bb/ui/crumbs/tasklistmodel: include package level information

Until now the hob UI has only contained recipe (pn) level targets in the
data store, this patch switches to including package level information.

This is slightly slower in all model related cases (more entries) but gives
much more flexibility for image customisation.

(Bitbake rev: 049927e99c8d1db7273fbd179b2614bd2ea9403b)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/crumbs/tasklistmodel: filter nativesdk packages out of views

(Bitbake rev: 0c62bbc5fb149c8baccffebd5f17e25d3e7dbf69)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake/logging: Overhaul internal logging process

At the moment it bugs me a lot that we only have one effective logging
level for bitbake, despite the logging module having provision to do
more advanced things. This patch:

* Changes the core log level to the lowest level we have messages of
  (DEBUG-2) so messages always flow through the core logger
* Allows build.py's task logging code to log all the output regardless
  of what output is on the console and sets this so log files now
  always contain debug level messages even if these don't appear
  on the console
* Moves the verbose/debug/debug-domains code to be a UI side setting
* Adds a filter to the UI to only print the user requested output.

The result is more complete logfiles on disk but the usual output to the
console.

There are some behaviour changes intentionally made by this patch:

a) the -v option now controls whether output is tee'd to the console.

Ultimately, we likely want to output a message to the user about where the
log file is and avoid placing output directly onto the console for every
executing task.

b) The functions get_debug_levels, the debug_levels variable, the
set_debug_levels, the set_verbosity and set_debug_domains functions are
removed from bb.msg.

c) The "logging" init function changes format.

d) All messages get fired to all handlers all the time leading to an
increase in inter-process traffic. This could likely be hacked around
short term with a function for a UI to only request events greater than
level X. Longer term, having masks for event handlers would be better.

e) logger.getEffectiveLevel() is no longer a reliable guide to what
will/won't get logged so for now we look at the default log levels instead.

[YOCTO #304]

(Bitbake rev: 45aad2f9647df14bcfa5e755b57e1ddab377939a)

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

lib/bb/providers: make "checking PREFERRED_PROVIDER_%s" a debug message

In verbose mode there are hundreds of these "checking PREFERRED_PROVIDER_%s"
messages, cluttering the output and obscuring the more important resulting
"selecting %s to satisfy runtime %s due to %s" messages. Individual "checking"
lines are more suited for debug mode, similar to "sorted providers for %s
are: %s", hence convert logger.verbose() to logger.debug().

(Bitbake rev: 85dfbec26abb5b944758f5c4749b7df16c0fb2e6)

Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rt-tests: use an explicit commit ID

Using a tag, like v0.73, forces network access to resolve a
commit ID. Use the corresponding commit ID in the recipe.

(From OE-Core rev: d126e22f6b3f27196144f87e22b36ebccd6dea65)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/bb/ui/crumbs/hobprefs: fix erroneous save/reparse

We need to unset the reload_required variable once we've triggered a reload
so that we don't cause a reload each time the preferences dialog is shown and
dismissed, regardless of whether anything has changed.

(Bitbake rev: b3ad7acebfad3063c3364f4492f53b25bf53cf81)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/crumbs/configurator: introduce writeConfFile method for all writes

Configuration files are written in several places, this refactors the code
to use a common method.

(Bitbake rev: 2843645755abb736220d7404dc6e853929093ff9)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/crumbs/hobeventhandler: adapt to reset -> reparse change

The API has changed from just a reparse call to a reset call followed by
a reparse call.

(Bitbake rev: c828ba571c3cf27d1bbd65184e287c45ef64de89)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/command|cooker: refactor the reparseFiles logic

Turn the reparseFiles logic into a command to reset the cooker's state
machine and a noop which triggers a cache rebuild.

The resetCooker command resets the cookers state machine such that a cache
update will be triggered by any async command which requires the cache.
The reparseFiles command remains as a noop async command that has the
needcache property set to True so that when called it ensures the cache is
built.

Patch from Richard with the addition of removing the force parameter from
the updateCache method.

CC: Richard Purdie <richard.purdie@linuxfoundation.org>
(Bitbake rev: a98f698fe9f38310024013e58475e6d1447ee154)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tune-ppce5500: Add a set of tune files for PowerPC e5500 core

The PPC e5500 is a 64-bit core so we add both a 32 and 64-bit set of
tune files to allow for:

* pure 32-bit build
* pure 64-bit build
* 32-bit base, 64-bit multilib
* 64-bit base, 32-bit multilib

(From OE-Core rev: 60286934715c5f7f27d539f4a43a7226488ef963)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tune-ppc: Update to pass glibc configure option to get cpu specific support

We need --with-cpu based to glibc to get proper support on 603e & e500mc
to pickup proper math libs to deal with sqrt.  These core do not
implement the fsqrt[s] instructions that the normal PPC math libs
utilize.

