buildtools-tarball: add python-git

python-git is needed by buildhistory-diff and the OE layer index update
script, and on machines where buildtools-tarball is necessary given we
are providing python we must provide this module as well.

Fixes [YOCTO #4747].

(From OE-Core rev: d9bf76b4fafc0ddcb19bf393e5e22678e49367b3)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-git: add recipe

This is used by buildhistory-diff and the OE layer index, and thus will
be useful in the context of the build appliance and buildtools-tarball.

(From OE-Core rev: 449ae5a70bc7d7d08f99de7d272b4cc3c83c3dcc)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-gitdb: add recipe

This is required by python-git.

(From OE-Core rev: bcbb34c26a235a103b46aa5c8ae7b3896904a215)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-smmap: add recipe

This is required by python-gitdb.

(From OE-Core rev: 37890014f406ca2ab24e0ec918af39613642afa0)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-async: add recipe

This is required by python-gitdb.

(From OE-Core rev: 31bc9fa8d53110a09ca048039d83213b34f4b7e4)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/insane: remove la2 check which no longer exists from ERROR_QA

This check was actually removed a very long time ago in
6656381714c5956f71ca634f5a5f4aa4661bbf7e, before we even changed the QA
checks to use names instead of numbers - the name was probably added
ERROR_QA mistakenly because it hadn't also been removed from the comment
listing the QA checks or the function that determined which ones were
errors.

(From OE-Core rev: ff65497cd9a96d5ab49b16ba1f7e30a216ff4a42)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Disable build qt related packagegroups on mips64 with 64 bits userspace

Because qt could not be built on mips64 with 64 bits userspace, set
COMPATIBLE_HOST for qt related packagegroups to disable them on mips64
with 64 bit userspace too.

(From OE-Core rev: 5f4c5de0b32d546ed28108d2403be41c685e27c7)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-multiprocessing: adding runtime dependencies

As python-multiprocessing requires python-threading and
python-pickle, this commit adds them as runtime dependency.

The observed behavior was:

When typing 'import multiprocessing' in the python shell on a
minimal image with only the python-multiprocessing recipe installed,
python reports at first:

Python 2.7.3 (default, Jun 27 2013, 08:26:25)
[GCC 4.7.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import multiprocessing;
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python2.7/multiprocessing/__init__.py", line 65, in <module>
    from multiprocessing.util import SUBDEBUG, SUBWARNING
  File "/usr/lib/python2.7/multiprocessing/util.py", line 38, in <module>
    import threading        # we want threading to install it's
ImportError: No module named threading

After adding python-threading as runtime dependency and rebuilding
the image, python reports:

Python 2.7.3 (default, Jun 27 2013, 08:26:25)
[GCC 4.7.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import multiprocessing;
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python2.7/multiprocessing/__init__.py", line 84, in <module>
    import _multiprocessing
ImportError: No module named cPickle

(From OE-Core rev: e913412ca0ff01cb654757c8199e8859f15b7cf7)

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@oss.bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-argparse: adding runtime dependency on python-codecs

When typing 'import argparse' in the python shell on a minimal image
with only the python-argparse recipe installed, python reports:

Python 2.7.3 (default, Jun 27 2013, 08:26:25)
[GCC 4.7.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import argparse;
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python2.7/site-packages/argparse.py", line 91, in <module>
    from gettext import gettext as _
ImportError: No module named gettext

The python-argparse recipe requires the python gettext module,
which is currently provided by python-codecs.
Hence, this commit adds python-codecs as runtime dependency to
resolve the issue.

(From OE-Core rev: 8d0e84bdfaf63ad61be7b015dd55dacccfa9132c)

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@oss.bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

init-live.sh: try to make a union mount when possible

This patch tries to make a union mount in live image. For example,
if aufs is enabled with the aufs-enable.scc configuration fragment,
the init-live.sh script will use aufs to make a union mount.

Although overlayfs is not supported by Yocto kernel yet, this patch
still takes it into consideration with the expect that the related
code should at least serve as a placeholder.

[YOCTO #1487]
[YOCTO #4761]

(From OE-Core rev: 3cd6fedd815688b2f3fd97a56feb5f8696ebeace)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

initrdscripts: unionfs cleanup

The unionfs has been disabled for more than a year and it's not going
to be used any more.

This patch cleans up the unionfs related code.

[YOCTO #4761]
[YOCTO #1487]

(From OE-Core rev: ba5e437bc7335468a70ea293496f78e1a9d66287)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt4*.bbclass: disable build qt on mips64 with 64 bits userspace

Qt/qt-embedded build fails on mips64 with 64 bits userspace. Set
COMPATIBLE_HOST in qt4e.bbclass and qt4x11.bbclass to disable build
qt/qt-embedded and packages which inherit these two classes on mips64
with 64 bits userspace.

(From OE-Core rev: 17890ebd637da0b3bf78804002d8b4f0ace078d2)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: cookerdata: Improve message if BBPATH is unset and bblayers.conf not found

If BBPATH isn't set and bblayers.conf isn't found, improve the message
shown to the user to help their understanding of what the problem might
be.

[YOCTO #3271]

(Bitbake rev: 0e639f5cbc813c8d4719019cfdd4287e9a429610)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
A

packagedata: Show error when trying to change PE/PV/PR from runtime/pkgdata

PN/PE/PV/PR should never change between do_package and the following do_package_write_*
tasks. If any do change you would see build failure due to the wrong WORKDIR being
used for example.

