iproute2: split out package for tc

tc (the IP traffic control utility) isn't often used and makes up a
reasonably large part of the iproute2 package as well as having a
runtime dependency on iptables, so split it out into its own package.

(From OE-Core rev: 1d353cb30b93cd08d7a0f743534c1cd712bbe018)

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

systemd-compat-units: don't mask dbus, the recipe does this itself

(From OE-Core rev: a5436fa039d1e3153fb724a3fa59c5955995aa2d)

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

dbus: mask the dbus-1 init script if using systemd

If the image is built with both systemd and sysvinit there'll be a dbus-1 init
script and a dbus service. This means systemd can try and launch both, which
won't work.

There's a systemctl mask to stop this in systemd-compat-units, but the logical
place for it to be is in the dbus recipe so it can't be left out of an image.

(From OE-Core rev: c910aa17689077362a25938aeebee7fb24057e30)

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

classes/buildhistory: write out files in UTF-8 format

If the package information contained unicode characters (for example, in
the package file listing) then writing them out as ASCII would fail, so
write them out using codecs.open() instead of open() using UTF-8. This
fixes ca-certificates failing in do_packagedata when buildhistory is
enabled.

(From OE-Core rev: dcf228fe69bfee4e22baad477ad407248c0f9cdb)

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

sstate: Improve funciton checksums

Now that bitbake is recursing into pre/postfuncs, we need to ensure
the dependencies of these functions is correct. We don't want dependencies
on MACHINE or other related variables. This patch adds in appropriate
variable exclusions to achieve this.

(From OE-Core rev: 8461283a648d7c5affd51971ebd9b35a8a4c625f)

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

bitbake: data: Account for pre/postfunc functions when calculating dependencies

pre/postfuncs were not being added to checksums. This meant that when reconfiguration
occurred, tasks were not always being rerun when they should. This include
sstate functions as well as systemd's do_install function in the OE metadata.

With the addition of postfuncs, its possible a shell task can have a python
pre/postfunc so we have to guard against this when generating shell output
in emit_func.

(Bitbake rev: b84d010144de687667cf855ddcb41c9b863c236e)

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

wic: Hook up --debug option

Hook up the existing --debug option to toggle the wic debug loglevel,
which is indispensible when things go wrong, and make it easy to use
from the command-line.

(From OE-Core rev: a5ece6f37656fa56b97fd8faf52917345238d015)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: Hook up BootimgEFIPlugin and BootimgPcbiosPlugin plugins

Remove all the Wic_PartData and DirectImageCreator code now
implemented by the BootimgEFIPlugin and BootimgPcbiosPlugin plugins,
as well as all the special-cased boot_type code, significantly
cleaning up the code.

Replace the calling code with general-purpose plugin invocations, in
essence calling the appropriate implementations at run-time based on
the --source value in effect.

Change the directdisk.wks and mkefidisk.wks scripts to make use of the
new plugins.

(From OE-Core rev: 43558610a5793888ff2b18bd3a27c7ab558e5ad0)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: Add BootimgEFIPlugin and BootimgPcbiosPlugin

Implement the BootimgPcbiosPlugin and BootimgEFIPlugin SourcePlugin
classes.  The configure/prepare_partition() methods are implemented
using code derived from similar code in the Wic_PartData class.

These classes have the corresponding names 'bootimg-pcbios' and
'bootimg-efi', which are the names that should be used in the --source
parameters of the .wks partition commands.

(From OE-Core rev: 6e147488b40f730e07f1e0f232083ed75388daa0)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: Add SourcePlugin class

Define the SourcePlugin class, which is the class that should be
subclassed to create a 'source' plugin.

'Source' plugins provide a mechanism to customize various aspects of
the image generation process in wic, mainly the contents of
partitions.

The initial version of wic defined a --source param for partitions,
which was in the first revision hard-coded to two possible values:
rootfs and bootimg.

This patch essentially removes the hard-coded --bootimg param and
replaces it with a plugin system that maps the value specified as
--source to a particular 'source' plugin instead.

A 'source' plugin is created as a subclass of SourcePlugin and the
plugin file containing it is added to scriptsl/lib/mic/plugins/source/
to make the plugin implementation available to the wic implementation.

When the wic implementation needs to invoke a partition-specific
implementation, it looks for the plugin that has the same name as the
--source param given to that partition.  For example, if the partition
is set up like this:

  part /boot --source bootimg-pcbios   ...

then the methods defined as class members of the plugin having the
matching .name class member would be used.

To be more concrete, here's the plugin definition that would match a
'--source bootimg-pcbios' usage, along with an example method that
would be called by the wic implementation when it needed to invoke an
implementation-specific partition-preparation function:

  class BootimgPcbiosPlugin(SourcePlugin):
      name = 'bootimg-pcbios'

  @classmethod
      def do_prepare_partition(self, part, ...)

If the subclass itself doesn't implement a function, a 'default'
version in a superclass will be located and used, which is why all
plugins must be derived from SourcePlugin.

This scheme is extensible - adding more hooks is a simple matter of
adding more plugin methods to SourcePlugin and derived classes.  The
code that then needs to call the plugin methods the uses
plugin.get_source_plugin_methods() to find the method(s) needed by the
call; this is done by filling up a dict with keys containing the
methon names of interest - on success, these will be filled in with
the actual methods. fPlease see the implementation for examples and
details.

Note that a source plugin need not restrict itself to methods that
apply directly to partitions - methods can also be defined for higher
level processing such as at the 'disk' level.  The
get_default_source_plugin() of DirectImageCreator allows the default
source plugin to be retrieved; by default this is set to be the same
plugin used for the /boot partition, but that can be overridden by
specifying a different --source and therefore different plugin on the
'bootloader' line.  This isn't ideal, but it avoids forcing a new
high-level object to be defined for that purpose.

