sshd
Otavio Salvador [Thu, 24 Sep 2020 19:58:00 +0000 (16:58 -0300)]
openssh: Allow enable/disable of rng-tools recommendation on sshd

We are adding a new PACKAGECONFIG option ('rng-tools') to control if we
wish the openssh-sshd to RRECOMMENDS the 'rng-tools' package. We are
enabling it by default so there is no behavior change.

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokernel-yocto: add KBUILD_DEFCONFIG search location to failure message
Bruce Ashfield [Thu, 24 Sep 2020 19:31:14 +0000 (15:31 -0400)]
kernel-yocto: add KBUILD_DEFCONFIG search location to failure message

It was pointed out that since the check for the existence of
KBUILD_DEFCONFIG knows where it is looking, that should be part of
the error/log message.

We only look in one location for the defconfigs, so add where we
checked to the message.

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agodhcpcd: set service to conflict with connman
Yi Zhao [Thu, 24 Sep 2020 02:41:14 +0000 (10:41 +0800)]
dhcpcd: set service to conflict with connman

As the default network manager, connman has its own internal DHCP
implement. If run dhcpcd and connman simultaneously, they may conflict
with each other.

Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosystemd: Upgrade v246.2 -> v246.6
Andrew Geissler [Wed, 23 Sep 2020 18:14:47 +0000 (13:14 -0500)]
systemd: Upgrade v246.2 -> v246.6

0001-Use-PREFIX-ROOTPREFIX-correctly.patch was fixed fix the following:
https://github.com/systemd/systemd/pull/17009