This patch ensures that if something is going wrong we see the error earlier and
with some better warning about what the real problem is.

[YOCTO #4102 partial]

This is a rewritten version of a patch from Martin Jansa <martin.jansa@gmail.com>

(From OE-Core rev: f5252fea11e13dbcec1c277cf1bf0d7e61b60690)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package: Ensure we iterate all the pool objects

There is the possibility that if we don't iterate through the multiprocessing
pool objects we might not catch return codes and this could lead to hung/zombie
processes either temproarily or on a wider scale. Adding this certainly doesn't
hurt anything and is better practise so we might as well do it.

Its not 100% clear if this fixes some issues or not.

(From OE-Core rev: 89c8493d4d85044cd72af2756569d15e87cd5947)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tcmode-default: Set GCC 4.8 as default

(From OE-Core rev: 7950a307bc7d4104e6cfb09bb2ea267c5da83f2a)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-4.8: Fix ICE on ppc/spe targets

Rename patches to make them easly to apply with git

(From OE-Core rev: 040a55d0b730bf78aad0f51e0018faa88655e279)

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: Check for the known broken version of make

See GNU Savannah bug 30612 -- make 3.82 is known to be broken.

A number of vendors are providing a modified version, so checking
for just the version string is not enough.  We also need to check
if the patch for the issue has been applied.  We use a modified
version of the reproduced to check for the issue.

(From OE-Core rev: dede532a980b0fabf0beae4519b89ec74a1c2474)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd: Ensure that we mount devtmpfs

Since systemd also used tmpfs we should make a similar patch for
the systemd-udev script

Fix for bug: https://bugzilla.yoctoproject.org/show_bug.cgi?id=4632
Since udev 172, the 'mknod' logic was removed from udev. Yocto Dylan
is now using udev 182. This means /dev is now required to be a
devtmpfs filesystem (maintained by the kernel).    If the root
filesystem is a ramdisk,  the kernel's auto-mount of /dev doesn't
activate since there is no rootfs to actually mount...  The bug causes
an unusable system as /dev doesn't contain even basic nodes required
to even get a login prompt.

The Yocto  udev/init script mounts tmpfs if it does not detect tmpfs
or devtmpfs mounted at /dev.  This appears to be  outdated logic that
is no longer correct. I believe the Yocto udev init script should be
checking and mounting only 'devtmpfs' on dev.

(From OE-Core rev: d3616f31617830cac9375e8f4aa33e344ac554ed)

Signed-off-by: Alex Olson <alex.olson+yocto@gmail.com>
[YOCTO #4632]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

autotools.bbclass: force copy Makefile.in.in to ${S}/po/

If a Makefile.in.in has existed under ${S}/po/ and is read-only, cp will fail.

(From OE-Core rev: 6e1b17f19411ed897c53ae0ef41a2d2972a9c113)

Signed-off-by: Roy.Li <rongqing.li@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

useradd.bbclass: Add members to a group

useradd.bbclass supports adding new users and new groups. But it does not
support adding existing users to existing groups.

There is a need of adding users to some groups (e.g. audio). The class was
extended to call groupmems utility with arguments passed via GROUPMEMS_PARAM.

(From OE-Core rev: 6b3bd34bf8c5e511bccfbb64bdd1236e1e7576e3)

Signed-off-by: Mikhail Durnev <mikhail_durnev@mentor.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

shadow-native: Add --root option in groupmems

    Patch add_root_cmd_groupmems.patch that we apply to shadow-native
    allows program groupmems from the shadow utility package to chroot()
    so it can be used to modify etc/passwd and etc/group if they are
    located in a sysroot.

    The --root option in groupmems is needed for class useradd.

(From OE-Core rev: ae7aa0ef68372c15224c0c518cb90ba7350137b4)

Signed-off-by: Mikhail Durnev <mikhail_durnev@mentor.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

adt_installer_internal:add sudo when permission deny on installation directory

When the user doesn't have rights to access the installation directory,
the sdk installation will fail, add sudo to fix this.

[YOCTO #4760]

(From OE-Core rev: 040010d04672c93f18d60308ecf3c26a26ec5fd3)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

adt_installer_internal:fix perl lib version mismatch

When run "autoreconf" in toolchain, there is an error if the host's perl's
version is not the same as the one in the SDK, the error says that the
executable perl mismatches the perl lib's version.

This is because most of the autotools' scripts use the "#!/usr/bin/perl -w"
which is host perl, but the gnu-configize uses "#! /usr/bin/env perl" which
invokes the perl wrapper in the SDK, and the wrapper will set the PERL5LIB to
the SDK which causes the mismatch. We can make all the perl scripts to use the
host perl or the SDK perl to fix this problem.

[YOCTO #4758]

(From OE-Core rev: 487d1fa7b79e89518494986461c157bace842613)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

adt_installer_internal: fix perl modules relocated failed on older distributions

The perl module for automake has an embedded path in it, this needs
to be relocated.

Older versions of 'file' do not return the "ASCII" text in the output for
a perl module file. Hence, the regex pattern didn't match perl module and
they were not getting relocated at all on older distributions.

1) On CentOS release 6.4, the version of file is 5.04
$ file /usr/lib/perl5/Config_heavy.pl
$ /usr/lib/perl5/Config_heavy.pl: Perl5 module source text

2) On Ubuntu 13.04, the version of file is 5.11
$ file /usr/lib/perl/5.14/Config_heavy.pl
$ /usr/lib/perl/5.14/Config_heavy.pl: C source, ASCII text, with very long lines

[YOCTO #4550]

(From OE-Core rev: 7cfab2c895bf4daa7716fb7509e367bf74f016e4)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bzip2: Add ptest

The existing bzip2 tests in the upstream Makefile are copied to Makefile.am
(yocto's) and modified to adopt to the ptest format.