Note that the '--source rootfs' param remains as its current
hard-coded value, which is just the rootfs to be used to populate the
partition - by default, that's just the value of the bitbake
ROOTFS_DIR variable (or whatever was passed in using the -r param).
Note that this also could also be overridden by creating a source
plugin using a different name; at this point, unlike with bootimg,
there's been no need to do so.

(From OE-Core rev: 663833d8ecccb36ab42150bc5c9c00be79fa5b93)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: Add wic-specific bootloader subclass

Add a new wic-specific bootloader subclass so we can add a --source
param to hang non-partition plugin off of.

By default, the bootloader gets the /boot partition source plugin, but
this can be overridden by the --source bootloader param if needed.

(From OE-Core rev: f90e4097c4e69d4f61c69923cb5d1ebb6b74d2ff)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: Create and use new functions for getting bitbake variables

Add get_bitbake_var() and bitbake_env_lines() functions for use by
plugins, which will need access to them for customization.

(From OE-Core rev: f0bb47b0d7ab6520c105ce131844269172de3efd)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: Move some common items to oe.misc

Move a couple items into a more common location since they're going to
need to be accessible from source plugins.

(From OE-Core rev: 95ca523949e838850b5afa090ba16f91b8557c12)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

binutils: Package additional file when using ld-is-gold

Fixes QA warnings like

WARNING: QA Issue: binutils: Files/directories were installed but not
shipped
  /usr/bin/ld.bfd
  /usr/bin/dwp

(From OE-Core rev: 6503ad0484d5a9cc3f8338eb4f176d65c2f3c13a)

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Acked-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/runtime/systemd: remove race in settle()

The settle() function had a race where services could still be activating at two
minutes but then when the final log is output, they've activated.

Remove this race and generally clean up the code.

(From OE-Core rev: 8d107e0a828868702cfe035104c1f0b51da4291e)

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

tcl-native: Depend on zlib-native

tcl can use its internal zlib. If it detects zlib during configure, then it
gets removed from the sysroot during the build (since its no in DEPENDS),
it causes build failures.

Worse, if the configure test fails to find zlib, it still appents -lz to
the other autoconf tests meaning several fail when they shouldn't. This results
in conflicts with system macros and other bizarre issues.

The easiest fix is to depend on zlib-native and make things determinstic.

(From OE-Core rev: b01db0424b9cf73e51808f57043710a1c665b2c5)

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

insane: Fix python function whitespace changes

(From OE-Core rev: 9dee6c2fe0d1882521c0f9a6998257a652845854)

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

classes/buildhistory: fix expansion of escape sequences

OE-Core commit 259b8718a31b886f8a158aeb5de164840c9a28b2 fixed UTF-8
errors but broke decoding of escape sequences in strings (e.g.
pkg_postinst scripts had \n \t in them instead of newlines and tabs.)
We need a second call to decode() here as specifying 'string_escape' as
the second parameter won't do anything.

(From OE-Core rev: 15e0cdff08b8b9b826bbb9f00192a27318a3ee65)

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

rootfs_ipk: Ensure that BAD_RECOMMENDATIONS are honoured for all architectures

Commit a5362de60c0051f16b88a40bd9cb41915bee0b0f restricted us to writing
only a single entry to the status file for any given package.  This
avoids writing garbage entries (with no Status) when multiple versions
of the same package exist in the feed.  However, in the case where the
same version exists for multiple architectures (for example, has been
built for both cortexa15 and generic armv7a), we do need to write out
status file entries for all of them since we can't trivially determine
which one opkg will decide to install.

Fix this by undoing a5362de60c0051f16b88a40bd9cb41915bee0b0f and writing
out an entry for everything that matches, but ensuring that we force the
Status to the correct value for each one.

(From OE-Core rev: 12d7b44c817bdf6eea425bda607379594057c3fb)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lttng-ust: Move lttng-gen-tp to ${PN}-bin

This helper script is used only during development and is not generally
useful on the target.  Inherit lib_package to move it to a different
package from the libraries.

(From OE-Core rev: f606f69254097dc6b3517416b43ddbf80027a8c1)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image.bbclass: Also uninstall update-alternatives and shadow for read-only-rootfs

If the rootfs is read-only then we aren't going to be updating any
alternatives or modifying the password file and these binaries will
be redundant.  In an ideal world we would be able to stop them from
being installed in the first place but this is non-trivial to arrange.
As a workaround in the meantime, let's just uninstall them once image
construction is finished.

(From OE-Core rev: d41097a970e9bfa553cd4bc3c9fad4b9073d7bd5)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base-passwd: Remove unnecessary DEPENDS

Commit 5cab5935d6924e0aca5da4b682bb83e54f20bc1d added dependencies on
docbook-utils-native and linuxdoc-tools-native, even though the same
commit ostensibly disabled building of the documentation and there was
no explanation of why these dependencies might be necessary.  It appears
that they don't serve any useful purpose so let's remove them again.

(From OE-Core rev: 435f0c2c86cec4bc40731eaddec6c59260dd1ef5)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wireless-tools: Upgrade 29 -> 30.pre9

This also fixes the underlinking problems that
are unearthed with newer binutils 2.24
where one of the wireless-tools libraries is asking
for symbols from libm but all the symbols it asks
for are unfortunately weak and they do not let new
linker convince enough to link libm even though -lm
is on cmdline since we are using --as-needed by default
in linker, which means we end up with errors like

| make: *** [iwconfig] Error 1
| libiw.so.30: undefined reference to `ceil'
| libiw.so.30: undefined reference to `pow'
| libiw.so.30: undefined reference to `log10'
| libiw.so.30: undefined reference to `floor'
| collect2: error: ld returned 1 exit status

(From OE-Core rev: e347b6b119f83a29802ae4499ddcb3ff6a23b317)

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

libcgroup: Add bison-native and flex-native in depends

* libcgroup requires native packages for bison and flex to configure
  itself successfully. Added those in its DEPENDS.

