bb/ui/crumbs/runningbuild: emit signal when command fails with exit signal

Emit the generic build-complete signal when a command fails with an exit
signal enabling the UI to update itself accordingly.

Addresses [YOCTO #1265]

(Bitbake rev: 25ec13075855f7e321a9763682a8cd4ca09203cd)

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

bb/ui/hob: disable editing in the build messages tree view

Addresses the second part of [YOCTO #1311]

(Bitbake rev: d790386112b0beb444713a12d766bd4bd3441f56)

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

ui/crumbs/runningbuild: add optional readonly mode, default off

In b947e7aa405966262c0614cae02e7978ec637095 Bob started to introduce code
for a right-click menu, whilst most of the code is non-invasive it does
enable the editable property of the gtk.TreeView which can be confusing.

This change adds a readonly parameter, defaulting to False, to the
RunningBuildTreeView which if True will prevent the editable property from
being set.

CC: Bob Foerster <robert@erafx.com>
(Bitbake rev: d1d6bfab1789d8c0a4faa109bcc06b3cefe62620)

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

bb/ui/hob: show build messages are displayed in the order they're received

Use the new sequential option of RunningBuild to ensure this.

Fixes the first part of [YOCTO #1311]

(Bitbake rev: 972769e6362ea3cf35c8fb7cdfbb147758d0a2f2)

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

bb/ui/crumbs/runningbuild: optionally create list entries sequentially

In b947e7aa405966262c0614cae02e7978ec637095 Bob changed the behaviour of
the RunningBuildModel such that the items are added to the model in a
non-sequential order.

The messages in the view being listed out of order from how they are
received is undesirable for users of the hob UI, therefore this patch adds
an optional sequential parameter to the RunningBuild initialiser which,
when set to True, will always append new messages so that the order shown
in the view is that the messages are received in. The parameter defaults to
to False such that the behaviour added by Bob is preserved.

Partially addresses [YOCTO #1311]

CC: Bob Foerster <robert@erafx.com>
(Bitbake rev: b16663e1919fddbf63d0ca7f9ad3ffdc7d1121fd)

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

bb/ui/hob: save changes to bblayers.conf when using Add Layer menu item

Fixes [YOCTO #1283]

(Bitbake rev: d1f1ebbe504f5567783461946c0eb45b1e923dd8)

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

gcc: Fix setting of GLIBC_DYNAMIC_LINKER

The sed regex in do_configure_prepend was producing the following result:

 #define GLIBC_DYNAMIC_LINKER64 SYSTEMLIBS_DIR "/lib64/ld-linux-x86-64.so.2"

instead of removing the leading "/lib" or "/lib64".

Now we have it do:

 #define GLIBC_DYNAMIC_LINKER64 SYSTEMLIBS_DIR "ld-linux-x86-64.so.2"

Additionally, with the regex fixed the manipulation of SYSTEMLIBS_DIR
needs to be removed.

(From OE-Core rev: b343436d03fa5690b16ee8ef5d4a738c5194b301)

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

automake: Fix issue with tar configure failing with large UID/GIDs

Add a fixup for trying to create ustar archives when the users
UID or GID is larger than the format allows.

(From OE-Core rev: 7a092818f901c5c860e82f573534cd505d9fb959)

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

package_rpm: Ensure alternatives links are reflected in rpm package dependencies

Currently, if a file is provided as an alternative link within the package, rpm
doesn't see the dependency. This works out badly for dependencies such as /bin/sh
which scripts might require.

Since rpm detects and adds these dependencies we do need to ensure the dependency
information in the packages is correct. This patch does so for the rpm backend
ensuring internal consistency whilst the approach for addressing this problem in
the other package backends is considered.

(From OE-Core rev: 102fc29b9c9eae8435b29edb41b5090ace2373c5)

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

autotools.bbclass: Fix m4 file race

If something removes .m4 files from the aclocal directory whilst aclocal is
running it gets upset. To avoid this we need to take a copy of the aclocal
directory and build against this instead.

[YOCTO #861]

(From OE-Core rev: 6a4778e21341f031569f561e7699d2c616b60af0)

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

bitbake.conf/qemux86-64: Automate TRANSLATED_TARGET_ARCH

(From OE-Core rev: 6def7129cf7580a935c05cc05b7f803812d5bb18)

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

libzypp: Correctly handle - in package architecture names

If package architectures contain - characters this needs to be mapped
to something else in (_) since its illegal in C variable names.

[YOCTO #1313 partially]

(From OE-Core rev: 4806a459880d0860b563d30d3fa8d7cde3461cc6)

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

hob: remove temporary directory on program shutdown

Move temp directory handling into the HobEventHandler and clean up the
temporary files on program close.

