When building mozjs the following error was observed
| Traceback (most recent call last):
| File "./config.status", line 9, in <module>
| from ConfigStatus import config_status
| File "/build/bitbake_build/tmp/work/corei7-64-wrs-linux/mozjs/17.0.0-r0/mozjs17.0.0/js/src/build/ConfigStatus.py", line 11, in <module>
| import sys, re, os, posixpath, ntpath
| File "/usr/lib/python2.7/ntpath.py", line 15, in <module>
| from genericpath import _unicode
| ImportError: cannot import name _unicode
| WARNING: exit code 1 from a shell command.
| ERROR: Function failed: do_configure (log file is located at /build//bitbake_build/tmp/work/corei7-64-wrs-linux/mozjs/17.0.0-r0/temp/do_configure/log.do_configure.1056
Note the host paths found in the error msg. Further this issue can be
explored using the host python, which will not error when attempting
the above import, and the python in the sysroot, which will fail when
we attempt the import.
By inheriting pythonnative we ensure the proper paths are set to avoid
this host contamination.
Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Fix QA warning:
WARNING: QA Issue: logwatch: /logwatch/usr/share/logwatch/scripts/services/resolver is owned by uid 1000,
which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated]
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
When building for qemux86 on a x86 host, there's a possibility that the build
will try to link with incorrect libraries. Due to problematic -rpath-link the
linker may try to use host libraries if they are present and are arch
compatible.
Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
* it was moved to meta-ruby (totherer with imsettings which depends on libgxim)
because of ruby-native dependency, now when ruby was moved from meta-ruby to
oe-core we can move libgxim back to meta-oe (which also allows imsettings
in meta-xfce to depend only on meta-oe as well)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* it's autodetected from sysroot:
ERROR: lcdproc: lcdd-driver-lirc package isn't created when building with minimal dependencies?
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Fix QA warning:
WARNING: QA Issue: tmux rdepends on libutempter, but it isn't a build
dependency? [build-deps]
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Fix QA issue when enable multilib:
ERROR: QA Issue: hplip: Files/directories were installed but not shipped in any package:
/usr/lib
/usr/lib/cups
/usr/lib/cups/filter
/usr/lib/cups/backend
/usr/lib/cups/filter/hpcac
/usr/lib/cups/filter/foomatic-rip-hplip
/usr/lib/cups/filter/pstotiff
/usr/lib/cups/filter/hpcups
/usr/lib/cups/filter/hplipjs
/usr/lib/cups/filter/hpps
/usr/lib/cups/filter/.debug
/usr/lib/cups/filter/.debug/hplipjs
/usr/lib/cups/filter/.debug/hpcups
/usr/lib/cups/filter/.debug/hpps
/usr/lib/cups/backend/hp
/usr/lib/cups/backend/.debug
/usr/lib/cups/backend/.debug/hp
Please set FILES such that these items are packaged. Alternatively if they are unneeded, avoid installing them or delete them within do_install.