(From OE-Core rev: f02258b304dc4544567601a1502080f3581c00fa)

Signed-off-by: Björn Stenberg <bjst@enea.com>
Signed-off-by: Anders Roxell <anders.roxell@enea.com>
Signed-off-by: Sona Sarmadi <sona.sarmadi@enea.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

groff: correct the install path of man.local

openvswitch build failed due to wrong install path of man.local which is
provided by groff.

Error log:
  /yocto/build/tmp/sysroots/x86_64-linux/usr/share/groff/1.22.2/tmac/an-old.tmac:690:
  warning: can't find macro file `man.local'

(From OE-Core rev: 5f2dd65e758ead8177a1cdda047bdb105b96e208)

Signed-off-by: Zhenhua Luo <zhenhua.luo@freescale.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types.bbclass: set 'filetype' ext4 feature

Generating filesystems with this flag allows more efficient directory
traversals because getdents() returns the filetype in 'd_type' which
allows to avoid an extra lstat() call.

Creating ext4 filesystems with 'mkfs.ext4' sets this flag by default
too.

(From OE-Core rev: 75e077025397f3bde84e60a9da2c0564ade09b39)

Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lsbtest: sync test suite packages version

Update file packages_list after sync test suite packages version with
upstream.
Check date: Jun 25, 2013

(From OE-Core rev: 8695a11c927e9ee130b8c9ddf6441f3cb7164fc2)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

boost: Limit PARALLEL_MAKE by -j 64

* greater paralelism isn't supported by bjam and causes segfault or ignoring -j
* PARALLEL_MAKE was enabled for boost in
  http://git.openembedded.org/openembedded-core/commit/?id=9b9cfc1dfe5e3b8f89b7a8508537166d0f23935e

(From OE-Core rev: c212f306934aa1c7c825e3bb060d4799be1efca1)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto-dev: bump version to 3.10+

The linux-yocto-dev kernel is at 3.10-rcX, so we should bump the version to
reflect reality.

(From OE-Core rev: 8be4d1314a1334f5218e4eb46d3f8b734d5516f5)

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.4: ltsi: sync to LTSI commit 5f05247ed

Updating the 3.4 branches to the latest LTSI baseline.

(From OE-Core rev: f53de5834559ed24b05f6bec8aaccdfc36f0a806)

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.8: add USB screen configuration and net sched options

Bumping the meta branch SRCREV for the followiong commits:

 meta: enable additional NET_SCHED options

    This change turns on NET_ACT_MIRRED (packet redirecting and mirroring)
    and NET_CLS_U32 (universal 32bit comparisons w/ hashing classification).

Signed-off-by: Michael Barabanov <michael.barabanov@windriver.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
    meta: add BSP-specific touchscreen support

    Add touchscreen-composite support to machines based on common-pc and
    common-pc-64, along with several other Atom boards that don't inherit
    from those, thus providing those machines with the out-of-the-box
    ability to make use of the set of USB touchscreen devices supported by
    the composite USB driver.

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
 meta: add usb/touchscreen-composite feature

    Add support for the 'composite' USB touchscreen driver.

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
 meta: add features/input/touchscreen

    Add a feature enabling basic support for touchscreen input devices.

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
(From OE-Core rev: 722f949f7defef62c4d258716cebc77c55edbbe2)

Signed-off-by: Michael Barabanov <michael.barabanov@windriver.com>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.4: allow kernel feature _appends to be overriden

Updating the linux-yocto 3.4 recipe's feature flags to match the 3.8
recipe, which has the following change:

    It was pointed out that the current way the KERNEL_FEATURES variable
    is appended in the base linux-yocto recipe doesn't allow the appended
    features to be prevented in a layer without using python code and
    a recipe finalize hook.

    To allow easier overriding of 'extra' or 'optional' features that are
    defined in the linux-yocto recipe, we create a KERNEL_EXTRA_FEATURES
    variable. This variable can be set in a layer to define extra features
    or cleared to prevent the recipe's extra features from being appended
    to the core functionality.

(From OE-Core rev: 3bd592e832c6ce10947882f37564c12f4fa7f8e2)

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev: only use devtmpfs for udev

Fix for bug: https://bugzilla.yoctoproject.org/show_bug.cgi?id=4632
Since udev 172, the 'mknod' logic was removed from udev. Yocto Dylan
is now using udev 182. This means /dev is now required to be a
devtmpfs filesystem (maintained by the kernel).    If the root
filesystem is a ramdisk,  the kernel's auto-mount of /dev doesn't
activate since there is no rootfs to actually mount...  The bug causes
an unusable system as /dev doesn't contain even basic nodes required
to even get a login prompt.

The Yocto  udev/init script mounts tmpfs if it does not detect tmpfs
or devtmpfs mounted at /dev.  This appears to be  outdated logic that
is no longer correct. I believe the Yocto udev init script should be
checking and mounting only 'devtmpfs' on dev.

(From OE-Core rev: 31ab19ab69bc6504df01cac7ee0670ca78d247ab)

Signed-off-by: Alex Olson <alex.olson+yocto@gmail.com>
[YOCTO #4632]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

alsa-tools: Pass ACLOCAL_FLAGS so aclocal uses the right params

The compile step ends regenerating the configure scripts included in
the source subdirs, for it to properly work we need to pass the
ACLOCAL_FLAGS or the .m4 files won't be found.