This causes use to not set AVAILTUNES specifically to the sub-arch only
as we arent generically compatiable.

(From OE-Core rev: 078699cb8c707830c86b55787fd535d87171388e)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

recipes-rt: add a README documenting the contents of recipes-rt

Document the non-obvious bits of using the recipes-rt recipes.

(From OE-Core rev: 2c86e636972bd02b35523b36bc81a2765f430f70)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Move meta-rt recipes to oe-core (meta)

Keeping the rt recipes in their own layer has led to maintenance issues,
particularly with the linux-yocto-rt recipes. As these kernel types are part of
the same linux-yocto source repository, it seems reasonable to include the rt
kernel recipes alongside the standard recipes. A new recipes-rt directory for
the other recipes provides adequate separation and eliminates the need for a
separate layer.

As there is no meta-rt/conf/layer.conf to force the kernel, users must now
specify the rt kernel in their local.conf or in the machine.conf:

PREFERRED_PROVIDER_virtual/kernel = "linux-yocto-rt"

The merging of the rt recipes into the core also eliminates complications with
multiple layer dependencies for new BSP layers. Having to either separate RT
BSPs from standard BSPs or force users to add meta-rt to bblayers even when not
building an RT BSP (because the RT BSPs in the same layer would fail to parse
without it) was sub-optimal at best.

(From OE-Core rev: bafaaad264fe3e745c714951ddcf2784d33b755a)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: default KMACHINE and SRCREV_machine to avoid parsing errors

Without a default KMACHINE and SRCREV_machine the git URL will fail to
expand to something usable and we get a parsing error. This approach was
also used in the non-rt version of this recipe.

(From OE-Core rev: 92aa6a587f4fabfadbebdb7ae24306d36f43910a)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: add an -rt-sdk image

Add images including the SDK without the addition of the grapical environment or
packages.

(From OE-Core rev: c594579fe1d9a8c97f2dcf8fbec9fc401e73c667)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: rename minimal-rt images to -rt, drop live image

Remove "minimal" from the image recipe names for -rt. Additional recipes types
will be added (-rt-sdk for example) and there is no need to keep so many image
type descriptors.

Drop the -live image as it is now obsolete.

(From OE-Core rev: e5ff4d0ec5752ccc76cfbf6cae827051cb402fc5)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: remove atom-pc support from the main layer

meta-rt resides in oe-core, as such it should only support the
qemu* machines. BSP layers should add support for specific
hardware platforms.

(From OE-Core rev: a9940d0660a1e056f399748fbee0835abe9627f5)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: add linux-yocto-rt_3.0 recipe

Add the initial linux-yocto-rt_3.0 recipe.

qemumips and qemuppc doesn't work quite yet, so they have been omitted from
COMPATIBLE_MACHINE.

(From OE-Core rev: 443533914c43c79d8b15185c32f75df450c6aae7)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: canonicalize directory structure

Use the more standard recipes-*/*/*.bb layout.

(From OE-Core rev: 3b2fb9c103c126aff1370f4b98a91649aa70e36b)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: rt-tests v0.73 instead of AUTOREV

AUTOREV is handy is a development aid, but it really shouldn't be used as a
default as it breaks the consistency and repeatability of a build. It also
causes additional fetches which may not be desired or expected.

(From OE-Core rev: 378368306435fda58cea8105670725feebfe6359)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-rt: refresh linux-yocto-rt recipe

Rename the recipe to match the linux-yocto recipe renames using the kernel
version in the PV portion of the filename.

Add the SRCREVs to the recipe as was done with the core linux-yocto recipes.

Add qemux86 as it is also present in the linux-yocto-2.6.34.git repository.

(From OE-Core rev: 6e9e19230ee731856639f71c5adfb01a92f620d8)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package_rpm: Fix attemptonly and suggest packages

[YOCTO #1325] [YOCTO #1366]

Packages that were in the PACKAGE_ATTEMPTONLY and SUGGESTS were not
being properly found, which was causing image creation failures.

In PACKAGE_ATTEMPTONLY, when an item was not found, it caused an
error.  This should have been a note, followed by skipping the
package.

The SUGGESTS processing was simply broken.  It was using a
non-existant function, due to an apparently typo.

In addition to the above, the MLPREFIX processing was not being
done properly, preventing multilib packages from working in this
with PACKAGE_ATTEMPTONLY.  (SUGGESTS doesn't need this as the names
are munged when creating the packages.)

