Age | Commit message (Collapse) | Author |
|
The files in the .pc directory won't have anything useful in them and are
often unreadable so lets just skip them.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
Acked-by: Khem Raj <raj.khem@gmail.com>
|
|
|
|
commit b6cc4bc217e695801db060a1366915a870cab5db moved work
from kernel_do_install to sysroot_stage_all_append
This change changed kerneldir and staged directy from
${SYSROOT_DESTDIR}${STAGING_KERNEL_DIR}. This works fine but
the patch removed the line kerneldir=${D}/kernel/
from do_install.
However, some kernel recipes have a do_install_append which
references ${kerneldir}.
This patch adds setting kerneldir to the end of do_install
so the value is available for everyone who appends
Signed-off-by: Frans Meulenbroeks <fransmeulenbroeks@gmail.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>
|
|
We want to get zoneinfo from the tzdata package.
Acked-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
copy the files directly from the work area to staging. That way
there are no unpackaged files in image/ so no NOTE messages.
Additional advantage is that it is faster as we save the copy to the image dir.
See also http://thread.gmane.org/gmane.comp.handhelds.openembedded/37487
Signed-off-by: Frans Meulenbroeks <fransmeulenbroeks@gmail.com>
Acked-by: Roman I Khimov <khimov@altell.ru>
Tested-by: Petr Stetiar <ynezz@true.cz>
Acked-by: Eric Benard <eric@eukrea.com>
Acked-by: Denys Dmytriyenko <denis@denix.org>
Tested-by: Paul Menzel <paulepanter@users.sourceforge.net>
|
|
This reverts commit 9a5382aad4e5a3e52fa52722cf5587fdba958e9f.
This one accidentally also dragged in classes/kernel.bbclass
I will redo the commit for wpa-supplicant later.
classes/kernel.bbclass will be se a separate commit
Thanks for Paul Menzel for spotting this
Signed-off-by: Frans Meulenbroeks <fransmeulenbroeks@gmail.com>
|
|
Signed-off-by: Frans Meulenbroeks <fransmeulenbroeks@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
libdir to be something else than '/lib'
* Add base_libdir_native as target libdir could be different e.g. lib32
lib64
* Dont assume lib to be system libdir during packaging.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Chris Larson <chris_larson@mentor.com>
|
|
bb.fetch.urldata contains cached SRC_URIs, but in case SRC_URI is altered by
amend.inc or machine override, the original SRC_URI is not removed from the
cache, leading to calling do_unpack() multiple times on original and modified
versions of the file. Don't rely on bb.fetch.urldata and traverse SRC_URIs
directly.
Some background and details:
http://thread.gmane.org/gmane.comp.handhelds.openembedded/36703
Signed-off-by: Denys Dmytriyenko <denis@denix.org>
Acked-by: Chris Larson <chris_larson@mentor.com>
Acked-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Frans Meulenbroeks <fransmeulenbroeks@gmail.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
This task is misleading and potentially quite harmful, as it wipes the entire
DL_DIR, yet tasks are supposed to be recipe specific. We have the proper
distclean task to wipe sources for a recipe + run do_clean, and the user can
always wipe all of DL_DIR manually if necessary.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Acked-by: Holger Hans Peter Freyther <holger@moiji-mobile.com>
Acked-by: Eric Bénard <eric@eukrea.com>
|
|
Some packages like package-stagemanager-native dont use autoools
and autotools_stage_dir can not be found for such recipes.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Chris Larson <chris_larson@mentor.com>
|
|
without this, launching a daemon can fail if postinst is supposed, for
example, to fix permissions of directories used by the daemon.
Signed-off-by: Eric Bénard <eric@eukrea.com>
Acked-by: Marcin Juszkiewicz <hrw@openembedded.org>
Acked-by: Frans Meulenbroeks <fransmeulenbroeks@gmail.com>
Acked-by: Henning Heinold <heinold@inf.fu-berlin.de>
|
|
To use it add the following to your local.conf:
TESTLABLOG = "remote"
TESTLABREMOTEDIR = "/OE/angstrom-testlab"
See http://gitorious.org/angstrom/angstrom-testlab/ for an example of the expected results
|
|
Signed-off-by: Eric Bénard <eric@eukrea.com>
Acked-by: Khem Raj <raj.khem@gmail.com>
|
|
* base.bbclass can set a package's PACKAGE_ARCH to MACHINE_ARCH if
it finds file:// in the SRC_URI and the corresponding file is machine
specific (recipes/foo/foo-1.0/<some machine>/)
* but is_machine_specific actually fails to detect these files as
urldata.path is used instead of urldata.localpath thus preventing the
comparison of the file's path with machinepaths
* tested with bitbake HEAD & 1.8.18, with OE's tree alone and OE's tree
plus an overlay.
Signed-off-by: Eric Bénard <eric@eukrea.com>
Acked-by: Chris Larson <clarson@kergoth.com>
Acked-by: Khem Raj <raj.khem@gmail.com>
|
|
* qt4e.bbclass did add a dependency on qt4-embedded for recipes
providing qt4-embedded. This breaks building of qt4-embedded-gles
in 2 ways:
1. PREFERRED_PROVIDER_qt4-embedded = qt4-embedded-gles
adds a circular dependency.
