Allow users to change the default cgroup mode at build time
and use the unified hierarchy mode.
Disabled by default - hybrid is the default upstream value.
Signed-off-by: Luca Boccassi <luca.boccassi@microsoft.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Fri, 26 Apr 2019 14:52:16 +0000 (10:52 -0400)]
linux-yocto/5.0: port RAID configuration tweaks from master
Porting the following three RAID config changes from master to
the 5.0 branch:
ffd8cf5baf8 intel-x86: add Intel VMD support 8edf951a15c cfg/efi.cfg: built-in CONFIG_EFIVAR_FS to support Intel VROC 041a6c04244 intel-x86: built-in nvme driver to support boot from nvme disk
Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It was a 4.19.31 backport of commit 6ea183d60c46 ("mm: handle
lru_add_drain_all for UP properly"). In summary, what that did
was to fix a possible harmless WARN_ON on non-SMP, introduced at
commit 4d43d395fed1 ("workqueue: Try to catch flush_work() without
INIT_WORK().") by adding non-SMP variants of lru functions.
The combination of that, with the -rt commit 473f14a9f234 ("mm:
perform lru_add_drain_all() remotely") at the merge of the two
results in the following build failure:
mm/swap.c:736:2: error: #endif without #if
since the -rt change wants RT specific lru and the stable backport
wants non-SMP specific lru, and a chunk of the backport with
an #ifdef CONFIG_SMP is missing.
However, before we add a four way cluster of ifdeffery to handle all
cases, we note 4d43d395fed1 was added to the v5.1 release, and it
was not (currently) backported to any 4.19.x stable release - so it is
unclear to me why this commit was ever backported to 4.19.31 at all.
Further, we note this change was to mm/swap.c -- and by definition,
any preempt-rt deployment that uses swap for anything other than a
failure contingency mitigation is broken by design.
Given all that, I decided that the best path forward was to revert
the two of the three chunks of the backport that remain in the -rt
branch, and return us to the pre-4.19.31 merge behaviour for -rt.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This provides git large file storage (lfs) extension.
Include git-lfs conditionally. If git-lfs is present on host and repo
has lfs pointers, then git-lfs will be used. If git-lfs is not present
on host, it will be ignored.
[YOCTO #13198]
Signed-off-by: Naveen Saini <naveen.kumar.saini@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Jaewon Lee [Thu, 25 Apr 2019 23:02:21 +0000 (16:02 -0700)]
Adding back wrapper and using OEPYTHON3HOME variable for python3
Adding back the python wrapper and adding a patch to use OEPYTHON3HOME
instead of PYTHONHOME if set, for python3.
If we add back the wrapper as is, we would see the following error that
we also see in Thud:
ImportError: No module named site
OpenEmbedded requires 'python' to be python v2 (>= 2.7.3), not python
v3.
Please upgrade your python v2
This is because python3 would've set PYTHONHOME to use nativesdk
python3 libraries but when the oe-buildenv-internal script tries to call
python2 for the py_v27_check, there will be no python2 libraries in the
PYTHONHOME directory.
In other words, bitbake needs host python2 and the env variable set from
the wrapper contaminates the env and host python2 won't be able to find
its libraries
Creating another variable OEPYTHON3HOME and using this in the python3
wrapper to allow for a way to set a different paths for python3 and
python2
[YOCTO #13208]
Signed-off-by: Jaewon Lee <jaewon.lee@xilinx.com> Signed-off-by: Alejandro Enedino Hernandez Samaniego <alejandr@xilinx.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
ERROR: bash-4.4.18-r0 do_package_write_ipk: Fatal errors occurred in subprocesses:
Command 'PATH="X" opkg-build -Z xz -a "--memlimit=50% --threads=88" Foobar /media/build1/poky/build/nodistro-glibc/work/core2-64-oe-linux/bash/4.4.18-r0/deploy-ipks/core2-64' returned non-zero exit status 1.: Traceback (most recent call last):
File "/media/build1/poky/meta/lib/oe/utils.py", line 272, in run
ret = self._target(*self._args, **self._kwargs)
File "/media/build1/poky/meta/classes/package_ipk.bbclass", line 230, in ipk_write_pkg
shell=True)
File "/usr/lib/python3.6/subprocess.py", line 336, in check_output
**kwargs).stdout
File "/usr/lib/python3.6/subprocess.py", line 418, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command 'PATH="X" opkg-build -Z xz -a "--memlimit=50% --threads=88" Foobar /media/build1/poky/build/nodistro-glibc/work/core2-64-oe-linux/bash/4.4.18-r0/deploy-ipks/core2-64' returned non-zero exit status 1.
Note how stdout/stderr from the failing command isn't shown.
After this patch:
ERROR: bash-4.4.18-r0 do_package_write_ipk: Fatal errors occurred in subprocesses:
Command 'PATH="X" opkg-build -Z xz -a "--memlimit=50% --threads=88" Foobar /media/build1/poky/build/nodistro-glibc/work/core2-64-oe-linux/bash/4.4.18-r0/deploy-ipks/core2-64' returned non-zero exit status 1.