,----[ Build error ]
| ./ac3dec
| aclocal: warning: autoconf input should be named 'configure.ac', not 'configure.in'
| configure.in:18: warning: macro 'AM_PATH_ALSA' not found in library
| automake: warning: autoconf input should be named 'configure.ac', not 'configure.in'
| configure.in:9: warning: AM_INIT_AUTOMAKE: two- and three-arguments forms are deprecated.  For more info, see:
| configure.in:9: http://www.gnu.org/software/automake/manual/automake.html#Modernize-AM_INIT_AUTOMAKE-invocation
| automake: warning: autoconf input should be named 'configure.ac', not 'configure.in'
| test/Makefile.am:1: warning: 'INCLUDES' is the old name for 'AM_CPPFLAGS' (or '*_CPPFLAGS')
| configure.in:18: error: possibly undefined macro: AM_PATH_ALSA
|       If this token and others are legitimate, please use m4_pattern_allow.
|       See the Autoconf documentation.
| make: *** [all] Error 1
| ERROR: oe_runmake failed
`----

Reported-by: Rogerio Nunes <rogerio.nunes@freescale.com>
(From OE-Core rev: dde80e6fac83ca55644cb1b56cb55b2ba01c6564)

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

dbus-ptest: Install missing files

Install files from EXTRA_DIST needed for some tests.
Add configure parameter with-dbus-test-dir to specify where the test
tools are located.

(From OE-Core rev: 330d594af9f14faf0e5770b6056f4457a033fedd)

Signed-off-by: Adrian Dudau <adrian.dudau@enea.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libx11: enable Xcms by default

The following do not work if Xcms is not enabled in libx11:
- starting xterm and doing ls --color / fails to color code the
  directories:
  xterm: Cannot allocate color "rgb:5c/5c/ff"
- xsetroot -solid rgb:5c/5c/ff fails with the following error:
  xsetroot:  unknown color "rgb:5c/5c/ff"
- xsetroot -solid rgbi:1.0/1.0/1.0 fails with the following error:
  xsetroot:  unknown color "rgbi:1.0/1.0/1.0"

More specifically, applications that pass Xcms color names to
XParseColor do not work properly.

[YOCTO #4576]

(From OE-Core rev: d860ee68208b84efb8049669ca18acc69f2f2d1b)

Signed-off-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

init-live.sh: fix automount failed occasionally

Reboot system repeatedly, occasionally found usb automount failed, a
low probability but it happens.
$ df
Filesystem           1K-blocks      Used Available Use% Mounted on
none                   1024972         4   1024968   0% /dev
/dev/sda3              7689384   3540940   3757840  49% /media/sda3
/dev/sda2            146127424   1238432 137466120   1% /media/sda2
/dev/sda1                17845     14570      2354  86% /media/sda1
/dev/sdb                293400    288560      4840  98% /media/sdb
/dev/sdc4               457632        32    457600   0% /media/sdc4
/dev/sdc1               475018      2321    447749   1% /media/sdc1
/dev/sdd               1382298   1382298         0 100% /media/sdd
/dev/sdc2               475694      2320    448374   1% /media/sdc2
/dev/loop0              270649    181249     75644  71% /
df: /media/sdc3: No such file or directory
tmpfs                  1029352         0   1029352   0% /dev/shm
tmpfs                  1029352      2816   1026536   0% /run
tmpfs                  1029352         0   1029352   0% /sys/fs/cgroup
tmpfs                  1029352         4   1029348   0% /tmp
tmpfs                  1029352         0   1029352   0% /media/ram
tmpfs                  1029352       116   1029236   0% /var/volatile

When boot media has been found, udev will be killed. If udev is busy
to mount other medias at the killing time (especially medias is many),
the above issue will occur occasionally.

Invoke `udevadm settle' before killing udev will resolve this
issue, it watches the udev event queue, and exits if all current
events are handled.

Use variable `_UDEV_DAEMON' to replace hardcoded `udevd' to keep
consistent with previous.

[YOCTO #4745]

(From OE-Core rev: 2f209a7045a93e7e42f90418a42f464827b4a7f8)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: Fix COREBASE sanity tests

We need to expand the COREBASE variable, no idea how these tests were
previously working at all...

(From OE-Core rev: 099063f353a7a18720c92d87400726a49eed432f)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: bitbake: python funcname can not include special character @

[YOCTO #4772]

When path:file change to python function, it maybe include '@' character.
So, add the special character to change to '_' for avoid error.

(Bitbake rev: 684bc6dcb11ecb1fd7a4d25c08909ad9879e8342)

Signed-off-by: Li Wang <li.wang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: bitbake: adding a new comment should be placed on a new line

In this case, the comment is appended to the end of the file.
Some text editors, do not place  a '\n' to the end of the file
after saving it.

[YOCTO #4636]
(Bitbake rev: 2beb9589b1bd9773f587b4dc08afdfe50f4ea913)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: Add check for @ character in build directory name

The @ character is not escaped properly in too many places within the system
to easily support it so add the character to the list of blacklisted characters.

Also tweak the other messages and ensure that all appropriate error messages
are disabled in one go.

