Age | Commit message (Collapse) | Author |
|
* Added the -h option to the tar command to make sure that any
patches that may have been applied by the recipe are actually
copied instead of made as symlinks to the recipe directory.
* This is so that people can see the actual patches and unapply
individual patches as desired.
Signed-off-by: Chase Maupin <Chase.Maupin@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
|
|
does, so fix that
Signed-off-by: Koen Kooi <koen@openembedded.org>
|
|
when installing icons
Also import a fix from OE-core for setting the loader dir.
Tested with gnome-icon-theme and libsoup recipes on angstrom.
Signed-off-by: Koen Kooi <koen@openembedded.org>
Acked-by: Steffen Sledz <sledz@dresearch-fe.de>
|
|
* This patch fixes fetching pre-built staging packages from PSTAGE_MIRROR having file:// uri
see: http://lists.linuxtogo.org/pipermail/openembedded-devel/2011-July/033938.html
(thanks to Tom Rini for helping me fixing this)
Signed-off-by: Fabrice Aeschbacher <fabrice.aeschbacher@siemens.com>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
Acked-by: Tom Rini <tom_rini@mentor.com>
Signed-off-by: Paul Menzel <paulepanter@users.sourceforge.net>
(minor improvement to commit message)
Acked-by: Paul Menzel <paulepanter@users.sourceforge.net>
|
|
The cpan build system unwrapps our perl -> perl.real shell script
and generates Makefile's that call perl.real directly and thus fail
to be relocable. So at the end, regex them back to calling perl.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Fixes "AttributeError: 'NoneType' object has no attribute 'split'" during
parsing with bitbake master. We should not be calling explode_deps with
None as the argument, so ensure the value isn't None.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Acked-by: Andrea Adami <andrea.adami@gmail.com>
|
|
* Add the include/crypto directory from the kernel to the list
of directories to be staged.
* This directory contains the cryptodev.h file required by other
drivers and applications.
* Not including this directory leads to build errors of being
unable to find the cryptodev.h file.
Signed-off-by: Greg Turner <gregturner@ti.com>
Signed-off-by: Chase Maupin <Chase.Maupin@ti.com>
Acked-by: Denys Dmytriyenko <denys@ti.com>
Acked-by: Roman I Khimov <khimov@altell.ru>
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
|
|
Removes legacy staging from QtE 2.3.10. Includes & libraries are now no
longer staged in a special location - they are as they would be installed
on the target system (/usr/include/qte and /usr/lib respectively).
qmake.bbclass has been updated to reflect this.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
* taken from OE-Core [1]
* thanks Khem for pointing out[2]
* fixes build failure perl
| env: ./perl5.10.1.real: No such file or directory
| make[1]: *** [pod/perlapi.pod] Error 127
[1] http://git.openembedded.org/cgit.cgi/openembedded-core/commit/?id=a10bd976f4cef54ac50b0c82f885c17a26e5989f
[2] http://lists.linuxtogo.org/pipermail/openembedded-core/2011-May/002975.html
Signed-off-by: Andreas Mueller <schnitzeltony@gmx.de>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
- Fixes build with certain vendor-supplied Montavista kernels.
- Pass the same set of variables to every invocation of make.
- Create KERNEL_*SUFFIX, TARGET_*_KERNEL_ARCH and HOST_*_KERNEL_ARCH
for the three variables, analogue to LD.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Acked-by: Phil Blundell <philb@gnu.org>
|
|
- KERNEL_VERSION may contain characters unsuitable for package
names, e.g. underscores. Use legitimize_package_name to replace
those characters.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Acked-by: Phil Blundell <philb@gnu.org>
|
|
that it shows up on the console
Signed-off-by: Koen Kooi <koen@openembedded.org>
Acked-by: Philip Balister <philip@balister.org>
Acked-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Philip Balister <philip@balister.org>
|
|
In the default configuration cmake does not set a rpath for its builded binaries. This
leads to errors at runtime when the binaries not find their needed runtime libraries.
Please note that this is only done for recipes inheriting the native class!
Signed-off-by: Simon Busch <morphis@gravedo.de>
Acked-by: Phil Blundell <philb@gnu.org>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
* It is more consistent.