(From OE-Core rev: 4fb6723ab1ee14cf539f0ef2df63a3ee1b978de6)

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

xcb-proto: update based on changes to python.m4 in automake

1. We dont need the aclocal patch as this will be handled by automake
2. We need to update xcb-proto.pc.in to know about ${libdir} since
   we'll end up with something like:

   pythondir=${libdir}/python2.6/site-packages

(From OE-Core rev: 4de2ba0eeaefbd6813443c6906108b815af9e109)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

automake: Update for python.m4 to respect libdir

As we tweak libdir we need python libs that utilize configure to respect
it setting.  By updating the python.m4 template, when we regen automake
files they will than respect the setting of libdir which is standard for
any autotools based recipe.

(From OE-Core rev: d300cfe603b00fc963f7ed874563016fcbcc47a8)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

init-install.sh: Fix make partition and make file system issues

[YOCTO #1151]

Change the offset from the beginning of the disk to aligne the blocks.

In this script, we use mkfs.ext3 to create file system after partition, but we use mkpartfs to create file system repeatly,
and get some warnings about choose another specific tools to create file system for reliability.

So use mkpart instead of mkpartfs and only use mkfs.ext3 to create file system.

(From OE-Core rev: 67c9804e0416defb7c3f69bd02bbae3710e5be0a)

Signed-off-by: Mei Lei <lei.mei@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package_rpm.bbclass: Update the platform config and --target

When constructing the /etc/rpm/platform file, we need to ensure that
the any, all, and noarch platforms will allow any "linux" variant to
be installable, not just matching variants.. i.e.

arm-oe-linux-gnueabi should be able to install noarch-oe-linux

Also ensure that we pass the full canonical arch via the --target=
parameter.  This allows us to define the proper platform settings
for all packages.

(From OE-Core rev: c6bb9445de90cfb46db9c40842dbe1bfbb5b8ce8)

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

rpm: Fix the canonical arch --target processing and cleanup

[YOCTO #1352]

Fix the canonical arch --target processor to work with the way OE
handles the GNU canonical arch.

Also cleanup a number of files that are no longer used, and refactor
the no-loop patch into the "no-loopmsg" patch.

(From OE-Core rev: 25904ad2bdeadda9189805bb7a210eafe5795024)

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

linux-yocto: update SRCREVs for preempt-rt support + build fixes

Updating the SRCREVs for linux-yocto to pickup a series of changes
that were made to support the build/boot of preempt-rt on 3.0.x.

7782fc6 routerstationpro: remove CONFIG_PREEMPT* definitions
b51300c meta/rt: sugarbay preempt-rt support
735711e meta/rt: common-pc[-64] preempt-rt no branch fix
53026bf meta/rt: jasperforest preempt-rt support
99bc260 meta/rt: fri2 preempt-rt support
e746ab8 meta/rt: fishriver preempt-rt support
fa1b73f meta/rt: emenlow preempt-rt support
1f686ab meta/rt: crownbay preempt-rt support
64a5069 meta/rt: arm-versatile-926ejs preempt-rt support
7f827a1 meta/rt: mti-malta32-be preempt-rt support
930a983 meta/rt: qemu-ppc32 preempt-rt support
367fc55 meta-rt: rsp preempt-rt support
b8a62ea meta/rt: mpc8315 preempt-rt support
b5f9daa meta/rt: beagleboard preempt-rt support
1e6e9eb meta/rt: use CONFIG_PREEMPT_RT_FULL for 3.0 kernels
eb5a06b boot: move -Os patch to standard branch
c692a23 meta/rt: remove explicit patch references
8a2ac6b meta/rt: updating to rt8
76b3679 meta: atom-pc update definition to re-use preempt-rt

(From OE-Core rev: 4cb49cc3212fb110bd90f29e3644df5e7030a359)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cross: also empty PACKAGES_DYNAMIC

(From OE-Core rev: 6135bfb1c462a6ce1d6f2e54c6d1b7609fb7c76d)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package: allow setting a section for locale packages

(From OE-Core rev: 521cab0239352e3b4316183c20e63cc06f5ae92b)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package: include SECTION in the pkgdata

(From OE-Core rev: 47664bb8309f5d791d7aa040a38c7324cb4f14e6)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity: shift the required utils into a variable

(From OE-Core rev: 4b766d82b15f9821a8c7299826799fad9bb88666)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nativesdk: inherit relocatable

(From OE-Core rev: 6a7a5381357861a098640c42ba2f894213086b41)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insane: add linux-gnu to the platform data

(From OE-Core rev: 206f1501b0ccdd971ee40f3d112b4756077a9af9)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: don't strip the main binary

With this, we get a useful busybox-dbg package.