(From OE-Core rev: b50bdeaea15fcd24fd2c03fd02d4a1a365948cef)

Signed-off-by: Noor Ahsan <noor_ahsan@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oeqa: allow multiple layers to provide their own TEST_TARGET class

Use a python module "folder" rather than a single module within
layers to ensure multiple layers can define a TEST_TARGET class.
Current implementation using controllers.py module will only allow
a single layer to define test targets.

Add a controllers folder as well as a TestTargetLoader class whose
job is to load the given TEST_TARGET class from any number of
python modules within the oeqa/controllers/ directory of any
layer.
The only condition will be that layers will need to ensure
the TEST_TARGET class name they provide is unique otherwise there
is no guarantee which class is instantiated. a bb.warn is used
to alude to this if it happens.

(From OE-Core rev: 3f25705f4a986e06cbd397aaea52b841c1a1e054)

Signed-off-by: Sipke Vriend <sipke.vriend@xilinx.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-2.0: fix handling of skipped tests

Backport a patch from upstream to fix skipped tests, essential for gdk-pixbuf to
pass without enabling all loaders.

(From OE-Core rev: 4bf996557409b63c2d783f175c6325c966aae236)

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

meta/recipes-core/base-passwd/base-passwd/noshadow.patch: Split it into two parts

The current change contains two different logics resulting that users like me
may need to partially override the behavior of it. It would be easier for end
users to keep one change while dropping the other if needed, like in our case.

(From OE-Core rev: 0ca984f32682151cfeff852167f0174aca20a8bc)

Signed-off-by: Laszlo Papp <lpapp@kde.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libpng: upgrade to 1.6.8

Note: license has not changed. MD5 sum of files changed because new
contributors were added to LICENSE and png.h file contains the
version of the new release.

(From OE-Core rev: f825549d18477bc2c02e96f7a879b36bad1be18d)

Signed-off-by: Marius Avram <marius.avram@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

binutils: Upgrade to 2.24

This builds and runs images for all qemu machines

(From OE-Core rev: 015eca84f1b0f25868b47d2480bb60cea698f70e)

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

nss: Add nativesdk to BBCLASSEXTEND and bugfix

Since shsignlibs is used
from the nss postinstall hook. It should be included in
nativesdk to make offline rootfs construction possible.

(From OE-Core rev: 42bc72d21226e76c9b013fc052f17d847dc6a97a)

Signed-off-by: David Nyström <david.nystrom@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nspr: Added nativesdk to BBCLASSEXTEND

nspr is a dependency of nss. Since shsignlibs is used
from the nss postinstall hook. It should be included in
nativesdk to make offline rootfs construction possible.

(From OE-Core rev: f7433d024baaa074f3e5f05a28ab22543bee7001)

Signed-off-by: David Nyström <david.nystrom@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nativesdk-packagegroup-sdk-host: Adding nativesdk-postinst-intercept to SDK

Adding ability to use postinstalls intercepts in the nativesdk env, and
making sure the correlate between repo + SDK.

This to enable rootfs generation from a package repository using only a
package repository and the toolchain tarball.

See https://github.com/nysan/rootfs-sandbox for examples.

(From OE-Core rev: a54bab492903322e2a2495ea9576ee6b3272700a)

Signed-off-by: David Nyström <david.nystrom@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

postinst-intercept: New recipe to include postinstall intercepts in nativesdk

Adding ability to use postinstalls intercepts in the nativesdk env, and
making sure the correlate between repo + SDK.

This to enable rootfs generation from a package repository using only a
package repository and the toolchain tarball.

See https://github.com/nysan/rootfs-sandbox for examples.

(From OE-Core rev: 5d5ef0f5620ec525491c53e8457e7c65c226bfb4)

Signed-off-by: David Nyström <david.nystrom@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-setuptools: Remove its provided by python-distribute

Use python-distribute-native in packagegroup-toolset-native

Also fixes warnings like

WARNING: The recipe python-setuptools is trying to install files into a
shared area when those files already exist. Those files and their
manifest location are:

/srv/ssd/sgw/builds/world/tmp/sysroots/qemux86-64/usr/lib/python2.7/site-packages/setuptools.pth
    Matched in manifest-qemux86-64-python-distribute

/srv/ssd/sgw/builds/world/tmp/sysroots/qemux86-64/usr/lib/python2.7/site-packages/site.pyc
    Matched in manifest-qemux86-64-python-distribute

/srv/ssd/sgw/builds/world/tmp/sysroots/qemux86-64/usr/lib/python2.7/site-packages/site.py
    Matched in manifest-qemux86-64-python-distribute

Conflicts:
meta/recipes-devtools/python/python-setuptools_1.4.bb

(From OE-Core rev: ddb0c70a7d4d51a1f404c194a562325137222146)

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

python-distribute: Add recipes for python2 and python3

in python3-distribute rename easy_install to easy3_install

(From OE-Core rev: 712d65335b9a92bba427949f8885102d0aefbeb4)

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

python_2.7.3.bb: Inherit python-dir

So it can get the correct python direcotries for python2

(From OE-Core rev: cfc780ec26474a61bd52a8f6a66c7280fa12a075)