Fixes [YOCTO #1307]

(Bitbake rev: 1009ca570a750a00b0e60afcc30ead070c7b310a)

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

ui/crumbs/layereditor: if layer adding fails show an explanatory dialog

When adding a layer fails try to explain why to the user.

(Bitbake rev: 2f7eadfdd710f84a299d6fc7be67ddb089f03ecc)

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

ui/crumbs/configurator: tweak addLayerConf return values

Use different return types when adding failed versus when the layer is
already enabled.

(Bitbake rev: 906854d75a12c8752efb1e4065eaf1a10b191612)

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

ui/crumbs/hobeventhandler: emit a signal when a command fails

If a CommandFailed event is seen emit a signal with the error message.

(Bitbake rev: 15cc4fe5051dcb6ac5915e10228b5f3d3ed845bd)

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

ui/hob: show error and exit if we receive the fatal-error signal

If we receive the fatal-error signal from the handler display a dialog
informing the user of an error and displaying the error message. Dismissing
the dialog quits the application.

Fixes [YOCTO #1279]

(Bitbake rev: 04dbd80bbc9ffe2a566b8cc488f0feebadcdde60)

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

ui/crumbs/hobeventhandler: emit a signal when there's a fatal-error

If the cooker encounters an error we're unable to proceed so emit a signal
so that UI's can notify the user and exit.

(Bitbake rev: b6f6edd5090e933ac97a65f93ea9d1b255fca811)

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

hob: unset busy cursor on exit

Prevent the busy cursor being shown after hob exits if quit is called
whilst the busy cursor is set.

(Bitbake rev: 7b977ff222d4a318efabd0ca5f37fa03b9acb996)

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

ui/crumbs/hobeventhandler: remove unused code

Remove some unused variables and methods.

(Bitbake rev: b1b02d523c6281615b72255774ed455b4cae4847)

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

tune/ppc: Fix various TUNE_PKGARCH issues

We need to ensure only one value ends up in TUNE_PKGARCH rather than several.
This change ensures consistency accross all the PPC tune files and that they
correctly inherit the core value but also allow it to be overwritten.

(From OE-Core rev: f9a8b719dd3fc7593a509c8f288caf1486add2f8)

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

eglibc: Add missing ${PN} and ${PKGSUFFIX} usages

This fixes warning messages such as:

The providers found were: ['virtual:nativesdk:eglibc_2.13.bb', 'eglibc_2.13.bb']
The PREFERRED_PROVIDER entries resulting in this conflict were: ['PREFERRED_PROVIDER_virtual/libc-nativesdk = eglibc-nativesdk', 'PREFERRED_PROVIDER_virtual/libc = eglibc']
NOTE: multiple providers are available for runtime glibc-utils (eglibc, eglibc-nativesdk, external-csl-toolchain, external-poky-toolchain)
NOTE: consider defining a PREFERRED_PROVIDER entry to match glibc-utils

This highlights how ugly adding to PN is when using BBCLASSEXTEND. We need to do this to
eglibc so the nativesdk class continues to work correctly for now and ensure consistency
in how the various eglibc packages are generated.

Longer term, if we switch nativesdk to use the multilib approach and MLPREFIX, we can
remove all the PKGSUFFIX mess. I'd rather not get into changing nativesdk right at this
time though given all the other changes that are going on.

(From OE-Core rev: 3a18aa65a4c123ea922b6cd09ca6a77090f12252)

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

package_ipk: SDK generation workaround

During SDK generation the multilib_sanity_check function is being called,
however it is not available unless we've been called from the image.bbclass.

Disable the check if MULTILIB_CHECK_FILE (also set in image.bbclass) doesn't
exist or is empty.

(From OE-Core rev: 0611cceca1595ea4656ebcb985cbd92bb9774ec8)

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

populate_sdk_*: Sync SDK and regular rootfs functions

In order for things to be easier to maintain in the future, sync up
the sdk and rootfs versions.

(From OE-Core rev: 3c78da15457d8d20c9964cd845f6e021d442aaae)

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

bitbake.conf: Add SDK_PACKAGE_ARCHS

Add SDK_PACKAGE_ARCHS to avoid the sed operations we were previously doing
inside of the variouns populate_sdk functions and related items.

Also add documentation to populate_sdk to explain when the various functions
are expected to be doing.

Finally fix a bug in populate_sdk_rpm where the wrong value was being set,
noticed while working on this change.