(From OE-Core rev: 95c9834a358118f322159139950dd9a92f561e88)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

local.conf.sample: Cleanup and improve

local.conf is the first thing anyone new to the project sees. Over time it
has built up a ton cruft and isn't even accurate in places.

This patch:

* Moves things to local.conf.sample.extended if a new user is unlikely
  to need to immediately care about the options
* Reorders the file to be more intuitive to a new user
* Moves certain default values to default-distrovars.inc in cases where
  most users wouldn't want to change the value
* Adds large blocks of text to explain what an option does. There have
  been too many cases of a user not realising what some of these
  settings do and how they can use them to their advantage (like DL_DIR
  for example).

(From OE-Core rev: a0ccb33ac5a2231c979c3369159f763dff1e0a2e)

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

gnome-doc-utils: respect python-dir setting EXTRA_OECONF

Don't just assign but append to EXTRA_OECONF so we maintain python-dir
setting fo EXTRA_OECONF.

This fixes issue if libdir is set to something like '/usr/lib64'.

(From OE-Core rev: a92d56058b21913570bb17ae416c3b00afce055e)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Magic file path should be given for rpmbuild

Magic file path should be given for rpmbuild in
_rpmfc_magic_path define so that build system default file
is not used by accident. Not doing this caused many
packages to fail building in several systems.

Fixes [YOCTO #1358]

Signed-off-by: Jani Uusi-Rantala <jani.uusi-rantala@nokia.com>

scripts/runqemu: Make it run on ubuntu 11.10

location of libGL has moved in ubuntu 11.10
so we look for it in the new locations

(From OE-Core rev: 8d80918995a15663db7cc9c0683d45ee8ba7d45c)

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

chkconfig: remove link for update-alternatives to disambiguate

[YOCTO #936]
"update-alternatives" installed: one in /usr/sbin from chkconfig \
(symlinked to "alternatives"), and the other in /usr/bin \
from update-alternatives-cworth.
It appears for whatever reason that the one from chkconfig is run \
during postinst processing but if you run the script from the command \
line it gets the other one.
this is due to differences in the PATH environment variable.

The following is the sequence to call scripts after kernel booting
inittab
  -->rcS
     -->./S40networking(PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin)
     -->S98configure(call rpm-postinstall)
        #"/usr/sbin" is found prior to "/usr/bin", so update-alternatives from chkconfig is run in this script
  -->rc5.d
  -->profile(profile:4:PATH="/usr/local/bin:/usr/bin:/bin" profile:15: PATH=$PATH:/usr/local/sbin:/usr/sbin:/sbin)
  #"/usr/bin/" is found prior to "/usr/sbin", so update-alternatives from update-alternatives-cworth is run in this script

So I remove the symlink (update-alternatives linked to chkconfig). The one from update-alternatives-cworth is left alone.

(From OE-Core rev: f3a52a73251b26ead15abd4910dc82c1011ab852)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/combo-layer: a simple way to script the combo-layer conf

This small patch introduces a a very simple and basic way to script
the combo-layer conf file. With that a combo can be shared with no
need to change its config - associated to the use of environment
variables for example.

*Similar* to bitbake it considers every value starting with @ to be
a python script. So local_repo could be easily configured as:

[bitbake]
local_repo = @os.getenv("LOCAL_REPO_DIR") + "/bitbake"

or any more sophisticated python syntax.

This version updates the config file description so users can be
aware of.

(From OE-Core rev: 62269642ce0e0e56d68d495b6c4d27327c9ed649)

Signed-off-by: Leandro Dorileo <ldorileo@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate: Add level 2 debug so its possible to see what sstate is doing

Currently its hard to figure out if/when sstate is checking for possible
packages to speed up builds. This patch adds level 2 debug output which
better indicates what files are being searched for an why.

[YOCTO #1259]

(From OE-Core rev: a752b23767189f2678367e47ca0b41f49c56a631)

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

gcc: Various fixups to ensure consistent gcc builds

We ensure that:
* the shared work directory contains PR and ensure PR values are consistent across gcc builds
* the regexp to handle library directories is in a specific task and run once

This avoids breakage that was seen in incremental builds after commit
be1f70d68b6b75772ebab8bdff683ddd7c42b0cd where the interpretor could
become corrupted. This was due to the sed expression corrupting
the source directory.

(From OE-Core rev: 970af6b09e1d69041b0d82fa56ace19543405eb1)

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

bitbake/usermanual: Update to be more in sync with bitbake codebase

(Bitbake rev: 6c382c2ee7740a5e78b4135648870e5c181ec23f)

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

bitbake: Fix -e when used with -b option

When using the -e and -b options together an expection was occuring.
This was due to incorrect initialisation and this patch adds in the
correct initialisation calls.