2ee1c57c4f (tag: v246.6) kernel-install/90-loaderentry: fix when /boot is not mountpoint
a2a8802abe homed: make clean that --storage=directory --image-path=/dev/some-block-device is not supported
e1df274a76 homed: make sure our worker processes finish before we exit
f2763ff6cd nspawn: fix fd leak on failure path
6f9c8af500 nspawn: check return of setsid()
31875b3cb0 dissect: is_loop_device() returns negative on error, don't mistake that is true
418cf0e78c dissect: always invalidate secondary arch partitions if we found primary arch
1d17c00260 util: wireguard is merged into upstream kernel
48a308b76a repart: don't mark image files generated with --empty=create executable
9353607ef3 man: substitute path to rc.local in the man page
f2b9345cd2 meson: RC_LOCAL_SCRIPT_PATH_START to RC_LOCAL_PATH
b0fff9012a socket-proxy: allow localhost addresses
5e8bf33de1 sd-bus: allow localhost addresses
ad77a988f2 fstab-generator: add 'nofail' when  NFS 'bg' option is used
d08f6fddc0 busctl: add missing shortopt -l
64d58480ac socket-util: reset length argument for second getsockopt call in fd_[gs]et_rcvbuf
0046c58334 test: use pclose() for popen()
689d2e061b core: propagate unit start limit hit state to triggering path unit
2f53d690ea core: propagate triggered unit in more load states
a06cf9a5ba bootctl: handle if LoaderSystemToken is invalid for some reason
e2c0023dd2 test-fileio: test test strings with shell, too
251b96c9b5 test-env-file: add test that tests our env file read + writer + shell against each other
4d1922d8bb env-file: bring our decoding of double-quoted strings in env files in line with shell
281581cd74 test-fileio: use test strings that are actually valid in shell
366d38d8ed man: document that RemainAfterElapse=no means that also the triggered unit needs to deactivate
d0f0f048ec (tag: v246.5) sd-path: use ROOTPREFIX without suffixed slash
e662cf6d51 hashmap: make sure to initialize shared hash key atomically
da310c6b45 socket: downgrade log warnings about inability to set socket buffer sizes
ab6fcd9135 core: fix securebits setting
4f6925484d capability-util: add new function for raising setpcap
771436884d network: do not add prefix to RA if radv is not configured
fb2afc5f30 man: document the random delay of persistent timers
b2006ddc8f test-network: add test for ENOBUFS issue #17012
8758580ef5 backlight: do not claim that ID_BACKLIGHT_CLAMP= property is not set
57fc184a6c fs-util,tmpfiles: fix error handling of fchmod_opath()
db0f031e70 bootctl: don't accidentally propagate errors in "bootctl status"
3e2c806681 ethtool-util: don't pass error value that isn't used to log_syntax
b671730edb network: don't fail on various config parse errors
0ad86030c5 man: document that sd_bus_message_close_container() may only be called at end of container
f3da018017 cryptsetup: Fix null pointer dereference (#16987)
6f65eaf9c2 core: fix set keep caps for ambient capabilities
08338a234e core: fix comments on ambient capabilities
f0e6d9876d network: make log_link_error() or friends return void
35766dc61b core: make log_unit_error() or friends return void
3ed10b2ee8 core/slice: explicitly specify return value
2f6406914b udev: do not discard const qualifier
07671aa4cc sd-device: make log_device_error() or friends return void
d4bea73972 udev: explicitly specify return value
7db399be1e udev: return negative errno for invalid EVDEV_ABS_XXX= property
8c8d188e85 udev: make log_rule_error() or friends return void
4921375fd3 socket: fix copy/paste error
0f7fd97749 udev: warn if failed to set buffer size for device monitor
fc763d38d8 network: increase receive buffer size for device monitor
3bf7797f1f network: do not start device monitor if /sys is read-only
ebc0729c6a network: honor the buffer size specified in networkd.socket
ef3d2e178b core/socket: use fd_set_{rcv,snd}buf()
5dd4cc4b10 sd-device-monitor: use fd_set_rcvbuf()
fe9b92e566 util: introduce fd_set_{snd,rcv}buf()
4dcae66688 util: try to set with SO_{RCV,SND}BUFFORCE when requested size is larger than the kernel limit
4b6b523946 util: refuse to set too large value for socket buffer size
b4be8edb45 network: ignore error on increasing netlink receive buffer size
5ce47fb491 tree-wide: if get_block_device() returns zero devno, check for it in all cases
8ea6ec18e7 btrfs: if BTRFS_IOC_DEV_INFO returns /dev/root generate a friendly error message
e1ff4947d2 basic/log: make log_{info,warning,...} return void
8019995e9a tree-wide: correct cases where return log_{error,warning} is used without value
932f4c3e8b test-execute/exec-dynamicuser-statedir.service: fix quoting
16b9426f70 man: fix quickhelp listing in sysusers.d(5)
bde903d9e9 network: fix NDisc handling for the case when multiple routers exist
c965063b64 network: expose route_{hash,compare}_func()
6d24a40669 network: expose address_{hash,compare}_func()
054838a2e0 util: expose in6_addr_{hash,compare}_func()
58bd4a70de network: fixes gateway assignment through DHCPv4
8ad5382fe3 bash-completion: resolvectl: support 'log-level' command
a98bd75072 resolvectl: add 'log-level' to help message
78262fe807 core/socket: we may get ENOTCONN from socket_instantiate_service()
fecb3f00c4 homed: remember the secret even when the for_state is FIXATING_FOR_ACQUIRE
dc2e82af33 (tag: v246.4) core: create per-user inaccessible node from the service manager
0b3c497347 nspawn,pid1: pass "inaccessible" nodes from cntr mgr to pid1 payload via /run/host
2239965c29 coredump: don't convert s â†’ Âµs twice
61d29b7f8c firstboot: fill empty color if ansi_color unavailable from os-release
9678a3daf6 varlink: do not parse invalid messages twice
4e516dcbc1 userdbctl: add forgotten --output mode in help
aee20dfbd8 shared/{user,group}-record-nss: adjust filtering of "valid" passwords
5933d77afe doc: cross link sd_listen_fd() docs a bit
97fdde3239 Rework how we cache mtime to figure out if units changed
0500968241 core: always try to reload not-found unit
8ae22f0d64 pid1: use the cache mtime not clock to "mark" load attempts
715507c277 core: rename manager_unit_file_maybe_loadable_from_cache()
20ad76d0a7 man: document fd ownership for sd-bus fd marshalling
38ae73fafd resolved: make sure we initialize t->answer_errno before completing the transaction
a1ba0fbef6 homed: fix log message to honor real homework path
d6b1e659b3 src/shared/dissect-image.c: fix build without blkdid (#16901)
e42f9add21 analyze: fix error handling in one case
4804ce1488 units: add missing usb-gadget.target
5ad4e68c37 man: extend on the usec/sec discrepancy
2fb612371d login/logind: Include sys/stat.h for struct stat usage
5e884e7ee0 partition/makefs: Include missing sys/file.h header
7bbc3807da network: dhcp6: logs only new address
2056429e0f Don't run test-repart when loop devices are not available
dcbea51c5a man: clarify that several networkctl commands takes device names
16e4cfcc82 networkctl: label command does not take any argument
2352921244 missing: Add new Linux capability
8b29c4a4f9 tty-ask-pw-agent: properly propagate error
f7ce2e9839 tty-ask-pw-agent: the message string might not be set
29cba5c9ef tty-ask-pw-agent: make sure "--list" works correctly
e1ce367d73 add "list" verb to autocompleted commands
1f4cb5da1e shell-completion/zsh: add missing verbs for networkctl
a4236a2764 path: Improve $PATH search directory case
b7cef386bd (tag: v246.3) path: Skip directories when finalising $PATH search
122945f315 rules: don't install 80-drivers.rules when kmod is disabled
42fab2d454 zsh: correct journalctl command completion parsing
ed3f97f962 basic/missing_syscall: fix syscall numbers for arm64 :(
ba6e7f7c46 shared/install: fix preset operations for non-service instantiated units
d39f139348 nss-resolve: treat BUS_ERROR_NO_SUCH_UNIT the same as SD_BUS_ERROR_SERVICE_UNKNOWN too
9bb3e64d71 various: treat BUS_ERROR_NO_SUCH_UNIT the same as SD_BUS_ERROR_SERVICE_UNKNOWN
6d802dd596 man: drop reference to long gone .busname unit type
a29656804b man: fix a fix of a typo in systemd.service example
21ce0f5b33 network: can: Fix CAN initialization
cab5cde8c9 man: update autogenerated dbus api lists
0d8000522b man: fix invalid tag place
ea94f218be man: add conditionals to more man pages
ef91325349 meson: add ENABLE_ANALYZE conditional
83f7c0a7ec core: add missing conditions/asserts to unit file parsing
716718155d analyze: rework condition testing
5c4c7581bc sd-bus: fix error handling on readv()
6cd058f305 user-runtime-dir: deal gracefully with missing logind properties
6a2d73638d shared/seccomp: do not use ifdef guards around textual syscall names
7355ac9689 machine-id-setup: don't use KVM or container manager supplied uuid if in chroot env
496a71f444 man: Fix typo in systemd-tmpfiles
6c5d216ad8 homework: downgrade chattr failure log message
1708f06a00 homework: explicitly close cryptsetup context, to not keep loopback device busy
a21eaa2a3a homework: correct error passed into log message
3a2d169f36 homework: sync everything to disk before we rename LUKS loopback file into place
84e1ab74d2 homed: downgrade quota message in containers
8b62cadf36 analyze-security: do not assign badness to filtered-out syscalls
29854a5437 NEWS: clarify two points
4cb4fb82f7 meson: add min version for libfdisk
76331f86f6 load-fragment: fix grammar in error messages
1e53c2d70f Fix function description in logind man page
669066564d network: do not fail if UseMTU=yes on DHCP lease lost
a2a3f16cdc missing_syscall: do not use function name that may conflict with glibc
4091dcd469 missing_syscall: fix pidfd_{send_signal,open} numbers for alpha
7875daf52b network: wait for previous address removal before configuring static addresses
120064b4a1 network: only process non-error message
f44ec1de15 test: accept that char device 0/0 can now be created witout privileges

Signed-off-by: Andrew Geissler <geissonator@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobash: fix CVE-2019-18276
De Huo [Thu, 24 Sep 2020 02:39:44 +0000 (10:39 +0800)]
bash: fix CVE-2019-18276

An issue was discovered in disable_priv_mode in shell.c in GNU Bash
through 5.0 patch 11. By default, if Bash is run with its effective UID
not equal to its real UID, it will drop privileges by setting its
effective UID to its real UID. However, it does so incorrectly. On Linux
and other systems that support "saved UID" functionality, the saved UID
is not dropped. An attacker with command execution in the shell can use
"enable -f" for runtime loading of a new builtin, which can be a shared
object that calls setuid() and therefore regains privileges. However,
binaries running with an effective UID of 0 are unaffected.

