| Age | Commit message (Collapse) | Author | Files |
|---|
|
* BASEPKG_HOST_SYS from bitbake.conf is set as if packages
were being built for target. So we need to override it
for cross packages.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Andrea Adami <andrea.adami <at> gmail.com>
|
|
* Bump the ABI_LAYOUT. Build from scratch will be needed.
* Replace using TARGET_OS from linux->linux-gnu.
* Replace using linux-uclibcgnueabi->linux-uclibceabi.
* Add 'eabi' to DISTRO_FEATURES (only for minimal and micro).
* Use eabi and BASE_PACKAGE_ARCH to compute real TARGET_OS.
* Fix the micro conf to get console image building.
* Fix the linux-uclibcgnueabi overrides in all recipes.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Phill Blundell <pb@reciva.com>
Acked-by: Michael 'Mickey' Lauer <mickey@vanille-media.de>
|
|
management
|
|
These are created with the package and get installed in
/var/dpkg/info. Afterward it's a great way to find modified files
for backup with a little shell script magic.
It feels a bit weird to still use MD5, but that seems to be the
convention in the Debian world.
Signed-off-by: Michael Smith <msmith@cbnco.com>
Acked-by: Chris Larson <clarson@kergoth.com>
|
|
On Mac OSX its not able to find md5sum if
path is not passed to os.popen shell.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Koen Kooi <koen@openembedded.org>
|
|
This class uses events to capture the state of the datastore when the task
starts, and after it completes. It diffs those captured states, and emits
messages showing which variables changed, and what their values were changed
to.
It provides a mechanism to blacklist variables you expect to change, both
globally and on a per-task basis.
Signed-off-by: Chris Larson <clarson@kergoth.com>
|
|
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
Can be particularly useful when using multiple srctree recipes.
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
This fixes issues seen with the use of kernel.bbclass, so in existing source
tree kernel builds work happily now.
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
This should be an opt-in feature.
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
into org.openembedded.dev
|
|
This provides a GITVER variable which is a (fairly) sane version, for use in
${PV}, extracted from the ${S} git checkout, assuming it is one. This is most
useful in concert with srctree.bbclass.
Signed-off-by: Chris Larson <clarson@kergoth.com>
|
|
This enables operation inside of an existing source tree for a project, rather
than using the fetch/unpack/patch idiom.
By default, it expects that you're keeping the recipe(s) inside the
aforementioned source tree, but you could override S to point at an external
directory and place the recipes in a normal collection/overlay, if you so
chose.
It also provides some convenience python functions for assembling your
do_clean, if you want to leverage things like 'git clean' to simplify the
operation.
Signed-off-by: Chris Larson <clarson@kergoth.com>
|
|
Signed-off-by: Chris Larson <clarson@kergoth.com>
|
|
into org.openembedded.dev
|
|
It can be selected by setting PATCHTOOL = "git".
This is useful because git-apply honors the permissions information
produced by git-format-patch.
Signed-off-by: Dale Farnsworth <dfarnsworth@mvista.com>
Signed-off-by: Chris Larson <clarson@mvista.com>
Acked-by: Denys Dmytriyenko <denis@denix.org>
|
|
Currently the only thing in EXTRA_OEMAKE is '-e MAKEFLAGS='. We don't
want to overide the kernel's Makefile variables from the environment.
It caused the passed -j<N> parameter from PARALLEL_MAKE to be ignored.
Signed-off-by: Dale Farnsworth <dfarnsworth@mvista.com>
Signed-off-by: Chris Larson <clarson@mvista.com>
Acked-by: Denys Dmytriyenko <denis@denix.org>
|
|
Signed-off-by: Chris Larson <clarson@mvista.com>
Acked-by: Michael Smith <msmith@cbnco.com>
|
|
python-native isn't in DEPENDS.
Signed-off-by: Chris Larson <clarson@mvista.com>
Acked-by: Michael 'Mickey' Lauer <mlauer@vanille-media.de>
|
|
Will allow reuse by xorg-xserver.
Signed-off-by: Michael Smith <msmith@cbnco.com>
|
|
Often gzip is reporting broken pipe errors with do_unpack of
tar.gz files.
If you use the commands described above to extract a tar.gz file, gzip
sometimes emits a Broken pipe error message. This can safely be ignored
if tar extracted all files without any other error message.
We do not let python install its SIGPIPE handler and use subprocess call
to invoke the command.
This is based on the following python bug report.
http://bugs.python.org/issue1652
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Some packages include <linux/types.h>, which requires <asm/types.h>.
Supplement b6220af33a70d96345dea130005ce842eeadfdf5 to copy files
from arch/x86/include/asm to include/asm-x86.
Signed-off-by: Michael Smith <msmith@cbnco.com>
Acked-by: Roman I Khimov <khimov@altell.ru>
|
|
Signed-off-by: Jeremy Puhlman <jpuhlman@mvista.com>
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
Works around src_distribute failures with the use of packaged staging.