[YOCTO #4772]

(From OE-Core rev: 008cb3c501c8313a0a1a0ebce2b0aa61239b548d)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk_base, adt_installer: abort install if path contains spaces

Spaces are not handled properly in some parts of oe-core and it's safer
to abort toolchain installation if path contains spaces. Even though
we fix space handling in the toolchain installation script, there are
various other parts in the toolchain (perl scripts, sysroot path passed to
toolchain binaries, shebang lines) that would need special handling. So,
for now, just bail out if path contains spaces.

The checking for spaces in the path is done after expanding relative
paths to absolute and tilde conversion.

[YOCTO #4488]

(From OE-Core rev: 8c35ba2d3048ce69f74f72cb2676e4bc162cfb63)

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gdk-pixbuf: add a wrapper for gdk-pixbuf-pixdata

gdk-pixbuf 2.26 added a new binary gdk-pixbuf-pixdata, but no wrapper script was
added for it.

(From OE-Core rev: 930a48b299c42803fe14185ed31df63c162d3ec6)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cleanup-workdir: fix typo

Fix typo.

(From OE-Core rev: 07b4682250694c5783aa374814c95ba9a2974a69)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base-files: create /usr/lib/locale dir

Lsbtest shows that /usr/lib/locale dir is lost, so create it

(From OE-Core rev: 6fb6934a590e5ecda864183c0be83e1b59fec8c7)

Signed-off-by: Roy.Li <rongqing.li@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ghostscript: upgrade to 9.07

Remove unuseful patch:
  * 0001-make-ghostscript-work-with-long-building-directory.patch

Port applicable patches:
  * ghostscript-9.02-genarch.patch
  * ghostscript-9.02-parallel-make.patch
  * ghostscript-9.02-prevent_recompiling.patch
  * ghostscript-9.05-NOT-check-endian.patch
  * ghostscript-native-fix-disable-system-libtiff.patch

(From OE-Core rev: aea8f29275fd7c2341c429b46c25e17afe532b3e)

Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

chrpath: Fix SRC_URI to correct location

(From OE-Core rev: 6b7ab43722a94e7a0ced72ffb7497e5d507a39cd)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

buildtools-tarball: Add nativesdk-make

Recently it was discovered that many Fedora hosts have a broken version of
make 3.82.  Add make to the buildtools-tarball, as well ad modify make to
support building a special nativesdk version.

(From OE-Core rev: ea972fc4ce2268f01be6beeafd27dd949d800f9d)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

make: Fix second part of bug Savannah 30612

The Savannah bug 30612 describes two different issue.  The first,
previously fixed, errors parsing multiple objects in parenthesis.
The second, (this issue), extra white space contained in the
parenthesis.

The fix was backported from the current make git tree:
http://git.savannah.gnu.org/cgit/make.git/commit/?id=b06b8c64a29a5ba3a8daecd829fa2f98d42cb285

(From OE-Core rev: 681b4c2cd9830cd523080aa830748d9c3367c7c9)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: Add ptest

Install busybox test suite and run it as ptest.

(From OE-Core rev: 0d29dc5b5f7742df60fdba90835ef77425963bde)

Signed-off-by: Anders Roxell <anders.roxell@enea.com>
Signed-off-by: Björn Stenberg <bjst@enea.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

buildtools-tarball: Change the way the SDK is sanitized

The previous method of shrinking down the scale of the environment-setup
file would not work properly when the target is configured with multilibs.

In addition, the configured machine name and similar settings could leak
into the SDK naming.  This was resolved by clearing the SDK generated
files and generating our own custom files.

Note, the name of the environment-setup now is suffixed with the SDK_SYS.

(From OE-Core rev: 778cd8d2110ef2db5ff4ae4f0b55c52a8f1ea334)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc: When cross compiling a target gcc, target flags may be used on the host

The original fix worked some of the time, but not on all machines.  Fix
this by applying the change in a different way.  Following the example of
the existing BUILD_CFLAGS.

Below is the commit message from the original change to help explain why
this is needed:

Configure identifies a number of warning flags (WARN_CFLAGS and
WARN_CXXFLAGS) from the $CC value.  The cross compiler may be different
from the host compiler and may not support the same set of flags.  This
leads to problems such as:

cc1plus: error: unrecognized command line option "-Wno-narrowing"
cc1plus: error: unrecognized command line option "-Wno-overlength-strings"

Work around this problem by removing the warning flags from the
BUILD_CXXFLAGS value, in a way similar to the BUILD_CFLAGS.

(From OE-Core rev: be21c6e8e4f810e826538337dac6e34ed96e1f6f)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemtap: Systemtap can not be build w -O0 optimization

Systemtap will fail with:

../usr/include/features.h:330:4: error: #warning _FORTIFY_SOURCE requires compiling with optimization (-O) [-Werror=cpp]
|  #  warning _FORTIFY_SOURCE requires compiling with optimization (-O)

Use the same technique that eglibc uses to catch and correct the optimization
level, changing to -O2.

(From OE-Core rev: 9ceebb966e5294a270222475ab6317dcae2952fa)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-firmware: Package some iwlwifi firmware separately

Taken from meta-intel, might as well merge this into the core recipe.

(From OE-Core rev: bb3201b9c21b29604eb06b03e935d53210a7b762)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gst-plugins-bad: element selection rationalisation

Using --with-plugins means you only get the elements you enable, so we were
dropping vast numbers of useful plugins such as the MPEG muxers.

Instead, follow gst-plugins-base by using PACKAGECONFIG to enable/disable
plugins with optional dependencies.