Get the patch from [1] to fix the issue.

[1] https://git.savannah.gnu.org/cgit/bash.git/commit/?h=devel&id=951bdaa

Signed-off-by: De Huo <De.Huo@windriver.com>
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agogettext: Fix ptest failure
Khem Raj [Thu, 24 Sep 2020 05:22:54 +0000 (22:22 -0700)]
gettext: Fix ptest failure

msgmerge-29 is a script without execute permissions but executed as
./msgmerge-29 which results in this test failing.

It is only one test which fails on gettext/qemux86-64

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoqemuboot: Add QB_RNG variable
Khem Raj [Thu, 24 Sep 2020 01:25:06 +0000 (18:25 -0700)]
qemuboot: Add QB_RNG variable

RNG passthru has been enabled on all qemu machines but its being added
to each one of them, with this patch its turned into QB variables which
defaults to host passthru, yet it can be overridden if needed via
machine or config metadata if needed.

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agopackagegroup: rrecommend perf also for musl on ARM
Christian Eggers [Thu, 24 Sep 2020 09:32:17 +0000 (11:32 +0200)]
packagegroup: rrecommend perf also for musl on ARM

The perf recipe builds fine for musl on ARM.

Fixes: d758a4445a ("packagegroup: Disable packages not available on musl")
Signed-off-by: Christian Eggers <ceggers@arri.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agomaintainers.inc: Add Zang Ruochen and Wang Mingyu for several recipes
Wang Mingyu [Thu, 24 Sep 2020 14:15:50 +0000 (22:15 +0800)]
maintainers.inc: Add Zang Ruochen and Wang Mingyu for several recipes

Zang Ruochen is responsible for the maintenance as follows:
elfutils
libvorbis
ell
python3-dbus
python3-pbr
python3-pip
python3-pycairo
python3-pygobject
python3-six

Wang Mingyu is responsible for the maintenance as follows:
btrfs-tools
dtc
iso-codes
libpciaccess
libpipeline
librepo
liburcu
libyaml
msmtp

Signed-off-by: Wang Mingyu <wangmy@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobuild-appliance-image: Update to master head revision
Richard Purdie [Wed, 23 Sep 2020 19:56:05 +0000 (20:56 +0100)]
build-appliance-image: Update to master head revision

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agossh-pregen-hostkeys: Add a recipe with pregenerated ssh host keys
Richard Purdie [Wed, 23 Sep 2020 11:53:12 +0000 (12:53 +0100)]
ssh-pregen-hostkeys: Add a recipe with pregenerated ssh host keys

Host keys are getting bigger and taking an ever increasing amount of time
to generate. Whilst we do need to test that works, we don't need to test
it in every image. Add a recipe which can be added to images with
pre-generated keys, allowing us to speed up tests on the autobuilder
where it makes sense to.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agooeqa/selftest/containerimage: Update to match assumptions in configuration
Richard Purdie [Wed, 23 Sep 2020 17:12:42 +0000 (18:12 +0100)]
oeqa/selftest/containerimage: Update to match assumptions in configuration

Assumptions in the test mean it doesn't work with ssh-pregen-hostkeys.
It also doesn't work with systemd. Update the configuration to make
sure neither of these effect the test.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agooeqa/selftest/incompatible_lib: Fix append usage
Richard Purdie [Wed, 23 Sep 2020 16:28:42 +0000 (17:28 +0100)]
oeqa/selftest/incompatible_lib: Fix append usage

It's pure luck this has worked so far, add a missing space to the append.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agodhcpcd: add dhcpcd user to support priviledge separation
Yi Zhao [Wed, 23 Sep 2020 09:45:47 +0000 (17:45 +0800)]
dhcpcd: add dhcpcd user to support priviledge separation

The dhcpcd enables privsep by default. It requires a user added to the
system. Add dhcpcd user and group to support it.

Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agodhcpcd: set --runstatedir to /run
Yi Zhao [Wed, 23 Sep 2020 09:45:46 +0000 (17:45 +0800)]
dhcpcd: set --runstatedir to /run

Set --runstatedir to /run/dhcpcd rather than /var/run/dhcpcd

Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agodhcpcd: pass --dbdir to EXTRA_OECONF to set database directory
Yi Zhao [Wed, 23 Sep 2020 09:45:45 +0000 (17:45 +0800)]
dhcpcd: pass --dbdir to EXTRA_OECONF to set database directory

Set --dbdir to /var/lib/dhcpcd rather than /var/db/dhcpcd to satisfy FHS
compliance

Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobuildtools-tarball: Fix conflicts with oe-selftest and other tooling
Richard Purdie [Tue, 22 Sep 2020 22:23:10 +0000 (23:23 +0100)]
buildtools-tarball: Fix conflicts with oe-selftest and other tooling

OECORE_NATIVE_SYSROOT is used by tools like oe-run-native and hence
we were seeing selftest failures when newer buildtools-tarballs that
use this were run on the autobuilder.