(From OE-Core rev: 587c1d5bac71fa6faa65ee3a271391cbf931e8f7)

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

rootfs_rpm: Cleanup and minor bug fixes

in packages_rpm.bbclass:

Cleanup the way package_update_index_rpm works.  Break this out into two
functions, one for common functionality and one for controlling target vs
sdk package processing.

Similar change to package_generate_rpm_conf.

Usage of IMAGE_ROOTFS should actually be "target_rootfs", this fixes a problem
with the SDK case.

in rootfs_rpm.bbclass:

Minor re-order of the values, and avoid multiple uses of IMAGE_ROOTFS in order
to make the merge with the SDK behavior easier.

(From OE-Core rev: 46559f36d8a39655060b45d3d235de44a2c4c09b)

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

oe-init-build-env, scripts/oe-buildenv-internal: add error detecting for $BDIR

[YOCTO #671]

"readlink -f" in Ubuntu 10.04 is buggy: it doesn't ignore a trailing / (e.g.,
"readlink -f /tmp/non-existent-dir/" returns nothing, but according to
http://www.gnu.org/s/coreutils/manual/coreutils.pdf it should do that --
hence we get bug 671. It seems Ubuntu 10.10 or even later Ubuntu 11.04,
and other Linux distributions(e.g., Open Suse 11.4) haven't such an issue.

So I think we should detect this and ask Ubuntu 10.04 users to avoid supply
a path with trailing slash here.

Moreever, I also add the detection of non-existent path, e.g.,
source oe-init-build-env /non-existent-dir/build
can be detected and we'll print an error msg.
And, if we get errors in oe-buildenv-internal, we should stop the script
and shouldn't further run.

(From OE-Core rev: 651ccb3b031d9ccb8331505a51171372002230d9)

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/runqemu: grep for line beginning with TMPDIR

Currently the grep regexp matches any occurance of
'TMPDIR=' but if you have another variable defined
e.g. OE_BUILD_TMPDIR=xxx then that gets picked up
too.

$ bitbake -e | grep TMPDIR=\"
TMPDIR="/home/kraj/work/angstrom/build/tmp-angstrom_2010_x-eglibc"
OE_BUILD_TMPDIR="/home/kraj/work/angstrom/build/tmp-angstrom_2010_x"

So we become a bit more stringent and look for
line starting with TMPDIR

$ bitbake -e | grep ^TMPDIR=\"
TMPDIR="/home/kraj/work/angstrom/build/tmp-angstrom_2010_x-eglibc"

make sure that it greps
only TMPDIR=xxx occurance and not values of other variables
whose names happens to end with TMPDIR

(From OE-Core rev: 12ddf6c6a7559d97d9b8f84fcc89ed02e30df85d)

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

arch-armv7a.inc: fix armv7a-vfp-neon -> armv7a compat case

Without this 'armv7a' is used as TUNE_ARCH but does *not* end up in PACKAGE_EXTRA_ARCHS:

arch all 1
arch any 6
arch noarch 11
arch arm 16
arch armv4 21
arch armv4t 26
arch armv5 31
arch armv5t 36
arch armv5-vfp 41
arch armv5t-vfp 46
arch armv5e 51
arch armv5te 56
arch armv5e-vfp 61
arch armv5te-vfp 66
arch armv6-vfp 71
arch armv6t-vfp 76
arch armv7-vfp 81
arch armv7t2-vfp 86
arch armv7a-vfp 91
arch armv7at2-vfp 96
arch armv7a-vfp-neon 101
arch armv7at2-vfp-neon 106
arch beagleboard 111

Which leads to a failing do_rootfs

(From OE-Core rev: 2a41a311ddda11713296391050f3c2c1b2c1d3d3)

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

hob: fix save/restore of toolchain preferences

Add some (namespaced) custom variables to the configuration file for sake
of this UI.

(Bitbake rev: c9dd2592434338bdddb3cc6f42e760c86fa9e6bb)

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

hob: more reliable disabling of GPLv3 packages

1. reflect GPLv3's presence in INCOMPATIBLE_LICENSE value in the UI

The hob UI currently only supports GPLv3 as a value for
INCOMPATIBLE_LICENSE but doesn't properly reflect whether the value is
already set. This patch rectifies this.

2. don't stomp over other INCOMPATIBLE_LICENSE values when disabling GPLv3

In case the user has other values set for INCOMPATIBLE_LICENSE we don't
want to overwrite the value, we want to modify it.