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

python3: Add target and native recipes

3.3.3 is latest bugfix release in 3.3 series

License text is changed to reflect year and prev releases

+++ ../Python-3.3.2/LICENSE     2013-05-15 09:32:54.000000000 -0700
@@ -74,7 +74,9 @@
     3.2.1           3.2         2011        PSF         yes
     3.2.2           3.2.1       2011        PSF         yes
     3.2.3           3.2.2       2012        PSF         yes
+    3.2.4           3.2.3       2013        PSF         yes
     3.3.0           3.2         2012        PSF         yes
+    3.3.1           3.3.0       2013        PSF         yes

 Footnotes:

@@ -110,8 +112,8 @@
 distribute, and otherwise use Python alone or in any derivative
version,
 provided, however, that PSF's License Agreement and PSF's notice of
copyright,
 i.e., "Copyright (c) 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008,
2009, 2010,
-2011, 2012 Python Software Foundation; All Rights Reserved" are
retained in Python
-alone or in any derivative version prepared by Licensee.
+2011, 2012, 2013 Python Software Foundation; All Rights Reserved" are
retained
+in Python alone or in any derivative version prepared by Licensee.

We use native python in disutils bbclass and this needs
to peek into target sysroot when building plugins in cross
environment. Otherwise anything that inherits distutils3.bbclass
will not build.

Fix host include contamination issue

Let compiler append sysroot to include path if it can

Fix the compiler invocation and linker flags when cross compiling

The details are in patch header. This should fix the QA errors about
host contamination

Add virtual/libintl to dep list

There is use of libintl.h but we dont
express the dependency

Add lzma dependency for python3-misc

Fixes

and from test-dependencies.sh run there is undeterministic lzma
dependency:

WARN: packages/armv5te-oe-linux-gnueabi/python3/python3-misc/latest lost
dependency on  liblzma

Fix staging warnings due to duplicate install

WARNING: The recipe python3 is trying to install files into a shared
area when those files already exist. Those files and their manifest
location are:

builds/world/tmp/sysroots/qemux86-64/usr/include/python3.3m/pyconfig.h
builds/world/tmp/sysroots/qemux86-64/usr/lib/libpython3.3m.so.1.0
builds/world/tmp/sysroots/qemux86-64/usr/lib/libpython3.3m.so
builds/world/tmp/sysroots/qemux86-64/usr/lib/python3.3/config/Makefile

Let compiler search includes relative to sysroot

This makes it cross compilable and assumption on hardcoded
paths is fixed.

Errors like below would appear if ncursesw is installed on build host

cc1: warning: include location "/usr/include/ncursesw" is unsafe for
cross-compilation [-Wpoison-system-directories]

(From OE-Core rev: 185fcfc3286b355394ae89300f22b8cb7aaa43c1)

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Fix linking not happening issue in modules

We need to tweak CONFIGURE_LDFLAGS for python3 instead of LDFLAGS
directly

Add second compile stage which uses original makefile

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

distutils: Introduce PYTHON_ABI variable

In python3 this has to be taken care of, it defines a variable to
denote ABI currently its at 'm' and it uses this to construct
the directory names for installing the python headers and library
names in sysroot. e.g. it will be something like ../python3.3m/...

We need this information when we are using distutils to cross build
python extentions and want to know the locations of python headers
and libraries install locations

(From OE-Core rev: 9a3e4ac4a4eeed64ec80d400130dff3d26daf336)

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

classes: Add distutils for python3

In line with python2 add distutils and setuptools
classes for handing python3

Use python-distribute instead of python-setuptools in setuptools bbclass

Remove --single-version-externally-managed since its setuptools
specific and we dont use it anymore

Do build_ext as separate step during compile

Add DISTUTILS_BUILD_EXT_ARGS for modules to pass flags to build_ext step
in setup.py

Add build_ext as sepate step during compile and add
the cross sysrooted library and headers since we are cross
compiling

Use ${PYTHON_PN} in place of hardcoding python name

Remove the /etc/share if its empty

Since OE-Core times we now have machine specific sysroots
for targets unlike before when we used arch specific sysroots
so reflect that here

Use MACHINE for sysroot when not building for build host

Python's machinery replaces directories in sysroot path to match OE's
staging area sysroots. Earlier we use to have HOST_SYS represent sysroot
always but now we use MACHINE to represent target sysroots but HOST_SYS
to represent host sysroot. This patch caters to that difference

(From OE-Core rev: 8bb0206ed67228c88dd5bc2d8b36ce28f48b78f4)

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

classes/distutils: Introduce PYTHON_PN

This is needed to accomodate python3 alongside
python2

(From OE-Core rev: ae931c2cf9e48e1fb74b4b727dbf668ea880023f)

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

package_rpm.bbclass: Trigger the regexp on exact keyword match

This causes issues when postinstalls have ERROR keywords
its interpreted as error and image build is cancelled

(From OE-Core rev: f5bc3cfac9545c402b415695c4e0f98ad38fb2b0)

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

python3native.bbclass: Add python3 abstraction class

(From OE-Core rev: ead30aa5e3046f450ee688e14f6d5268d9ed0bd8)

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

python-3.3-manifest: Add python3 manifest file

Like python2 mostly except

This will package collections into python3-core instead
of python3-misc fixes errors like

Traceback (most recent call last):
File "/usr/lib/python3.3/site.py", line 69, in

     import os
   File "/usr/lib/python3.3/os.py", line 659, in <module>
     from collections.abc import MutableMapping
 ImportError: No module named 'collections'

(From OE-Core rev: e314404876e62c05c7ea5f5e79b2b05c3ed9ab84)

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

generate-manifest-3.3.py: Add script to generate python 3.3 manifests

Bases on python 2.7.x generator

Package collections/ in python-core

(From OE-Core rev: 468115573275d6c32924e56bff660b9f6d38de84)