Unset the variable after use to avoid these issues.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agometa: add/fix invalid Upstream-Status tags
Ross Burton [Tue, 22 Sep 2020 14:42:26 +0000 (15:42 +0100)]
meta: add/fix invalid Upstream-Status tags

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agolinux-yocto/beaglebone: Switch to sdhci-omap driver
Bruce Ashfield [Tue, 22 Sep 2020 13:11:33 +0000 (09:11 -0400)]
linux-yocto/beaglebone: Switch to sdhci-omap driver

Integrating the following commit(s) to linux-yocto/.:

    541066c5b1e beaglebone: Switch to sdhci-omap driver

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokea: fix conflict between multilibs
Kai Kang [Tue, 22 Sep 2020 07:02:33 +0000 (15:02 +0800)]
kea: fix conflict between multilibs

There are conflict of config files between kea and lib32-kea:

| Error: Transaction test error:
|  file /etc/kea/kea-ctrl-agent.conf conflicts between attempted installs of
     lib32-kea-1.7.10-r0.core2_32 and kea-1.7.10-r0.core2_64
|  file /etc/kea/kea-dhcp4.conf conflicts between attempted installs of
     lib32-kea-1.7.10-r0.core2_32 and kea-1.7.10-r0.core2_64

Because they are all commented out, replace the expanded libdir path with
'$libdir' in the config files to avoid conflict.

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosystemd: disable xdg-autostart generator by default
Kai Kang [Tue, 22 Sep 2020 07:02:32 +0000 (15:02 +0800)]
systemd: disable xdg-autostart generator by default

xdg-autostart generator is an opt-in feature which is imported by commit:

470c58ba45 systemd: Upgrade v245.6 -> v246

The generator outputs unexpect log info on terminal when log in a image
which includes an X desktop such as Xfce. So add an package config and
disable it by default.

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agogettext: no need to depend on bison-native
Ross Burton [Mon, 21 Sep 2020 20:27:55 +0000 (21:27 +0100)]
gettext: no need to depend on bison-native

This was needed because of a backported patch, this patch was dropped
in oe-core c9f8b48 but the removal of the dependency was missed.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agocurl: Change SRC_URI from http to https
Randy MacLeod [Mon, 21 Sep 2020 18:24:25 +0000 (14:24 -0400)]
curl: Change SRC_URI from http to https

The official links on:
   https://curl.haxx.se/download.html
use https now and we're seeing this warning:
   WARNING: curl-native-7.72.0-r0 do_fetch: Failed to fetch URL http://curl.haxx.se/download/curl-7.72.0.tar.bz2, attempting MIRRORS if available

Signed-off-by: Randy MacLeod <Randy.MacLeod@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agomesa: update 20.1.6 -> 20.1.8
Anibal Limon [Mon, 21 Sep 2020 16:11:22 +0000 (11:11 -0500)]
mesa: update 20.1.6 -> 20.1.8

Signed-off-by: Aníbal Limón <anibal.limon@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agopackagegroups: remove strace and lttng-tools for rv32/musl
Khem Raj [Thu, 17 Sep 2020 06:48:37 +0000 (23:48 -0700)]
packagegroups: remove strace and lttng-tools for rv32/musl

These tools are not yet ported to rv32/musl

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoncurses: Create alternative symlinks for st and st-256color
Khem Raj [Thu, 17 Sep 2020 00:23:01 +0000 (17:23 -0700)]
ncurses: Create alternative symlinks for st and st-256color

Adjust for other st implementations

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agotestexport: rename create_tarball method
Konrad Weihmann [Mon, 21 Sep 2020 14:29:44 +0000 (16:29 +0200)]
testexport: rename create_tarball method

as otherwise when using it in combination with archiver.bbclass we would
have 2 methods of the same name but with different signatures
leading to various hard to understand exceptions

Signed-off-by: Konrad Weihmann <kweihmann@outlook.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobind: remove -r option for rndc-confgen
Mingli Yu [Mon, 21 Sep 2020 06:37:58 +0000 (23:37 -0700)]
bind: remove -r option for rndc-confgen

The named service fail to start as below:
 # systemctl  status  named.service
   named.service - Berkeley Internet Name Domain (DNS)
     Loaded: loaded (/lib/systemd/system/named.service; enabled; vendor preset: enabled)
     Active: failed (Result: exit-code) since Wed 2020-09-16 06:07:49 UTC; 9s ago
    Process: 134206 ExecStartPre=/usr/sbin/generate-rndc-key.sh (code=exited, status=1/FAILURE)

Sep 16 06:07:49 intel-x86-64 systemd[1]: Starting Berkeley Internet Name Domain (DNS)...
Sep 16 06:07:49 intel-x86-64 generate-rndc-key.sh[134206]: Generating /etc/bind/rndc.key:
Sep 16 06:07:49 intel-x86-64 generate-rndc-key.sh[134207]: rndc-confgen: The -r option has been deprecated.
Sep 16 06:07:49 intel-x86-64 generate-rndc-key.sh[134208]: chown: cannot access '/etc/bind/rndc.key': No such file or directory
Sep 16 06:07:49 intel-x86-64 generate-rndc-key.sh[134209]: chmod: cannot access '/etc/bind/rndc.key': No such file or directory
Sep 16 06:07:49 intel-x86-64 systemd[1]: named.service: Control process exited, code=exited, status=1/FAILURE
Sep 16 06:07:49 intel-x86-64 systemd[1]: named.service: Failed with result 'exit-code'.
Sep 16 06:07:49 intel-x86-64 systemd[1]: Failed to start Berkeley Internet Name Domain (DNS).

It is because fail to execute "/usr/sbin/generate-rndc-key.sh" as
-r is deprecated since bind 9.13.x and the random function changes
in [1], so remove -r option to fix the above issue.