Fixes [#1286]

(Bitbake rev: 68b992922bc7148d657a1c706c6acc67812a87c0)

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

package_rpm.bbclass: fix one typo

fix the MLPREFIX referrence typo which cause multilib rpm
do_rootfs failure

CC: Mark Hatle <mark.hatle@windriver.com>
(From OE-Core rev: 9653089620ad5aef7876b582708561c3a315a064)

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

package_{deb, rpm, ipk}.bbclass: fix 'lingusa' typo

(From OE-Core rev: 5e22d7d1fea7dfb14380de5692e4b2940781b518)

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

rootfs_rpm.bbclass: Fix multilib configurations to use DEFAULTTUNE

Earlier iterations of the multilib patch used "TUNENAME", which was later
renamed to DEFAULTTUNE.  This file was missed in the rename.

(From OE-Core rev: fdd39925df8268ad41c1ff643a0761b03fad8e08)

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

sanity.bbclass: Add sanity check that TUNE_PKGARCH appears in PACKAGE_ARCHS

Its possible we get duplications if we explicity add TUNE_PKGARCH to
PACKAGE_ARCHS so instead just add a sanity check to verify it.

(From OE-Core rev: b2ae0444eb1984b86c02e571bb1efeaacdeafcf1)

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

tune-ppce500mc: Adjust PowerPC e500mc tune file to correctly set hard fpu.

(From OE-Core rev: 1456cdef9e1a28f7abe1385b10caa5069ba7afac)

Signed-off-by: Malcolm Crossley <malcolm.crossley@ge.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tune/arch-powerpc64: Remove support for soft-float from ppc64

All 64-bit PPC processors support hard-float so no need to support
soft-float.

(From OE-Core rev: 54c7d1faf5376c8fb9b19f4e192ce959c8442782)

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

tune/arch-powerpc64: Fix typo with 64-bit TUNE_CCARGS handling

When figuring out how to set TUNE_CCARGS we should look for 'm64' not
'n64' in TUNE_FEATURES.

(From OE-Core rev: 7a9ea28e69e8121a559f610dd2330edd33f0a907)

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

sgml-common-native: Read a catalog backup to repopulate sgml-docbook.cat

This reads the sgml-docbook.bak and adds the entries into sgml-docbook.cat since
this occurs out of order during shared state sysroot populate.

(From OE-Core rev: 87d68520c190007869ed1877fbb40eef23763524)

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

openjade-native: Write a bak file for shared state

This patch writes a sgml-docbook.bak that is read by the sgml-common during a shared
state sysroot populate, because it comes out of order.

(From OE-Core rev: b372198fe3c9d8be18bd15545fb0bc4dd9d5aa68)

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

docbook-dsssl-stylesheets-native: Write a bak file for shared state

This patch writes a sgml-docbook.bak that is read by the sgml-common during a shared
state sysroot populate, because it comes out of order.

(From OE-Core rev: 4d4047c7b997a9cee96f170c203315cbd82ff084)

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

docbook-sgml-dtd-native: Write a bak file for shared state

This patch writes a sgml-docbook.bak that is read by the sgml-common during a shared
state sysroot populate, because it comes out of order.

(From OE-Core rev: 457f30a389031947f037bd7ea1217c190c2089c1)

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

libzypp: Fix variable substitution problem in do_archgen step

The do_archgen step creates a script that utilizes the variable name
${ARCH}.  However, we also utilize and define ${ARCH} so instead of
having the following in the script:

COMPAT_WITH="${ARCH},${COMPAT} $COMPAT_WITH"

We get something like:

COMPAT_WITH="powerpc,${COMPAT} $COMPAT_WITH"

Just renaming the variable in the script to not conflict with ${ARCH}
fixes the issue.

[PR bump from Saul Wold]
(From OE-Core rev: 09573a276456122b408c5bcebc537455fd1c5f29)

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

glibc: deleted

Glibc 2.10.1 is outdated now and eglibc seems to provide a superset of
its functionality.

(From OE-Core rev: 1d9866a2c4fe93202749a56c7ad89a3cbd07d9f8)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-environment: set correct *MULTIMACH_TARGET_SYS.

Fixed [BUGID #1299]. OLD_MULTIMACH_TARGET_SYS is no longer available.
Use new recipe-scope variable REAL_MULTIMACH_TARGET_SYS instead.

(From OE-Core rev: b69a2be2f30dc633597399d42d6c87d9f0910c1a)

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

feature-arm-thumb: respect ARM_INSTRUCTION_SET

(From OE-Core rev: e23f9ce928353c9da2b9c4bfa9a1a125a7d160f2)

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

tune-xscale: fix xscale/xscale-be confusion

Currently tune-xscale.inc has options wrt. setting of xscale/xscale-be tunes.
Fix that.