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

sstate-cache-management.sh: don't remove all packagedata sstate archives

* packagedata task was introduced in:
  commit 6107ee294afde395e39d084c33e8e94013c625a9
  Author: Richard Purdie <richard.purdie@linuxfoundation.org>
  Date:   Wed Jan 23 14:27:33 2013 +0000
  Subject: Split do_packagedata task from do_package
* rm_by_stamps wasn't using do_packagedata or do_packagedata_setscene
  stamp files to find which sstate archives to keep, so it was removing
  all of them

(From OE-Core rev: 66ef2e62a7b5fe36c718f56a8ea9d7f6fd77c393)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate-cache-management.sh: Fix available architectures

* grep for AVAILTUNES isn't enough in cases where AVAILTUNE doesn't
  match exactly with TUNE_PKGARCH, e.g. AVAILTUNE "cortexa8thf-neon"
  and TUNE_PKGARCH "cortexa8t2hf-vfp-neon", instead of trying to find
  dynamically every available TUNE_PKGARCH (we have _a lot_ of them
  even with oe-core only), add parameter --extra-archs where user can
  define extra architectures he supports in given build
* Don't replace '-' with '_' for extra-archs, it does apply to MACHINE
  names and some AVAILTUNES, but e.g. cortexa8thf-neon shouldn't be
  converted to cortexa8thf_neon
* Add empty architecture for populate_lic sstate archives
* Add ${build_arch}_${arch} combinations for toolchain recipes (e.g.
  gcc-cross is using x86_64_i586

(From OE-Core rev: a27cc54fb2d0e59f3a800893c1848cb26a7c5335)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate-cache-management.sh: Show total number of files when showing how many will be deleted

* it's good to see some the ratio of delted files until now it was
  showing only when all or none files were to be removed

(From OE-Core rev: 54e6e25f1a369fa6c21ce0f9db3479b1a481825f)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate-cache-management.sh: Fix rm_by_stamps() to work after SSTATE_SWSPEC change

* format of filenames for sstate archives was changed in:
  commit 6f823a23c5f1d0ffa0a27db1c1bc1907de788505
  Author: Richard Purdie <richard.purdie@linuxfoundation.org>
  Date:   Fri Dec 20 12:06:02 2013 +0000
  Subject: sstate: Convert to use ':' as a filename sperator and use SSTATE_SWSPEC globally
* this one doesn't need special care for old sstate- names
  they will be removed automatically as they don't match with
  any checksum in rigth format from stamps directory

(From OE-Core rev: aa36f9c9b5abac58de899f98803d1c4375678044)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate-cache-management.sh: Fix remove_duplicated() to work after SSTATE_SWSPEC change

* format of filenames for sstate archives was changed in:
  commit 6f823a23c5f1d0ffa0a27db1c1bc1907de788505
  Author: Richard Purdie <richard.purdie@linuxfoundation.org>
  Date:   Fri Dec 20 12:06:02 2013 +0000
  Subject: sstate: Convert to use ':' as a filename sperator and use SSTATE_SWSPEC globally
* remove_duplicated() wasn't able to find available architectures and
  duplicate files since this change
* add extra step to remove old sstate archives starting with sstate-
  (instead of sstate:)

(From OE-Core rev: ddb26341611c3dff41ea92a73d93ec01ae2865de)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/cp-noerror: Avoid a race

Its possible something can delete $1 (since it may be empty) whilst cp-noerror
is starting. Add an exception to handle this issue since if this happens, we
shouldn't return an error.

(From OE-Core rev: 13061ed1e1f347589d6955d5cc50a50574b00218)

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

apmd.service: Fix typo (not mandatory EnvironmentFile prefix)

Prefix to EnvironmentFile should be preciding the filenamn.

(From OE-Core rev: 1f694e4cb493b0737b6009382c0957e6837ebbed)

Signed-off-by: Tobias Blom <tobias.blom@techne-dev.se>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

minicom: fix for multilib and x32 build crash

For source URI, the use of ${PN}-${PV} leads to
build failures for multilib and x32.
${BP} is properly used instead.

Partial fix for [YOCTO #5781].

(From OE-Core rev: 2e0dbfaed17fb89381e3bf155a29cd606b698c7c)

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

oeqa/runtime: syslog: update --help test

busybox 1.22 now returns exitcode 0 instead of 1 for
--help options, so this test needs to be updated when
busybox gets upgraded to 1.22.

https://bugs.busybox.net/show_bug.cgi?id=5612
http://git.busybox.net/busybox/commit/?id=efd0698f74caab0a0c8a51228b923ee142e8e278

(From OE-Core rev: 1e560d234b6d3040a7a9f0eb023f1e4a654be1ea)

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

procps: make the 'watch' command register to /bin/watch

This solves the integration problem with busybox.

Previously, there was a patch in busybox to move 'watch' to /usr/bin.
Such patch is not accepted by upsteam and really not necessary as
our ALTERNATIVE system can easily solve such intergration problem.

(From OE-Core rev: 9e6e2aee21ff59687ddfd0e23fd0add9ee81d397)

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>

coreutils: move 'stat' to base_bindir_progs

To correctly integrate with busybox in our system, we should add
'stat' to base_bindir_progs so that the 'stat' commands from busybox
and coreutils both register to /bin/stat.

Previously there was a patch in busybox to move 'stat' to /usr/bin.
But as we can easily solve this integration problem by modifying the
coreutils recipe, this patch has been removed. After all, maintaining
a patch that's not accepted by upsteam should not be our No.1 choice.

(From OE-Core rev: d98d6122bdfd84faaa37912ca66dabebc7eb9da6)

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>

meta-skeleton: rename busybox's bbappend file

As busybox has been upgraded, rename this bbappend file to make it
match the current version of busybox.