* Fixes subtle binconfig breakages:
* binconfig were populated with a different name that upstream intended
* populated binconfig files that were not intended to be populated
* See http://lists.linuxtogo.org/pipermail/openembedded-devel/2011-April/031972.html
Signed-off-by: Stanislav Brabec <utx@penguin.cz>
Acked-by: Andreas Oberritter <obi@opendreambox.org>
Acked-by: Paul Menzel <paulepanter@users.sourceforge.net>
Acked-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
|
|
* reported by Otavio Salvador
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Acked-by: Koen Kooi <koen@dominion.thruhere.net>
|
|
later
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
|
|
The host download.osgeo.org is not reachable from some network/country
because of some routing issue, and this makes the do_fetch task of the
tiff recipe to fail.
Add a reachable mirror for it, so that libtiff can build fine.
Signed-off-by: Antonio Ospite <ospite@studenti.unina.it>
Acked-by: Phil Blundell <philb@gnu.org>
|
|
* 4f883343de0427769de3df96ed8adaa589279112 could move
bb_persist_data.sqlite3 from default CACHE tmp/cache/machine/
to global tmp/cache/
* because I have 1st report of wrong SRCPV behavior, add sanity check from
http://patches.openembedded.org/patch/823/
even without moving it outside TMPDIR, in case of CACHE != PERSISTENT_DIR
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
|
|
This patchset enables fennec to build again
Upgrades to latest release tag
Bunch of patches to fix uclibc build
Disable few options notably --disable-elf-hack
Add -L<sysroot>/usr/lib to LDFLAGS otherwise
make tried to find the libs in /usr/lib
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
* suggested by kergoth on #oe as response to similar patch limited to gtk+
http://lists.linuxtogo.org/pipermail/openembedded-devel/2011-March/031074.html
15:03 < kergoth_> JaMa: pretty sure we just need PACKAGES_DYNAMIC = ""
and PACKAGES_DYNAMIC_virtclass-native = "" in native.bbclass.
everyone i've talked to so far as agreed that it's a sane approach,
including RP.
* otherwise couple of errors like this are shown
NOTE: multiple providers are available for runtime gdk-pixbuf-loader-gif (gtk+-native, gtk+, gtk+-fastscaling)
NOTE: consider defining a PREFERRED_PROVIDER entry to match gdk-pixbuf-loader-gif
ERROR: Trying to resolve runtime dependency gdk-pixbuf-loader-xpm resulted in conflicting PREFERRED_PROVIDER entries being found.
The providers found were: ['/OE/dev/recipes/gtk+/gtk+_2.20.1.bb', 'virtual:native:/OE/dev/recipes/gtk+/gtk+_2.20.0.bb']
The PREFERRED_PROVIDER entries resulting in this conflict were: ['PREFERRED_PROVIDER_gtk+ = gtk+', 'PREFERRED_PROVIDER_gdk-pixbuf-csource-native = gtk+-native']
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Acked-by: Otavio Salvador <otavio@ossystems.com.br>
|
|
originally I had another couple of utility functions in this class but
I removed them before pushing upstream. Rename now so its purpose is
clearer.
Signed-off-by: Graeme Gregory <dp@xora.org.uk>
|
|
Use the DISTRO variable for the name of the distro making this class
distro neutral.
Signed-off-by: Graeme Gregory <dp@xora.org.uk>
|
|
The current image link-creation code will unconditionaly create a link
from .<type> pointing to the .rootfs.<type> output.
This is not compatible with the UBIFS images produced which have
.<type>.img extension since they are not considered to be valid rootfs
images when they are not included in a UBI container.
Check for existence of the link target .rootfs.<type> before creating the
link and fallback to a check for the .<type>.img target after that.
Signed-off-by: Ben Gardiner <bengardiner@nanometrics.ca>
Acked-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
|
|
With BBCLASSEXTEND recipes which also modify DEPENDS_virtclass-native
gettext.bbclass is otherwise not working and gettext-native isn't
added. So we need to have a DEPENDS_virtclass-$foo_prepend to
make sure GETTEXT_DEPENDS gets in.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
This should be handled via virtual/libintl already.
Acked-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Virtually all gnome recipes pass this already, this just fixes
a few such as zenity.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
In a few functions we checked if pstagingactive was set and if so, grabbed
the lock. But really, we should always be locking here. And inside of
package_do_shlibs we need to keep the lock a little longer than we had
been otherwise we can run into a race..