(Bitbake rev: e7ade6dcd646b478df79f968934a0edcb63254f7)

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

bb/cache: rename confusing variable

The bNeedUpdate variable doesn't reflect its purpose, and doesn't match
coding style (type encoded in variable name, camel case) - rename to
cache_ok.

(Bitbake rev: cc4158db215493ac270e27f92d977bcbe337f3db)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/crumbs/tasklistmodel: don't include an item in its own depends

This causes the simple removal algorithm to perform needless circular logic

(Bitbake rev: 396bbc220604c19ced4add30dd17f7f22d5e0022)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/crumbs/tasklistmodel: correctly uniquify dependency list

Fix thinko - the squish method returns a uniquified list, it doesn't modify
the list in place. Therefore the call to squish() was useless as its return
value was never assigned.

(Bitbake rev: 93bae8f223cdeb7b7e31c309b5d785b97ae1c2ac)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/crumbs/tasklistmodel: fix some typos and add comments to mark()

Two similarly named variables in the mark() method resulted in the wrong
variable being used in a couple of places. This patch adresses this in
several ways:
1) Renames the variables to be less similar
2) Uses the correct variables
3) Adds some coments to document the methods intent

Partially addresses [YOCTO #1355]

(Bitbake rev: ba9f2fe496eec0a221b563ffc9bb76eca592192f)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/cooker: only emit ConfigFilePathFound for files which were parsed

When the requested configuration file is found on disk check the
against the configuration files in __depends/__base_depends to ensure the
file was parsed before emitting the ConfigFilePathFound event.
If the requested file wasn't parsed just return (and don't emit).

Fixes [YOCTO #1246]

(Bitbake rev: 705d14d1e1108e0544c7eab827f1242f0839add9)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

fetcher2: show warning message when checksum is mismatch

Currently, if checksum mismatch, fetcher will try mirror,
and if mirror fetching fail, it will show error "Unable
to fetch URL xxx from from any source" which actually
hide the real reason and make user confuse.

so showing warning message will make it more clear.

Fix [YOCTO #1256]

(Bitbake rev: b3e924297d670963714343d02f7898798fec84fd)

Signed-off-by: Yu Ke <ke.yu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake/ui/knotty: Ensure previous failures aren't masked by a final successful command

When running bitbake in -k mode, the last command might succeed but we
still need to preserve any previous failure codes.

(Bitbake rev: 3ad3a53545b37ee8b26b22f4a0e00b19615b449a)

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

bitbake/ui/knotty: Ensure 'No Provider' errors set an exit code

If a No Provider error is shown we need to set the exit code to
show an error occured too.

[YOCTO #1322]

(Bitbake rev: 4d4e01aa3e38d809617e8c1bbaa081182a2308a0)

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

eglibc: force GLIBC_EXTRA_OECONF to not be uset for nativesdk

Don't use GLIBC_EXTRA_OECONF for the nativesdk builds.  Easiest solution
is to just override it settings.  Otherwise we might pickup configure
options ment for the target build.

(From OE-Core rev: a1f87ec65fa1a6d5ce9a010548dbe7c01ab9b711)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe.terminal: improve how we spawn screen

- Name the screen session 'devshell', to avoid confusion if running bitbake
  itself under a screen session.
- Display a warning message when spawning screen, so it's clear to the user
  that screen has been run (otherwise do_devshell just appears to hang).

(From OE-Core rev: 13e01d29d14e7e7403d0c45c5699ea9160243868)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Rework how the devshell functions

In the new implementation, each known terminal is defined as a class in
oe.terminal, as a subclass of bb.process.Popen.  terminal.bbclass wraps this
functionality, providing the metadata pieces.  It obeys the OE_TERMINAL
variable, which is a 'choice' typed variable.  This variable may be 'auto',
'none', or any of the names of the defined terminals.

When using 'auto', or requesting an unsupported terminal, we attempt to spawn
them in priority order until we get one that's available on this system (and
in the case of the X terminals, has DISPLAY defined).  The 'none' value is
used when we're doing things like automated builds, and want to ensure that no
terminal is *ever* spawned, under any circumstances.

Current available terminals:

    gnome
    konsole
    xterm
    rxvt
    screen

(From OE-Core rev: 69f77f80965fa06a057837f8f49eda06855c4086)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

core-image: adjust to use the new IMAGE_FEATURES support

(From OE-Core rev: fa7fbc9695829f7fa981b16d9559e9e984494e09)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image: add support for generally useful {dev,doc,dbg}-pkgs features

(From OE-Core rev: 7fd8fe53072d3bebb31da4e556878d3c25a10478)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image: implement IMAGE_FEATURES

IMAGE_FEATURES is analagous to DISTRO_FEATURES and MACHINE_FEATURES, for
root filesystem construction. Currently, the only supported features are
any defined package groups, as used by the oe.packagegroup python
module.