(From OE-Core rev: 12e33bd0dc02720122eb8088b08a181b13b6ef0e)

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: upgrade to stable 1.22.1

Upgrade busybox to the stable release 1.22.1.

During this upgrade, 9 patches are removed. Reasons are detailed below.

The following 6 patches are removed as they have been merged.
    meta/recipes-core/busybox/busybox/busybox-lineedit-initialize-delptr.patch
    meta/recipes-core/busybox/busybox/busybox-list-suid-and-non-suid-app-configs.patch
    meta/recipes-core/busybox/busybox/busybox-sed-fix-sed-clusternewline-testcase.patch
    meta/recipes-core/busybox/busybox/busybox-sulogin-empty-root-password.patch
    meta/recipes-core/busybox/busybox/find-get-rid-of-nested-functions.patch
    meta/recipes-core/busybox/busybox/testsuite-du-du-k-works-fix-false-positive.patch

The following three patches are removed because they are mainly about moving
binaries from /bin to /usr/bin to make the update-alternative work correctly
at rootfs time. We can easily solve this problem by changing recipes. There's
no compelling reason why such patches are needed.
    meta/recipes-core/busybox/busybox/run-parts.in.usr-bin.patch
    meta/recipes-core/busybox/busybox/stat-usr-bin.patch
    meta/recipes-core/busybox/busybox/watch.in.usr-bin.patch

(From OE-Core rev: fa282186d82e8fe3c590d4ea79e464116e5ceea2)

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: check that TMPDIR is not located on nfs

There would be some unexpected errors when the whole TMPDIR is located
on nfs, so add a test for it in sanity.bbclass.

Note:
The better way to get the filesystem id should be get f_fsid from struct
statvfs, but there is no f_fsid in os.stat() or os.statvfs(), so we use
'stat -f -c "%t"' here.

BTW., s/tmpdir/TMPDIR/ in the previous comment message to make them have
a uniform.

[YOCTO #5442]

(From OE-Core rev: ee4061b43522c4893b41c3be63d06be1ee7e3c70)

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

sysvinit: add initscripts-functions to RDEPENDS

As sysvinit doesn't inherit update-rc.d, we need to add this dependency
manually.

(From OE-Core rev: 4ea9ca2bbacb20c3ed454b80e2c020c1073f3299)

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>

update-rc.d: automatically rdepend on initscripts-funtions

Automatically add 'initscripts-functions' to packages when needed.

This extra dependency is added only when the package's init script
would be used at system start-up. In other words, in a systemd based
image, if the package has already got its service files, it will not
have this extra dependency.

(From OE-Core rev: de875cb8893fd8372dd77babaef54ab2d790693d)

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>

Revert "Add missing RDEPENDS of initscripts-functions"

Instead of manually adding initscripts to RDEPENDS of each package,
we should make it automatically handled by the update-rc.d.bbclass.
This solution would have the benefit of backward compatibility. In
other words, users need not modify their recipes.

This reverts commit 16080a3485bd793edd66ed8361f1e8b86a9e19ea.

(From OE-Core rev: f9f193219bd510160b6b09bae652a9dc8ea01e7b)

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>

parted: enable ptest support

Install parted test suite and run it as ptest.

(From OE-Core rev: 425a3fb935f40325baeff88d2519b601a87c418d)

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

curl:Fix multilib header conflict - curl/curlbuild.h

curl/curlbuild.h conflicts between 32-bit and 64-bit versions.

(From OE-Core rev: 8b2e163338331fde05e47a4843f6bd1c9b4f9333)

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

Add Cortex A7 support for NEONv2 & FPv4

[YOCTO #5710]

Add tuning options for Cortex-A7 with NEONv2 & FPv4:
- cortexa7hf-neon-vfpv4
- cortexa7thf-neon-vfpv4

(From OE-Core rev: e97d152ca13556b41a236c1a4cfb11e77ff857d7)

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

mdadm: fix build on qemumips64

The mips64 uses long long for u64 in the kernel, but powerpc's asm/types.h
prevents 64-bit userland from seeing this definition, instead defaulting to u64
== long in userspace. Define __SANE_USERSPACE_TYPES__ to get int-ll64.h
included.

We had a similar fix on ppc64, use it for mips64 will fix the problem.

[YOCTO #5758]

(From OE-Core rev: 97bc0d1575b08830e3953933cbbc5732b1a82b31)

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

e2fsprogs:fix build native package error

e2fsprogs requires pkgconfig to build native package

(From OE-Core rev: c9036a6d2ba4a246ec7216be1971de7eb91aad5f)

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

oeqa/utils: sshcontrol: rewrite the SSHControl class

Split the class in two, one to handle the process and the
timeout based on output and one for the actual ssh/scp commands.
The ssh/scp methods now use the same run method.
It does the same thing as before but:
- it looks cleaner.
- adds support for using a different user than root
- optionally, raises an exception when exit code != 0
(that's useful for code outside of tests, where you wouldn't want
to check the return code every time as the tests do)

(From OE-Core rev: bb14a7598d3c0636dc249f719bde0d9d65b2694d)

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

quilt: add ac_cv_path_BASH to CACHED_CONFIGUREVARS also for native

A issue was fixed in target recipe with commit: 95893404

[
    quilt: added ac_cv_path_BASH to CACHED_CONFIGUREVARS

    On Fedora 17, bash has moved to /usr/bin/bash and the configure process finds it
    on the host machine there, this ensures that it is set correctly for the target.

    [YOCTO #2363]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
]

But it's also impacting native package in the case that the sstate is used
between Fedora 17(or later) and other hosts.