(From OE-Core rev: 90ce6e4b7ffaaf1e3af39583ca62b20f08e78959)

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

arch-armv6, arch-armv5-dsp: correct endianness confusion

PACKAGE_EXTRA_ARCHS_tune-armv5eb needs to be defined in terms of
the non-e with the same endianness, i.e. PACKAGE_EXTRA_ARCHS_tune-armv5b
not PACKAGE_EXTRA_ARCHS_tune-armv5, otherwise PACKAGE_EXTRA_ARCHS will
end up containing a semi-random mixture of endiannesses and disaster
will ensue.  Likewise for the vfp and armv6 variants.

This is all a bit confusing because TUNE_FEATURES is done the opposite
way around, i.e. TUNE_FEATURES_tune-armv5eb is derived by taking the
armv5e version and adding bigendian.  But fixing that is probably
a subject for a separate patch.

(From OE-Core rev: 391c0102a81455c76244d13b6878e3a76cca65dc)

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

tune-cortex{m1, m3, r4}: correct spelling of "cortex"

Otherwise the test in TUNE_CCARGS will never match.

(From OE-Core rev: 3b7784021259ac745c80043bec16189fa8f4e45e)

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

bitbake.conf: remove PACKAGE_EXTRA_ARCHS_tune-XXX default

Because of the way BitBake handles ??= under certain circumstances, this
default setting ends up stepping all over the real setting from the arch
include file. Since virtually all arch include files or tune files define
a real value for this we shouldn't need to have a default (or it needs to
be done in a different way).

(From OE-Core rev: 6c43ca6fc6a7fffc84cf28684cac0c0eb4129902)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rootfs_rpm: Disable debug messages

(From OE-Core rev: 59c26a280f909470e19a2be04b8ea9dd3aa5c24b)

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

package_rpm.bbclass: Fix mistake reported by Mark Hatle

(From OE-Core rev: 78a1a915d660d6039ac532239e25ead8507bc2df)

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

bitbake/providers: list PREFERRED_VERSION candidates when unavailable

If the specified PREFERRED_VERSION is not available then list the
available versions in the output. (PR is omitted.)

(Bitbake rev: eea5ff9f34bb9b2e29f5fa43deb80d4aa6ef7ddc)

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

bitbake/taskdata: fix incorrect usage of rdependees instead of dependees

This looked like a copy-paste error - the code around is dealing with
depends and not rdepends.

(Bitbake rev: bb688635c2050da3cbbaa5aa5b00e882887695de)

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

bitbake: show more information for NoProvider errors

"Nothing PROVIDES" errors often come up when a recipe has been skipped
for some reason, and therefore it is useful to print out that reason
information when showing the error so that the user understands why the
error has occurred.

Given that we already feed the reason information into the skiplist for
various situations (COMMERCIAL_LICENSE, COMPATIBLE_MACHINE etc.) this
should now output a useful error message for skipped recipes.

Fixes [YOCTO #846], [YOCTO #1127]

(Bitbake rev: 6765218430e31c165888f26fbc75023c89a6eab2)

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

ui/crumbs/hobprefs: trigger reparse when package format changed

reload_data after package format change to make the change take effects in
next build.

Fixes [YOCTO #1287].

(Bitbake rev: fb47c7452455f3f8d943b21dd61300ec55eea141)

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>

cooker: populate rdepends-pkg in generatePkgDepTreeData

The rdepends-pkg field of the generated depend_tree model was not populated
in the original implementation of this method, this series adds in the
loop to populate the rdepends-pkg column of the model.

(Bitbake rev: 4f9a6f6f43cf2ef38115285897fbbde01964e892)

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

hob: fix loading customised image recipe

The signal handler of the 'Base image' combo was still connected during load
such that updating the UI to reflect the loaded base image triggered a change
of the model. Fix this by disconnecting the signal handler when updating the
displayed 'Base image'.

Fixes [YOCTO #1282]

(Bitbake rev: 58036a79cb79d1dff307e2cfed0e684493178507)

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

ui/crumbs/tasklistmodel: work around overly aggressive package removal

The mark() method, which removes dependent and rdependent items, is overly
aggressive removing items which are actually required by user selected
items and then causing a removal of those items. Because the data
structures used are not fine grained enough to do more intelligent
dependency tracking the simplest "fix" is to track removals which are
marked as "User Selected" and re-add those (and therefore their
dependencies) once the aggressive removal is completed.

Because the aggressive removal already ignores images and tasks this should
make the removal behave as expected though certainly leaves area for
improvement in future.

Fixes [YOCTO #1280].

(Bitbake rev: 1e1055262450de994202fc3e5943b8b19f628681)

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

ghostscript: fix parallel build issue

ghostscript fails some time on autobuilder, it seems a parallel build issue.
Add patch to fix it.