Example usage:

    PACKAGE_GROUP_myfeature = "vim iptables"
    IMAGE_FEATURES += "myfeature"

(From OE-Core rev: f9f4416a8cfbd37c7d3a8eb19ee82820e2e6b38c)

Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/oe-buildenv-internal: improve the error detecting for $BDIR

The previous fix for a bug in Ubuntu 10.04 readlink,
be2a2764d8ceb398d81714661e6f199c8b11946c, notified the user when a trailing
slash was used. As there is no semantic difference, simply remove any
trailing slashes and proceed without nagging the user.

See [YOCTO #671] for more details.

(From OE-Core rev: 074ca832c0274e0e92698b4d006ef2708be105b8)

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Acked-by: Darren Hart <dvhart@linux.intel.com>
Cc: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tune-x86-64: Fix DEFAULTTUNE order so weak default to overrides the arch-ia32 version

(From OE-Core rev: 34f2b2a207df8013f70a6de5a5f7e911ee2a8d71)

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

sqlite3: Ensure nativesdk package extension is handled correctly

This avoids the warning:

NOTE: multiple providers are available for runtime libsqlite3-dev (sqlite3, sqlite3-nativesdk)
NOTE: consider defining a PREFERRED_PROVIDER entry to match libsqlite3-dev

until such times as we convert nativesdk to use a prefix like multilib.

(From OE-Core rev: 8fb81f538c80735d32db86a20e7f46b3902b4e75)

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

base.bbclass: Add MULTI_PROVIDER_WHITELIST manipulation for multilib

(From OE-Core rev: 4dbc340c63a86879fa727908cfb502f299f5d21b)

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

default-distrovars.inc: Allow world builds to be successful

Currently we exclude some packages with license issues from world builds
but we don't exclude packages that depend on them leading to errors
when trying a "bitbake world". This patch also blacklists the dependees
so that a world build doesn't show horrible errors and requires the -k
option.

[YOCTO #1262]

(From OE-Core rev: c69453fe94a649c518b0e6d79616f05579b864ce)

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

defaultsetup: Use .= to add TCLIBCAPPEND to TMPDIR

Lazily appending causes a bug where wrong cache is
cleared when BB_SRCREV_POLICY = "clear".

Tested with qemuarm on uclibc/eglibc in same build dir

(From OE-Core rev: 391286cfae94f979bc60ecca2e73501781abdd37)

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

qemu: Poke more paths for presence of libgl

On ubuntu 11.10 libGL is not in
/usr/lib/`uname -i`-linux-gnu/ directory
so we search this dir too.

(From OE-Core rev: ced947e989dfbca8055fe57e14207cb6f1357430)

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

bb/ui/hob: Restore toolchain relevant preference settings for build

Fixes [#YOCTO 1354]

(Bitbake rev: c3827690b9ba3625d1df32d9517efbe13d7d9a1e)

Signed-off-by: Jessica Zhang <jessica.zhang@intel.com>
Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/hob: move some code around to avert a race

The data-generated and model-updated signals are different, the model
should only be accessed *after* the model-updated signal. Move code
setting the image combo's backing model to the model-updated callback to
ensure the combo is accurately set when changing the machine with an
image selected.

(Bitbake rev: 46953d3b4595667935d43fe7b20264ae0efd1bdf)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/crumbs/tasklistmodel: prevent hang when removing item

It's possible to trigger an infinite recursion when removing a package
where many of the dependencies share their dependencies. Prevent this by
keeping a list of removed item names and only removing the item when it's
not in the list.

Addresses [YOCTO #1319]

(Bitbake rev: 0ed58f28e869cceeefa930a3f1f46c5f3e9ffe21)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uclibc_0.9.32: Sync mount.h from eglibc

mount.h has few more defines that are being used by newer
applications such as systemd

(From OE-Core rev: 866a34dd80228a0c10cbea5d7715e2acd6cea131)

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

base.bbclass: Ensure PREFERRED_PROVIDER and PREFERRED_VERSION values are set for multilibs

(From OE-Core rev: b12cd9f378eeafb6e23f379d5b005baa6e9a1a62)

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

module-init-tools-cross: Clean the RDEPENDS value

cross recipes should not RDEPENDS on any package. Cross recipe has no
"PACKAGES" define, thus clean RDEPENDS_${PN} value to ensure correctness
in multilib.

