The -dev has a runtime dependency against ${PN} which is empty; we
allow it to be empty to satisfy the dependency instead of mangingling
the -dev dependencies.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* This updates to the 3.4.2 release and rework the packaging splitting
the C bindings in another binary package to reduce footprint in
rootfs; this also moves geos-config to the development package as it
should.
* Fix -dev installation and upgrade path
Using 'lib_package' renames the binaries and we need a transitional
package to be used as upgrade path. This is done using an empty 'geos'
package which installs the new ones as runtime dependency.
This same package is used to satisfy -dev dependency of geos package
allowing for its inclusion into SDK.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
The pointercal-xinput has been moved to OE-Core so we should handle it
here.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Includes a patch from Zeroc to help with ice builds.
Signed-off-by: Philip Balister <philip@balister.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
xfconf-4.10.0 does not use OE sysroot as is.
--with-sysroot=yes
will make configure run CC --print-sysroot
and use that.
When searching the PATH for perl, no target perl is found,
since PATH does not contain any reference.
Instead native perl is found, and if certain perl packages
are installed on the host machine, the xfconf perl support
is enabled which may cause the build to fail on 64 bit hosts.
If the perl packages are not installed on the host, then
perl support in xfconf is disabled, and the xfconf build
will succeed, even on 64 bit hosts.
If native perl is found, and the perl packages are available
on the host, the perl support is enabled.
Since native perl is used, the native linker is used to link the
resulting *target* code.
The perl support requires some libraries, which are built for the target.
If the host is 64 bit (x86_64) and the target is 32 bit (ARM?),
there is a mismatch which is detected by the linker,
and the libraries found are rejected.
If the host does not have the correct libraries,
then the xfconf build fails.
If the host has the correct libraries, then the linker will try
to link host libraries with the target application.
The patch will define PERL as the target perl.
This means that configure will look for the required perl
packages in the target sysroot, instead of in the native filesystem.
The patch has been tested in Angstrom master, which does not
build the required perl packages, and thus perl support
in xfconf is disabled, and the build succeeds.
A simpler way to do this, proposed by Charles Ash <ashcharles@gmail.com>
is to simply disable the perl-support in xfconf.
This patch will be better, if ever the support for the needed
perl packages is added to the OpenEmbedded/Angstrom build.
Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
Tested-by: Ash Charles <ashcharles@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
This recipe follows gnome-3.2 release and SHA
in SRCREV also belongs to same branch.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Below errors only occurs on Ubuntu 13.10:
$arch-linux-libtool: link: g++ ... -o .libs/mysqltest_embedded \
../../libmysqld/.libs/libmysqld.so -ldl
../../libmysqld/.libs/libmysqld.so: undefined reference to `dlopen'
../../libmysqld/.libs/libmysqld.so: undefined reference to `dlclose'
../../libmysqld/.libs/libmysqld.so: undefined reference to `dlerror'
../../libmysqld/.libs/libmysqld.so: undefined reference to `dlsym'
GCC/ld verion on the host:
gcc (Ubuntu/Linaro 4.8.1-10ubuntu8) 4.8.1
GNU ld (GNU Binutils for Ubuntu) 2.23.52.20130913
This is a strange behavior on Ub13.10, it fails even '-ldl' in the
link command line. This patch will append '-ldl' to dependency_libs
in libmysqld.la.
Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Add lua 5.2.2 recipe.
Update:
* remove bitwise_operators.patch that lua uses bit32 library to
support for bitwise operations.
* update uclibc-pthread.patch
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
When enable multilib on x86-64, it fails to build lib32-rsyslog with
error:
| ERROR: Function failed: SYSTEMD_SERVICE_lib32-rsyslog value lib32-rsyslog.service does not exist
It uses wrong value ${PN}.service to set SYSTEMD_SERVICE. Correct it by
using BPN.
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Previously, modphp estimates endian on host rather than checks it on
target. If the host is little-endian and the target is big-endian,
modphp claims that endian is little. As a result, a memory location
that it is not allowed to access when calling libphp5.so module on
target. It will occur segmentation fault.
This patch enables endian check support for modphp.
Signed-off-by: Chong Lu <Chong.Lu@windriver.com>
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Use /bin/false as the login shell, just like what Ubuntu does,
otherwise there might be secure issue; add /var/lib/ftp as user
ftp home-dir.
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
Initial recipe created by Steve Arnold. Original build patch and 1.0.10
recipe graciously contributed by bencoh (in #oe on irc.freenode.net).
New recipe and init script contributed by this author. Built and
tested on master branches using author's fork of meta-raspberrypi.