Fixes [Yocto #1202]

(From OE-Core rev: 90c0eabcce04e8358ce8df9cd9ed60fdeea68cf5)

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

opkg svn: bump SRCREV to 625

tested on beagleboard/angstrom and qemuarm/angstrom

(From OE-Core rev: 19145d0cd27c5c7e732c06dee9fce4cd60cbd0b3)

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

base bbclass: add TUNE_FEATURES to the default banner

This makes debugging the new tune code easier since it doesn't involve staring at 'bitbake -e' output anymore.

(From OE-Core rev: c1903b1221e9b419aefe49e40a8acd61575de797)

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

libpam 1.1.4: fix packaging

| ERROR: QA Issue: non debug package contains .debug directory: libpam-xtests path /work/armv7a-angstrom-linux-gnueabi/libpam-1.1.4-r0/packages-split/libpam-xtests/usr/share/Linux-PAM/xtests/.debug/tst-pam_dispatch4

(From OE-Core rev: 4acbbefd67ac02322bb2ca455e6aaf5b78652dff)

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

connman: Change hard coded package name to support multilib

connman plugins have dependency on bluez4, wpa-supplicant, and ofono.
These names are hardcoded, fix the issue to support multilib.

(From OE-Core rev: a27503dc76652afef6eaf7330c8e6a421d72a6fc)

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

package(rootfs)_ipk.bbclass: support multilib in opkg backend.

Support install multiple multilib in opkg backend.

The installation is done in 3 phases.

Phase 1: install normal packages to IMAGE_ROOTFS.

Phase 2: install multilib packages under MULTILIB_TEMP_ROOTFS. Packages
belongs to the same multilib arch would be installed to a unique
directory.

Phase 3: check file confliction between IMAGE_ROOTFS and
MULTILIB_TEMP_ROOTFS, install multilib packages to IMAGE_ROOTFS only if
the sanity check passed.

(From OE-Core rev: 9d81c2166be7ed388e5616e01ca2b4059e524c8e)

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

image.bbclass: Added variables for multilib support.

1. Added MULTILIB_PACKAGE_INSTALL for multilib instances of packages to
be installed in the rootfs.

2. MULTILIBRE_ALLOW_REP contains the regular expression to match the
files allow to be replaced by the conflicting files.

3. MULTILIBRE_FORCE_SAME contains the regular expression to match the
files allow to be replaced only if the conflicting files are identical.

4. Added shell function multilib_sanity_check() to check whether the
overwring for multilib situation is allowed.

(From OE-Core rev: 137a4626a7e8107fc8a71724d5124f44236293b9)

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

package/rootfs_rpm: Implement RPM multilib package handling

This is a first pass at adding multilib support to the RPM package
and image handling code.

(From OE-Core rev: 1939a4395997098862912e013a2b13ed2f385f9f)

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

multilib: Add missing files from broken patch merge

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

base.bbclass: Fix PACKAGE_ARCH typo

(From OE-Core rev: 22a42df1843c8a95cf3be214a413d660adaadb32)

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

module-init-tools-cross: Drop static binaries patch as a better fix has been merged

(From OE-Core rev: 4f6783b83d722d040707d0b7cd69f27b133bd562)

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

gcc: Drop part of the 64bithack patch which is no longer used

Since we now handle GLIBC_DYNAMIC_LINKER in gcc-configure-common.inc:

's#\(GLIBC_DYNAMIC_LINKER[^ ]*\)\( *"/lib.*\)#\1 SYSTEMLIBS_DIR\2#'

we can drop the patch which changes a hardcoded value for this.

No PR bump since there is no code change resulting from this, its just
a cleanup.

There is still a valid question over the remainder of this patch and its
interaction with multilib configurations.

(From OE-Core rev: 036faf66c3889cd8bf4cd3c9b97c80f008f3c6e2)

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

linux-yocto: revise the dependency for multilib

(From OE-Core rev: b9a0fd48570a603692c8a177b2b0583429639592)

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

eglibc: fix for multilib RPROVIDES issue

in eglibc-package.inc, the "PACKAGES" variable use the ${PN} reference,
while RPROVIDES does not use ${PN}. This will lead multilib.bbclass
not mapping its RPROVIDES, and cause "NO RPROVIDER: lib64-glibc-utils"
error.