Subprocess output:Foobar
*** Error: Package name Foobar contains illegal characters, (other than [a-z0-9.+-])
opkg-build: Please fix the above errors and try again.
We suddenly get a much more usable error message. The traceback is supressed
as its distracting from the real problem in this case.
Ideally python itself would handle this but it doesn't so we have to
wrap the exception. We already do this in bitbake itself for the same reason.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Andreas Müller [Thu, 25 Apr 2019 00:10:54 +0000 (02:10 +0200)]
vte: move shell auto scripts into seperate package
It bugged me for a while that xfce4-terminal sets (tab-)title to 'Untitled'.
This is caused by bash variable PROMPT_COMMAND not set. vte ships /etc/profile.d/vte.sh
adding PROMPT_COMMAND and a function to handle.
Since
* not all consumers of 'libvte' are interested in this feature
* executable shipped by package 'vte' is not necessary for PROMPT_COMMAND support
create a seperate package.
Once applied patches for xfce4-terminal and gnome-terminal will be send.
Signed-off-by: Andreas Müller <schnitzeltony@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
kernel-module-split.bbclass: support CONFIG_MODULE_COMPRESS=y
In case, kernel config enables compressed modules, support of
splitting via split_kernel_module_packages won't find any module.
So, first expand module pattern regex to recognize compressed
modules and then objcopy on temporary extacted to extract module
information.
Signed-off-by: Jens Rehsack <sno@netbsd.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Martin Jansa [Thu, 28 Mar 2019 10:21:39 +0000 (10:21 +0000)]
glibc: always use bfd linker
* Work around broken ld-2.29.so when gold is used
causing qemu-arm to segfault during e.g. gobject-introspection
or postinst at do_rootfs time, more details in:
http://lists.openembedded.org/pipermail/openembedded-devel/2019-March/198937.html
https://sourceware.org/bugzilla/show_bug.cgi?id=24148
https://sourceware.org/bugzilla/show_bug.cgi?id=10937
https://sourceware.org/bugzilla/show_bug.cgi?id=18103
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
resulttool/manualexecution: Enable test case configuration option
Current manualexecution required user to exceute all test cases defined inside a "modulename.json" file in oeqa/manual
There are cases when all test cases all not required to run for a module on specific DUT.
Enable manualexecution to have the optional feature where it will use pre-defined json format test case configuration file
where user will be able to select test cases from the "modulename.json" instead of running all of them. This will help
in reducing testing time and reporting unneccesary skip or failures.
Example pre-defined json format test case configuration file (for build-applince):
FAT16 volumes do not have a minimum size of 8250 blocks
(== slightly over 4MB). Exhibit A: floppy disks.
The original commit message suggests this was a workaround for a bug in
parted - in which case we should fix it there, or use the wic
`--fixed-size` option to pass down an enforced minimum from a more
context-aware point in the callstack.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
core-image-rt-sdk: make sure that we append to DEPENDS
This also fix postinst intercept hook 'update_gio_module_cache' failed warnings
that are now flagged as errors after the recent chagnes at lib/oe/package_manager.py
Signed-off-by: Naveen Saini <naveen.kumar.saini@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
core-image-rt: make sure that we append to DEPENDS
This also fix postinst intercept hook 'update_gio_module_cache' failed warnings
that are now flagged as errors after the recent chagnes at lib/oe/package_manager.py
Signed-off-by: Naveen Saini <naveen.kumar.saini@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
oeqa/core/runner: dump stdout and stderr of each test case
Some CI pipelines might perform further processing of the test output
(for instance, to plot some metrics into a chart). However, Since `thud`
we switched away from the XML-based jUnit reporting, and at the same
time we lost the ability of collecting the stdout and stderr of the
various tests.
We now restore this functionality by adding `stdout` and `stderr` keys
to the JSON reports. This behavior is off by default; in order to enable
it, one must set the `TESTREPORT_FULLLOGS` variable in the bitbake
configuration.
Signed-off-by: Alberto Mardegan <amardegan@luxoft.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
icecc: Don't generate recipe-sysroot symlinks at recipe-parsing time
the set_icecc_env function depends on KERNEL_CC variable even for recipes
like linux-libc-headers
* KERNEL_CC variable depends on STAGING_KERNEL_DIR:
Variable KERNEL_CC value is ${CCACHE}${HOST_PREFIX}gcc ${HOST_CC_KERNEL_ARCH} -fuse-ld=bfd ${DEBUG_PREFIX_MAP} -fdebug-prefix-map=${STAGING_KERNEL_DIR}=${KERNEL_SRC_PATH}
List of dependencies for variable KERNEL_CC is {'STAGING_KERNEL_DIR', 'HOST_PREFIX', 'DEBUG_PREFIX_MAP', 'HOST_CC_KERNEL_ARCH', 'KERNEL_SRC_PATH'}
which depends on MACHINE:
Variable STAGING_KERNEL_DIR value is ${TMPDIR}/work-shared/${MACHINE}/kernel-source
List of dependencies for variable STAGING_KERNEL_DIR is {'MACHINE'}
Drop yy_scan_string patch, this only affects builds with flex 2.6.2. We
currently have 2.6.0 and when we upgrade it will be to at least 2.6.3, which
fixes the regression.