(From OE-Core rev: 7722fd48995d5d430d58b94ecf69a6ad9f1c741b)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gst-plugins-base: element selection rationalisation

Use PACKAGECONFIG to control the X11 elements, and add statements for the
elements that we don't always enable.

Remove the freetype dependency as it's apparently (but not really) only needed
by the examples.

(From OE-Core rev: d40606de85a42c48327dead5d6e4c9a1de2cc39c)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gst-plugins-bad: tremor was moved from here to -base

(From OE-Core rev: 6e149551a36825ac5e25b14dbaa62b794154c11d)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gst-plugins-bad: use correct option when enabling librsvg

(From OE-Core rev: 82eab38f2bf1dac3e3414a5a20b8e51f871e07ce)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bash ptest: Sed away the Makefile dependency to remove error messages.

The bash Makefile defines a dependency on itself and tries to run configure,
causing error messages when running ptest on target:

make: *** No rule to make target `configure.in', needed by `configure'.
make: *** No rule to make target `aclocal.m4', needed by `configure'.
make: *** No rule to make target `config.h.in', needed by `configure'.
make: *** No rule to make target `Makefile.in', needed by `Makefile'.
make: Failed to remake makefile `Makefile'.

This patch edits out this dependency in the Makefile installed for ptest,
to get rid of these messages.

(From OE-Core rev: b5fe8c9ff330105337b003be0de2f970545d13ef)

Signed-off-by: Sona Sarmadi <sona.sarmadi@enea.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt-mobility: remove /usr/lib from ld rpath-link option

Remove /usr/lib from ld rpath-link option to fix the below build error:

       bitbake_build/tmp/sysroots/x86-kvm-guest/usr/lib/libpng16.so.16:
        undefined reference to `inflateReset2@ZLIB_1.2.3.4'

since sysroot seems not work for rpath-link, and "rpath-link,/usr/lib" makes
ld to search host libraries for target libraries, once host has different
version zlib, the error will happen.

qmake uses QT_MOBILITY_LIB to generate "rpath-link,/usr/lib" when do_configure
but we can not add sysroot into QT_MOBILITY_LIB, since QT_MOBILITY_LIB is dir
which libraries will be installed to, so I remove this dir from rpath-link
before do_compile

(From OE-Core rev: f7409a9fe83ba2535a43f39ed57cd78242a88557)

Signed-off-by: Roy.Li <rongqing.li@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libbsd: Add recipes for libbsd - utility functions from BSD systems

When building BSD programs this library provides common
BSD functions that are missing on other OSes e.g. Linux

This library is elemental for porting large set of BSD applications
current consumer of this are in meta-networking/openbsd-netcat
but once we have it in OE-Core more recipes depending on it in different layers
can be added.

(From OE-Core rev: 2df53911f25234d2724bc8163ac9406af0bdad06)

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gst-ffmpeg: fix libav config error for ppc

Pass --cpu to libav config when we're building ppc, or else there are
errors like:

    You need a compiler that supports {} in AltiVec vector declarations.

Also patch libav configure to have knowledge of more ppc CPUs.

(From OE-Core rev: a825781fc822f4630bc29906ca1ca79b8fad4836)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemctl: Support all unit types in the directives.

The Alias and WantedBy directives can accept all valid unit types when
using the systemctl from systemd. And since the systemctl script should
match the behavior of systemd as much as possible, add the current set
of unit types listed at
http://www.freedesktop.org/software/systemd/man/systemd.unit.html
to the Alias and WantedBy directives.

The deficiency was exposed when trying to use:
    Alias=default.target
in a foo.target. No symlink was created by running
"systemctl enable foo.target" during the package's postinst.

(From OE-Core rev: 374b9c37b3310cf2a3373633197ca7ba21f6d1bd)

Signed-off-by: Randy Witt <rewitt@declaratino.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ltp: update to new release

(From OE-Core rev: 7f804ccd2a1e8ccfec1481ef757ce35b6edcbacf)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: fix to use udev-cache correctly

Previouly, the udev-cache has no real effect even if it's installed
into the system. The key problem here is that at first boot, the
/etc/dev.tar is not present, thus resulting /dev/shm/udev.cache not
created on first boot even if udev-cache is enabled.

This patch fixes this problem. The /dev/shm/udev.cache will be created
if necessary, that is, on first boot or when some part of the system is
changed. In the latter case, the udev cache may not be valid.

[YOCTO #4738]

(From OE-Core rev: 84e0ec2e677fb0236a38478372cdd75797cf5a2e)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libpam: check if PAM is enabled when building

(From OE-Core rev: fd9bad3e48a605e9fd28c129413300ff6b548788)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libpam: fix whitespace in shell function

(From OE-Core rev: 1b4b25d3cebab90398db208281d54e7442d43bcd)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

logrotate: fix for CVE-2011-1548

If a logfile is a symlink, it may be read when being compressed, being
copied (copy, copytruncate) or mailed. Secure data (eg. password files)
may be exposed.

Portback nofollow.patch from:
http://logrotate.sourcearchive.com/downloads/3.8.1-5/logrotate_3.8.1-5.debian.tar.gz

(From OE-Core rev: d0e3fc1b28fc16200adbe690aa27124041036ba3)

Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rt-tests: fix error check in hackbench

Add a patch to fix when an unsigned number is taken as a negative error
code.