DNSSEC validation is now active by default after bind upgrade to 9.16.x,
but it is not in 9.11.x. So disable DNSSEC validation explicitly to
silence below message.
Sep 18 03:21:37 intel-x86-64 named[23272]: managed-keys-zone: Unable to fetch DNSKEY set '.': timed out

[1]: https://gitlab.isc.org/isc-projects/bind9/-/commit/3a4f820d625c214cfb21f5e6d18ce9160d2a193b

Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokea: create /var/lib/kea and /var/run/kea folder
Mingli Yu [Fri, 18 Sep 2020 09:25:10 +0000 (02:25 -0700)]
kea: create /var/lib/kea and /var/run/kea folder

Create /var/lib/kea and /var/run/kea folder if they don't
exist to fix below error:
 # keactrl start
 INFO/keactrl: Starting /usr/sbin/kea-dhcp4 -c /etc/kea/kea-dhcp4.conf
 INFO/keactrl: Starting /usr/sbin/kea-dhcp6 -c /etc/kea/kea-dhcp6.conf
 INFO/keactrl: Starting /usr/sbin/kea-ctrl-agent -c /etc/kea/kea-ctrl-agent.conf
 Unable to use interprocess sync lockfile (No such file or directory): /var/run/kea/logger_lockfile
 Service failed: Launch failed: Unable to open PID file '/var/run/kea/kea-ctrl-agent.kea-ctrl-agent.pid' for write
 [snip]
 ERROR [kea-dhcp4.dhcp4/615.140641792751488] DHCP4_CONFIG_LOAD_FAIL configuration error using file: /etc/kea/kea-dhcp4.conf, reason: Unable to open database: unable to open '/var/lib/kea/kea-leases4.csv'
 [snip]

Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobash : include patch 17 & 18
Chee Yang Lee [Fri, 18 Sep 2020 14:37:21 +0000 (22:37 +0800)]
bash : include patch 17 & 18

patch 17 :-
https://ftp.gnu.org/gnu/bash/bash-5.0-patches/bash50-017
There were cases where patch 16 reaped process substitution file
descriptors (or FIFOs) and processes to early. This is a better
fix for the problem that bash50-016 attempted to solve.

path 18 :-
https://ftp.gnu.org/gnu/bash/bash-5.0-patches/bash50-018
In certain cases, bash does not perform quoted null removal on
patterns that are used as part of word expansions such as
${parameter##pattern}, so empty patterns are treated as non-empty.

Signed-off-by: Chee Yang Lee <chee.yang.lee@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobuildtools-tarball: Add python3-jinja2
Richard Purdie [Mon, 21 Sep 2020 11:56:53 +0000 (12:56 +0100)]
buildtools-tarball: Add python3-jinja2

This is used by various pieces of the result handling code, particularly the
performance testing and we're seeing autobuilder failures that are
easiest resolved using this.

Acked-by: Tim Orling <ticotimo@gmail.com>
Signed-off-by: Tim Orling <ticotimo@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agopython3-jinja2: Import from meta-oe/meta-python
Richard Purdie [Mon, 21 Sep 2020 11:55:28 +0000 (12:55 +0100)]
python3-jinja2: Import from meta-oe/meta-python

This is used by some of the results handling code and needed as part of
buildtools tarball on various autobuilder worker for testing.

ptest is disabled for OE-Core, at least for now since it depends on
python3-pytest which in turn has may other dependencies.

Acked-by: Tim Orling <ticotimo@gmail.com>
Signed-off-by: Tim Orling <ticotimo@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agopython3-markupsafe: Import from meta-oe/meta-python
Richard Purdie [Mon, 21 Sep 2020 11:54:12 +0000 (12:54 +0100)]
python3-markupsafe: Import from meta-oe/meta-python

Needed as a dependency of python3-jinja2. ptest is disabled for OE-Core, at
least for now since it depends on python3-pytest which in turn has may other
dependencies.

Acked-by: Tim Orling <ticotimo@gmail.com>
Signed-off-by: Tim Orling <ticotimo@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agooeqa/qemurunner: Increase serial timeout
Richard Purdie [Fri, 18 Sep 2020 05:57:10 +0000 (06:57 +0100)]
oeqa/qemurunner: Increase serial timeout

Increase the serial login timeout from 60 to 120s. This seems like a
long time, however for a qemumips image with systemd+PAM and openssh,
(e.g. core-image-sato-sdk + DISTRO=poky-altcfg), the getty connects
to systemd's pam module which waits on logind and 45s for all this
to happen at the same time as things like ssh key generation happens
is not unknown.

Increase the timeout to match the longer times we know these things
can take in the worst case scenarios since we're tired of intermittent
issues related to the serial login affecting the autobuilder.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agodropbear/openssh: Lower priority of key generation
Richard Purdie [Fri, 18 Sep 2020 05:52:57 +0000 (06:52 +0100)]
dropbear/openssh: Lower priority of key generation

Where we have images with PAM+systemd, serial login can be extremely
slow. The load generated by key generation does slow down the rest
of the boot process.

Lower the priority level of these systemd services, since we'd
prefer to have the rest of the system boot more effectively.

This doesn't "solve" the slow systemd boot issues but does help.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoimage-artifact-names: make variables overridable
Jens Rehsack [Thu, 17 Sep 2020 14:45:34 +0000 (16:45 +0200)]
image-artifact-names: make variables overridable

Since the newly introduced image-artifact-names class is inherited at later
processing stage, individual IMAGE_NAME settings are overridden (hopefully
by accident instead on purpose).

Allow derived distributions define their own name schema by setting defaults
iwth question mark.