Drop manipulate_fds_instead-of-FILE.patch, the original problem was fix upstream
in May 2015[1] so the 20170303 upgrade should have dropped this patch instead of
rebasing.
Call the upstream install target with variables set appropriately, instead of
hand-coding an install.
Sets the WAFLOCK environment variable. This controls the name of the
lock file that waf uses to pass the build configuration from 'configure'
to 'build' and 'install'. Using a uniquely generated name based on the
parameters passed to 'configure' ensures that the source directory can
be configured for multiple different builds without conflicting (since
the lock file is stored in ${S})
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Steven Hung (洪于玉) <Steven.Hung@mediatek.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Joslyn [Sat, 20 Apr 2019 22:05:16 +0000 (15:05 -0700)]
qemu: Add PACKAGECONFIG for snappy
In addition to toggling libsnappy support, this allows a qemu-native
build to explicitly disable snappy support. When building qemu-native,
the configure script will detect libsnappy on the build machine and
improperly enable snappy support. This can corrupt the sstate cache if
the cache is populated with a qemu-native build with snappy improperly
enabled, and the cached qemu-native gets used on another build machine
that does not have libsnappy.
Signed-off-by: Robert Joslyn <robert.joslyn@redrectangle.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
License checksum from quota.c was removed since according to the
project, copyrighted code in question has been replaced with own
implementation (see @bcbc0d08e5cd).
Removed patches were upstreamed.
Signed-off-by: Oleksandr Kravchuk <open.source@oleksandr-kravchuk.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When DEBUG_BUILD = "1" added in local.conf, there
comes below build error when "bitbake gcc-sanitizers":
| ./../../../../../../../../work-shared/gcc-8.3.0-r0/gcc-8.3.0/libsanitizer/libbacktrace/../../libbacktrace/elf.c: In function 'elf_is_symlink':
| ../../../../../../../../../work-shared/gcc-8.3.0-r0/gcc-8.3.0/libsanitizer/libbacktrace/../../libbacktrace/elf.c:772:21: error: 'st.st_mode' may be used uninitialized in this function [-Werror=maybe-uninitialized]
| return S_ISLNK (st.st_mode);
After commit[16643b0322 bitbake.conf: Use -Og in DEBUG_OPTIMIZATION]
introduced, "-Og" added to compiler when debug
build enabled.
Per https://gcc.gnu.org/ml/gcc-patches/2019-04/msg00315.html,
the gcc upstream thinks the warning is a false
positive and suggests to use -O2 rather than -Og
or -O1 when compiling that file, so pass -Wno-error
to compiler when -Og is used to silence the error.
Signed-off-by: Mingli Yu <Mingli.Yu@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since we want to ultimately move this recipe to that name, add a PROVIDES
to allow things to work whilst we fix the bugs preventing the move. This
avoids having to keep changing the autobuilder.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Create a common include file which lists recipes that have ptests divided
into 'fast' and 'slow' groups. This allows us to include ptests which otherwise
may not get included in images and allows us to test the faster running things
more regularly.
The new image allows access to these faster executing tests. 'fast' is defined
as tests which execute in under 30s roughly speaking.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix below build failure with musl when ptest
enabled.
| In file included from ../../elfutils-0.176/tests/dwfl-proc-attach.c:33:
| ../../elfutils-0.176/lib/system.h:63:35: error: called object 'err' is not a function or function pointer
| #define error(status, errno, ...) err(status, __VA_ARGS__)
| ^~~
| ../../elfutils-0.176/tests/dwfl-proc-attach.c:92:5: note: in expansion of macro 'error'
| error (-1, 0, "dwfl_linux_proc_attach pid %d: %s", pid,
| ^~~~~
| ../../elfutils-0.176/tests/dwfl-proc-attach.c:79:7: note: declared here
| int err;
| ^~~
The root cause is because the conflicts between
vairable and function name, so change the variable
name to workaround it.
Signed-off-by: Mingli Yu <Mingli.Yu@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add testsuite-ignore-elflint.diff from debian
(http://ftp.de.debian.org/debian/pool/main/e/elfutils/elfutils_0.176-1.debian.tar.xz)
to fix below ptest failure:
| ./run-strip-strmerge.sh: line 33: testrun_on_self_skip: command not found
| FAIL: run-strip-strmerge.sh
Add check for gcc for two test cases such as
run-strip-nothing.sh and run-strip-g.sh which
depends on gcc at run time.
Signed-off-by: Mingli Yu <Mingli.Yu@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