(From OE-Core rev: 7538a9cd0c6f0216ef95956ad86e2f88ebd4c8ea)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libproxy_0.4.7 do_unpack failed of qemuppc_world

The tarball from upstream shows many lines of....

   tar: Ignoring unknown extended header keyword
      `LIBARCHIVE.xattr.security.selinux'
   tar: Ignoring unknown extended header keyword `SCHILY.dev'
   tar: Ignoring unknown extended header keyword `SCHILY.ino'
   tar: Ignoring unknown extended header keyword `SCHILY.nlink'

Replacing it with the .zip file from upstream

(From OE-Core rev: 1cad5073346bcccbe5bafa3c8876890a0a62615c)

Signed-off-by: Amy Fong <amy.fong@windriver.com>
Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

core-image-weston: add clutter examples

Now that Clutter supports Wayland too, add the Clutter example app to the image.

(From OE-Core rev: 74a1ace0c942f25a5e2278795fee6cfc523a3b77)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

clutter: add Wayland support

Add PACKAGECONFIG stanzas for Wayland client and server, and respect the x11 and
wayland DISTRO_FEATURES to enable the relevant backends as appropriate by
default.

(From OE-Core rev: aa01a4a2c74ee2ee9f629ea07a71b06bc2fdda99)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cogl: add Wayland support

Add PACKAGECONFIG stanzas for Wayland client and server, and respect the x11 and
wayland DISTRO_FEATURES to enable the relevant backends as appropriate by
default.

(From OE-Core rev: 84ffd212f9ca083301d19b9c7a9720d05c733e5c)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

weston: move mtdev dependency to the DRM backend, where it belongs

(From OE-Core rev: 2d1e8b554e10173377bfe231d5999561adb62321)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mesa: remove a redundant do_install tweak

Mesa isn't incorrectly installing GLU headers anymore, so we don't need to
delete them.

(From OE-Core rev: 96b91dd669d2701fcb4c71fb482392029ce7cd7f)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

weston: add patches to make weston-launch work

(From OE-Core rev: 7b4f2c7c305ef4c0848311712f17cab76232c4bf)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

weston-init: fix a typo in a user-visible message

This is the weston init script, not X.

(From OE-Core rev: 8d30f9c1fe27795c4a1a3e64229cd1cc243e50be)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-pygtk: fix parallel compile issue

defs.c dependes on gdk-types.defs and gtk-types.defs. When parallel
compile, it may fails with:
"IOError: [Errno 2] No such file or directory: 'gtk-types.defs'"

Add them to dependences of defs.c to fix this issue.

(From OE-Core rev: edf278eec71552bcd3ac661dce8e8b7489463f6a)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-pygobject: disable parallel install

The installation of __glib.so is invoked by install-data-am target
which are generated by automake. installing libpyglib-2.0-python is
invoked by install-exec-am.

"make install" will firstly install libpyglib-2.0-python, then install
__glib.so, the sequence should not be broken, since _glib.so has
dependence on libpyglib-2.0-python. But when enable parallel install,
the sequence maybe break then installation fails with:
 "/usr/bin/ld: cannot find -lpyglib-2.0-python"

Disable parallel install to fix this issue.

(From OE-Core rev: e87a0c81c77d11f892a34c2d14ffbeb4342d32dd)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

openjade-native: fix build failure

Fix openjade-native build failure when build directory name contains the
characters ".a".

(From OE-Core rev: f005670ee8f6d02e0b0517a48b47b364f0bddf2d)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tinylogin: remove recipe

tinylogin has been deprecated and the functionality ported into busybox.
We now use busybox as the login manager, so the tinylogin recipe could
be deleted.

[YOCTO #4207]

(From OE-Core rev: 2762ff976a3473be4259889029e048ab8b3be5ab)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mingetty: lower the ALTERNATIVE_PRIORITY

mingetty doesn't work with serial consoles. Currently, it has the
same ALTERNATIVE_PRIORITY with busybox. So if it is installed with
busybox together, it's possible that the getty is linked to the
mingetty, causing failures when we login to the serial consoles.

Lower the its ALTERNATIVE_PRIORITY to solve this problem.

[YOCTO #4207]

(From OE-Core rev: 8cbae825c8a90874868bfe8cbf4e2b2f0d11b0de)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

packagegroup-core-basic: set the default login manager

Set the default login manager to 'busybox', drop the mingetty in
the RDEPENDS, use ${VIRTUAL-RUNTIME_login_manager} instead.

mingetty doesn't work with serial consoles, so if the login console
is ttyS0 for example, we get error messages on screen and cannot login
on ttyS0.

The login manager, no matter it's tinylogin or busybox, provides
getty, so we can just rdepend on it.

[YOCTO #4207]

(From OE-Core rev: 878596c0943c015f9995997cdaf894c2c236f9a0)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

packagegroup-core-boot: use busybox as the default login manager

tinylogin has been deprecated and the functionality ported into busybox,
so we switch to using busybox as the default login manager.

[YOCTO #4207]

(From OE-Core rev: 54fc2a698d49f7acc93c8e4b6f6c4e7b3b6ffcc8)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: add the ability to split the busybox binary

This patch enables us to split the busybox into two binaries, one
containing suid applications, and the other containing nosuid apps.

Add a variable, BUSYBOX_SPLIT_SUID, to control whether to split the
busybox binary into two parts. We default it to "1" to enable the
splitting, but users could still override it to disable the splitting.
After all, busybox has no internal support for this suid apps splitting,
so there might be users out there who want just one busybox binary.

The basic idea here is to build the busybox twice, each with the correct
configuration items. We extract the non-app part of the original .config
file, and merge this part with the suid-app part to form a .config which
contains only suid apps. The same strategy applies to the non-suid apps.