hplip: 17 installed and not shipped files. [installed-vs-shipped]
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Fix the do_configure function to suite the latest version to install
python libraries into ${libdir}.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
* see:
http://lists.openembedded.org/pipermail/openembedded-devel/2015-September/103271.html
* fixes:
ERROR: tipcutils different signature for task do_configure.sigdata between qemux86copy and qemux86
runtaskdeps changed from ['autoconfautoconf_2.69.bb.do_populate_sysroot:virtual:native', 'automakeautomake_1.15.bb.do_populate_sysroot:virtual:native', 'gccgcc-cross_5.2.bb.do_populate_sysroot', 'gccgcc-runtime_5.2.bb.do_populate_sysroot', 'glibcglibc_2.22.bb.do_populate_sysroot', 'gnu-configgnu-config_git.bb.do_populate_sysroot:virtual:native', 'libtoollibtool-cross_2.4.6.bb.do_populate_sysroot', 'libtoollibtool-native_2.4.6.bb.do_populate_sysroot', 'linuxlinux-yocto_4.1.bb.do_populate_sysroot', 'tipcutilstipcutils_2.0.6.bb.do_patch']
to
['autoconfautoconf_2.69.bb.do_populate_sysroot:virtual:native', 'automakeautomake_1.15.bb.do_populate_sysroot:virtual:native', 'gccgcc-cross_5.2.bb.do_populate_sysroot', 'gccgcc-runtime_5.2.bb.do_populate_sysroot', 'glibcglibc_2.22.bb.do_populate_sysroot', 'gnu-configgnu-config_git.bb.do_populate_sysroot:virtual:native', 'libtoollibtool-cross_2.4.6.bb.do_populate_sysroot', 'libtoollibtool-native_2.4.6.bb.do_populate_sysroot', 'linuxlinux-yocto_4.1.bb.do_populate_sysroot', 'tipcutilstipcutils_2.0.6.bb.do_patch']
openembedded-core/meta/recipes-kernel/linux/linux-yocto_4.1.bb.do_populate_sysroot with hash 1a8aa7a50a6cf3ebf5401af43dd6bf3b
changed to
openembedded-core/meta/recipes-kernel/linux/linux-yocto_4.1.bb.do_populate_sysroot with hash 143d2686a37f592891fb2aa6fa9b752e
Hash for dependent task linuxlinux-yocto_4.1.bb.do_populate_sysroot changed from 1a8aa7a50a6cf3ebf5401af43dd6bf3b to 143d2686a37f592891fb2aa6fa9b752e
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Wipe is a little command for securely erasing files from magnetic media.
We do not support native, because wipe needs root permission.
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Since cpan.bbclass has evolved, the old wrapper simulation needs some adoption.
Use as much as possible from cpan.bbclass instead of copying code from there.
Signed-off-by: Jens Rehsack <sno@netbsd.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
A library to provide support for making real-mode calls x86 calls. On
x86 hardware, vm86 mode is used. On other platforms, x86 emulation is
provided.
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* otherwise it fails with:
| CMake Error at sysroots/x86_64-linux/usr/share/cmake-3.3/Modules/FindPackageHandleStandardArgs.cmake:148 (message):
| Could NOT find SWIG (missing: SWIG_EXECUTABLE SWIG_DIR)
* drop cmake-native, because it's already added by cmake.bbclass
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
An additional layer for Perl modules from cpan (https://github.com/rehsack/meta-cpan)
provides additional modules for dealing with rrd files: RRDTool::OO from Michael Schili.
This extension requires an executable version of rrdtool.
Signed-off-by: Jens Rehsack <sno@netbsd.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
1. Remove PNBLACKLIST[libuio] line, since libuio compiles fine with gcc-5
with this patch.
2. By default, gcc-5 uses C99 inline semantics, this semantics doesn't
generate externally visible function for inline functions. This results in
below error, when an another translation unit (TU) tries to link with the
inline function,
-- snip --
| lsuio.o: In function `main':
| <...>/libuio/0.2.1-r0/git/tools/lsuio.c:85: undefined reference to `uio_mmap'
| collect2: error: ld returned 1 exit status
| make[2]: *** [lsuio] Error 1
-- CUT --
To solve this error and make libuio to compile with both 4.x and 5.x,
1. We can remove 'uio_mmap' inline function definition in uio_mmap.c, and move
that definition into uio_helper.h file (which is included by lsuio.c) and
replace inline with "static inline". Similarly it can be done to
uio_single_munmap and uio_munmap inline functions
2. Add 'extern' keyword in front of uio_mmap inline function declaration , to
make uio_mmap as a externally visible function, and so that it can be
linked with other TUs.
Going with option 1.
Signed-off-by: Jagadeesh Krishnanjanappa <jkrishnanjanappa@mvista.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
When building rrdtool with multilib enabled, we would meet the following error.
ERROR: QA Issue: rrdtool: Files/directories were installed but not shipped in any package:
/usr/lib
/usr/lib/python2.7
/usr/lib/python2.7/site-packages
/usr/lib/python2.7/site-packages/rrdtool.so
....
Fix this problem by using nonarch_libdir instead of libdir.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
When building rrdtool in a building directory where systemd is configured
to be the init system, we would meet the following error.