(From OE-Core rev: b91fbc597c182dcbd514639a24c575b0cccc8741)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gnome-doc-utils: Avoid RDEPENDS on bash in native case.

gnome-doc-utils-native should not RDEPENDS on bash, thus add RDEPENDS
override to fix the issue.

(From OE-Core rev: 0676c235cdd85143a3a7e0699f343ebf5fb51e3d)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

x86 tune inc files: add x32 abi tune parameters

(From OE-Core rev: 19252e0592c59ed0fb06ca510d11e564518f746d)

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insane.bbclass: add entries for linux-gnux32

For x86_64 new ABI : x32, there is a new
name for the TARGET_OS: linux-gnux32

(From OE-Core rev: 9dd0a7bf64ca2ca3613ab45dc779e9dcc1592342)

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

siteinfo.bbclass: add entries for new x86_64 ABI x32 target

(From OE-Core rev: 1d14c146e31d2d4111d8daafbfe01d23f036fd7d)

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc: remove PACKAGES from eglibc.inc

* PACKAGES were defined in eglibc.inc as well as eglibc-package.inc, definition
  from eglibc.inc was overriden from recipes including eglibc.inc only

(From OE-Core rev: ae97ce68a739dfcf6e7de3feda444b1466503588)

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

multilib.conf: Supported multilib extend in more recipes.

Added more recipes supporting multilib extend.

(From OE-Core rev: 844b2db35ba5b2e263960aa79dee0d4eb2db3f2e)

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lsb: Fix hard code libdir issue for multilib

(From OE-Core rev: 722aa1eb162b80f57290dd924fd7b41654af3a2b)

Signed-off-by: Mei Lei <lei.mei@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kernel-yocto: add CCACHE_DIR dependency to do_kernel_configme

[YOCTO #1350]

Since do_kernel_configme is added before the standard do_configure task
we needed to add CCACHE_DIR so when the kernel builds it's host configure
tools the CCACHE_DIR exists.

(From OE-Core rev: 564d5fd5a289be0ca47d53608e43f9b78d5c6ee0)

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

linux-yocto: merge v3.0.1

3.0.1 -stable has been released. This now becomes the baseline for
the 3.0 linux-yocto tree. As was the policy in the 2.6.34 and
2.6.37 kernels, the version stays at 3.0 in the recipe.

Build and boot tested on qemu* targets.

(From OE-Core rev: 015a74ca21b06d08c04d18f77b8c2a74ffab5a66)

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

linux-yocto: update atom-pc preempt-rt definition

The atom-pc preempt-rt configuration wasn't pulling in the common-pc
settings, and wasn't defining re-use of the preempt-rt/base branch
properly.

(From OE-Core rev: f60c20322018afe9d27aa2d3b085c58ff0dee867)

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

linux-yocto: allow configuration of arbitrary branches

When building an external tree or bootstrapping a BSP the
external branch may not have been checked out. The tools now ensure
that the tree is ready for configuration, so we no longer need to
force the checkout of the external branch.

This change is coupled with some kern tools tweaks as follows:

  40d9bab updateme: allow the location of board descriptions based on defines
  59859ca createme: use branch name when creating meta data
  91b4275 configme: determine meta branch based on directories, not branch naming
  f5a915c kgit-meta: make branch creation and renaming more robust

(From OE-Core rev: 02c82bf5255b09bb27a0a4509861a3d2bd84b09f)

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

linux-yocto: pass KMACHINE to updateme, not MACHINE

To support the mapping of any oe/yocto MACHINE to a kernel
branch that may not share that naming structure we have
KMACHINE and KBRANCH. To allow the mapping to work, we
actually have to pass KMACHINE into updateme and not MACHINE.

(From OE-Core rev: 658f2cc721b121c0a9a9e7aa8b41463952314c0b)

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

lttng-control: Update to 0.89

(From OE-Core rev: 220c1e5a3aeb20cf0657995d0af4a6eb811f2e24)

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

lttng-ust: Update to 0.15

This address [YOCTO #1005] which was waiting for upstream updates
from the lttng and GDB teams.

(From OE-Core rev: 0d240f56ea0128c1228f7cfe1471039a7b602a88)

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

liburcu: Update to 0.6.4

(From OE-Core rev: a2f09cb149121ed2fddd7d602b286711ce1acecf)

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

gdb: Update to 7.3

This has one upstream patch removed and a couple rebased

(From OE-Core rev: e1e9ee24e82e0ad45cf18387e131d2365c8e77b4)

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

gtk-icon-cache bbclass: don't assing to global RDEPENDS

Only the package with the icons needs it

[Squashed laster RDEPENDS fix from koen into commit]
(From OE-Core rev: 34fca51d84d4629d82880aa0f47123db86de256b)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

apr-util: disable pqsql support to avoid configure error

The pqsql config script will check host path and add host paths to compiler and linker options:
adding "-I/usr/include/postgresql" to CPPFLAGS
adding "-L/usr/lib" to LDFLAGS

Disable pqsql support since we didn't use this feature in other recipes.