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
- Kill the big shell function with the case statement used in _local in
favor of a set of bitbake variables, with the appropriate one selected
as SRC_DISTRIBUTECOMMAND.
- Add and use SRC_DISTRIBUTE_DLONLY to toggle distribution of all sources
versus just those downloaded to DL_DIR. This variable expects to be set
to an integer, anything but 0 will enable dlonly operation. This is
disabled by default.
- Add a more sanely named distribute_sources_all, and make distsrcall
depend on it, for compatibility.
- By default, distributes the sources into
${SRC_DISTRIBUTEDIR}/${LIC}/${PN}/.
- Move some _local specific bits, like use of SRC_DISTRIBUTEDIR, out of
the main class.
- Set a "LIC" var along with the "SRC" when running SRC_DISTRIBUTECOMMAND,
to facilitate the above.
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
DBUS_SESSION_BUS_ADDRESS.
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
In the past, it only avoided adding the rpm-native/fakeroot-native deps if
PACKAGES is empty, but unfortunately some native recipes have PACKAGES set to
non-empty, so let's explicitly check for native/cross.
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
- Add base_ifelse convenience function to base.bbclass.
- Replace all conditionals in the recipes relating to hostap and madwifi to
look for their name in COMBINED_FEATURES rather than looking for specific buses (pci, pcmcia).
- Change the default COMBINED_FEATURES to enable:
- madwifi, when:
- distro has pci, wifi, and madwifi in its features
- machine has pci in its features
- hostap, when:
- 'wifi' and 'hostap' are in distro features
- either 'pci' or 'pcmcia' are in both distro and machine features
Signed-off-by: Chris Larson <clarson@mvista.com>
|
|
Updating recipes and checksums.ini for vanishing upstream hosts or simple
restructuring can quickly become a chore for OE devs if taken seriously.
The OE user experience suffers severely when at any given point in time
4 to 5 recipes on average fail do_fetch for a simple image.
The solution is for OE to provide its own caching of upstream sources.
mirrors.openembedded.org is a swarm of possibly incomplete source mirrors
provided by the community and load-balanced via DNS round robin. They
act as a first line of defense and lower bandwidth load on melo which
hosts sources.openembedded.org. The melo mirror aims to have a complete
copy of free-to-publish sources used by the current org.oe.dev branch.
More information at http://wiki.openembedded.net/index.php/Sources_mirror
Acked-by: Holger Freyther <zecke@selfish.org>
Acked-by: Chris Larson <clarson@kergoth.com>
|
|
image.bbclass is trying to inherit package_${IMAGE_PKGTYPE} and
this will fail because a rootfs_tar.bbclass doesn't exist. When
planning to build images one needs to inherit package_ipk,
package_rpm or package_deb in addition to INHERIT += "package_tar".
Signed-off-by: Holger Hans Peter Freyther <zecke@selfish.org>
Acked-by: Philip Balister <philip@balister.org>
|
|
* Acked-By: Holger Freyther <zecke@selfish.org>
* Acked-by: Phil Blundell <philb@gnu.org>
|
|
Similar to KERNEL_IMAGE_BASE_NAME, allow overwriting from a recipe
Signed-off-by: Denys Dmytriyenko <denis@denix.org>
Acked-by: Henning Heinold <heinold@inf.fu-berlin.de>
|
|
We were using EXTRA_PACKAGE_ARCHS which was sort of misleading
if you started to search for this variable. It was a typo
hence fixed as obvious.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
fr-fr and thus locale-fr packages were bing omitted)
|
|
pkg_postinst_append is executed after the class postinst.
This is important since recipes like gpanel need to to execute gconftool-2 --load after the gconf class code has processed the schema info
|
|
|
|
This is fixing the compilation of libtool-sdk_2.6.6a
which is currently the only user of BPN.
|
|
and do_install as discussed on ML.
linux.inc : removed uImage creation, its now in kernel.bbclass
Signed-off-by: Graham Gower <graham.gower@gmail.com>
Acked-by: Graeme Gregory <dp@xora.org.uk>
Acked-by: Koen Kooi <koen@openembedded.org>
|
|
|
|
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
|
|
This reverts commit 736c06e8d8efa79d3d2bc512f13a51f0f63412e2
as this change breaks future versions of bitbake.
See the OE mailing list, in the "Bitbake runqueue performance improvement"
thread for a discussion on why this isn't the best way to solve
multi-image problem and that other better solutions exist.
|
|
* Also, how do I get the build banner back? This bug went unnoticed due to it being gone
|
|
|
|
multiple-machine-share-same-kernel problem
|
|
a recipe blacklister
|
|
|
|
Allow a package to be built that contains only dotfiles or
dot directories under '/'.
Signed-off-by: Michael Smith <msmith@cbnco.com>
|