[YOCTO #4207]

(From OE-Core rev: e5a1442819dfb74e86a6f69da008ba6908c8bbc7)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: enable to list suid and non-suid app configs

This patch, written by Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>,
adds the ability to busybox to list configuration items of suid apps
and non-suid apps separately.

`make busybox.cfg.suid' generates a file containing config items of the
suid apps.

'make busybox.cfg.nosuid' generates a file containing config items of
the non-suid apps.

This patch helps to separate busybox into two binaries, the suid one and
the non-suid one.

[YOCTO #4207]

(From OE-Core rev: 832d1b5575c76f61623f2e0337554287d056422b)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: add a config fragment to enable login utilities

Create a config fragment to enable the login/passwd utilities of busybox.

[YOCTO #4207]

(From OE-Core rev: 2a37490d4d55a1a6e4c4d268231a9d8904d38f8b)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: add support for CONFIG_FEATURE_INDIVIDUAL

Previously, if CONFIG_FEATURE_INDIVIDUAL was enabled for busybox,
yocto-based systems could start correctly.

This is because if busybox is built as individual apps, '/bin/busybox'
may not be present, so setting the default ALTERNATIVE_TARGET to
'/bin/busybox' is not appropriate and could lead to errors.

This patch fixes this problem by checking the existence of '/bin/busybox'
before setting the ALTERNATIVE_TARGET to '/bin/busybox'.

After this change, if busybox is built as individual apps, we'll have
links like '/bin/ls -> /bin/ls.busybox', otherwise, we'll have links
like '/bin/ls -> /bin/busybox'.

Note there's a grep expression change in this patch. The old expression
doesn't work well, it has an unwanted underscore, so I changed it to make
it work.

[YOCTO #4570]

(From OE-Core rev: 87efa83191f0b4f4cec3a821b6dfe425c781c077)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: remove the postinst part of the recipe

Remove the pkg_postinst_${PN} from this recipe, as it's redundant.
It basically wants to do the same thing as the update-alternatives
does. But it doesn't do it well.

(From OE-Core rev: 5b961afc784747eb8240540e542a789849118596)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: Various improvements

I started trying to make some changes in sanity.bbclass and couldn't stand
the current code. This patch splits out the one big function into a number
of smaller units and adds a status class which handles queuing of messages
and issues like network errors and whether a reparse is required. It also
cleans up some syntax and obsolete code.

The other fundamental change is the tests are split into ones which run
once and those which run at every build. This was always the way the
class was intended to work but that got lost somewhere along the way.
This patch fixes that.

Its still far from perfect but it is hopefully an improvement
and sets the scene for other improvements to be built on top of this.

(From OE-Core rev: 4e6de6d5f0454024eec2ec775a938c5dab70610c)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: Drop horrible obsolete minversion hack

We once needed to do this, things seem to work fine without this now,
thankfully.

(From OE-Core rev: ac090ace11d654dafd642fd93c94091a164476fa)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Sync with PRINC removal from meta-intel

PRINC usage is removed from meta-intel, bump core PR values to ensure
no version regressions.

(From OE-Core rev: c6108ebe116eafaf0a22b5f22a3eec87c86859f4)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

yocto-bsp: Drop PRINC usage

We no longer need to manually bump PR values so lets not generate
code using this.

(From meta-yocto rev: ed23b0eee9791b06b0bae1ad17595e72ccaa86cf)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

psplash: Bump PR to allow removal of PRINC from meta-yocto

(From OE-Core rev: b2df9eef5a99e1d5a9dd262c1091b409a8c8174d)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

strace: add configure options libaio and acl

Add strace configure options libaio and acl to enable or disable build
with them. Then PACKAGECONFIG can be used to handle dependency.

Set "libaio" as a default feature of PACKAGECONFIG.

(From OE-Core rev: 9b6b7973a5e49fe9cd283c5b4af31e8a7940e7b5)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

strace: update to 4.8

Update strace to 4.8:

* Update License file.
* Remove the backport patches which are already in version 4.8.
* Add file git-version-gen from git repo. Without this file configure
  fails.

(From OE-Core rev: 3ae2592b40900caaebe2452c875912f82f9e5200)

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: data_smart: Fix bug with overrides and weak default values

(aka pay the cookie monster for weak defaults)

If you have code like:

MYVAR = "a"
MYVAR_override ??= "b"

then MYVAR will get the value "a" even when override is in OVERRIDES. The
reason is that the value of ??= is set as a flag not a value and the cookie
monster isn't paid.

The fix is to ensure appropriate payment is made for a defaultval varflag
matching the usual setVar case.

(Bitbake rev: 3d8044bc79c482c5ea008ddf12a8128dcd1527ee)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: data_smart: Ensure variable flags are accounted for in config data hash

Currently if the flags set against a variable in the base data store
change, it doesn't automatically trigger a reparse when it really
should. For example with the blacklist class setting:

PNBLACKLIST[qemu] = "bar"
PNBLACKLIST[bash] = "foo"

will not trigger a reparse if only one entry is changed and a
blacklisted recipe can still be built.

I did consider using BB_SIGNATURE_EXCLUDE_FLAGS in here however it
doesn't make sense, we want to trigger a reparse when any of the
flags change too (which is different to the sstate signatures which
we wouldn't want to change in those cases).

[YOCTO #4627]

(Bitbake rev: ed74ea50043f6feb698c891e571feda2b9f8513d)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>