ERROR: QA Issue: rrdtool: Files/directories were installed but not shipped in any package:
/lib
/lib/systemd
/lib/systemd/system
/lib/systemd/system/rrdcached.socket
/lib/systemd/system/rrdcached.service
Fix this problem by adding systemd support to the rrdtool recipe.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
When we build rrdtool, we would meet the following QA issue.
QA Issue: /usr/share/rrdtool/examples/stripes.py_rrdtool-perl contained
in package rrdtool-perl requires /usr/bin/python, but no providers found
in its RDEPENDS [file-rdeps]
The problem is that /usr/share/rrdtool/examples not only contain perl examples,
but also contains python examples.
Besides, ${datadir}/examples don't exist, fix it to be ${datadir}/rrdtool/examples.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
When building rrdtool, we would meet the following error.
ERROR: QA Issue: rrdtool-python: work/core2-64-poky-linux/rrdtool
/1.5.4-r0/packages-split/rrdtool-python/usr/lib/python2.7/site-packages
/rrdtool.so contains probably-redundant RPATH /usr/lib [useless-rpaths]
Fix this issue by removing 'LD_RUN_PATH=$(libdir)' when compiling rrdtool.so.
As we use '--disable-rpath' in EXTRA_OECONF, we don't need to the to remove
'--rpath=$(libdir)' in Makefile.am.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
It was still using deprecated lua contructs
also fix for musl while at it
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Upgrade efibootmgr from 0.6.0 to 0.12.
* update HOMEPAGE
* drop obsolete patches that source files have been removed
* add dependency efivar
* update CFLAGS to fix print wrong version issue
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Upgrade efivar from 0.7 to 0.21.
* update HOMEPAGE and git repo.
* update license file checksum. It made FSF addresses in COPYING be URLs.
Ref: https://github.com/rhinstaller/efivar/commit/ca88012
* add -native recipe to provide command makeguids
and add patch for target package to use this native makeguids command
* add patch to build efivar-native with lower version gcc which doesn't
support options '-Wmaybe-uninitialized' and '-std=gnu11'
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Polkit is currently broken in images built with multilib and systemd.
This is because the patch, 0001-do-not-hardcoded-libdir.patch, applied on
top of the polkit source code modifies where the polkitd binary is
installed, but it does not modify the polkit.service file to start the
binary from its new location.
At first it seemed reasonable to modify the systemd service file to
search for the binary in the correct place. This change, as well as what
the patch (0001-do-not-hardcoded-libdir.patch) already does was proposed to
the polkit maintainers at https://bugs.freedesktop.org/show_bug.cgi?id=92094
During the discussion with the polkit maintainers it became apparent that the
change to support multilib polkit should not be done with a patch to
the polkit source code, but instead a change to the polkit recipe.
Polkit correctly installs libraries when multilib is in use without any
changes to its source code. What is being changed by
0001-do-not-hardcoded-libdir.patch is not where the polkit libraries are
installed but where the binaries are installed.
Installing binaries in /usr/lib when baselib is lib64 is acceptable (see
http://refspecs.linuxfoundation.org/FHS_3.0/fhs/ch04s06.html ).
So, instead of patching polkit to install its binaries under the same
library directory as its libraries we maintain the design of the polkit
installer to install the binaries in /usr/lib. This is the same as what is
done in distros like Fedora that supports multilib.
With this patch the polkit package, when built with multilib, installs
files into /usr/lib* as follows:
polkit/usr/lib64/libpolkit-agent-1.so.0
polkit/usr/lib64/libpolkit-gobject-1.so.0
polkit/usr/lib64/libpolkit-gobject-1.so.0.0.0
polkit/usr/lib64/libpolkit-agent-1.so.0.0.0
polkit/usr/lib
polkit/usr/lib/polkit-1
polkit/usr/lib/polkit-1/polkitd
polkit/usr/lib/polkit-1/polkit-agent-helper-1
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* to resolve QA warnings:
sblim-sfcCommon-1.0.1: sblim-sfcCommon-dev doesn't match the [a-z0-9.+-]+ regex [pkgname]
sblim-sfcCommon-1.0.1: sblim-sfcCommon-staticdev doesn't match the [a-z0-9.+-]+ regex [pkgname]
sblim-sfcCommon-1.0.1: sblim-sfcCommon doesn't match the [a-z0-9.+-]+ regex [pkgname]
sblim-sfcCommon-1.0.1: sblim-sfcCommon-doc doesn't match the [a-z0-9.+-]+ regex [pkgname]
sblim-sfcCommon-1.0.1: sblim-sfcCommon-dbg doesn't match the [a-z0-9.+-]+ regex [pkgname]
sblim-sfcCommon-1.0.1: sblim-sfcCommon-locale doesn't match the [a-z0-9.+-]+ regex [pkgname]
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Add PACKAGECONFIG for x11 and enable it when x11 is
included in DISTRO_FEATURES, and specify the correct
path for x-includes and x-libraries to avoid
searching the host libraries.
Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
After rsyslog updated to 7.6.1, ptest fails to build & run since the
serial-tests config patch was dropped:
- rsyslog-use-serial-tests-config-needed-by-ptest.patch
This patch involved serial test harness:
+ AM_INIT_AUTOMAKE([serial-tests])
Which is deprecated in favour of parallel test harness. Automake
generated a parallel test harness by default. It features automatic
collection of the test scripts output in .log files ...
More details please refer to:
http://www.gnu.org/software/automake/manual/html_node/Parallel-Test-Harness.html
To enable the ptest with new changes, we should:
* add target 'buildtest-TESTS' for building test components;
* fix 'top_srcdir' and install required script 'test-driver';
* run testcases with target 'check-TESTS'.
Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* remove trailing whitespaces
* run through oe-stylize
Signed-off-by: Andrea Galbusera <gizero@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
use 0644 instead of 0755 to install configuration file
Signed-off-by: Andrea Galbusera <gizero@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
1. fix a floating dependence on openjdk
2. fix the host contamination, since the default value of java home
is /usr/lib/jvm:
warning: library search path "/usr/lib/jvm/jre-1.7.0/lib/amd64/server" is unsafe for cross-compilation
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Meta-virtualization and meta-luv both has iasl_*.bb, which is a subset of this recipe. Extend this recipe to provide the native tools and add the necessary runtime conflict options.
Also address style issues by moving around a few bits and adding some whitespace.
Signed-off-by: Koen Kooi <koen.kooi@linaro.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Even though the PACKAGECONFIG of libvirt is used to fix the floating
dependence, but enable-libvirt/disable-libvirt become useless after
collectd upgrade, and libvirt is checked unconditionally, and lead to
a floating dependence on libvirt.
check if libvirt is available only when a user wants to use libvirt
to fix this issue
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
remove use-pkg-config-for-ncurses-detection.patch, the new version is
using the pkg-config to check ncurses
configuration assumes the cf_have_ncuconfig is unknown for cross-compile,
but ncurses is added into DEPENDS, so change it as yes
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
It fails to build efibootmgr with gcc 5.2:
src/lib/disk.o: In function `disk_get_scsi_pci':
src/lib/disk.c:154: undefined reference to `get_scsi_pci'
src/lib/efi.o: In function `make_edd30_device_path':
src/lib/efi.c:474: undefined reference to `get_scsi_idlun'
collect2: error: ld returned 1 exit status
Because functions get_scsi_pci and get_scsi_idlun are declared as
inline. It has different semantics for inilne function between gcc 4
and gcc 5.
Pass flag '-fgnu89-inline' to gcc to fix the errors.
Ref:
https://gcc.gnu.org/gcc-5/porting_to.html
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
not use github archives anymore, they are regenerated from time to time with different checksums.
Signed-off-by: Lei Maohui <leimaohui@cn.fujitsu.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Mark Asselstine
Li xin
Koen Kooi
Martin Jansa
Yi Zhao
Maciej Borzecki
Qi.Chen@windriver.com
Christopher Larson
Hongxu Jia
Dan McGregor
Jens Rehsack
Roy Li
Joe Slater
Jagadeesh Krishnanjanappa
Khem Raj
Kai Kang
Reinette Chatre
Jackie Huang
Wenzong Fan
Andreas Müller
Andrea Galbusera
leimaohui
Bian Naimeng