(From OE-Core rev: 34bd53c657f82a402723fcf2640b0511a68b6af5)

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

ltp: fix ffsb ftbfs on cross-compiled environment

Add ffsb-remove-hardcoded-configure.patch: get rid of the hardcoded configure call preventing cross-compilation
Add do_configure_prepend to explicitly force regeneration the configure file and pass configure options as appropriate.

(From OE-Core rev: ce48f4ebfeae0bca599c6e39958ec8dec11e09dd)

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

ltp: fix comment to refer to oe-core instead of poky

(From OE-Core rev: 3312774a257ccec6adcc77534579526ef45732c3)

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

ltp: update ffsb-6.0-rc2 license file path

(From OE-Core rev: 9c6e76effaf94f01a577cbb6cfa08a35ca92172c)

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

ltp: drop do_unpack_append/do_extract_tarball

we don't use tarball anymore and switched to ltp github mirror.

(From OE-Core rev: 2af5bcd2e68388ac0844cf6bd14d96823cf3deb9)

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

ltp: fix style using oe-stylize (no changes in the content of the recipe itself)

(From OE-Core rev: 0fa8c2619e08326b4913950e278a3a2b2a55d929)

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

ltp: update SRC_REV to LTP 20140115 release

(From OE-Core rev: 30b24980ea67a95daed5534632fd6605e076f65a)

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

kernel.bbclass: Fix empty modules directory QA issue

If a kernel is built without any external modules (aka no CONFIG_*=m),
then during a modules_install of the kernel an empty directory is
created at /lib/modules/${KERNEL_VERIONS}/kernel. This is behaviour of
the kernel infrastructure, the directory would normally be populated
with the modules that were built.

However because of the expectations of kernel-modules-split, no packages
are created when there are no modules and an empty directory lingers.
This raises QA issues as warning or errors (depending on the distro).

The following patch changes the kernel_do_install task to check if the
directory is empty and if so removes it.

(From OE-Core rev: 63d3070b7e1207164891b154a5b9017731e75872)

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

syslinux: fix booting memtest86+ using pxelinux regression

After syslinux was updated from 4.07 to 6.01, booting memtest86+ using
pxelinux would fail:
PXELINUX 6.01 2013-07-04 Copyright (C) 1994-2013 H. Peter Anvin et al
Loading memtest86... ok
Booting kernel failed: invalid argument

This backports the necessary upstream patches to allow memtest86+ to
boot using pxelinux again.

[YOCTO #5501]

(From OE-Core rev: ed9ccb8622b347173602be8b2126324d4fdf54d8)

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

unfs3: Fix compile regression from unfs3 port

A line got dropped from the original patch when porting to the latest
unfs3.  The regression was introduced in commit 7d8075c64 (unfs3: Fix
dependencies and allow target builds).

This patch restores the missing line from the original which had been
working fine for over a year's time.

 --- a/Config/Makefile.in
 +++ b/Config/Makefile.in
 @@ -12,7 +12,9 @@ all: lib.a
  lib.a: $(OBJS)
        $(AR) crs lib.a $(OBJS)

 -y.tab.h y.tab.c: exports.y
 +y.tab.h: y.tab.c
 +
 +y.tab.c: exports.y
         $(YACC) -d exports.y

(From OE-Core rev: 9edbd9d872c128038fd7b56fac713256fd69c6c5)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: runqueue: Fix race against tasks sharing stamp files

Shared work directories work by assuming bitbake will not run
more than one task with a specific stamp name. Recent runqueue optimisations
accidentally broke this meaning there could be races. This fixes the code.

(Bitbake rev: b1628b1a260ddf43fc9985535b1ddcfcebbb1e5b)

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

gcc: Include patch scheduled for GCC 4.8.3 to fix epilogue on ARM

GCC 4.8.0, 4.8.1 and 4.8.2 can generate broken epilogues for the
ABI used by the kernel. Apply the patch that is included for GCC
4.8.3 from http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58854.

The issue was found on Yocto/Dora and the patch should be backported
to this branch. A kernel built with Dora's GCC 4.8.1 misbehaved on:

 while true;
 do
    (for i in `seq 1 100`;
        do
            echo "Log message... $RANDOM";
        done) | logger;
 done

busybox's syslogd would from time to read a huge negative value and
then exit, strace would get stuck waiting on a syscall. After this
patch it appears to work better.

(From OE-Core rev: 3004eb3b7ee5fd8dfe9c4e5749b4e125d0bd4b59)

Signed-off-by: Holger Hans Peter Freyther <holger@moiji-mobile.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libtool: Set CONFIG_SHELL for nativesdk

We're seeing similar failures that we saw from libtool-native and libtool-cross
where /bin/sh changed from bash to dash on different machines after sstate
reusage. This patch fixes nativesdk to avoid the same issue.

(From OE-Core rev: 7994b2870dac8fd5f6db6d47043378534b644515)

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

db: Fix libtool linking against local libstdc++.so

Building the C++ bindings in a484b35b818768487ff27cf06b8c5d4e128126af
introduced this error on systems with /usr/lib/libstdc++.so present:

    /usr/lib/libstdc++.so: file not recognized: File format not recognized

The shipped libtool is sysroot aware, so pass --with-sysroot so it will
extract the sysroot from the compiler.

(From OE-Core rev: db8a2c29b936e5252970c85def927d9cc56a8376)

Signed-off-by: Tyler Hall <tylerwhall@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cogl: Place the patch in the correct directory

Otherwise this causes parsing warnings and build failures.

(From OE-Core rev: 6de220c2ae8effb9ecb61b9f55d3bacc01abedf9)

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

toaster.bbclass: save original package name

If we can get the original package name when reading
package infos in toaster.bbclass, we save it for proper
referincing in the toater ui.