Signed-off-by: Jens Rehsack <sno@netbsd.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoscripts/oe-build-perf-report: Use python3 from the environment
Richard Purdie [Sun, 20 Sep 2020 11:46:43 +0000 (12:46 +0100)]
scripts/oe-build-perf-report: Use python3 from the environment

This means ot matches the rest of the system rather than causing confusing errors
where multiple python versions are mixed.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobuild-appliance-image: Update to master head revision
Richard Purdie [Thu, 17 Sep 2020 09:12:39 +0000 (10:12 +0100)]
build-appliance-image: Update to master head revision

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobitbake.conf: use ${TCMODE}-${TCLIBC} directory for CACHE
Martin Jansa [Wed, 16 Sep 2020 13:28:52 +0000 (15:28 +0200)]
bitbake.conf: use ${TCMODE}-${TCLIBC} directory for CACHE

* move TCMODE and TCLIBC from defaultsetup.conf to bitbake.conf
* set CACHE as it was in defaultsetup.conf and drop it from defaultsetup.conf
* most if not all DISTROs are now including defaultsetup.conf and
  TCLIBC is pretty much expected to be always set correctly, e.g.:
  meta/recipes-core/systemd/systemd_243.2.bb:    if d.getVar('TCLIBC') == "musl":
  meta/recipes-devtools/gcc/gcc-runtime.inc:        if [ "${TCLIBC}" != "glibc" ]; then
  meta/recipes-devtools/gcc/libgcc.inc:     if [ "${TCLIBC}" != "glibc" ]; then
  meta/recipes-devtools/icecc-toolchain/nativesdk-icecc-toolchain_0.1.bb:ENV_NAME="${DISTRO}-${TCLIBC}-${SDK_ARCH}-@TARGET_PREFIX@${DISTRO_VERSION}.tar.gz"
  meta/recipes-devtools/valgrind/valgrind_3.15.0.bb:RRECOMMENDS_${PN} += "${TCLIBC}-dbg"
  meta/recipes-kernel/linux/kernel-devsrc.bb:RDEPENDS_${PN} = "bc python3 flex bison ${TCLIBC}-utils"

  meta/classes/buildhistory.bbclass:BUILDHISTORY_DIR_IMAGE = "${BUILDHISTORY_DIR}/images/${MACHINE_ARCH}/${TCLIBC}/${IMAGE_BASENAME}"
  meta/classes/cross-canadian.bbclass:    if d.getVar("TCLIBC") in [ 'baremetal', 'newlib' ]:
  meta/classes/kernel.bbclass:    tclibc = d.getVar('TCLIBC')
  meta/classes/toaster.bbclass:    BUILDHISTORY_DIR_IMAGE_BASE = e.data.expand("%s/images/${MACHINE_ARCH}/${TCLIBC}/"% BUILDHISTORY_DIR)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoxinput-calibrator: change SRC_URI to branch with libinput support
Steve Sakoman [Wed, 16 Sep 2020 16:17:31 +0000 (06:17 -1000)]
xinput-calibrator: change SRC_URI to branch with libinput support

Since "conf: Use xf86-input-libinput by default" [1] there are
reports [2] of xinput-calibrator failing because it expects
xf86-input-evdev and with the above patch xf86-input-libinput
takes precedence.

Fix this issue by using a branch of xinput calibrator which supports
xf86-input-libinput.

[1] https://git.openembedded.org/openembedded-core/commit/meta/recipes-graphics/xorg-xserver/xserver-xorg.inc?id=2d005faff6341a81a2afae28860101ba9db51ae8
[2] https://www.yoctoproject.org/pipermail/yocto/2018-December/043487.html

Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agovalgrind: fix memcheck vgtests remove fullpath-after flags
Stacy Gaikovaia [Wed, 16 Sep 2020 23:23:04 +0000 (19:23 -0400)]
valgrind: fix memcheck vgtests remove fullpath-after flags

Previously in:
    f75792b28e valgrind: make a few more ptests pass
the vagrind test runner was adjusted to suppress part
of a path that only exists when run in the yocto ptest
environment. Unfortunately this change includes the
valgrind version and when valgrind was last updated,
the patch was not changed. Rather than continually updating
the patch or even generating the version dynamically,
we can simply change the expected output for two tests.

The reason the option: --fullpath-after=foo was
introduced into the effected valgrind ptests was to
deal with builds where ccache is used. Compiling with
ccache enabled sometimes causes the source file absolute
name to be found in a full path that is not the same as $PWD.

See commit c80f32e662dfa2a4f046960a25d5b8b7a8821bea in
valgrind for more information about changes to the
arguments that test badfree3 and varinfo5 run with.

There is also a minor fix to add the missing overloading.pm
perl package and put the dependencies in alphabetic order.

Signed-off-by: Stacy Gaikovaia <Stacy.Gaikovaia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobuildhistory-diff: use BUILDDIR to know where buildhistory is
Ross Burton [Tue, 15 Sep 2020 12:55:48 +0000 (13:55 +0100)]
buildhistory-diff: use BUILDDIR to know where buildhistory is

Instead of assuming that the buildhistory is stored in ./buildhistory
unless told otherwise, use the environment variable BUILDDIR if present
(which is set by oe-init-build-env) to locate the buildhistory
directory.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agobison: remove the parallel build patch
Mingli Yu [Wed, 16 Sep 2020 02:41:31 +0000 (19:41 -0700)]
bison: remove the parallel build patch

Remove the parallel build patch and related fix for parallel
build as the issue have already been fixed in bison 3.7.2
(f7b642cf build: fix incorrect dependencies).

Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosystemd: avoid failing if no udev rules provided
Peter Bergin [Tue, 15 Sep 2020 19:33:05 +0000 (21:33 +0200)]
systemd: avoid failing if no udev rules provided

In the case when no udev rules are provided by the recipe the do_install
step of systemd is failing with:

    install: cannot stat '<WORKDIR>/*.rules': No such file or directory

By default the systemd recipe provides touchscreen.rules but in the
case this is unwanted on target the install step fails. This solution
is dynamic and search for *.rules files in $WORKDIR. If found they are
installed.