Signed-off-by: stephen.arnold42 <stephen.arnold42@gmail.com>
Signed-off-by: Sébastien Mennetrier <s.mennetrier@innotis.org>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
* oe-core version of xinput-calibrator installs
30xinput_calibrate.sh and ls -X used in Xsession script sorts it as
last (not respecting the numbers) and xinput-calibrator is executed
too late (after 90xXWindowManager)
* don't forget to update all your components installing Xsession.d
files as well
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Pass LIB_SUFFIX to CMake during a multilib build so that the generated library folder has the correct suffix.
Signed-off-by: Nick D'Ademo <nickdademo@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
The 'config' script which gets run returns '../scripts/CONFIG.<machine>', so
the current value was '${datadir}/lmbench/../scripts/CONFIG.<machine>', which
is obviously not ideal. Remove the $SHAREDIR prefix, as the relative path will
work from the scripts dir where lmbench-run runs. With this, we can install
a custom CONFIG.qemux86 into ${datadir}/lmbench/scripts/, for example.
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Rather than hardcoding ${datadir}/rubygems, we can let it find the gems
relative to its own location, to make it relocatable.
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Pass libdir to EXTRA_OECONF so that recipe generates the correct library folder in a multilib build.
Install shared library symbolic link to image.
Signed-off-by: Nick D'Ademo <nickdademo@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Explicitly set the path to the udev dir to ${base_libdir} in EXTRA_OECONF. Otherwise, the multilib build will fail as it will look for udev in the default location of /lib/udev/.
Signed-off-by: Nick D'Ademo <nickdademo@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* has runtime dependency on TUNE_PKGARCH usb-modeswitch
Hash for dependent task usb-modeswitch_1.2.5.bb.do_packagedata changed from 5709ee415d286847b58e7b438b5b9f75 to fbef5eee3bb2bacb805a0bead2095b52
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* runtime dependencies are TUNE_PKGARCH causing do_package_write_*
task to have different signature for MACHINEs with different
TUNE_PKGARCH
* move kernel-module-zram to RRECOMMENDS, because signature handler
is excluding kernel-module-* dependencies only from RRECOMMENDS and
also because some machine can have zram enabled in kernel and
kernel-module-zram won't exist.
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
* runtime dependencies are TUNE_PKGARCH causing do_package_write_*
task to have different signature for MACHINEs with different
TUNE_PKGARCH
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
WARNING: QA Issue: inetutils: configure was passed unrecognised options: --with-path-login --with-path-procnet-dev
Neither of these appear to be valid for inetutils-1.9.1, so remove them.
Signed-off-by: Joe MacDonald <joe@deserted.net>
Correct the usage of alternative link and target:
ALTERNATIVE_LINK_NAME[name] = "target"
ALTERNATIVE_TARGET[name] = "target"
Signed-off-by: Li Wang <li.wang@windriver.com>
Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
The default ntp.conf file does the right thing but one of the comments is
a bit misleading based on the active parts of the file changing a few
times. Update the comment to accurately describe what is happening in the
configuration file and what a user should do next.
Signed-off-by: Joe MacDonald <joe@deserted.net>
When ntpd starts, it will create a daemon to connect this invalid server,
and fail, then ntpd will create other daemon to connect the local-only
(127.127.1.0) server.
The users should be aware that they need to configure the correct ntp
servers, a invalid server will lead to a redundant daemon, so it is better
to comment this example server to wait user to configure
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
tcpreplay - a suite of GPLv3 licensed tools for UNIX operating systems
which gives you the ability to use previously captured traffic in libpcap
format to test a variety of network devices.
Signed-off-by: Zongchun Yu <Zongchun.Yu@freescale.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
/var/run/ is more suitable to store pidfile for OE, and this fixes
"/etc/init.d/proftpd stop" failure too, since this script assumes
the pidfile is under /var/run
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
According to DISTRO_FEATURES to add pam support for quagga, and import
configure file from Fedora.
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
Change the default config file for making watchquagga can startup successful
Signed-off-by: Bopeng Liu <Bopeng.Liu@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
commit 27ba970b9("quagga/ripd: Fix two bugs after received SIGHUP signal")
introduces an regression: ifp->connected list is cleaned up when ripd is
restarting, however, for interface addresses which are not specified in
ripd configuration file, they are never to be added into ifp->connected
again, this will lead to some abnormal behavior for route advertising.
Instead of cleaning up the ifp->connected list to avoid duplicated
connected address being added into this list, we can check this
condition during interface address adding process and return early
when an identical address has already been added.
Signed-off-by: Xufeng Zhang <xufeng.zhang@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
Use /bin/false as the login shell, just like what Ubuntu does,
otherwise there might be secure issue.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Joe MacDonald <joe@deserted.net>