This patch unify the recipe to fix this issue

(From OE-Core rev: 37ff0fea8f7180b1a9d91d24dfe1735730427497)

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

do_split_packages: revise for multilib case

in multilib case, the PACKAGE_DYNAMIC is overrided with multilib
prefix. Take multilib:lib64-perl as example. the "perl-module-*"
will become "lib64-perl-module-*"

the output_pattern in do_split_packages is designed to work with
PACKAGE_DYNAMIC, so it should be applied with the same logic, i.e.
overriding with multilib prefix. otherwise the do_split_package will
split incorrect files

this patch implements the mulitlib override logic for do_split_packages

We also need to rename the extra_depends to support multilib case
(from Dongxaio Xu).

(From OE-Core rev: a4bc86713892502aeefbbdb3053e8cf1e1fc0bdb)

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

lib/oe/util.py: Add MLPREFIX support to prune_suffix()

prune_suffix() also needs be aware of MLPREFIX when changing PN for
use in BPN.

(From OE-Core rev: 20dd241052afa5ff80b91ebf09b0b48765bc3412)

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

multilib: Add support for compiling recipes against multiple ABIs

This patch adds the core multilib class which can be used along with a
parameter specifying the mutlilib to use in BBCLASSEXTEND.

The MLPREFIX variable is added and can be used in cases where its too
difficult to dynmaically work out where a mutltilib prefix is needed
to be added to a variable.

This includes:
  * SHLIBSDIR and PACKAGE_ARCH fixes from Lianhao Lu.
  * PACKAGE_DYNAMIC mapping from Yu Ke
  * PACKAGE_INSTALL mapping from Yu Ke
  * RPROVIDES mapping from Yu Ke
  * TARGET_VENDOR fix from Mark Hatle
  * Ignorning *-native-runtime dependnecies as well as *-native from Yu Ke
  * Map PKG and ALLOW_EMPTY from Dongxiao Xu
  * Ensure RCONFLICTS and PKG field dependencies are remapped (from Dongxiao Xu)
  * Ensure PN and MLPREFIX are set at the same time to ensure consistent BPN values (Yu Ke)

(From OE-Core rev: 5d9453049915db48ec4b5972e12287417ebb61a2)

Signed-off-by: Yu Ke <ke.yu@intel.com>
Signed-off-by: Xu Dongxiao <dongxiao.xu@intel.com>
Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: update SRCREVs

Updating the linux-yocto SRCREVs to pickup changes from the
continuing uprev to korg 3.0. With this set of update, the
meta/feature audits are complete.

qemumips and qemuppc are fixed with this update and can now
boot to a prompt.

(From OE-Core rev: b4ec5c27b51c15e8bde7ca2597329c4f5b801240)

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

base.bbclass: Add compatibility package name mapping handler

This means if PKGARCHCOMPAT_ARMV7A is set, "armv7a-vfp-neon" is renamed
to be "armv7a". Other compatibility mappings can be added as needed.

(From OE-Core rev: 08c0b7060009113e8dffdef51ff6b9b4b7f28894)

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

cmake.bbclass: add ${libdir} for multilib case

    (From OE-Core rev: e198b72bc03c7545417944dcb3c930aaf2376488)

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

RPM: multilib file class/color

We need to include file class and file color in order to do the conflict
resolution required with a multilib design.  This is normally disabled when
we don't use the internal dependency generator, patch enables it in
all cases.

(From OE-Core rev: 4c43ea3440ecca02d2945b02017f56a8de9ff808)

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

distcc: fix makefile parameter order

When linking with gcc, the parameter is read from left to write, if *.o
calls a function contained in libxxx, the parameter order should be

"gcc ... *.o -lxxx".

This commit fixes the problem in distcc.

(From OE-Core rev: 782499eaa817f1beecc0aea5bfa075196f8e2a53)

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

Add basic PowerPC core tune config

(From OE-Core rev: 3212029f0967dd353fec8cc147d2b95031d1018a)

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

Add basic Mips core tune config

Acked-by: Mark Hatle <mark.hatle@windriver.com>
(From OE-Core rev: b33cdc23c04acfa55df0f764e1648bf030ae5e9c)

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

Add ARM tune file overhaul based largely on work from Mark Hatle

(From OE-Core rev: 789dcb8e68a2ab9784ac10ab36815010c61af2fc)

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

package.bbclass: fixup_perms - change symlink processing

We switch to using os.lchown in order to avoid following a symlink.

We also now check if an item is a symlink, if so we avoid the
os.chmod as a symlink inherits the mode of it's target.

(From OE-Core rev: c64d075b3d367e6c76aafa17782585d026b1f31e)

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

ncurses: Resolve header and man page multilib conflicts

By default man pages are compressed, the way ncurses compresses
them causes differences (timestamp and name) to be embedded into
the gziped file.  So each build will end up with a slightly
different file.  Avoid this, by not compressing the man pages.

Avoid header conflict with curses.h using the multilib header
helper.