Signed-off-by: Peter Bergin <peter@berginkonsult.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years ago*-initramfs: don't use .rootfs IMAGE_NAME_SUFFIX
Martin Jansa [Wed, 16 Sep 2020 12:54:49 +0000 (14:54 +0200)]
*-initramfs: don't use .rootfs IMAGE_NAME_SUFFIX

* fixes the issue when image-live.bbclass expects the image
  ending with just INITRAMFS_FSTYPES:
  image-live.bbclass:INITRD_LIVE ?= "${DEPLOY_DIR_IMAGE}/${INITRD_IMAGE_LIVE}-${MACHINE}.${INITRAMFS_FSTYPES}"
  while by default it now was with .rootfs suffix:
  -rw-r--r-- 2 bitbake bitbake 1.5K Oct 25 16:12 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs.env
  -rw-r--r-- 4 bitbake bitbake  11M Oct 25 16:13 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64-1.0-r0-20191025154349.cpio.gz
  -rw-r--r-- 4 bitbake bitbake 1.2K Oct 25 16:11 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64-1.0-r0-20191025154349.manifest
  -rw-r--r-- 4 bitbake bitbake 1.3K Oct 25 16:12 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64-1.0-r0-20191025154349.qemuboot.conf
  -rw-r--r-- 4 bitbake bitbake 196K Oct 25 16:11 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64-1.0-r0-20191025154349.testdata.json
  -rw-r--r-- 4 bitbake bitbake 118M Oct 25 16:13 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64-1.0-r0-20191025154349.wic
  -rw-r--r-- 4 bitbake bitbake 3.1K Oct 25 16:13 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64-1.0-r0-20191025154349.wic.bmap
  -rw-r--r-- 4 bitbake bitbake 1.3K Oct 25 16:12 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64.qemuboot.conf
  -rw-r--r-- 4 bitbake bitbake  11M Oct 25 16:13 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64.rootfs.cpio.gz
  -rw-r--r-- 4 bitbake bitbake 1.2K Oct 25 16:11 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64.rootfs.manifest
  -rw-r--r-- 4 bitbake bitbake 118M Oct 25 16:13 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64.rootfs.wic
  -rw-r--r-- 4 bitbake bitbake 3.1K Oct 25 16:13 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64.rootfs.wic.bmap
  -rw-r--r-- 4 bitbake bitbake 196K Oct 25 16:11 /OE/build/oe-core/tmp/deploy/images/genericx86-64/core-image-minimal-initramfs-genericx86-64.testdata.json

  initramfs images aren't normally used for rootfs, so just set
  the suffix to empty, people using different artifact names might
  still need to set INITRD_LIVE (e.g. when their images don't end
  with "-${MACHINE}" as well)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokernel.bbclass: use camelCase notation for bash variables in do_deploy
Martin Jansa [Wed, 16 Sep 2020 12:54:48 +0000 (14:54 +0200)]
kernel.bbclass: use camelCase notation for bash variables in do_deploy

* to match other variables there like deployDir imageType

[YOCTO #12937]

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokernel.bbclass: eliminate (initramfs_)symlink_name variables
Martin Jansa [Wed, 16 Sep 2020 12:54:47 +0000 (14:54 +0200)]
kernel.bbclass: eliminate (initramfs_)symlink_name variables

* they are used only once, we can use the value directly
* notice that .bin extension isn't part of the variable values

[YOCTO #12937]

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agokernel.bbclass: use bash variables like imageType, base_name without {}
Martin Jansa [Wed, 16 Sep 2020 12:54:46 +0000 (14:54 +0200)]
kernel.bbclass: use bash variables like imageType, base_name without {}

* just to make sure it looks like bash variable not bitbake variable in
  run.do_* scripts

[YOCTO #12937]

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoimage-artifact-names: introduce new bbclass and move some variables into it
Martin Jansa [Wed, 16 Sep 2020 12:54:45 +0000 (14:54 +0200)]
image-artifact-names: introduce new bbclass and move some variables into it

* similar to kernel-artifact-names for other recipes/bbclasses which
  need to use some deployed artifacts

* bitbake.conf: move IMAGE_BASENAME, IMAGE_VERSION_SUFFIX, IMAGE_NAME,
  IMAGE_LINK_NAME variables

* image_types.bbclass: move IMAGE_NAME_SUFFIX variable

* currently IMAGE_NAME_SUFFIX is used only by image.bbclass,
  image_types.bbclass and meta/recipes-core/images/build-appliance-image_15.0.0.bb
  but if it's needed by some recipe which isn't itself an image, then
  it's useful in bitbake.conf, e.g. we have a recipe for creating
  VirtualBox appliances which combines .wic.vmdk with .ovf file to
  create .zip with appliance, but for that we need the filename of
  .wic.vmdk which now contains IMAGE_NAME_SUFFIX
  https://github.com/webOS-ports/meta-webos-ports/blob/4980ce52a43ac6897657602810313af359f0b839/meta-luneos/recipes-core/images/luneos-emulator-appliance.inc#L24

* we were hardcoding .rootfs suffix where needed, but for quite long
  time it's configurable with IMAGE_NAME_SUFFIX since:

  commit 380ee36811939d947024bf78de907e3c071b834f
  Author: Patrick Ohly <patrick.ohly@intel.com>
  Date:   Mon Mar 7 18:07:52 2016 +0100

    image creation: allow overriding .rootfs suffix

  and might not match with hardcoded .rootfs, so make it easier to
  use IMAGE_NAME_SUFFIX where needed even without inheritting whole
  image_types.bbclass

[YOCTO #12937]

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoqemu: override DEBUG_BUILD
Mingli Yu [Wed, 16 Sep 2020 08:09:04 +0000 (01:09 -0700)]
qemu: override DEBUG_BUILD