2. bitbake qt4-embedded-gles
first builds and installs qt4-embedded to sysroots/ and then
fails to link the qt4-embedded-gles build against the libs in
sysroots which don't provide the gl specific symbols
* The fix avoids adding the dependency for package names _beginning_
with 'qt4-embedded' (as it is done in the x11 version).
Signed-off-by: Koen Kooi <koen@openembedded.org>
|
|
openmoko_two_get_subdir
Usually openmoko inheriting recipes have SECTION like openmoko/libs
etc then the code works well but for packages e.g. -dev -dbg same
section is not true and the functions fail
File "openmoko2.bbclass", line 2, in openmoko_two_get_subdir
ValueError: need more than 1 value to unpack
This patch makes sure to split the string only if SECTION contains '/'
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Due to a bitbake bug to be fixed in 1.10.1, EXPORT_FUNCTIONS isn't
remapping kernel_do_compile[depends] to do_compile[depends], so
make a comment about cleaning this up once we depend on a working
bitbake. Next, it needs to add to do_configure (like the users
have manually set a dependancy on) rather than do_compile so that
the initramfs is copied into place in time. Since we're fixing the
logic in kernel.bbclass, we can drop the workarounds that linux-kexecboot
and linux-preboot have added.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
libtool-cross must not inherit cross, so wasn't getting the right
mangling and saving done. In addition, we want a slightly broader grep
to be used in these cases.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
This fixes a problem where ncurses 5.7 (and readline) wasn't having
ipks stored. package_ipk.bbclass uses PKGV not PV which means
packaged-staging should too.
Signed-off-by: Tom Rini <tom_rini@mentor.com>
|
|
for now pushed this patch as it fixes the breakage.
If it is decided not to have SOC_FAMILY we can always revert or overwrite
see http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-September/023680.html
Signed-off-by: Frans Meulenbroeks <fransmeulenbroeks@gmail.com>
Acked-by: Chase Maupin <chase.maupin@ti.com>
and got a looks good from Phil.
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
* Make the inclusion of the README and recipe files optional
for source ipks. For some packages the sources owner may not
want these additional files added to their sources.
Signed-off-by: Chase Maupin <chase.maupin@ti.com>
Signed-off-by: Denys Dmytriyenko <denis@denix.org>
|
|
This ensures that the output of the stage manager is swallowed, rather than
shown unnecessarily to the user.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
For now, just ensures its inherited. In the future, we can merge / simplify
staging.bbclass with packaged-staging.bbclass as appropriate.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Robert Schuster <robertschuster@fsfe.org>
|
|
- Consolidate 'is this file a patch' logic
- Move unpack functions from classes into oe.unpack
- Move the unpacking message printing into do_unpack
- Move the destination directory determination into do_unpack
- Use subprocess's ability to pass in PATH and cwd rather than mangling the cmd
- Use shutil.copy2/copytree for ordinary file "unpack"
- Use the existing urldata from bb.fetch.init rather than re-decodeurl'ing the urls
- Make handling of globs in url paths explicit rather than implicit, calling
oe_unpack on each one, so showing an unpacking message to the user for each
globbed file, rather than the entirety
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Acked-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Tom Rini <tom_rini@mentor.com>
|
|
Signed-off-by: Graham Gower <graham.gower@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
The code which used host/path from the url to construct the destination of the
copy was using os.path.join. os.path.join treats a "/" in an element as an
indication to drop all previous elements, so the full path resulted in a
destination exactly the same as the source, resulting in a no-op unpack.
Fixed by using oe.path.join instead.
The previous fix missed two instances of os.path.join -- there's code
duplication in oe_unpack_file, really need to resurrect my do_unpack cleanup
branch.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Drop this for now, as it breaks other things. Of course, without this, you
once again can no longer have absolute file:// paths the way we used to be
able to. Will look into it further.
This reverts commit 599b03fe999d41114d124ce663b6c59fa57256d8.
|
|
The code which used host/path from the url to construct the destination of the
copy was using os.path.join. os.path.join treats a "/" in an element as an
indication to drop all previous elements, so the full path resulted in a
destination exactly the same as the source, resulting in a no-op unpack.
Fixed by using oe.path.join instead.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
This one is intended to be used from python snippets in variables. It returns
the stdout of the subprocess and raises an exception if the exit code isn't 0.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
This reverts commit 5678affd4e1b6453d6802b8cfb7823fec6f47269.
|
|
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Use the 'patchdir' flag. If relative, it's assumed to be relative to ${S}.
It creates a patchset class per patch directory, so for quilt this results in
multiple .pc/patches directories to manage each of the patchsets.
Signed-off-by: Chris Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
|
|
First seen in libclamav.pc, may be relevant for other packages too
Signed-off-by: Roman I Khimov <khimov@altell.ru>
|
|
Signed-off-by: Krzysztof Koroscik <soltys@szluug.org>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
|