(From OE-Core rev: ae5269039511d0f3d8aa6a87cb8d9fcf01d79ddd)

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

eglibc_2.13: Resolve multilib header conflicts

Due to mismatches between various header definitions in i386 and x86_64
we need to overwrite the i386 versions of the headers with their x86_64
equivalents.

This can not be done for the syscall.h as it's dynamically generated,
so use the header conflict resolution helper for that.

(From OE-Core rev: 019a33236f76aacb989e8f37b09b81599c27f296)

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

linux-libc-headers: Fix file conflict -- ..install.cmd

The ..install.cmd conflicts between various configure runs.

This isn't used anywhere, so remove it to avoid the conflict.

(From OE-Core rev: f7b1ccfaa15e2dd4ff25f2a5f56ac3aff63c7288)

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

beecrypt: Fix multilib header conflict - beecrypt/gnu.h

beecrypt/gnu.h conflicts between 32-bit and 64-bit versions.

(From OE-Core rev: f520b91f878bb16b4b351ddb5f2240e318181044)

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

apr: Fix multilib header conflict - apr.h

apr.h conflicts between 32-bit and 64-bit versions

(From OE-Core rev: 6e280bf3e2edf3a9e7dfe0e86bcc5c763d52211a)

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

binutils: Fix multilib header conflict - bfd.h

bfd.h conflicts between 32-bit and 64-bit versions.

(From OE-Core rev: cd2975f7ad380aaac8c88bc151a2c34a42d4c540)

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

multilib_header.bbclass: Add oe_multilib_header wrapper

This helper function and associated header will allow us to
resolve two/three header files that conflict due to contents
that change based on wordsize and ABI.

(From OE-Core rev: 1fe66d01b7bce70a37245d47b1abce155fae926e)

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

classes/image*: Revamp creation of live images

This creates a live image as an IMAGE_FSTYPES, thus removing the
need to have additional -live.bb recipes.  To create a live image
one just needs to add live to the IMAGE_FSTYPES list

(From OE-Core rev: b3ff63796cd6629975ff0a726ba18cc168e0a2b2)

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

utils.bbclass: Don't forget to pass the cmdline options

This ensure that the command line options from the creation of the wrapper
are actaully passed into the wrapper.

(From OE-Core rev: 190b4f17f5fc9879e680278dd0e31e26e5f5bca0)

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

Remove -directdisk.bb recipes

(From OE-Core rev: bdcd377eb54b5af34151b7c20459771901a65f4b)

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

Remove -live.bb recipes

(From OE-Core rev: 3b29e2c94cb696a22602f99d7714950a38b4eb8e)

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

siteinfo: Rework the siteinfo for powerpc to fix uclibc fallout

I believe that powerpc-linux is now a common file across 32bit/64bit
linux for powerpc be it uclibc or glibc. I compared the differences
between powerpc-linux-uclibc and powerpc-linux files and it
powerpc-linux was more uptodate and all the new stuff it had was needed
for uclibc anyway so we do not need to keep exact copy of powerpc-linux
as powerpc-linux-uclibc instead we use powerpc-linux for powerpc/uclibc
targets.

Secondly linux specific files were added in archinfo dictionary
which I think logically belongs to targetinfo dictionary therefore
moved them to targetinfo

now uclibc/powerpc is buildable again

(From OE-Core rev: 0fc798640ba60cbb9074d617ba18518b113b7186)

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

siteinfo: Move (getpgrp|setpgrp|setgrent|*get{pwuid,grgid}) to common-libc

These functions are determined by the C library so move them to common-glibc
and common-uclibc

(From OE-Core rev: 9226635ab28b5ca44ec04931d27c5c3313d56e61)

Signed-off-by: Tom Rini <tom_rini@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

siteinfo: Move general realloc/malloc values to common-$libc

These are tests for glibc behavior which we have enabled in uclibc.
Note that if we ever disable MALLOC_GLIBC_COMPAT the uclibc tests
will need to be changed (but I believe this would also entail massive
patching to the rest of userspace so this should be unlikely).

(From OE-Core rev: cab512d7ee50fb55b9412c0e1a52a589de09e0a7)

Signed-off-by: Tom Rini <tom_rini@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

siteinfo: Move certain db entries to common-linux

Note that we had a number of unused and ignored settings wrt mutex
support.  These have been dropped as they weren't being used and
implied some odd things to boot.  This is also a partial resync with
oe.dev which had some, but not all of these changes.

(From OE-Core rev: 7cc52fca921259e331a34263d8113e29b9e0dee9)

Signed-off-by: Tom Rini <tom_rini@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>