Override DEBUG_BUILD for qemu as the qemu upstream states it
doesn't work without optimization [1] to fix below build failure
when debug build enabled.
 | /usr/lib/gcc/x86_64-wrs-linux/10.1.0/../../../../x86_64-wrs-linux/bin/ld.bfd: /mnt/build/tmp/work/x86_64-linux/qemu-system-native/5.1.0-r0/qemu-5.1.0/fsdev/qemu-fsdev-throttle.c:25: undefined reference to `unknown_lock_type'
 | /usr/lib/gcc/x86_64-wrs-linux/10.1.0/../../../../x86_64-wrs-linux/bin/ld.bfd: ../fsdev/qemu-fsdev-throttle.o: in function `fsdev_co_throttle_request':
 | /mnt/build/tmp/work/x86_64-linux/qemu-system-native/5.1.0-r0/qemu-5.1.0/fsdev/qemu-fsdev-throttle.c:103: undefined reference to `unknown_lock_type'
 | /usr/lib/gcc/x86_64-wrs-linux/10.1.0/../../../../x86_64-wrs-linux/bin/ld.bfd: ../fsdev/qemu-fsdev-throttle.o:/mnt/build/tmp/work/x86_64-linux/qemu-system-native/5.1.0-r0/qemu-5.1.0/fsdev/qemu-fsdev-throttle.c:103: more undefined references to `unknown_lock_type' follow
 | collect2: error: ld returned 1 exit status

[1]: https://lists.nongnu.org/archive/html/qemu-devel/2020-09/msg03873.html

Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoinsane: check for missing update-alternatives inherit
Andrey Zhizhikin [Mon, 14 Sep 2020 23:31:59 +0000 (23:31 +0000)]
insane: check for missing update-alternatives inherit

Add a package QA check to test if the recipe sets ALTERNATIVE variable
for any of its packages, and does not inherit update-alternatives class.
This causes the do_rootfs to fail since a proper alternative cannot be
created.

Signed-off-by: Andrey Zhizhikin <andrey.z@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agosato-screenshot: improve .la removal
Ross Burton [Tue, 15 Sep 2020 11:05:32 +0000 (12:05 +0100)]
sato-screenshot: improve .la removal

Don't assume that the .la files are installed, because some libtool
alternatives don't install these.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoalsa-plugins: improve .la removal
Ross Burton [Tue, 15 Sep 2020 11:05:31 +0000 (12:05 +0100)]
alsa-plugins: improve .la removal

Don't assume that the .la files are installed, because some libtool
alternatives don't install these.

Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4 years agoRevert "core-image-sato: qemumips use 512 mem"
Richard Purdie [Tue, 15 Sep 2020 10:55:01 +0000 (11:55 +0100)]
Revert "core-image-sato: qemumips use 512 mem"

This reverts commit 45da2fc4f7272366ffc7cde24d9040c919466201.

qemumips is unstable with more than 256MB memory, the original issue
this limit was added for remains.

We'll have to find other solutions to the systemd out of memory issues.

4 years agoRevert "qemu: always define unknown_lock_type"
Richard Purdie [Tue, 15 Sep 2020 10:54:17 +0000 (11:54 +0100)]
Revert "qemu: always define unknown_lock_type"

This reverts commit e442924d9ac4e202aec571b5d26bb959b5faaadd.

This was rejected upstream and its been pointed out qemu doesn't work
without optimization.

Instead we should just error if the user attempts to build it without
optimization.

4 years agolttng-modules: backport patches from 2.12.x to fix 5.4.64+ and 5.8.9+ builds
Jens Rehsack [Mon, 14 Sep 2020 07:52:22 +0000 (09:52 +0200)]
lttng-modules: backport patches from 2.12.x to fix 5.4.64+ and 5.8.9+ builds

Backporting the 10 patches since the lttng 2.12.2 release. We'll drop
them once .3 is released, but for now, we need the fixes to build
against the latest 5.4, 5.8 and 5.9 kernels.

We also bump the devupstream SRCREV to pickup the same changes.

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
4 years agolinux-yocto/5.4: update to v5.4.65
Bruce Ashfield [Mon, 14 Sep 2020 16:13:17 +0000 (12:13 -0400)]
linux-yocto/5.4: update to v5.4.65

Updating linux-yocto/5.4 to the latest korg -stable release that comprises
the following commits:

    6c3d34dea2fc Linux 5.4.65
    ddb279d64b72 net: disable netpoll on fresh napis
    09c45065257b tipc: fix shutdown() of connectionless socket
    b95eb482c288 taprio: Fix using wrong queues in gate mask
    20f8c874789a sctp: not disable bh in the whole sctp_get_port_local()
    8b61bb0b2d57 net: usb: dm9601: Add USB ID of Keenetic Plus DSL
    346fefa82325 netlabel: fix problems with mapping removal
    3f73dbf94f8f ipv6: Fix sysctl max for fib_multipath_hash_policy
    263e463358d3 ipv4: Silence suspicious RCU usage warning

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
4 years agolinux-yocto/5.8: update to v5.8.9
Bruce Ashfield [Mon, 14 Sep 2020 15:21:07 +0000 (11:21 -0400)]
linux-yocto/5.8: update to v5.8.9

Updating linux-yocto/5.8 to the latest korg -stable release that comprises
the following commits:

    f82f36b2fb1d Linux 5.8.9
    cdba995d2f19 mptcp: free acked data before waiting for more memory
    655e1af6aecd net: disable netpoll on fresh napis
    6f0e276cda3e tipc: fix using smp_processor_id() in preemptible
    aae250a26893 tipc: fix shutdown() of connectionless socket
    44c3fca3de64 taprio: Fix using wrong queues in gate mask
    db16256c95ac sctp: not disable bh in the whole sctp_get_port_local()
    05a8237c08f0 net: usb: dm9601: Add USB ID of Keenetic Plus DSL
    6540bd18bf2f netlabel: fix problems with mapping removal
    089097476c4c ipv6: Fix sysctl max for fib_multipath_hash_policy
    d68c3397de48 ipv4: Silence suspicious RCU usage warning
    bd15d2192261 RDMA/cma: Execute rdma_cm destruction from a handler properly
    d4794085738d RDMA/cma: Remove unneeded locking for req paths
    c2064ac30331 RDMA/cma: Using the standard locking pattern when delivering the removal event
    5a0c4cbd099b RDMA/cma: Simplify DEVICE_REMOVAL for internal_id
    87d8175e9c3d io_uring: fix linked deferred ->files cancellation
    f5fea75f4ea4 io_uring: fix cancel of deferred reqs with ->files

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>