(From OE-Core rev: c3f3fed1b397b6ee6d44e7d39e63d084e4a88e30)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

documentation.conf: update task descriptions

This patch updates the task descriptions in documentation.conf

It also has a bunch of grammar fixes for the variable descriptions.

(From OE-Core rev: 3f96e97b65bfb2505fb0127a4d6a585e9b14a3e4)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package.bbclass: Show which files require given dependency in debug output

* when log.do_package shows some unexpected dependency, people usually
  need to grep package directory to find which binary was creating that
  dependency, show it directly in the debug output

(From OE-Core rev: cf0696a39b811b13bb6e7dd06a2dad607e93a643)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package.bbclass: Don't search for providers of PRIVATE_LIBS

* split PRIVATE_LIBS and don't use find(), so that libfoo cannot be
  found in PRIVATE_LIBS = "libfoobar"

(From OE-Core rev: 2dec075478f977b554061dd9a4b2b8ff4af3597a)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cogl: kms: include necessary header for size_t

Fixes compilation in cases where stddef.h is not already
included by EGL headers.

(From OE-Core rev: 5df33fc62f2d3a5a15dc387ed26cb7da8d9fbbe1)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

unzip: Pass LDFLAGS to the linker

Change Makefile to use LDFLAGS

(From OE-Core rev: 4f211322eb1179db62c03616b4c113114c612cf8)

Signed-off-by: Mikhail Durnev <Mikhail_Durnev@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/runtime: smart: fix adding channels

Without this it won't add core2-64 and similar channels, as the
directory name in deploy_dir/rpm uses _ not - as the package arch.

(From OE-Core rev: 14e3a8f58e309121760fec70619633281dd9d88c)

Signed-off-by: Stefan Stanacar <stefanx.stanacar@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/runtime: drop restartTarget method

Restart is a property of a specific target, not
of a test class, should a test really need to restart
the target the direct method should be called.
Also some tests used this to enforce more ram, which
makes sense only for qemu targets only (and the inital
reason this was needed isn't valid anymore, qemu machines had
the default ram size bumped a while ago).

(From OE-Core rev: 333a4326082e500bdbcd323af37e183e74adf617)

Signed-off-by: Stefan Stanacar <stefanx.stanacar@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemu: Allow building without X11 support

To enable the construction of minimal hypervisors based on OE, it is
important to be able to build qemu without it requiring X support. By
checking that the distro is built with the x11 feature before adding
virtual/x11 to the dependencies, this is made possible.

(From OE-Core rev: e88b85406f2a6722a507b5db3485358a88d950f7)

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

unfs3: Fix dependencies and allow target builds

Fixed in this patch:
  * All patches marked as submitted to the upstream
  * Remove the pseudo dependency because unfs3 can fully stand alone
    or be used with pseudo and it does not link against pseudo
  * Dependencies to flex for nativesdk and target builds are fixed
    such that unfs3 can be deployed into an image
  * Add unfs3 references in separatebuilddir.inc because unfs3
    works correctly with autotools.

(From OE-Core rev: 7d8075c64bd0734cb70d16acef36c1a17276b359)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pkgconfig: Drop glibconfig from SRC_URI to unbreak builds after previous commit

(From OE-Core rev: 53ac749f02e112061f4d2506f2bc26a73e89a345)

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

pkgconfig: remove out-dated configure options and dependencies

0.27 onwards uses a snapshot of glib 2 not glib 1, so drop the
do_configure_prepend and update the internal/installed options as appropriate.

Legacy scripts were removed in 0.26.

Use of popt was removed in 0.28.

(From OE-Core rev: 99b263285f0ee60b4336992251840e418aa695c4)

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

python-numpy: don't search host directories for libraries

The build process hard-codes searches of various directories in /usr when
looking for libraries, delete this to avoid host contamination.

(From OE-Core rev: 9febe2d184ef76b7cacace15cbe17968e8c37617)

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

scripts/runqemu-internal: use -cpu core2duo for qemux86-64

Now that the tune for qemux86-64 changed to core2-64 we need to
tell the emulator to use a proper CPU model. With the default setting
of qemu64 we'll get things like:

root@qemux86-64:~# smart --help
traps: python[758] trap invalid opcode ip:7f2af01f6be7 sp:7fff49466ef0 error:0 in strop.so[7f2af01f5000+6000]
Illegal instruction

If the tune for qemux86 changes, that needs to be updated too.

(From OE-Core rev: e6ade33a6f52434e884dd97549b8ac731347d9ad)

Signed-off-by: Stefan Stanacar <stefanx.stanacar@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/runtime: rpm: fix finding the test rpm file

Translate TUNE_PKGARCH to find the right file (this
used to work because tune for qemux86-64 used to be
x86_64 now it's core2-64)
Also, while using packagedata was nice, it's harder to make
the test exportable and runnable outside of the
build system. (where oe.packagedata isn't available)

(From OE-Core rev: fd0e9ad4d295ca11b33c3e3e11069421dee834e8)

Signed-off-by: Stefan Stanacar <stefanx.stanacar@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: user-manual: Fix issue with previous revert

(Bitbake rev: 2d052690d46506ba3f4180ad5cdfa835ce85e59a)

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

bitbake: Revert "usermanual: Updated CC License to version 3.0"

This reverts commit 0546bc19557de9263b448ceb1707884543d2de56.

This change was not discussed with the copyright holders so the license
needs to remain at version 2.5.

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

udev: fix typo

Fix the typo in the RDEPENDS statement to make it have real effect.

(From OE-Core rev: aa1224e4d83a4273848ba7601162157f5e415e30)

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