(From OE-Core rev: a0034efba2c5960ef946de318438794acb13f454)

Signed-off-by: Mei Lei <lei.mei@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-native: Fix a compiler finding issue

The CC variable sometimes add option information after compiler name, but python can't get the real compiler name if those information added.
Fix this issue by dropping the option information when finding compiler name.

(From OE-Core rev: 461df79fbb655e4c6894d9a9e9ce52656b96303d)

Signed-off-by: Mei Lei <lei.mei@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gst-plugins: partially sync packaging with OE .dev

Gst-plugins get 2 extra packages:

${PN}-apps: helper apps in ${bindir}
${PN}-meta: meta package that will drag in all plugins, libs and apps ${PN} generates

And all libs are split out and run through debian style renaming if enabled.

The packaging include was split out to be reused by external plugins (e.g. gst-plugin-gl, gst-fluendo-*).

The new package list looks like:

libgstapp-0.10-0_0.10.32-r1_armv7a.ipk
libgstfft-0.10-0_0.10.32-r1_armv7a.ipk
[..]
gst-plugins-base-videorate_0.10.32-r1_armv7a.ipk
gst-plugins-base-videoscale_0.10.32-r1_armv7a.ipk
[..]
gst-plugins-base-meta_0.10.32-r1_armv7a.ipk
[..]
gst-plugins-base-apps_0.10.32-r1_armv7a.ipk

(From OE-Core rev: 4bfd595934c1f6839ae848b09ecca439d92bf014)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc: use ${base_lib} to match gcc default configuration

Rather than tweaking MULTILIB_DIRNAMES & MULTILIB_OSDIRNAMES like is
done for x86-64 via 64bithack.patch.  We can just go with gcc defaults
and utilize ${base_lib} for where to find gcc libs.

(From OE-Core rev: cfcc282efef30ea1988ccbe14835254a999584ea)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake.conf/powerpc64: Set baselib to 'lib64' for ppc64

(From OE-Core rev: 7a278238d9b08e0315e92d386282cb488cc0c7b4)

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/cooker: mark parsed conf files as potential cache invalidators

The include parameter to _parse() tells the parser whether to include the
file in __base_depends, the contents of which have their mtime tested to
invalidate the cache when they have changed.

As I understand it we would want all of the configuration files that are
currently handled with the _parse() method to invalidate the cache to set the
default value of the include parameter to True.

(Bitbake rev: 5cacdc4f1641eda1b5707c96f7c40924a9db6174)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bb/ui/hob: reset event handler once build complete

The HobEventHandler passes events on to RunningBuild when a build is in
progress. This patch resets the HobEventHandler to handle events when the
build is complete.

(Bitbake rev: 3d66d4e64a6352c01662a24ee957eddcd3fde7fc)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

task-core-boot.bb: Add PACKAGE definition to facilitate multilib

If there is no PACKAGES defined in bb file, its default definition would
be "${PN}-dbg ${PN} ${PN}-doc ${PN}-dev ${PN}-staticdev ${PN}-locale".
In multilib case, ${PN} is extended with ${MLPREFIX}, so there is no
rename occurred for the PACKAGES. Therefore for task-core-boot,
RDEPENDS_lib32-task-core-boot will be empty, resulting busybox,
tinylogin, etc. are not bulit out.

This fixed the issue that /bin/sh is not provided while do_rootfs.

(From OE-Core rev: f3d8fd28d779ff841410e5bd22289680f04022a2)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc-locale: Added ${MLPREFIX} for task dependency.

[YOCTO #1338] Added ${MLPREFIX} to the task dependency between
eglibc-locale:do_install and the corresponding virtual/libc's
do_populate_sysroot.

(From OE-Core rev: c96bf9197e1e6566ec7a4cd6b92499fb82c9c697)

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ghostscript: Fixed the incorrect DEPENDS.

[YOCTO #1337]
Using ghostscript-native instead of ${PN}-native in DEPENDS to correct
the invalid DEPENDS in multilib cases.

(From OE-Core rev: d8d412ff1beee4c820cc6e7b6124280db920cccd)

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Misc: Corrected the dependency on non-native for -native package.

[YOCTO #1336] Corrected the wrong dependency on non-native for -native
packages for the following recipes:

- util-macros
- intltool

(From OE-Core rev: e33bd6024c5f4cdd5c16d51e9b5c75ec4e1f1fb8)

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>