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
A distro can require conf/licenses.conf or set SRC_DISTRIBUTE_LICENSES
selecting which licenses are wanted to be distributed. In case it is
empty or undefined we default to distribute all licenses to not break
backward compatibility.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
This is nice in theory, but there's a risk that a broken source tree could try
to modify the files through the symlinks, screwing up everybody else using
them. Of course, per-recipe sysroots would resolve this, so we can revisit
this if/when that gets implemented.
This reverts commit 302395545185b4e85bf76a8f319bc71bd3a7be1d.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Commit 9c5b9db29738a89837917b1a2d8ee5a1cae0b4cc (bitbake.conf: Simplify
TARGET_CPPFLAGS/LDFLAGS) potentially breaks external toolchains, when their
default sysroot is different from our sysroot/staging location.
Pass default --sysroot to compiler/linker through TOOLCHAIN_OPTIONS, avoiding
native and nativesdk packages. This matches the corresponding changes in Poky.
cross.bbclass was already doing it, since it originated from Poky, AFAIK.
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Acked-by: Tom Rini <tom_rini@mentor.com>
|
|
* Allow precompiled modules to depend on a specific kernel version.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Until now the Packages.gz generation was hardcoded in rootfs_deb_do_rootfs.
(Tom Rini: While in here, trailing whitespace cleanup)
Signed-off-by: Filip Zyzniewski <filip.zyzniewski@gmail.com>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
This writes a qt.conf inside WORKDIR to properly configure projects
based on CMake. This is required since qmake variables (returned by
-query command) are fixed into the binary and can only be changed
using a qt.conf file.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
dpkg-deb does not like version numbers without digits:
NOTE: Running task 428 of 604 (ID: 14,
[...]/openembedded/recipes/linux-firmware/linux-firmware_git.bb,
do_package_write_deb)
dpkg-deb - error: (upstream) version (`git') doesn't contain any digits
dpkg-deb: 1 errors in control file
NOTE: package
linux-firmware-git-r1+gitr2fab48391fda8a914b55c8ac265eb3b6ee10ace8:
task do_package_write_deb: Started
This commit causes version numbers without digits to be appended with '0':
linux-firmware-agere_git0-r1+gitr2fab48391fda8a914b55c8ac265eb3b6ee10ace8_all.deb
Signed-off-by: Filip Zyzniewski <filip.zyzniewski@gmail.com>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
* This patch removes the absolute path in sysroots/<arch>/usr/lib/libstdc++.la
*
* It was build tested wih clean tmp for xfce46-image. Additional known
* QA-RPATH-error-candidates were checked for successful build:
*
* aspell-0.60.6-r0
* curl-7.21.2-r4.0
* clutter-box2d-0.10.0-r2.0+git4e72b9c135a9439b82d801c62422531db611f98b
* enchant-1.6.0-r2
* gnome-vfs-2.24.1-r3
* libcdio-0.82-r3.0
* libsoup-2.2.100-r1:
* neon-0.25.5-r4
* schroedinger-1.0.9-r0
* tiff-3.9.4+4.0.0beta6-r0
*
* Build environment was (RPATH QA errors treated as fatal / acl fixed manually)
*
* BB_VERSION = "1.10.2"
* METADATA_BRANCH = "cross-bbclass-test"
* METADATA_REVISION = "a44300e"
* TARGET_ARCH = "arm"
* TARGET_OS = "linux-gnueabi"
* MACHINE = "overo"
* DISTRO = "angstrom"
* DISTRO_VERSION = "2010.7-test-20110219"
* TARGET_FPU = "hard"
*
* Notes:
* - To have effect, gcc-cross has to be rebuild (changing all gcc packed
* was not an option)
* - This patch does not affect libtool-2.4 distos
* - The story is found at
* http://lists.linuxtogo.org/pipermail/openembedded-devel/2011-February/030115.html
Signed-off-by: Andreas Mueller <schnitzeltony@gmx.de>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
A future TODO is to see if there's anything left that must
build with ARCH=ppc in the kernel that can't be upgraded to a
newer kernel with ARCH=powerpc.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
* see comment for reason why we need this
* more info:
http://lists.linuxtogo.org/pipermail/openembedded-devel/2011-February/029877.html
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
|